SPB Git
spboucher.ai

spb/wp45_uqo

Public

WP45: parser+QA, ontology annotation, profiles, hybrid index, writing engine, audits, Claude Code integration; preregistration frozen

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Simon-Pierre Boucher committed (Nov 3, 2026) parent feb6c9e Browse files

240 changed files +19,991 −5

240 changed files
Unified Split
added .claude/agents/jof-corpus-auditor.md +7 −0
@@ -0,0 +1,7 @@
1 +---
2 +name: jof-corpus-auditor
3 +description: Corpus Auditor — verifies the JoF-250 corpus inventory, manifest integrity, duplicates and bibliographic records
4 +model: haiku
5 +tools: Bash, Read, Glob, Grep
6 +---
7 +You audit the JoF-250 corpus. Use `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 corpus audit` and the files under `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/data/manifests/`. Report counts, exclusions, Crossref failures and any inconsistency between the frozen manifest and the folder. Never modify PDFs or the manifest by hand.
added .claude/agents/jof-econometric-referee.md +7 −0
@@ -0,0 +1,7 @@
1 +---
2 +name: jof-econometric-referee
3 +description: Econometric Referee — identification, inference, robustness critique
4 +model: opus
5 +tools: Bash, Read, Glob, Grep
6 +---
7 +You referee empirical finance work. Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 review --paper <path> --role referee --output <path>/jof250_reviews/referee.md`; focus on estimand, identifying assumptions, threats, standard errors/clustering, multiple testing, economic vs statistical significance, and whether causal language matches the design. Tie each comment to a section/table/equation. Do not invent results.
added .claude/agents/jof-integrity-auditor.md +7 −0
@@ -0,0 +1,7 @@
1 +---
2 +name: jof-integrity-auditor
3 +description: Citation/Integrity Auditor — citations, DOIs, unsupported claims, numeric consistency, overlap
4 +model: sonnet
5 +tools: Bash, Read, Glob, Grep
6 +---
7 +Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 audit --paper <path> --output <path>/jof250_reviews/audit.md` and `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 review --paper <path> --role integrity --output <path>/jof250_reviews/integrity.md`. Report unknown citation keys, unresolved DOIs, prose numbers absent from tables, claims without anchors, causal overreach, and corpus overlap. Severity-ranked, minimal fixes, no speculation.
added .claude/agents/jof-replication-auditor.md +7 −0
@@ -0,0 +1,7 @@
1 +---
2 +name: jof-replication-auditor
3 +description: Replication Auditor — what a replicator needs and what is missing
4 +model: sonnet
5 +tools: Bash, Read, Glob, Grep
6 +---
7 +Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 review --paper <path> --role replication --output <path>/jof250_reviews/replication.md`. Produce a present/absent checklist (data access, sample filters, variable definitions, specifications, software, seeds, code/data availability) with locations and the questions the authors must answer.
added .claude/agents/jof-rhetorical-analyst.md +7 −0
@@ -0,0 +1,7 @@
1 +---
2 +name: jof-rhetorical-analyst
3 +description: Rhetorical Analyst — explains the rhetorical structure of a section/manuscript against the measured JoF profiles
4 +model: sonnet
5 +tools: Bash, Read, Glob, Grep
6 +---
7 +You analyse scholarly writing structure. Load the relevant profile from `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/profiles/` and, for the user's section, label each paragraph with its rhetorical move (motivation, tension, gap, preview of approach/results, contribution, roadmap, …), compare the sequence and placement with the profile's measured patterns, and recommend changes as proposals. Use `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 retrieve --topic "..."` to show comparable published patterns with provenance. Never copy source sentences.
added .claude/agents/jof-table-figure-editor.md +7 −0
@@ -0,0 +1,7 @@
1 +---
2 +name: jof-table-figure-editor
3 +description: Table/Figure Editor — captions, notes, units, uncertainty
4 +model: sonnet
5 +tools: Bash, Read, Glob, Grep
6 +---
7 +Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 review --paper <path> --role table_editor --output <path>/jof250_reviews/table_editor.md` and `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 audit --paper <path>`. Propose corrected captions and notes for each table/figure (sample, unit, period, SEs, significance conventions, uncertainty in figures) as a patch; numbers never change.
added .claude/agents/jof-writing-editor.md +7 −0
@@ -0,0 +1,7 @@
1 +---
2 +name: jof-writing-editor
3 +description: Finance Writing Editor — proposes rewrites that keep every fact, number and citation unchanged
4 +model: sonnet
5 +tools: Bash, Read, Glob, Grep
6 +---
7 +You edit academic finance prose. Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 review --paper <path> --role editor --output <path>/jof250_reviews/editor.md` and then propose paragraph-level rewrites as a patch file (never edit the manuscript directly). Every number, sign, unit and citation key must be preserved; run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 audit --paper <path>` afterwards and report numeric preservation.
added .claude/commands/jof-audit.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Deterministic manuscript audit: tables, equations, citations/DOIs, unsupported claims, corpus overlap
3 +argument-hint: <paper-path> [--tables] [--equations]
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Parse `$ARGUMENTS` as `<paper-path> [flags]`. Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 audit --paper <path> --output /tmp/jof-audit.md` and present the report grouped by severity: broken references/labels and unknown citation keys first, then DOI resolution failures, then unsupported-claim sentences, then table/figure caption-and-notes issues, then overlap. Offer concrete fixes, as a patch proposal, not as silent edits.
added .claude/commands/jof-benchmark.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Run (or dry-run) the preregistered JoF-250 evaluation protocol
3 +argument-hint: [--protocol heldout_v1] [--dry-run]
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Parse `$ARGUMENTS`. Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 benchmark --config /Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/config/evaluation.yaml <args>` (use `--dry-run` to list tasks/conditions/cost estimate without API calls; a real run spends money — ask the user to confirm the printed budget first). Report the summary JSON and where outputs were written (`/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/experiments/outputs/`). Never present results that were not actually run.
added .claude/commands/jof-corpus-audit.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Re-run the JoF corpus inventory/verification and summarize the audit
3 +argument-hint:
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 corpus audit` (Crossref verification included; add `--no-verify` if the user is offline), then read `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/data/manifests/corpus_audit.md` and summarize: included/excluded counts by year and type, duplicates, Crossref status, and any records needing review.
added .claude/commands/jof-evidence.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Locate the evidence for a claim inside a manuscript (tables, numbers, sentences)
3 +argument-hint: <paper-path> --claim "..."
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Parse `$ARGUMENTS` as `<paper-path> --claim "<text>"`. Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 evidence --paper <path> --claim "<text>"` and report the status (supported_by_table / mentioned_in_text / no_evidence_found) with the hits (table labels, sentences, numbers found). If no evidence is found, say so plainly and suggest what table or analysis would be needed.
added .claude/commands/jof-outline.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Propose a section outline for a manuscript using JoF-250 (profiles + structured retrieval)
3 +argument-hint: <paper-path> --section <name>
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Parse `$ARGUMENTS` as `<paper-path> [--section <name>]` (default section: introduction). Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 write --paper <path> --section <name> --mode outline --output /tmp/jof-outline-<name>.tex`. Read the resulting `.tex` and `.report.md`; present the outline, the rhetorical rationale (change log), the flags, and the retrieval provenance. Never edit the manuscript.
added .claude/commands/jof-profile.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Explain a measured JoF writing profile (empirical_asset_pricing, corporate_finance, financial_econometrics, theoretical_finance, …)
3 +argument-hint: [profile-name]
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Read `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/profiles/$ARGUMENTS.yaml` (if `$ARGUMENTS` is empty, list `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/profiles/`). Present the profile as a short structured brief: sample size and selection rule, length/structure medians with IQRs, abstract and introduction move patterns, contribution placement, identification/robustness/table conventions, and the caveat that observed frequencies are descriptive, not editorial rules.
added .claude/commands/jof-review.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Produce a referee-style review of a manuscript (roles: referee, editor, table_editor, integrity, replication)
3 +argument-hint: <paper-path> [--role referee|editor|table_editor|integrity|replication] [--full]
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Parse `$ARGUMENTS` as `<paper-path> [--role R] [--full]` (default role referee). If `--full`, run roles referee, editor, table_editor, integrity and replication in sequence. For each: `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 review --paper <path> --role <R> --output <paper-path>/jof250_reviews/<R>_<date>.md`. Then read each report and give the user a prioritized synthesis (major issues first), keeping every number as written. Do not modify the manuscript.
added .claude/commands/jof-status.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Show JoF-250 pipeline status (corpus, parsing, annotations, profiles, index, spend)
3 +argument-hint:
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 status` with Bash and report the output to the user verbatim, then one sentence on what is missing (if anything).
added .claude/commands/jof-upgrade.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Rewrite/upgrade an existing section while preserving every result, number and citation
3 +argument-hint: <paper-path> --section <name> [--preserve-results]
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Parse `$ARGUMENTS` as `<paper-path> [--section <name>]`. Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 write --paper <path> --section <name> --mode upgrade --output <paper-path>/jof250_proposals/<name>_upgrade_<date>.tex`. Read the `.report.md`: if the deterministic audit reports numbers missing from the output or new numbers not in the manuscript, say so prominently and do not recommend adopting the text until fixed. Show the diff summary and change log. The manuscript itself is never modified by the command; offer to apply the patch only if the user explicitly asks.
added .claude/commands/jof-write.md +6 −0
@@ -0,0 +1,6 @@
1 +---
2 +description: Draft a manuscript section with the full JoF-250 engine (profiles, structured retrieval, verification, specialist referee)
3 +argument-hint: <paper-path> --section <name> [--condition A|B|C|D|E]
4 +allowed-tools: Bash(/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250:*), Bash(ls:*), Bash(cat:*), Bash(mkdir:*), Read, Glob, Grep
5 +---
6 +Parse `$ARGUMENTS` as `<paper-path> [--section <name>] [--condition X]` (defaults: introduction, E_full_engine). Run `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250 write --paper <path> --section <name> --mode write --condition <X> --output <paper-path>/jof250_proposals/<name>_<date>.tex` (create the folder; this is a NEW file, the manuscript is never modified). Then read the `.report.md` next to it and show: the proposed text location, the change log, every FLAG (facts the author must verify), the deterministic audit summary (numbers not in the manuscript, unknown citation keys, unsupported claims, corpus overlap), and retrieval provenance.
added .claude/skills/jof250/SKILL.md +20 −0
@@ -0,0 +1,20 @@
1 +---
2 +name: jof250
3 +description: JoF-250 Writing Engine (UQO WP45) — draft, outline, upgrade, review and audit scholarly manuscripts (LaTeX/Markdown) using measured writing conventions from 276 pre-ChatGPT Journal of Finance articles. Use when the user asks to write or improve a paper section, get a referee report, audit tables/equations/citations, locate evidence for a claim, or inspect JoF writing profiles. Commands: /jof-status /jof-corpus-audit /jof-profile /jof-outline /jof-write /jof-review /jof-upgrade /jof-audit /jof-evidence /jof-benchmark.
4 +---
5 +# JoF-250 Writing Engine
6 +
7 +Backing CLI: `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv/bin/jof250` (Python 3.12 venv at `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/.venv`). Project root: `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250`. Docs: `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/docs/USER_GUIDE.md`.
8 +
9 +## What it does
10 +- **Profiles** (`/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/profiles/*.yaml`): measured distributions (medians, IQRs, n) of section lengths, abstract/introduction rhetorical moves, contribution placement, identification language, robustness and table conventions, built from the train split of the corpus. Descriptive, not prescriptive.
11 +- **Retrieval** (`jof250 retrieve --topic "..."`): hybrid BM25 + local embeddings over structural annotations and ≤400-char excerpts; every hit carries provenance (doc_id, DOI, year, page, section, confidence). Full texts are never exposed.
12 +- **Engine** (`jof250 write --paper P --section S --mode outline|write|upgrade --condition E_full_engine --output F`): inspects the manuscript (LaTeX with \input resolution, or Markdown), detects genre/subfield, selects a profile, retrieves rhetorical patterns, drafts, runs deterministic checks (numeric preservation, unknown citation keys, unsupported claims, corpus overlap) and a specialist-referee revision. Output = new file + `.report.md` (change log, flags, provenance, audit, diff). **The manuscript is never modified.**
13 +- **Reviews** (`jof250 review --paper P --role referee|editor|table_editor|integrity|replication`), **audits** (`jof250 audit --paper P`), **evidence lookup** (`jof250 evidence --paper P --claim "..."`).
14 +
15 +## Rules for the assistant
16 +1. Always pass an explicit `--output` path under the user's paper folder (`jof250_proposals/`, `jof250_reviews/`) or `/tmp`; present proposals and diffs; apply changes only on explicit request.
17 +2. Read and relay the FLAGS and the deterministic audit; never present a generated number or citation as verified.
18 +3. For theory, methods or non-finance manuscripts, say which JoF conventions do not apply (the engine prints a genre note).
19 +4. API calls cost money (ledger: `/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250/data/annotations/api_ledger.jsonl`; `jof250 status` shows spend). The benchmark command spends more: dry-run first.
20 +5. If the venv or index is missing, run `cd /Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/wp45-jof250 && make install && make index` (see `docs/INSTALL.md`).
modified config/models.yaml +12 −5
@@ -1,10 +1,17 @@
1 1 # Model configuration (no secrets here; API key read from JOF250_ANTHROPIC_KEY_FILE or ANTHROPIC_API_KEY)
2 −annotation_model: "claude-sonnet-5-5"
3 −writing_model: "claude-sonnet-5-5"
4 −judge_model: "claude-sonnet-5-5"
2 +annotation_model: "claude-opus-5-5"
3 +writing_model: "claude-opus-5-5"
4 +judge_model: "claude-sonnet-5-5" # deliberately a different model from the generator to limit self-preference bias
5 5 embedding_model: "sentence-transformers/all-MiniLM-L6-v2" # local, CPU/MPS, 384-d
6 6 max_tokens_annotation: 4000
7 7 max_tokens_writing: 6000
8 −temperature: 0.0
8 +effort_annotation: "medium"
9 +effort_writing: "high"
10 +effort_judge: "medium"
11 +# first-party API list prices (USD per 1M tokens) as of the claude-api skill cache 2026-10-06; cache reads ~0.1x input
12 +pricing:
13 + claude-opus-5-5: {input: 4.0, output: 20.0, cache_read: 0.20, cache_write: 5.0}
14 + claude-sonnet-5-5: {input: 2.0, output: 10.0, cache_read: 0.20, cache_write: 2.5}
15 + claude-haiku-5-5: {input: 0.10, output: 0.50, cache_read: 0.01, cache_write: 0.125}
9 16 crossref_mailto: "simon-pierre.boucher@uqo.ca"
10 −budget_usd_cap: 60.0 # amendment A-001 (docs/PROGRESS.md): self-fixed envelope for annotation + pilot experiments
17 +budget_usd_cap: 90.0 # amendment A-001 (docs/PROGRESS.md): executor-fixed envelope for annotation (~25 USD at 0.09 USD/doc on Opus 5.5) + pilot experiments
added data/annotations/_llm_cache/03432647aa61462a9220c698add3503dbded4858f9b51fc851331e5ae0b90656.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of ETF ownership, the share of a stock's capitalization held by ETFs, on stock-level daily volatility\",\"strategy\":\"OLS with controls and stock fixed effects. The annual Russell 1000/2000 reconstitution (index switching) serves as an instrument for ETF ownership in a quasi-natural experiment.\",\"key_assumptions\":[\"Index switching affects volatility only through ETF ownership (exclusion restriction)\",\"The mechanical index assignment rule is as good as random near the cutoff\",\"Liquidity shocks revert while fundamental news is permanent\"],\"threats_addressed\":[\"Endogeneity of ETF ownership / omitted variables\",\"Effects of other institutional owners (mutual funds, hedge funds) on prices\",\"Price discovery as an alternative explanation\",\"Reverse ex post demand for ETFs explaining alphas\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":4},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_sample_period\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes state the sample (S&P 500 vs Russell 3000), the unit of observation (stock-month, stock-day, stock-quarter) and how variables are scaled or standardized. Panels split samples by period or show first-stage results. Portfolio tables report raw returns and factor alphas.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests a causal interpretation'\",\"'consistent with'\",\"'appears to'\",\"'admittedly indirect'\",\"acknowledging the local nature of the IV estimates\"]},\"claim_evidence\":[{\"claim\":\"Stocks with higher ETF ownership have higher volatility in OLS regressions.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The Russell reconstitution IV confirms a positive effect of ETF ownership on volatility.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"ETF flow price impact largely reverts within about 40 days, indicating nonfundamental demand.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"ETF ownership increases deviations from a random walk and makes return autocorrelation more negative.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"ETF ownership attracts higher-turnover traders.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The volatility effect is weaker when limits to arbitrage are higher.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"High-ETF-ownership portfolios earn positive alphas of up to 56 bps per month.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"The introduction motivates the question with ETF growth and sketches an arbitrage-based propagation mechanism. It then previews OLS and IV results with percentile magnitudes and addresses endogeneity through the Russell reconstitution instrument. Next it poses a competing price-discovery explanation and describes reversal-based tests to separate the two. It closes by positioning the paper in the derivatives-and-price-quality debate, followed by a roadmap. The body proceeds through institutional details and data, hypothesis development (illustrated by schematic figures), main volatility results, channel tests, and asset pricing implications. The conclusion restates the findings with hedges and lists open questions.\",\"writing_profile_keywords\":[\"quasi-natural experiment\",\"competing hypotheses\",\"mechanism exploration\",\"percentile-based magnitudes\",\"asset pricing implications\",\"acknowledged limitations\"],\"confidence\":0.8,\"notes\":\"The format for standard errors and significance is not visible in the excerpts. Section IV, labeled robustness by the rule, functions as channel results.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of ETF ownership, the share of a stock's capitalization held by ETFs, on stock-level daily volatility", "strategy": "OLS with controls and stock fixed effects. The annual Russell 1000/2000 reconstitution (index switching) serves as an instrument for ETF ownership in a quasi-natural experiment.", "key_assumptions": ["Index switching affects volatility only through ETF ownership (exclusion restriction)", "The mechanical index assignment rule is as good as random near the cutoff", "Liquidity shocks revert while fundamental news is permanent"], "threats_addressed": ["Endogeneity of ETF ownership / omitted variables", "Effects of other institutional owners (mutual funds, hedge funds) on prices", "Price discovery as an alternative explanation", "Reverse ex post demand for ETFs explaining alphas"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 4}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_sample_period", "alternative_specification"]}, "tables": {"convention_notes": "Table notes state the sample (S&P 500 vs Russell 3000), the unit of observation (stock-month, stock-day, stock-quarter) and how variables are scaled or standardized. Panels split samples by period or show first-stage results. Portfolio tables report raw returns and factor alphas.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests a causal interpretation'", "'consistent with'", "'appears to'", "'admittedly indirect'", "acknowledging the local nature of the IV estimates"]}, "claim_evidence": [{"claim": "Stocks with higher ETF ownership have higher volatility in OLS regressions.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The Russell reconstitution IV confirms a positive effect of ETF ownership on volatility.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "ETF flow price impact largely reverts within about 40 days, indicating nonfundamental demand.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "ETF ownership increases deviations from a random walk and makes return autocorrelation more negative.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "ETF ownership attracts higher-turnover traders.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The volatility effect is weaker when limits to arbitrage are higher.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "High-ETF-ownership portfolios earn positive alphas of up to 56 bps per month.", "evidence_type": "table", "evidence_ref": "Table X"}], "structural_summary": "The introduction motivates the question with ETF growth and sketches an arbitrage-based propagation mechanism. It then previews OLS and IV results with percentile magnitudes and addresses endogeneity through the Russell reconstitution instrument. Next it poses a competing price-discovery explanation and describes reversal-based tests to separate the two. It closes by positioning the paper in the derivatives-and-price-quality debate, followed by a roadmap. The body proceeds through institutional details and data, hypothesis development (illustrated by schematic figures), main volatility results, channel tests, and asset pricing implications. The conclusion restates the findings with hedges and lists open questions.", "writing_profile_keywords": ["quasi-natural experiment", "competing hypotheses", "mechanism exploration", "percentile-based magnitudes", "asset pricing implications", "acknowledged limitations"], "confidence": 0.8, "notes": "The format for standard errors and significance is not visible in the excerpts. Section IV, labeled robustness by the rule, functions as channel results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.21, "usage": {"input": 7466, "output": 2066, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071978, "request_id": "req_011Cggq6marPSEsMNDeTpav9", "cached": false, "ts": "2026-11-03T23:04:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/05028c35361907c71442407bc74ba778a9f8f64668709441e4700ac646c215e0.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Average effect of treatment on the treated (ATT) of a sovereign downgrade on investment, debt/equity issuance and bond yields for firms rated at or above the sovereign ceiling (bound firms), relative to matched nonbound firms\",\"strategy\":\"Abadie-Imbens bias-corrected matching combined with difference-in-differences. Bound firms are treated and matched nonbound firms from the same country and year serve as controls. The source of variation is the sovereign ceiling policy of rating agencies. Linear DiD regressions are used as an alternative.\",\"key_assumptions\":[\"Nonbound firms have similar but slightly lower credit quality, so fundamentals shocks affect both groups similarly or affect controls more\",\"Asymmetric corporate downgrades arise from the ceiling rule rather than from changes in fundamentals\",\"Exposure to government is similar across treated and control firms\"],\"threats_addressed\":[\"common macro shocks and spillovers\",\"changes in fundamentals and credit risk\",\"differential government exposure and government support or ownership\",\"recessions and crisis periods unrelated to downgrades (placebo)\",\"credit demand versus supply\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"other\"},{\"paragraph_index\":8,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":7,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"subsample\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Notes define the treated group (rating at or above sovereign in year before downgrade), define each variable as a ratio to lagged assets or PP&E, and describe the sample source and period. Most tables report DiD matching estimators, with linear regression tables for the alternative and government-exposure tests. Inference details are not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'likely to be due to'\",\"'cannot completely rule out'\",\"explicit limitations paragraph on quasi-natural experiments\",\"'interpreted with caution due to small sample size and lack of statistical power'\",\"'consistent with'\",\"'can have'\"]},\"claim_evidence\":[{\"claim\":\"Following sovereign downgrades, bound firms are more likely to be downgraded than nonbound firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figures 2-3\"},{\"claim\":\"Treated firms cut investment by about 8.9 pp of capital (ATT), relative to matched controls\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 4\"},{\"claim\":\"Treated firms reduce net debt issuance more than controls (ATT -5.5 pp), with some increase in equity issuance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII; Figure 5\"},{\"claim\":\"Bond yields of bound firms rise about 34 bp more over 3 months and 61 bp over 6 months\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X; Figure 6\"},{\"claim\":\"Effects do not appear in recessions without downgrades or in other placebo windows\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Exposure to government appears similar across treated and control firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VII\"},{\"claim\":\"Effects are stronger when downgrades move sovereigns across Basel II risk-weight bins\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"}],\"structural_summary\":\"The introduction opens with recent sovereign downgrades and poses a direct question. It then presents the ceiling-based identification and defends it on credit-quality grounds. A regulatory mechanism is sketched, followed by a preview of the matching estimator and quantified results, a government-exposure caveat, three contributions positioned against the ratings literature, and an explicit limitations paragraph. The body describes the institutional background and matching design, then sequential outcome sections: ratings, investment and financing (with placebo, linear and robustness subsections), and cost of debt. A mechanism section uses Basel II bins. The conclusion restates findings and draws a public debt management implication.\",\"writing_profile_keywords\":[\"quasi-natural experiment\",\"sovereign ceiling\",\"matching estimator\",\"difference-in-differences\",\"credit supply\",\"real effects\",\"explicit limitations\"],\"confidence\":0.85,\"notes\":\"The intro paragraph on government exposure tests appears partially truncated. Significance reporting conventions are not visible in the table notes. Section V is labelled robustness by the rule but is actually a mechanism or results section.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "international_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Average effect of treatment on the treated (ATT) of a sovereign downgrade on investment, debt/equity issuance and bond yields for firms rated at or above the sovereign ceiling (bound firms), relative to matched nonbound firms", "strategy": "Abadie-Imbens bias-corrected matching combined with difference-in-differences. Bound firms are treated and matched nonbound firms from the same country and year serve as controls. The source of variation is the sovereign ceiling policy of rating agencies. Linear DiD regressions are used as an alternative.", "key_assumptions": ["Nonbound firms have similar but slightly lower credit quality, so fundamentals shocks affect both groups similarly or affect controls more", "Asymmetric corporate downgrades arise from the ceiling rule rather than from changes in fundamentals", "Exposure to government is similar across treated and control firms"], "threats_addressed": ["common macro shocks and spillovers", "changes in fundamentals and credit risk", "differential government exposure and government support or ownership", "recessions and crisis periods unrelated to downgrades (placebo)", "credit demand versus supply"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "other"}, {"paragraph_index": 8, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 7, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_specification", "subsample", "falsification"]}, "tables": {"convention_notes": "Notes define the treated group (rating at or above sovereign in year before downgrade), define each variable as a ratio to lagged assets or PP&E, and describe the sample source and period. Most tables report DiD matching estimators, with linear regression tables for the alternative and government-exposure tests. Inference details are not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'likely to be due to'", "'cannot completely rule out'", "explicit limitations paragraph on quasi-natural experiments", "'interpreted with caution due to small sample size and lack of statistical power'", "'consistent with'", "'can have'"]}, "claim_evidence": [{"claim": "Following sovereign downgrades, bound firms are more likely to be downgraded than nonbound firms", "evidence_type": "table", "evidence_ref": "Table III; Figures 2-3"}, {"claim": "Treated firms cut investment by about 8.9 pp of capital (ATT), relative to matched controls", "evidence_type": "table", "evidence_ref": "Table IV; Figure 4"}, {"claim": "Treated firms reduce net debt issuance more than controls (ATT -5.5 pp), with some increase in equity issuance", "evidence_type": "table", "evidence_ref": "Table VIII; Figure 5"}, {"claim": "Bond yields of bound firms rise about 34 bp more over 3 months and 61 bp over 6 months", "evidence_type": "table", "evidence_ref": "Table X; Figure 6"}, {"claim": "Effects do not appear in recessions without downgrades or in other placebo windows", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Exposure to government appears similar across treated and control firms", "evidence_type": "table", "evidence_ref": "Tables VI-VII"}, {"claim": "Effects are stronger when downgrades move sovereigns across Basel II risk-weight bins", "evidence_type": "table", "evidence_ref": "Table XI"}], "structural_summary": "The introduction opens with recent sovereign downgrades and poses a direct question. It then presents the ceiling-based identification and defends it on credit-quality grounds. A regulatory mechanism is sketched, followed by a preview of the matching estimator and quantified results, a government-exposure caveat, three contributions positioned against the ratings literature, and an explicit limitations paragraph. The body describes the institutional background and matching design, then sequential outcome sections: ratings, investment and financing (with placebo, linear and robustness subsections), and cost of debt. A mechanism section uses Basel II bins. The conclusion restates findings and draws a public debt management implication.", "writing_profile_keywords": ["quasi-natural experiment", "sovereign ceiling", "matching estimator", "difference-in-differences", "credit supply", "real effects", "explicit limitations"], "confidence": 0.85, "notes": "The intro paragraph on government exposure tests appears partially truncated. Significance reporting conventions are not visible in the table notes. Section V is labelled robustness by the rule but is actually a mechanism or results section."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.35, "usage": {"input": 8177, "output": 2314, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079782, "request_id": "req_011CggpSEWmGLpptcvkQCqXk", "cached": false, "ts": "2026-11-03T22:55:25+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/052b21227ee450698be9622092c52eff7ace73e07efbfe0d00b67e8aefb8928a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Parameters of affine stochastic volatility models with self-exciting price/volatility cojumps, fit to intradaily S&P 500 futures returns, plus model fit to returns, realized variance and options\",\"strategy\":\"Maximum likelihood estimation with recursive Bayesian filtration of latent diffusive variance and jump intensity every 15 minutes; nested model comparison via likelihood ratio tests; out-of-sample evaluation\",\"key_assumptions\":[\"Affine model specification is adequate\",\"15-minute realized variance approximates quadratic variation\",\"Pricing kernel specification for change of measure\"],\"threats_addressed\":[\"model misspecification via normalized residual diagnostics\",\"intradaily and overnight seasonality\",\"price rounding\",\"out-of-sample overfitting\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"},{\"idx\":13,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Notes define the diagnostic statistic (e.g., normalized returns or residuals that should be i.i.d. standard normal under correct specification), report units such as squared percentages and annualization, and compare several model variants side by side. Heteroskedasticity-consistent standard errors are shown in parentheses where reported.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"robust\"},\"figures\":{\"uncertainty_shown\":\"no\",\"no_figures\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'reasonably well' qualifiers\",\"'does not appear to'\",\"'largely consistent with'\",\"'typically'\"]},\"claim_evidence\":[{\"claim\":\"Intradaily jumps in futures prices are small; the largest-jump component in the three-factor model has roughly a 1% standard deviation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Appendix B jump parameter tables\"},{\"claim\":\"Multifactor variance and jump-intensity models improve likelihood fit substantially over one-factor models\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Appendix B log likelihood tables\"},{\"claim\":\"The 1987 and 2008 crashes developed through runs of self-exciting intradaily volatility spikes rather than single jumps\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5, 6\"},{\"claim\":\"Models capture realized variance outliers but underpredict inliers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 9\"},{\"claim\":\"Bipower variation studies suggest no jumps occurred on October 19, 1987\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Tauchen and Zhou (2011)\"},{\"claim\":\"Option-pricing fit depends on how risk premia are specified\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI; Figure 10\"}],\"structural_summary\":\"The paper opens with a provocative question about what a crash is and uses the 1987 crash to show the Merton model's shortcomings. It then positions itself against nonparametric realized-variance work and daily-data stochastic volatility models, and states its contribution: intradaily dynamic estimation of self-exciting cojumps. A combined data-and-models section covers estimation, filtration and seasonals, followed by realized variance diagnostics and option pricing implications. A short summary restates the volatility-spike interpretation. Appendices give characteristic functions, parameter tables and the change of measure.\",\"writing_profile_keywords\":[\"affine jump-diffusion\",\"intradaily volatility\",\"self-exciting jumps\",\"filtration\",\"likelihood comparison\",\"specification diagnostics\",\"S&P 500 futures\",\"option pricing\"],\"confidence\":0.8,\"notes\":\"Single-author structural time-series econometrics paper. The roadmap is not visible in the truncated introduction. The 'four issues' enumeration begins in paragraph 4 but is cut off in the excerpt.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "structural_estimation", "identification": {"estimand": "Parameters of affine stochastic volatility models with self-exciting price/volatility cojumps, fit to intradaily S&P 500 futures returns, plus model fit to returns, realized variance and options", "strategy": "Maximum likelihood estimation with recursive Bayesian filtration of latent diffusive variance and jump intensity every 15 minutes; nested model comparison via likelihood ratio tests; out-of-sample evaluation", "key_assumptions": ["Affine model specification is adequate", "15-minute realized variance approximates quadratic variation", "Pricing kernel specification for change of measure"], "threats_addressed": ["model misspecification via normalized residual diagnostics", "intradaily and overnight seasonality", "price rounding", "out-of-sample overfitting"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "preview_of_approach"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}, {"idx": 13, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_sample_period"]}, "tables": {"convention_notes": "Notes define the diagnostic statistic (e.g., normalized returns or residuals that should be i.i.d. standard normal under correct specification), report units such as squared percentages and annualization, and compare several model variants side by side. Heteroskedasticity-consistent standard errors are shown in parentheses where reported.", "units_stated_in_notes": true, "standard_errors_description": "robust"}, "figures": {"uncertainty_shown": "no", "no_figures": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'reasonably well' qualifiers", "'does not appear to'", "'largely consistent with'", "'typically'"]}, "claim_evidence": [{"claim": "Intradaily jumps in futures prices are small; the largest-jump component in the three-factor model has roughly a 1% standard deviation", "evidence_type": "table", "evidence_ref": "Appendix B jump parameter tables"}, {"claim": "Multifactor variance and jump-intensity models improve likelihood fit substantially over one-factor models", "evidence_type": "figure", "evidence_ref": "Figure 3; Appendix B log likelihood tables"}, {"claim": "The 1987 and 2008 crashes developed through runs of self-exciting intradaily volatility spikes rather than single jumps", "evidence_type": "figure", "evidence_ref": "Figures 5, 6"}, {"claim": "Models capture realized variance outliers but underpredict inliers", "evidence_type": "table", "evidence_ref": "Table IV; Figure 9"}, {"claim": "Bipower variation studies suggest no jumps occurred on October 19, 1987", "evidence_type": "prior_literature", "evidence_ref": "Tauchen and Zhou (2011)"}, {"claim": "Option-pricing fit depends on how risk premia are specified", "evidence_type": "table", "evidence_ref": "Tables V, VI; Figure 10"}], "structural_summary": "The paper opens with a provocative question about what a crash is and uses the 1987 crash to show the Merton model's shortcomings. It then positions itself against nonparametric realized-variance work and daily-data stochastic volatility models, and states its contribution: intradaily dynamic estimation of self-exciting cojumps. A combined data-and-models section covers estimation, filtration and seasonals, followed by realized variance diagnostics and option pricing implications. A short summary restates the volatility-spike interpretation. Appendices give characteristic functions, parameter tables and the change of measure.", "writing_profile_keywords": ["affine jump-diffusion", "intradaily volatility", "self-exciting jumps", "filtration", "likelihood comparison", "specification diagnostics", "S&P 500 futures", "option pricing"], "confidence": 0.8, "notes": "Single-author structural time-series econometrics paper. The roadmap is not visible in the truncated introduction. The 'four issues' enumeration begins in paragraph 4 but is cut off in the excerpt."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.29, "usage": {"input": 7909, "output": 2157, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07557, "request_id": "req_011CggqAtLWt3NMW7h3UgJS2", "cached": false, "ts": "2026-11-03T23:05:02+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/0558df49948e0ed9c1bd18fffe9215b398c1b0470c0ca43e425fda8c1b98cb7f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"mutual fund performance / institutional herding\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Predictive association between fund-level herding tendency and future fund performance (alphas, returns), with skill as the posited driver\",\"strategy\":\"Construct a fund-level herding measure from cross-sectional regressions of fund trades on past institutional trades; run portfolio sorts and predictive panel and Fama-MacBeth regressions; add auxiliary tests (stocks with light institutional trading, time-varying opportunities, persistence, anticipating crowd trades, career concerns)\",\"key_assumptions\":[\"The herding measure filters out common styles, momentum trading and common information signals\",\"Predictability reflects skill rather than chance or price pressure from institutional herds\",\"Risk adjustment through factor models is adequate\"],\"threats_addressed\":[\"price pressure from institutional herds (test on stocks not heavily traded)\",\"chance (performance persistence)\",\"overlap with past performance and other skill measures (regression controls)\",\"stock characteristics and serial dependence in a fund's own trades (alternative herding measures)\",\"choice of performance measure (holdings-based DGTW CS and Grinblatt-Titman)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes describe variable construction in detail, often referring back to Table I. Results use decile and quintile portfolio sorts, Fama-MacBeth and panel predictive regressions, and a logit regression. Returns are mostly monthly, with alphas from factor models such as Carhart.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggests\",\"may/might in theory discussion\",\"is related to\",\"in line with theory\"]},\"claim_evidence\":[{\"claim\":\"The highest-herding decile of funds underperforms the antiherding decile by about 2.28% per year, and by 1.92% on a four-factor alpha basis\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Herding predicts performance beyond past performance, other fund characteristics and other skill measures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Among stocks not heavily traded by institutions, antiherding funds' large bets outperform herding funds' holdings by about 38 bps per month in Carhart alpha\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The herding-antiherding performance gap widens when investment opportunities are greater\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The performance gap persists for up to two years\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Trades of antiherding funds predict future institutional trades\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Herding reveals skill mainly among less experienced managers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX-XI\"},{\"claim\":\"Fund flows show an insignificant negative relation to herding\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XV\"}],\"structural_summary\":\"The paper motivates the question with theoretical herding models that link ability to deviation from the crowd, and notes the lack of empirical evidence. It then builds a new dynamic fund-level herding measure and shows that it negatively predicts fund returns, using portfolio sorts and predictive regressions. Four sets of tests attribute the predictability to skill. A further section ties herding to career concerns and managerial experience. Robustness checks use alternative herding and performance measures, followed by a short analysis of investor flows and a summary conclusion.\",\"writing_profile_keywords\":[\"new measure\",\"predictability\",\"skill attribution\",\"theory-motivated tests\",\"portfolio sorts\",\"mutual funds\",\"multi-pronged evidence\"],\"confidence\":0.82,\"notes\":\"Abstract sentence 4 bundles the mechanism tests; the abstract text also runs into the opening of the introduction. Standard error conventions and significance notation are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "mutual fund performance / institutional herding", "primary_method": "reduced_form_regression", "identification": {"estimand": "Predictive association between fund-level herding tendency and future fund performance (alphas, returns), with skill as the posited driver", "strategy": "Construct a fund-level herding measure from cross-sectional regressions of fund trades on past institutional trades; run portfolio sorts and predictive panel and Fama-MacBeth regressions; add auxiliary tests (stocks with light institutional trading, time-varying opportunities, persistence, anticipating crowd trades, career concerns)", "key_assumptions": ["The herding measure filters out common styles, momentum trading and common information signals", "Predictability reflects skill rather than chance or price pressure from institutional herds", "Risk adjustment through factor models is adequate"], "threats_addressed": ["price pressure from institutional herds (test on stocks not heavily traded)", "chance (performance persistence)", "overlap with past performance and other skill measures (regression controls)", "stock characteristics and serial dependence in a fund's own trades (alternative herding measures)", "choice of performance measure (holdings-based DGTW CS and Grinblatt-Titman)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_measure", "alternative_specification"]}, "tables": {"convention_notes": "Table notes describe variable construction in detail, often referring back to Table I. Results use decile and quintile portfolio sorts, Fama-MacBeth and panel predictive regressions, and a logit regression. Returns are mostly monthly, with alphas from factor models such as Carhart.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggests", "may/might in theory discussion", "is related to", "in line with theory"]}, "claim_evidence": [{"claim": "The highest-herding decile of funds underperforms the antiherding decile by about 2.28% per year, and by 1.92% on a four-factor alpha basis", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Herding predicts performance beyond past performance, other fund characteristics and other skill measures", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Among stocks not heavily traded by institutions, antiherding funds' large bets outperform herding funds' holdings by about 38 bps per month in Carhart alpha", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "The herding-antiherding performance gap widens when investment opportunities are greater", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The performance gap persists for up to two years", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Trades of antiherding funds predict future institutional trades", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Herding reveals skill mainly among less experienced managers", "evidence_type": "table", "evidence_ref": "Tables IX-XI"}, {"claim": "Fund flows show an insignificant negative relation to herding", "evidence_type": "table", "evidence_ref": "Table XV"}], "structural_summary": "The paper motivates the question with theoretical herding models that link ability to deviation from the crowd, and notes the lack of empirical evidence. It then builds a new dynamic fund-level herding measure and shows that it negatively predicts fund returns, using portfolio sorts and predictive regressions. Four sets of tests attribute the predictability to skill. A further section ties herding to career concerns and managerial experience. Robustness checks use alternative herding and performance measures, followed by a short analysis of investor flows and a summary conclusion.", "writing_profile_keywords": ["new measure", "predictability", "skill attribution", "theory-motivated tests", "portfolio sorts", "mutual funds", "multi-pronged evidence"], "confidence": 0.82, "notes": "Abstract sentence 4 bundles the mechanism tests; the abstract text also runs into the opening of the introduction. Standard error conventions and significance notation are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.73, "usage": {"input": 7815, "output": 2165, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075354, "request_id": "req_011Cggq54qsG6DMdXyTrK64P", "cached": false, "ts": "2026-11-03T23:03:43+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/0743830246f87c71fda4767d01bb822d368b32f489aa559d5f1133b9f76e5d92.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"industrial organization / antitrust (common ownership, product market competition)\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of common ownership concentration (MHHI delta) on route-level airline ticket prices relative to separate ownership.\",\"strategy\":\"Within-route fixed-effect panel regressions of log fares on MHHI delta and HHI, plus a DiD/IV design using cross-route variation in common ownership implied by the BlackRock–BGI combination; supplemented by lead-lag regressions and passenger-volume tests.\",\"key_assumptions\":[\"The BlackRock–BGI acquisition was unrelated to expected route-level airline price changes because airline stocks were a small share of the merging portfolios\",\"Fixed effects absorb confounders at the firm, route, firm-route and firm-time levels\",\"Control rights are proportional to votes held (relaxed in later tests)\",\"The MHHI framework, based on a model of competition, captures anticompetitive incentives\"],\"threats_addressed\":[\"omitted variables such as fuel price changes\",\"reverse causality from anticipated demand\",\"endogeneity of ownership and market shares\",\"functional form\",\"market definition\",\"confounding airline mergers and bankruptcies\",\"proportional control assumption\",\"model of competition\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"separate_section\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"placebo\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes give the data period (2001Q1–2014Q4), the sample filter (routes with at least 20 passengers per day), data sources and variable definitions. They open with 'This table shows'. Tables run from illustrative ownership cases to summary statistics, then baseline, robustness, distributed-lag, IV second-stage and investor-type tables.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"conditional framing ('if robust')\",\"'arguably' when defending identification\",\"'may be' for estimated magnitudes\",\"'suggest' for welfare implications\",\"explicit disclaimer that the raw correlation is not causal\"]},\"claim_evidence\":[{\"claim\":\"Common ownership concentration on the average route is over ten times the antitrust threshold presumed likely to enhance market power.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Figure 1; Table II\"},{\"claim\":\"Within-route increases in common ownership concentration are associated with 3–7% higher ticket prices.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"The baseline price effect survives alternative specifications and market definitions.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Variation from the BlackRock–BGI combination implies 10–12% higher prices due to common ownership.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 2\"},{\"claim\":\"Lead-lag patterns and falling passenger volume argue against reverse causality from anticipated demand.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Effects differ by investor type (high- vs. low-churn) and by shareholder rank.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Theory predicts that common ownership softens competition among portfolio firms.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section I; Section II\"}],\"structural_summary\":\"The introduction contrasts IO theory, which predicts anticompetitive effects of common ownership, with an empirical literature that assumes those effects away. It then poses two questions: how large common ownership is, and whether it raises prices. Results are previewed with magnitudes, a DiD design based on an asset-manager merger, and a robustness list, followed by a roadmap. Separate sections cover related literature and hypotheses (an intuitive two-firm sketch) and data, including construction of the MHHI measure. One long empirical section then runs baseline panel regressions, robustness, reverse-causality tests (leads and lags, BlackRock–BGI DiD/IV, passenger volume) and relaxations of the control assumption. A discursive section on governance mechanisms follows. The conclusion restates the findings, hedges conditionally, and draws research and policy implications, framed as a trilemma among diversification, good governance and competitive markets.\",\"writing_profile_keywords\":[\"common ownership\",\"antitrust\",\"airline industry\",\"MHHI\",\"fixed-effects panel\",\"natural experiment\",\"policy-oriented\",\"reverse causality defense\"],\"confidence\":0.8,\"notes\":\"Primary method could be classed as reduced-form regression, since the baseline is a fixed-effects panel; DiD/IV is the main identification strategy for the causal claim. Section V is labeled 'other' because it discusses institutional mechanisms rather than reporting results. Standard-error treatment is not visible in the excerpts, and no explicit contribution paragraph appears in the introduction. The contribution claim of being 'the first' appears in Section I.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "industrial organization / antitrust (common ownership, product market competition)", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of common ownership concentration (MHHI delta) on route-level airline ticket prices relative to separate ownership.", "strategy": "Within-route fixed-effect panel regressions of log fares on MHHI delta and HHI, plus a DiD/IV design using cross-route variation in common ownership implied by the BlackRock–BGI combination; supplemented by lead-lag regressions and passenger-volume tests.", "key_assumptions": ["The BlackRock–BGI acquisition was unrelated to expected route-level airline price changes because airline stocks were a small share of the merging portfolios", "Fixed effects absorb confounders at the firm, route, firm-route and firm-time levels", "Control rights are proportional to votes held (relaxed in later tests)", "The MHHI framework, based on a model of competition, captures anticompetitive incentives"], "threats_addressed": ["omitted variables such as fuel price changes", "reverse causality from anticipated demand", "endogeneity of ownership and market shares", "functional form", "market definition", "confounding airline mergers and bankruptcies", "proportional control assumption", "model of competition"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "separate_section"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "placebo", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Table notes give the data period (2001Q1–2014Q4), the sample filter (routes with at least 20 passengers per day), data sources and variable definitions. They open with 'This table shows'. Tables run from illustrative ownership cases to summary statistics, then baseline, robustness, distributed-lag, IV second-stage and investor-type tables.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["conditional framing ('if robust')", "'arguably' when defending identification", "'may be' for estimated magnitudes", "'suggest' for welfare implications", "explicit disclaimer that the raw correlation is not causal"]}, "claim_evidence": [{"claim": "Common ownership concentration on the average route is over ten times the antitrust threshold presumed likely to enhance market power.", "evidence_type": "descriptive_statistic", "evidence_ref": "Figure 1; Table II"}, {"claim": "Within-route increases in common ownership concentration are associated with 3–7% higher ticket prices.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "The baseline price effect survives alternative specifications and market definitions.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Variation from the BlackRock–BGI combination implies 10–12% higher prices due to common ownership.", "evidence_type": "table", "evidence_ref": "Table VI; Figure 2"}, {"claim": "Lead-lag patterns and falling passenger volume argue against reverse causality from anticipated demand.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Effects differ by investor type (high- vs. low-churn) and by shareholder rank.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Theory predicts that common ownership softens competition among portfolio firms.", "evidence_type": "prior_literature", "evidence_ref": "Section I; Section II"}], "structural_summary": "The introduction contrasts IO theory, which predicts anticompetitive effects of common ownership, with an empirical literature that assumes those effects away. It then poses two questions: how large common ownership is, and whether it raises prices. Results are previewed with magnitudes, a DiD design based on an asset-manager merger, and a robustness list, followed by a roadmap. Separate sections cover related literature and hypotheses (an intuitive two-firm sketch) and data, including construction of the MHHI measure. One long empirical section then runs baseline panel regressions, robustness, reverse-causality tests (leads and lags, BlackRock–BGI DiD/IV, passenger volume) and relaxations of the control assumption. A discursive section on governance mechanisms follows. The conclusion restates the findings, hedges conditionally, and draws research and policy implications, framed as a trilemma among diversification, good governance and competitive markets.", "writing_profile_keywords": ["common ownership", "antitrust", "airline industry", "MHHI", "fixed-effects panel", "natural experiment", "policy-oriented", "reverse causality defense"], "confidence": 0.8, "notes": "Primary method could be classed as reduced-form regression, since the baseline is a fixed-effects panel; DiD/IV is the main identification strategy for the causal claim. Section V is labeled 'other' because it discusses institutional mechanisms rather than reporting results. Standard-error treatment is not visible in the excerpts, and no explicit contribution paragraph appears in the introduction. The contribution claim of being 'the first' appears in Section I."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 21.89, "usage": {"input": 8489, "output": 2452, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08379, "request_id": "req_011CggpyQvT4x5LbudWjdLoi", "cached": false, "ts": "2026-11-03T23:02:31+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/095173e62259d89082fb0fc424aa1fea9f7ca3ae9a728e332260a18c01487542.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theory paper with descriptive motivating evidence on predictability of forecast errors and revisions)\",\"strategy\":\"Analytical model of diagnostic expectations embedded in a neoclassical investment model; motivating time-series regressions of forecast errors and revisions on current credit spreads\",\"key_assumptions\":[\"Beliefs follow diagnostic expectations derived from the representativeness heuristic\",\"No financial frictions in the baseline model\",\"Fundamentals follow a persistent stochastic process\"],\"threats_addressed\":[\"Lucas critique of mechanical backward-looking belief models\",\"Distinguishing which predictions also hold under rational expectations\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":4},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"literature_gap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_theory_or_model\",\"synthesis\",\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"theory\"},{\"idx\":7,\"role\":\"theory\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"No table notes are visible. The text refers to an econometric test in Table II, but table contents are not shown in the excerpts.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"framing evidence as preliminary or motivating\",\"stating that data are insufficient for a stringent test\",\"'can account for' phrasing\",\"explicitly noting which predictions also arise under rational expectations\"]},\"claim_evidence\":[{\"claim\":\"Credit expansions and low spreads forecast later reversals and weaker real activity\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Intro para 1 (Schularick and Taylor; Greenwood and Hanson; Lopez-Salido et al.)\"},{\"claim\":\"Errors in analyst forecasts of the Baa–Treasury spread are predictable from the current spread\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Section I.A\"},{\"claim\":\"Forecast revisions of spreads are predictable: upward when current spreads are low, downward when high\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2 and Table II, Section I.B\"},{\"claim\":\"Diagnostic expectations generate both neglect of tail risk and extrapolation\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3, Section II\"},{\"claim\":\"In the model, spreads are excessively volatile, overreact to news and revert predictably\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"}],\"structural_summary\":\"The paper opens with crisis and Minsky-style motivation and reviews empirical credit-cycle facts. It sketches the representativeness-based belief mechanism and claims to unify extrapolation with risk neglect. Four model predictions are listed, and the authors separate those that also hold under rational expectations from those unique to their mechanism. They then position the work against financial-friction models. The body presents motivating survey-forecast evidence, derives diagnostic expectations from first principles, embeds them in a frictionless macro model and solves for equilibrium spreads. A safe-debt extension follows, and a short conclusion stresses portability and robustness to the Lucas critique.\",\"writing_profile_keywords\":[\"behavioral macro-finance\",\"microfounded beliefs\",\"enumerated predictions\",\"motivating stylized facts\",\"unification claim\",\"Lucas critique defense\",\"model extension\"],\"confidence\":0.8,\"notes\":\"Mostly theoretical; the empirical Section I is explicitly motivating rather than a formal test. Sections idx 2–3 are figure-title artifacts of Section I. No roadmap paragraph is visible in the truncated introduction.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "macro_finance", "subfield_secondary": "behavioral_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theory paper with descriptive motivating evidence on predictability of forecast errors and revisions)", "strategy": "Analytical model of diagnostic expectations embedded in a neoclassical investment model; motivating time-series regressions of forecast errors and revisions on current credit spreads", "key_assumptions": ["Beliefs follow diagnostic expectations derived from the representativeness heuristic", "No financial frictions in the baseline model", "Fundamentals follow a persistent stochastic process"], "threats_addressed": ["Lucas critique of mechanical backward-looking belief models", "Distinguishing which predictions also hold under rational expectations"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 4}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "literature_gap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_theory_or_model", "synthesis", "new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "theory"}, {"idx": 7, "role": "theory"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "No table notes are visible. The text refers to an econometric test in Table II, but table contents are not shown in the excerpts.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["framing evidence as preliminary or motivating", "stating that data are insufficient for a stringent test", "'can account for' phrasing", "explicitly noting which predictions also arise under rational expectations"]}, "claim_evidence": [{"claim": "Credit expansions and low spreads forecast later reversals and weaker real activity", "evidence_type": "prior_literature", "evidence_ref": "Intro para 1 (Schularick and Taylor; Greenwood and Hanson; Lopez-Salido et al.)"}, {"claim": "Errors in analyst forecasts of the Baa–Treasury spread are predictable from the current spread", "evidence_type": "figure", "evidence_ref": "Figure 1, Section I.A"}, {"claim": "Forecast revisions of spreads are predictable: upward when current spreads are low, downward when high", "evidence_type": "figure", "evidence_ref": "Figure 2 and Table II, Section I.B"}, {"claim": "Diagnostic expectations generate both neglect of tail risk and extrapolation", "evidence_type": "figure", "evidence_ref": "Figure 3, Section II"}, {"claim": "In the model, spreads are excessively volatile, overreact to news and revert predictably", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV"}], "structural_summary": "The paper opens with crisis and Minsky-style motivation and reviews empirical credit-cycle facts. It sketches the representativeness-based belief mechanism and claims to unify extrapolation with risk neglect. Four model predictions are listed, and the authors separate those that also hold under rational expectations from those unique to their mechanism. They then position the work against financial-friction models. The body presents motivating survey-forecast evidence, derives diagnostic expectations from first principles, embeds them in a frictionless macro model and solves for equilibrium spreads. A safe-debt extension follows, and a short conclusion stresses portability and robustness to the Lucas critique.", "writing_profile_keywords": ["behavioral macro-finance", "microfounded beliefs", "enumerated predictions", "motivating stylized facts", "unification claim", "Lucas critique defense", "model extension"], "confidence": 0.8, "notes": "Mostly theoretical; the empirical Section I is explicitly motivating rather than a formal test. Sections idx 2–3 are figure-title artifacts of Section I. No roadmap paragraph is visible in the truncated introduction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.23, "usage": {"input": 5653, "output": 1982, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.063046, "request_id": "req_011CggpoJAs4f1ifBATuiYyU", "cached": false, "ts": "2026-11-03T23:00:08+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/0d21fdb91550bc31a988c9d97b023eefcdfc335590edcad60539edb23bb26b89.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"mutual fund performance and skill\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Within-fund time-series slope of subsequent benchmark-adjusted fund return on lagged fund turnover, interpreted as evidence of skill in timing profit opportunities\",\"strategy\":\"Panel regressions with fund fixed effects isolate the time-series turnover-performance relation, contrasted with cross-sectional slopes; model-derived predictions are tested by splitting funds into categories; turnover of similar funds is used as an additional predictor\",\"key_assumptions\":[\"Profit opportunities vary over time and jointly determine turnover and performance\",\"Funds choose turnover optimally given trading costs (relaxed via the suboptimal-trading extension)\",\"More-skilled funds charge higher fees\",\"Measured turnover is a noisy proxy for true discretionary turnover\"],\"threats_addressed\":[\"Cross-sectional confounds between fund types, handled by fund fixed effects and the time-series design\",\"Alternative ways of measuring turnover\",\"Alternative benchmark models\",\"Out-of-sample validity\",\"Measurement error in turnover\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable precisely (net return plus expense ratio minus the Morningstar benchmark), describe the regressors and regression variants, and report correlation and summary-statistics tables by fund category. How standard errors are computed is not visible in the excerpts.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"suggesting\",\"appears to be related\",\"under the plausible assumption\",\"seems unique\"]},\"claim_evidence\":[{\"claim\":\"A fund's turnover positively predicts its subsequent benchmark-adjusted return in the time series\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"The time-series turnover-performance slope exceeds the cross-sectional slope, as the model predicts\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"The relation is stronger for small-stock funds, small funds, and high-fee funds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Fund turnover comoves strongly across funds, especially among similar funds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Table V, Figure 1\"},{\"claim\":\"Average turnover rises with investor sentiment and return dispersion and falls with market liquidity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 2\"},{\"claim\":\"A one-SD rise in similar funds' average turnover is associated with about 0.43% per year higher performance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The model implies a positive time-series turnover-performance relation\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I, equation (11)\"}],\"structural_summary\":\"The paper opens with motivation about active fund skill and immediately sketches the model's intuition. The introduction then previews results (main relation, heterogeneity, commonality, predictive power of peer turnover), offers two model-based interpretations (measurement error and suboptimal trading), and positions the work against three literatures before a roadmap. Section I presents an analytical model of optimal turnover. Section II estimates time-series versus cross-sectional slopes and includes robustness subsections. Section III tests heterogeneity across fund categories. Section IV studies common turnover, its links to mispricing proxies, and its predictive power. The conclusion restates the findings, and the appendices hold the model extension and the econometric and proof derivations.\",\"writing_profile_keywords\":[\"model-then-test\",\"prediction-confirmation framing\",\"time-series vs cross-section\",\"fund fixed effects\",\"mutual fund skill\",\"commonality\",\"mispricing proxies\",\"appendix derivations\"],\"confidence\":0.85,\"notes\":\"Results are framed as confirming model predictions rather than as causal estimates. Robustness appears as subsections within Section II (turnover measurement, benchmark models, out-of-sample evidence). The out-of-sample check is mapped to the alternative_sample_period type, which is an approximation. Significance conventions are not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "mutual fund performance and skill", "primary_method": "reduced_form_regression", "identification": {"estimand": "Within-fund time-series slope of subsequent benchmark-adjusted fund return on lagged fund turnover, interpreted as evidence of skill in timing profit opportunities", "strategy": "Panel regressions with fund fixed effects isolate the time-series turnover-performance relation, contrasted with cross-sectional slopes; model-derived predictions are tested by splitting funds into categories; turnover of similar funds is used as an additional predictor", "key_assumptions": ["Profit opportunities vary over time and jointly determine turnover and performance", "Funds choose turnover optimally given trading costs (relaxed via the suboptimal-trading extension)", "More-skilled funds charge higher fees", "Measured turnover is a noisy proxy for true discretionary turnover"], "threats_addressed": ["Cross-sectional confounds between fund types, handled by fund fixed effects and the time-series design", "Alternative ways of measuring turnover", "Alternative benchmark models", "Out-of-sample validity", "Measurement error in turnover"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_measure", "alternative_specification", "alternative_sample_period"]}, "tables": {"convention_notes": "Table notes define the dependent variable precisely (net return plus expense ratio minus the Morningstar benchmark), describe the regressors and regression variants, and report correlation and summary-statistics tables by fund category. How standard errors are computed is not visible in the excerpts.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "suggesting", "appears to be related", "under the plausible assumption", "seems unique"]}, "claim_evidence": [{"claim": "A fund's turnover positively predicts its subsequent benchmark-adjusted return in the time series", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "The time-series turnover-performance slope exceeds the cross-sectional slope, as the model predicts", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "The relation is stronger for small-stock funds, small funds, and high-fee funds", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Fund turnover comoves strongly across funds, especially among similar funds", "evidence_type": "table", "evidence_ref": "Table IV, Table V, Figure 1"}, {"claim": "Average turnover rises with investor sentiment and return dispersion and falls with market liquidity", "evidence_type": "table", "evidence_ref": "Table VI, Figure 2"}, {"claim": "A one-SD rise in similar funds' average turnover is associated with about 0.43% per year higher performance", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The model implies a positive time-series turnover-performance relation", "evidence_type": "equation_or_proof", "evidence_ref": "Section I, equation (11)"}], "structural_summary": "The paper opens with motivation about active fund skill and immediately sketches the model's intuition. The introduction then previews results (main relation, heterogeneity, commonality, predictive power of peer turnover), offers two model-based interpretations (measurement error and suboptimal trading), and positions the work against three literatures before a roadmap. Section I presents an analytical model of optimal turnover. Section II estimates time-series versus cross-sectional slopes and includes robustness subsections. Section III tests heterogeneity across fund categories. Section IV studies common turnover, its links to mispricing proxies, and its predictive power. The conclusion restates the findings, and the appendices hold the model extension and the econometric and proof derivations.", "writing_profile_keywords": ["model-then-test", "prediction-confirmation framing", "time-series vs cross-section", "fund fixed effects", "mutual fund skill", "commonality", "mispricing proxies", "appendix derivations"], "confidence": 0.85, "notes": "Results are framed as confirming model predictions rather than as causal estimates. Robustness appears as subsections within Section II (turnover measurement, benchmark models, out-of-sample evidence). The out-of-sample check is mapped to the alternative_sample_period type, which is an approximation. Significance conventions are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.69, "usage": {"input": 7288, "output": 2326, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076466, "request_id": "req_011CggpdBTWWQeqXj48BGve5", "cached": false, "ts": "2026-11-03T22:57:53+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/114afe74e671995a8c1aa4b9fac91a3759b8afabe1235e38bf4f6a7c9493979e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Causal effect of the local same-ethnicity resident population share near firm headquarters on the firm's export/import ratio with that country\",\"strategy\":\"Fixed-effects panel regressions of trade ratios on Connected Population, followed by IV using the WWII Japanese internment camp locations as exogenous shocks to local Japanese population; placebo tests on other Asian populations\",\"key_assumptions\":[\"Internment camp locations affect current firm trade with Japan only through local Japanese population\",\"Camp siting was unrelated to factors driving later firm trade with Japan\"],\"threats_addressed\":[\"Reverse causality: firms bringing in foreign populations when planning trade\",\"Omitted common factors driving co-location of ethnic groups and trading firms\",\"Familiarity bias versus value-creating information channel\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"separate_section\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes write out the regression equation, define dependent and key variables, and list the fixed effects. IV tables report first and second stages in separate panels. Descriptive tables give dollar values in billions.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"plausibly exogenous\",\"consistent with\",\"point to\"]},\"claim_evidence\":[{\"claim\":\"Firms trade more with countries whose ethnic group is well represented near headquarters\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Internment camps caused persistent local Japanese population shifts that raise trade with Japan\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Placebo checks support the instrument; no effect on other Asian populations\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Strategic exporters see about 11% higher future profitability than nonstrategic exporters\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Connected networks raise cross-border M&A and foreign segment sales\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"Network effects are stronger with low tariffs and differentiated goods; connected trades get better prices\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X, XI\"},{\"claim\":\"Connected board members are associated with 68% more exports to the connected country\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III (board member variable)\"}],\"structural_summary\":\"Opens with the global-markets motivation and introduces local resident networks. The intro previews the correlational trade results, flags the endogeneity concern and the internment-camp instrument, then previews value effects, M&A and segment extensions, the board channel and mechanism tests, and closes with a roadmap. The body runs: literature section, data, baseline fixed-effects results, a long IV section (natural experiment, placebos, corroboration, pre-war firms), real effects, other corporate decisions, and mechanism tests. A short conclusion restates the contribution.\",\"writing_profile_keywords\":[\"natural experiment\",\"instrumental variables\",\"novel trade data\",\"ethnic networks\",\"mechanism tests\",\"economic magnitudes in intro\"],\"confidence\":0.8,\"notes\":\"Standard error treatment is not visible in the excerpts. Robustness evidence appears in the Internet Appendix (state-level measure), in a placebo subsection, and in the mechanism section, which is why location is marked mixed.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "international_finance", "primary_method": "instrumental_variables", "identification": {"estimand": "Causal effect of the local same-ethnicity resident population share near firm headquarters on the firm's export/import ratio with that country", "strategy": "Fixed-effects panel regressions of trade ratios on Connected Population, followed by IV using the WWII Japanese internment camp locations as exogenous shocks to local Japanese population; placebo tests on other Asian populations", "key_assumptions": ["Internment camp locations affect current firm trade with Japan only through local Japanese population", "Camp siting was unrelated to factors driving later firm trade with Japan"], "threats_addressed": ["Reverse causality: firms bringing in foreign populations when planning trade", "Omitted common factors driving co-location of ethnic groups and trading firms", "Familiarity bias versus value-creating information channel"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence", "new_data_or_measure", "new_identification"], "roadmap_present": true, "literature_positioning_location": "separate_section"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Table notes write out the regression equation, define dependent and key variables, and list the fixed effects. IV tables report first and second stages in separate panels. Descriptive tables give dollar values in billions.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["plausibly exogenous", "consistent with", "point to"]}, "claim_evidence": [{"claim": "Firms trade more with countries whose ethnic group is well represented near headquarters", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Internment camps caused persistent local Japanese population shifts that raise trade with Japan", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Placebo checks support the instrument; no effect on other Asian populations", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Strategic exporters see about 11% higher future profitability than nonstrategic exporters", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Connected networks raise cross-border M&A and foreign segment sales", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "Network effects are stronger with low tariffs and differentiated goods; connected trades get better prices", "evidence_type": "table", "evidence_ref": "Tables X, XI"}, {"claim": "Connected board members are associated with 68% more exports to the connected country", "evidence_type": "table", "evidence_ref": "Table III (board member variable)"}], "structural_summary": "Opens with the global-markets motivation and introduces local resident networks. The intro previews the correlational trade results, flags the endogeneity concern and the internment-camp instrument, then previews value effects, M&A and segment extensions, the board channel and mechanism tests, and closes with a roadmap. The body runs: literature section, data, baseline fixed-effects results, a long IV section (natural experiment, placebos, corroboration, pre-war firms), real effects, other corporate decisions, and mechanism tests. A short conclusion restates the contribution.", "writing_profile_keywords": ["natural experiment", "instrumental variables", "novel trade data", "ethnic networks", "mechanism tests", "economic magnitudes in intro"], "confidence": 0.8, "notes": "Standard error treatment is not visible in the excerpts. Robustness evidence appears in the Internet Appendix (state-level measure), in a placebo subsection, and in the mechanism section, which is why location is marked mixed."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.04, "usage": {"input": 6624, "output": 1972, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06673, "request_id": "req_011CggpSEVX1d38giR4Rjxbb", "cached": false, "ts": "2026-11-03T22:55:23+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/12564936cfa421bf29098a6256c4be0cdf436047fe7f6272a9d491e8f31af073.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Conditional associations between household portfolio value loadings and investor characteristics (age, wealth, leverage, human capital, labor income risk, industry macroeconomic exposure)\",\"strategy\":\"Pooled panel regressions of portfolio value loadings on household characteristics with year, industry and county fixed effects, using Swedish administrative data. Cohort plots and industry sorts complement the regressions. A twin subsample controls for latent family-level fixed effects.\",\"key_assumptions\":[\"Industry sensitivity to national income growth proxies systematic labor income risk\",\"Twin fixed effects absorb latent background and risk attitudes\",\"Observed characteristics are not proxies for other stock characteristics\"],\"threats_addressed\":[\"popular widely held stocks driving results\",\"investor experience\",\"stock characteristics other than value loading (dividend yield, size, firm age, skewness, taxes, professional proximity)\",\"latent investor fixed effects (twins)\",\"passive versus active changes in loadings\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"theory\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes describe the sample, the portfolio level (risky, stock or fund), the fixed effects included and the reference date (end of 2003). A dedicated table translates life-cycle changes into economic significance. A variable definitions table appears in the appendix.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"seem evident\",\"appear remarkably consistent\",\"to the best of our knowledge\",\"unlikely to be due to\"]},\"claim_evidence\":[{\"claim\":\"Households shift from growth to value as they age (the value ladder)\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Value loading rises with age and balance sheet strength and falls with human capital and income risk\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"A single national income factor explains on average 88% of time-series variation in industry per-capita income\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Households in highly cyclical industries hold lower value loadings, especially when young\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; Figure 3\"},{\"claim\":\"Results are not driven by the most widely held popular stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Households actively rebalance their value loadings\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"}],\"structural_summary\":\"The paper frames the value premium puzzle as a contest between risk-based and behavioral explanations and argues that holdings data can help discriminate between them. It then describes Swedish administrative data and the construction of local factors. The core findings follow: life-cycle and balance-sheet determinants of portfolio value tilts, followed by labor income risk across industries. An equilibrium model relates the evidence to risk-based theories, with a shorter section on sentiment explanations. Robustness checks, mostly in the Internet Appendix, precede a conclusion that restates the findings and suggests modeling extensions.\",\"writing_profile_keywords\":[\"administrative panel\",\"household portfolios\",\"value premium\",\"life cycle\",\"hedging demand\",\"consistency with theory\",\"Internet Appendix heavy\"],\"confidence\":0.8,\"notes\":\"Section idx 5 is a parsing artifact (equation fragment) within the model section. The equilibrium model is derived in the Internet Appendix. Standard error type is not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "asset_pricing", "primary_method": "reduced_form_regression", "identification": {"estimand": "Conditional associations between household portfolio value loadings and investor characteristics (age, wealth, leverage, human capital, labor income risk, industry macroeconomic exposure)", "strategy": "Pooled panel regressions of portfolio value loadings on household characteristics with year, industry and county fixed effects, using Swedish administrative data. Cohort plots and industry sorts complement the regressions. A twin subsample controls for latent family-level fixed effects.", "key_assumptions": ["Industry sensitivity to national income growth proxies systematic labor income risk", "Twin fixed effects absorb latent background and risk attitudes", "Observed characteristics are not proxies for other stock characteristics"], "threats_addressed": ["popular widely held stocks driving results", "investor experience", "stock characteristics other than value loading (dividend yield, size, firm age, skewness, taxes, professional proximity)", "latent investor fixed effects (twins)", "passive versus active changes in loadings"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "theory"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes describe the sample, the portfolio level (risky, stock or fund), the fixed effects included and the reference date (end of 2003). A dedicated table translates life-cycle changes into economic significance. A variable definitions table appears in the appendix.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "seem evident", "appear remarkably consistent", "to the best of our knowledge", "unlikely to be due to"]}, "claim_evidence": [{"claim": "Households shift from growth to value as they age (the value ladder)", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Value loading rises with age and balance sheet strength and falls with human capital and income risk", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "A single national income factor explains on average 88% of time-series variation in industry per-capita income", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Households in highly cyclical industries hold lower value loadings, especially when young", "evidence_type": "table", "evidence_ref": "Table IX; Figure 3"}, {"claim": "Results are not driven by the most widely held popular stocks", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Households actively rebalance their value loadings", "evidence_type": "table", "evidence_ref": "Table VII"}], "structural_summary": "The paper frames the value premium puzzle as a contest between risk-based and behavioral explanations and argues that holdings data can help discriminate between them. It then describes Swedish administrative data and the construction of local factors. The core findings follow: life-cycle and balance-sheet determinants of portfolio value tilts, followed by labor income risk across industries. An equilibrium model relates the evidence to risk-based theories, with a shorter section on sentiment explanations. Robustness checks, mostly in the Internet Appendix, precede a conclusion that restates the findings and suggests modeling extensions.", "writing_profile_keywords": ["administrative panel", "household portfolios", "value premium", "life cycle", "hedging demand", "consistency with theory", "Internet Appendix heavy"], "confidence": 0.8, "notes": "Section idx 5 is a parsing artifact (equation fragment) within the model section. The equilibrium model is derived in the Internet Appendix. Standard error type is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.85, "usage": {"input": 7843, "output": 1979, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071746, "request_id": "req_011CggpVD8FAv98BkAEmNqQN", "cached": false, "ts": "2026-11-03T22:56:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/13439cad7752a6bfe1e44cee2d43deedf217ab4cb6b42e014d47e97b88dcd43a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"high-frequency trading / market intermediation\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How second-by-second inventory changes of intraday intermediaries (HFTs, Market Makers) comove with contemporaneous and lagged price changes, and whether that relationship shifts during the Flash Crash.\",\"strategy\":\"Classify accounts into trader categories using audit-trail data. Regress inventory changes on contemporaneous and 20 lagged price changes, with dummy interactions for the crash down and up periods, comparing May 6 with May 3–5. Complemented by an event-type analysis of trading-volume shares around price-level changes.\",\"key_assumptions\":[\"The account classification correctly identifies intermediaries and HFTs.\",\"May 3–5 serve as a valid baseline for normal-day behavior.\"],\"threats_addressed\":[\"Change in behavior during the crash is tested with interaction dummies.\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":41,\"move\":\"motivation\"},{\"paragraph_index\":42,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":43,\"move\":\"preview_of_results\"},{\"paragraph_index\":44,\"move\":\"preview_of_approach\"},{\"paragraph_index\":44,\"move\":\"preview_of_results\"},{\"paragraph_index\":45,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":45,\"move\":\"preview_of_results\"},{\"paragraph_index\":45,\"move\":\"contribution_statement\"},{\"paragraph_index\":46,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":45,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes state the regression equation explicitly, give units (contracts, ticks), define the sample windows (May 3–5 versus May 6, 8:30–15:15 CT), and report descriptive panels by trader category.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' theory phrasing\",\"'our results suggest'\",\"'theory suggests'\"]},\"claim_evidence\":[{\"claim\":\"Combined net inventories of intraday intermediaries stayed far below the size of the large sell program, consistent with limited risk-bearing capacity.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table II; Figure 3\"},{\"claim\":\"The relationship between HFT inventory changes and price changes did not change during the Flash Crash, whereas Market Makers' relationship did.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"HFT inventory changes comove positively with contemporaneous price changes; Market Maker inventory changes comove negatively.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Prices rise after seconds in which HFTs are net buyers and remain higher for about 20 seconds.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"HFTs lift a disproportionate share of the final best-ask depth before price increases, consistent with stale-quote sniping.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Flash Crash price and volume dynamics are documented as the motivating event.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"Opens with a long quoted regulatory description of the event, then sketches inventory-theory mechanisms (limited risk-bearing capacity, latency arbitrage) and previews findings for each. The data section presents the market and trader classification. A single combined methods-and-results section covers trader categories, inventory–price regressions with crash interactions, and an HFT trading-pattern analysis. A short conclusion restates the findings and extends them to later flash events.\",\"writing_profile_keywords\":[\"event-driven\",\"theory-consistency framing\",\"audit-trail data\",\"high-frequency regressions\",\"regulatory report quotation\",\"short conclusion\"],\"confidence\":0.72,\"notes\":\"The introduction paragraph list is fragmented by the block quote: indices 1–40 are quote lines and are not annotated as separate moves. The abstract appears truncated and duplicates the opening of the introduction. Neither standard-error type nor robustness checks are visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "high-frequency trading / market intermediation", "primary_method": "reduced_form_regression", "identification": {"estimand": "How second-by-second inventory changes of intraday intermediaries (HFTs, Market Makers) comove with contemporaneous and lagged price changes, and whether that relationship shifts during the Flash Crash.", "strategy": "Classify accounts into trader categories using audit-trail data. Regress inventory changes on contemporaneous and 20 lagged price changes, with dummy interactions for the crash down and up periods, comparing May 6 with May 3–5. Complemented by an event-type analysis of trading-volume shares around price-level changes.", "key_assumptions": ["The account classification correctly identifies intermediaries and HFTs.", "May 3–5 serve as a valid baseline for normal-day behavior."], "threats_addressed": ["Change in behavior during the crash is tested with interaction dummies."], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 41, "move": "motivation"}, {"paragraph_index": 42, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 43, "move": "preview_of_results"}, {"paragraph_index": 44, "move": "preview_of_approach"}, {"paragraph_index": 44, "move": "preview_of_results"}, {"paragraph_index": 45, "move": "related_literature_positioning"}, {"paragraph_index": 45, "move": "preview_of_results"}, {"paragraph_index": 45, "move": "contribution_statement"}, {"paragraph_index": 46, "move": "roadmap"}], "contribution_paragraph_index": 45, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes state the regression equation explicitly, give units (contracts, ticks), define the sample windows (May 3–5 versus May 6, 8:30–15:15 CT), and report descriptive panels by trader category.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' theory phrasing", "'our results suggest'", "'theory suggests'"]}, "claim_evidence": [{"claim": "Combined net inventories of intraday intermediaries stayed far below the size of the large sell program, consistent with limited risk-bearing capacity.", "evidence_type": "descriptive_statistic", "evidence_ref": "Table II; Figure 3"}, {"claim": "The relationship between HFT inventory changes and price changes did not change during the Flash Crash, whereas Market Makers' relationship did.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "HFT inventory changes comove positively with contemporaneous price changes; Market Maker inventory changes comove negatively.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Prices rise after seconds in which HFTs are net buyers and remain higher for about 20 seconds.", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "HFTs lift a disproportionate share of the final best-ask depth before price increases, consistent with stale-quote sniping.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Flash Crash price and volume dynamics are documented as the motivating event.", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "Opens with a long quoted regulatory description of the event, then sketches inventory-theory mechanisms (limited risk-bearing capacity, latency arbitrage) and previews findings for each. The data section presents the market and trader classification. A single combined methods-and-results section covers trader categories, inventory–price regressions with crash interactions, and an HFT trading-pattern analysis. A short conclusion restates the findings and extends them to later flash events.", "writing_profile_keywords": ["event-driven", "theory-consistency framing", "audit-trail data", "high-frequency regressions", "regulatory report quotation", "short conclusion"], "confidence": 0.72, "notes": "The introduction paragraph list is fragmented by the block quote: indices 1–40 are quote lines and are not annotated as separate moves. The abstract appears truncated and duplicates the opening of the introduction. Neither standard-error type nor robustness checks are visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.29, "usage": {"input": 6018, "output": 1903, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062926, "request_id": "req_011CggpbvDTrBjcqYfxnxnAv", "cached": false, "ts": "2026-11-03T22:57:33+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/16924f337f3237e02506879130ce92a08a596be46796efc91d3cd1c165bd3531.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"executive compensation and risk-taking\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Overall net effect of an increase in the value of newly granted at-the-money options on CEO risk-taking, measured by changes in equity volatility and in firm outcomes such as leverage and investment\",\"strategy\":\"Two IV designs built on multiyear compensation plans. IV1 instruments option grants with a predicted first year of a new fixed-value cycle, predicted from the length of the previous cycle. IV2 instruments with the interaction of fixed-number plan type and industry returns.\",\"key_assumptions\":[\"Predicted cycle first years, based only on past cycle length, are unrelated to risk except through option grants.\",\"Sorting of firms into fixed-number versus fixed-value plans is quasi-arbitrary.\",\"Industry returns are beyond the executive's control and hard to predict, and plan type is not otherwise related to risk.\"],\"threats_addressed\":[\"Endogenous renegotiation of when new cycles start (handled by using predicted rather than actual first years)\",\"Endogeneity of option grants\",\"Nonlinear and interacted effects of industry returns (polynomial and interacted controls)\",\"Placebo test on CEOs who are not currently on cycles\",\"Validity tests that decompose the predicted first-year indicator\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_identification\",\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"subsample\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables are organized by IV strategy, each with first-stage, outcome and validity tables. Panels separate IV estimates from reduced-form estimates. Observations are at the CEO-year level. Notes define the sample restrictions and the variables. One illustrative numerical table is adapted from Hall (1999).\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"describing evidence as suggestive\",\"restricting findings to the range of grant changes observed in the sample\",\"noting that the effect may be nonmonotonic\",\"framing results as conditional on average grant levels\"]},\"claim_evidence\":[{\"claim\":\"Predicted fixed-value first years strongly predict roughly a 15% increase in the value, delta and vega of new option grants.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"A 10% increase in new options leads to a 2.8–4.2% rise in equity volatility.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, VIII\"},{\"claim\":\"The increase in risk is driven largely by increased leverage, while the effect on investment is modest and generally insignificant.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, IX\"},{\"claim\":\"The placebo instrument for CEOs not currently on cycles yields no comparable effect, supporting IV2 validity.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Effects are larger when new grants are large relative to existing holdings and in the financial and high-tech sectors.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"},{\"claim\":\"Firms sort into plan types somewhat arbitrarily.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Hall (1999)\"}],\"structural_summary\":\"The introduction motivates the paper with the prevalence of option pay and frames the net effect of options on risk-taking as theoretically ambiguous. It then lays out two IV strategies based on multiyear plans, defending each against endogeneity concerns, and previews the results with magnitudes. It closes by positioning the paper against the vega–risk literature. Later sections cover data and cycle detection, the empirical strategy for each IV, results split by IV along with heterogeneity and robustness, and a short conclusion that bounds the findings to moderate grant changes.\",\"writing_profile_keywords\":[\"natural experiment\",\"two instruments\",\"cross-validation\",\"executive compensation\",\"institutional detail\",\"theoretical ambiguity\",\"bounded inference\"],\"confidence\":0.85,\"notes\":\"The excerpts do not show how statistical significance or standard errors are reported. Robustness appears both within the Results section (subsection D) and in an internet appendix.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "executive compensation and risk-taking", "primary_method": "instrumental_variables", "identification": {"estimand": "Overall net effect of an increase in the value of newly granted at-the-money options on CEO risk-taking, measured by changes in equity volatility and in firm outcomes such as leverage and investment", "strategy": "Two IV designs built on multiyear compensation plans. IV1 instruments option grants with a predicted first year of a new fixed-value cycle, predicted from the length of the previous cycle. IV2 instruments with the interaction of fixed-number plan type and industry returns.", "key_assumptions": ["Predicted cycle first years, based only on past cycle length, are unrelated to risk except through option grants.", "Sorting of firms into fixed-number versus fixed-value plans is quasi-arbitrary.", "Industry returns are beyond the executive's control and hard to predict, and plan type is not otherwise related to risk."], "threats_addressed": ["Endogenous renegotiation of when new cycles start (handled by using predicted rather than actual first years)", "Endogeneity of option grants", "Nonlinear and interacted effects of industry returns (polynomial and interacted controls)", "Placebo test on CEOs who are not currently on cycles", "Validity tests that decompose the predicted first-year indicator"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}], "contribution_paragraph_index": 5, "contribution_types": ["new_identification", "new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_specification", "subsample", "falsification", "other"]}, "tables": {"convention_notes": "Tables are organized by IV strategy, each with first-stage, outcome and validity tables. Panels separate IV estimates from reduced-form estimates. Observations are at the CEO-year level. Notes define the sample restrictions and the variables. One illustrative numerical table is adapted from Hall (1999).", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["describing evidence as suggestive", "restricting findings to the range of grant changes observed in the sample", "noting that the effect may be nonmonotonic", "framing results as conditional on average grant levels"]}, "claim_evidence": [{"claim": "Predicted fixed-value first years strongly predict roughly a 15% increase in the value, delta and vega of new option grants.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "A 10% increase in new options leads to a 2.8–4.2% rise in equity volatility.", "evidence_type": "table", "evidence_ref": "Tables IV, VIII"}, {"claim": "The increase in risk is driven largely by increased leverage, while the effect on investment is modest and generally insignificant.", "evidence_type": "table", "evidence_ref": "Tables V, IX"}, {"claim": "The placebo instrument for CEOs not currently on cycles yields no comparable effect, supporting IV2 validity.", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Effects are larger when new grants are large relative to existing holdings and in the financial and high-tech sectors.", "evidence_type": "table", "evidence_ref": "Table XII"}, {"claim": "Firms sort into plan types somewhat arbitrarily.", "evidence_type": "prior_literature", "evidence_ref": "Hall (1999)"}], "structural_summary": "The introduction motivates the paper with the prevalence of option pay and frames the net effect of options on risk-taking as theoretically ambiguous. It then lays out two IV strategies based on multiyear plans, defending each against endogeneity concerns, and previews the results with magnitudes. It closes by positioning the paper against the vega–risk literature. Later sections cover data and cycle detection, the empirical strategy for each IV, results split by IV along with heterogeneity and robustness, and a short conclusion that bounds the findings to moderate grant changes.", "writing_profile_keywords": ["natural experiment", "two instruments", "cross-validation", "executive compensation", "institutional detail", "theoretical ambiguity", "bounded inference"], "confidence": 0.85, "notes": "The excerpts do not show how statistical significance or standard errors are reported. Robustness appears both within the Results section (subsection D) and in an internet appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.45, "usage": {"input": 7030, "output": 2096, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070834, "request_id": "req_011CggpkpE3CkeqHV3TSvy3N", "cached": false, "ts": "2026-11-03T22:59:35+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/17b3a5e1159d8f3551100a95a70b527a120b13b0f02af56db8da2d5812363c2d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"contract theory / costly enforcement\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; analytical costly-enforcement model with correlated borrower payoffs and no lender commitment, plus numerical examples\",\"key_assumptions\":[\"lenders cannot commit to enforcement ex ante\",\"borrower payoffs are correlated through an unobserved aggregate state\",\"enforcement is costly\",\"risk-neutral agents\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"motivation\"},{\"paragraph_index\":3,\"move\":\"motivation\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Tables specify model primitives, giving payoff probability distributions by state, and report algebraic value-per-loan expressions; no estimates are shown.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'proposes a mechanism that would explain'\",\"modal verbs (may, can)\",\"'consistent with' framing for empirical regularities\",\"acknowledging model simplifications\"]},\"claim_evidence\":[{\"claim\":\"When lenders cannot commit to enforce and borrower payoffs are correlated, intermediation dominates direct lending.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; proofs in Internet Appendix\"},{\"claim\":\"A large correlated portfolio provides ex post incentives to enforce because learning one loan's state informs on the others.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.A\"},{\"claim\":\"Higher correlation raises value per loan, and with increasing enforcement costs the optimal portfolio size is finite and varies with correlation.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-2\"},{\"claim\":\"Returns to scale are strongest for small portfolios and diminish quickly, consistent with mixed empirical evidence on bank scale economies.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Hughes and Mester (2013); Laeven, Ratnovski, and Tong (2014)\"},{\"claim\":\"Intermediaries issue risk-free deposits, earn positive profits, and hold systemic default risk.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A\"}],\"structural_summary\":\"The introduction opens with stylized facts about banks holding correlated, hard-to-value risks while still issuing safe deposits. It then states the key contribution together with the commitment mechanism and gives an extended informal motivation for the model's assumptions. Next it links the model to empirical regularities on returns to scale and on hard versus soft information, positions the paper against the delegated-monitoring, commitment-device, and safe-asset literatures, and closes with a roadmap. The body builds progressively: a one-borrower model with and without commitment; then many borrowers with correlated default plus numerical examples; then many lenders, where the main intermediation result is proved and empirical implications are drawn. The conclusion restates the mechanism and ends with an anecdotal banking illustration.\",\"writing_profile_keywords\":[\"theory\",\"costly enforcement\",\"commitment device\",\"correlated risk\",\"delegated monitoring\",\"numerical examples\",\"internet appendix proofs\",\"stylized facts motivation\"],\"confidence\":0.85,\"notes\":\"Sections idx 3 and 4 are parsing artifacts: 'A' is empty, and 'B' is a continuation of Section II containing the many-borrower equilibrium. Proofs are in the Internet Appendix. There is no empirical analysis; empirical regularities are cited only as consistent with the model.\"}", "json": {"genre": "theoretical", "subfield": "financial_intermediation_banking", "subfield_secondary": "contract theory / costly enforcement", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; analytical costly-enforcement model with correlated borrower payoffs and no lender commitment, plus numerical examples", "key_assumptions": ["lenders cannot commit to enforcement ex ante", "borrower payoffs are correlated through an unobserved aggregate state", "enforcement is costly", "risk-neutral agents"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "motivation"}, {"paragraph_index": 3, "move": "motivation"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Tables specify model primitives, giving payoff probability distributions by state, and report algebraic value-per-loan expressions; no estimates are shown.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'proposes a mechanism that would explain'", "modal verbs (may, can)", "'consistent with' framing for empirical regularities", "acknowledging model simplifications"]}, "claim_evidence": [{"claim": "When lenders cannot commit to enforce and borrower payoffs are correlated, intermediation dominates direct lending.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; proofs in Internet Appendix"}, {"claim": "A large correlated portfolio provides ex post incentives to enforce because learning one loan's state informs on the others.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.A"}, {"claim": "Higher correlation raises value per loan, and with increasing enforcement costs the optimal portfolio size is finite and varies with correlation.", "evidence_type": "figure", "evidence_ref": "Figures 1-2"}, {"claim": "Returns to scale are strongest for small portfolios and diminish quickly, consistent with mixed empirical evidence on bank scale economies.", "evidence_type": "prior_literature", "evidence_ref": "Hughes and Mester (2013); Laeven, Ratnovski, and Tong (2014)"}, {"claim": "Intermediaries issue risk-free deposits, earn positive profits, and hold systemic default risk.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A"}], "structural_summary": "The introduction opens with stylized facts about banks holding correlated, hard-to-value risks while still issuing safe deposits. It then states the key contribution together with the commitment mechanism and gives an extended informal motivation for the model's assumptions. Next it links the model to empirical regularities on returns to scale and on hard versus soft information, positions the paper against the delegated-monitoring, commitment-device, and safe-asset literatures, and closes with a roadmap. The body builds progressively: a one-borrower model with and without commitment; then many borrowers with correlated default plus numerical examples; then many lenders, where the main intermediation result is proved and empirical implications are drawn. The conclusion restates the mechanism and ends with an anecdotal banking illustration.", "writing_profile_keywords": ["theory", "costly enforcement", "commitment device", "correlated risk", "delegated monitoring", "numerical examples", "internet appendix proofs", "stylized facts motivation"], "confidence": 0.85, "notes": "Sections idx 3 and 4 are parsing artifacts: 'A' is empty, and 'B' is a continuation of Section II containing the many-borrower equilibrium. Proofs are in the Internet Appendix. There is no empirical analysis; empirical regularities are cited only as consistent with the model."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.04, "usage": {"input": 6884, "output": 2035, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06903, "request_id": "req_011CggpajDPXTVamw2kLieaY", "cached": false, "ts": "2026-11-03T22:57:17+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/199908f8bc7d399578f8ab034e7a91453de4a21bbada406029f92625c441cdf5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"household_finance; financial_intermediation_banking; corporate_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Cross-country association between how PAYGO-oriented a pension system is (public pension replacement rate) and financial-system structure outcomes (capital market size, bank loan share, firm size, insider ownership, household debt, finance share).\",\"strategy\":\"A model built on Diamond (2018) generates predictions. These are compared with country-level scatter plots and OLS regressions across about 23 OECD countries. Omitted factors (legal origin, trust, wartime destruction) and historical financial development are then added as controls to probe alternative explanations.\",\"key_assumptions\":[\"Household savings are exogenously higher in prefunded-pension economies\",\"The public pension replacement rate is a valid proxy for PAYGO orientation\",\"The control variables capture the main omitted factors\"],\"threats_addressed\":[\"omitted variable: legal origin (civil vs common law)\",\"omitted variable: faith in finance / trust\",\"omitted variable: wartime destruction\",\"reverse causality: prior financial development determining pension system choice\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables open with summary statistics of pension and financial variables for about 23 OECD countries in 2014, reported in percentages. Later tables regress each financial outcome on the public pension replacement rate plus one competing explanatory variable per table (civil law dummy, trust, wartime GDP ratio, 1950 market cap). Notes define each variable and cite its source.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"robust\"},\"figures\":{\"uncertainty_shown\":\"yes\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"broadly consistent with\",\"explicit statement that correlation does not prove causation\",\"it is possible that\",\"presumably\",\"calls for future work on causality using pension reforms\"]},\"claim_evidence\":[{\"claim\":\"Countries differ widely in pension design, e.g., private pension assets relative to GDP vary hugely between Italy and Denmark.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I; OECD (2015)\"},{\"claim\":\"More PAYGO-oriented countries have smaller capital markets.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"More PAYGO-oriented countries rely more on bank loans relative to bonds for corporate debt.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Countries with less PAYGO orientation show higher household debt and greater mortgage exposure.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 8, 10, 13\"},{\"claim\":\"The pension–finance relationship survives controls for legal origin, trust, wartime experience, and historical financial development.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III–VI\"},{\"claim\":\"In the model, PAYGO-oriented systems are more bank-based and involve less risk-taking.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II\"}],\"structural_summary\":\"Presidential address combining a stylized intermediation model with cross-country descriptive evidence. The introduction motivates with cross-country variation in pension systems, notes thin theory and evidence on how pensions shape financial structure, sketches the model and its departures from Diamond (2018), and positions the work between public economics and finance before giving a roadmap. A descriptive section on pension systems precedes the model section. The empirical section plots and regresses country-level outcomes on the public pension replacement rate. A dedicated section confronts omitted-variable and reverse-causality explanations. The conclusion acknowledges the causal limits, suggests reform-based designs, and draws stability and policy implications.\",\"writing_profile_keywords\":[\"presidential address\",\"first-person singular\",\"theory-then-facts\",\"cross-country scatter plots\",\"alternative explanations section\",\"small-N OECD sample\",\"policy-oriented conclusion\"],\"confidence\":0.8,\"notes\":\"Primary method could arguably be reduced_form_regression, since the empirical work is simple cross-sectional OLS. The model is central to the paper, so theoretical_model_analytical was chosen. The abstract's 'contribution' move is attributed to the model-presentation sentence. Figures report robust standard errors and significance stars.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "macro_finance", "subfield_secondary": "household_finance; financial_intermediation_banking; corporate_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Cross-country association between how PAYGO-oriented a pension system is (public pension replacement rate) and financial-system structure outcomes (capital market size, bank loan share, firm size, insider ownership, household debt, finance share).", "strategy": "A model built on Diamond (2018) generates predictions. These are compared with country-level scatter plots and OLS regressions across about 23 OECD countries. Omitted factors (legal origin, trust, wartime destruction) and historical financial development are then added as controls to probe alternative explanations.", "key_assumptions": ["Household savings are exogenously higher in prefunded-pension economies", "The public pension replacement rate is a valid proxy for PAYGO orientation", "The control variables capture the main omitted factors"], "threats_addressed": ["omitted variable: legal origin (civil vs common law)", "omitted variable: faith in finance / trust", "omitted variable: wartime destruction", "reverse causality: prior financial development determining pension system choice"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 2, "move": "literature_gap"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 8, "contribution_types": ["new_theory_or_model", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "other"]}, "tables": {"convention_notes": "Tables open with summary statistics of pension and financial variables for about 23 OECD countries in 2014, reported in percentages. Later tables regress each financial outcome on the public pension replacement rate plus one competing explanatory variable per table (civil law dummy, trust, wartime GDP ratio, 1950 market cap). Notes define each variable and cite its source.", "units_stated_in_notes": true, "standard_errors_description": "robust"}, "figures": {"uncertainty_shown": "yes"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["broadly consistent with", "explicit statement that correlation does not prove causation", "it is possible that", "presumably", "calls for future work on causality using pension reforms"]}, "claim_evidence": [{"claim": "Countries differ widely in pension design, e.g., private pension assets relative to GDP vary hugely between Italy and Denmark.", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I; OECD (2015)"}, {"claim": "More PAYGO-oriented countries have smaller capital markets.", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "More PAYGO-oriented countries rely more on bank loans relative to bonds for corporate debt.", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Countries with less PAYGO orientation show higher household debt and greater mortgage exposure.", "evidence_type": "figure", "evidence_ref": "Figures 8, 10, 13"}, {"claim": "The pension–finance relationship survives controls for legal origin, trust, wartime experience, and historical financial development.", "evidence_type": "table", "evidence_ref": "Tables III–VI"}, {"claim": "In the model, PAYGO-oriented systems are more bank-based and involve less risk-taking.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II"}], "structural_summary": "Presidential address combining a stylized intermediation model with cross-country descriptive evidence. The introduction motivates with cross-country variation in pension systems, notes thin theory and evidence on how pensions shape financial structure, sketches the model and its departures from Diamond (2018), and positions the work between public economics and finance before giving a roadmap. A descriptive section on pension systems precedes the model section. The empirical section plots and regresses country-level outcomes on the public pension replacement rate. A dedicated section confronts omitted-variable and reverse-causality explanations. The conclusion acknowledges the causal limits, suggests reform-based designs, and draws stability and policy implications.", "writing_profile_keywords": ["presidential address", "first-person singular", "theory-then-facts", "cross-country scatter plots", "alternative explanations section", "small-N OECD sample", "policy-oriented conclusion"], "confidence": 0.8, "notes": "Primary method could arguably be reduced_form_regression, since the empirical work is simple cross-sectional OLS. The model is central to the paper, so theoretical_model_analytical was chosen. The abstract's 'contribution' move is attributed to the model-presentation sentence. Figures report robust standard errors and significance stars."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.19, "usage": {"input": 8882, "output": 2329, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082902, "request_id": "req_011CggpxzyHNGegZam58fwRe", "cached": false, "ts": "2026-11-03T23:02:22+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/1d6c0149aff0d147ced718c9f83c97460f2e17f64f4bb2c79946fbbab4ddbfef.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Separate associations of adoptees' equity participation, risky share and portfolio volatility with those of their biological parents (prebirth/nature) and adoptive parents (postbirth/nurture)\",\"strategy\":\"Adoption design using Swedish administrative and multigenerational registers that link adoptees to both biological and adoptive parents; probit and linear regressions estimated separately for adoptees and own-birth children\",\"key_assumptions\":[\"Adoption occurs soon after birth, so biological parents contribute little postbirth environment\",\"Adoptees are as good as randomly assigned to adoptive families, or nonrandom placement is controlled for\",\"Results from adoptees and their parents generalize to the wider population\"],\"threats_addressed\":[\"nonrandom assignment of adoptees\",\"timing of adoption\",\"cohort and age composition\",\"timing of when equity holding is measured\",\"external validity / representativeness\",\"omitted parental characteristics (wealth, income, education, entrepreneurship)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"results\"},{\"idx\":9,\"role\":\"conclusion\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_specification\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes identify the estimator (probit marginal effects or linear regression), the 1999–2006 estimation window, observation and adoptee counts, and the controls included. Own-birth and adopted children are estimated separately in different panels. Adjusted R2 decompositions are reported in percentages. Appendix tables cover sample construction, reweighting and subsample splits.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggesting\",\"may play an important role\",\"is associated with\",\"it is plausible that\",\"not likely working through\"]},\"claim_evidence\":[{\"claim\":\"Adoptees' equity participation relates to the participation of both their biological and adoptive parents\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Adoptive-parent (nurture) effects are larger than biological effects for the risky share and portfolio volatility\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV\"},{\"claim\":\"Estimates are robust to addressing nonrandom placement of adoptees\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V-VII\"},{\"claim\":\"Controlling for parental education, earnings, wealth and entrepreneurship barely changes the intergenerational coefficients\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"Parental risk-taking variables add materially to the adjusted R2 of cross-sectional portfolio regressions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X, XI, A.X\"},{\"claim\":\"Twin studies attribute little role to shared environment in financial risk-taking\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Cesarini et al. (2010); Barnea, Cronqvist, and Siegel (2010)\"}],\"structural_summary\":\"Opens with the documented intergenerational correlation in risk-taking and frames a nature-versus-nurture question, motivated by welfare and inequality concerns. Introduces the Swedish adoption design, in which both biological and adoptive parents are observed. Positions the paper against twin-study and adoption literatures, stressing that it targets a different estimand. Moves through institutional background, empirical specification and data, then main results for three outcomes. Follows with extensive assumption tests and external-validity checks, a mechanisms section and an explanatory-power analysis. Closes by reconciling the findings with twin studies and acknowledging limits related to age.\",\"writing_profile_keywords\":[\"adoption design\",\"nature vs nurture\",\"administrative registers\",\"intergenerational transmission\",\"external validity\",\"variance decomposition\",\"household portfolio choice\"],\"confidence\":0.82,\"notes\":\"No figures are present. Standard error type is not visible in the table excerpts. The Institutional Background section is assigned the role 'other' rather than literature. The Mechanisms section functions as results/extensions.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Separate associations of adoptees' equity participation, risky share and portfolio volatility with those of their biological parents (prebirth/nature) and adoptive parents (postbirth/nurture)", "strategy": "Adoption design using Swedish administrative and multigenerational registers that link adoptees to both biological and adoptive parents; probit and linear regressions estimated separately for adoptees and own-birth children", "key_assumptions": ["Adoption occurs soon after birth, so biological parents contribute little postbirth environment", "Adoptees are as good as randomly assigned to adoptive families, or nonrandom placement is controlled for", "Results from adoptees and their parents generalize to the wider population"], "threats_addressed": ["nonrandom assignment of adoptees", "timing of adoption", "cohort and age composition", "timing of when equity holding is measured", "external validity / representativeness", "omitted parental characteristics (wealth, income, education, entrepreneurship)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "literature_gap"}, {"paragraph_index": 4, "move": "related_literature_positioning"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "results"}, {"idx": 9, "role": "conclusion"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_specification", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Table notes identify the estimator (probit marginal effects or linear regression), the 1999–2006 estimation window, observation and adoptee counts, and the controls included. Own-birth and adopted children are estimated separately in different panels. Adjusted R2 decompositions are reported in percentages. Appendix tables cover sample construction, reweighting and subsample splits.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggesting", "may play an important role", "is associated with", "it is plausible that", "not likely working through"]}, "claim_evidence": [{"claim": "Adoptees' equity participation relates to the participation of both their biological and adoptive parents", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Adoptive-parent (nurture) effects are larger than biological effects for the risky share and portfolio volatility", "evidence_type": "table", "evidence_ref": "Tables III, IV"}, {"claim": "Estimates are robust to addressing nonrandom placement of adoptees", "evidence_type": "table", "evidence_ref": "Tables V-VII"}, {"claim": "Controlling for parental education, earnings, wealth and entrepreneurship barely changes the intergenerational coefficients", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "Parental risk-taking variables add materially to the adjusted R2 of cross-sectional portfolio regressions", "evidence_type": "table", "evidence_ref": "Tables X, XI, A.X"}, {"claim": "Twin studies attribute little role to shared environment in financial risk-taking", "evidence_type": "prior_literature", "evidence_ref": "Cesarini et al. (2010); Barnea, Cronqvist, and Siegel (2010)"}], "structural_summary": "Opens with the documented intergenerational correlation in risk-taking and frames a nature-versus-nurture question, motivated by welfare and inequality concerns. Introduces the Swedish adoption design, in which both biological and adoptive parents are observed. Positions the paper against twin-study and adoption literatures, stressing that it targets a different estimand. Moves through institutional background, empirical specification and data, then main results for three outcomes. Follows with extensive assumption tests and external-validity checks, a mechanisms section and an explanatory-power analysis. Closes by reconciling the findings with twin studies and acknowledging limits related to age.", "writing_profile_keywords": ["adoption design", "nature vs nurture", "administrative registers", "intergenerational transmission", "external validity", "variance decomposition", "household portfolio choice"], "confidence": 0.82, "notes": "No figures are present. Standard error type is not visible in the table excerpts. The Institutional Background section is assigned the role 'other' rather than literature. The Mechanisms section functions as results/extensions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.61, "usage": {"input": 9999, "output": 2165, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08409, "request_id": "req_011CggpiNCK2no8ULgMj8aPY", "cached": false, "ts": "2026-11-03T22:59:01+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/1f5b570c5679d291d7b01413a321df38a39bce66444c619b7ab81f687e0b4c08.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Time-series alpha of volatility-scaled factor returns relative to the unmanaged factors, plus Sharpe ratio and utility gains.\",\"strategy\":\"Scale monthly factor returns by the inverse of the prior month's realized variance and regress the managed returns on the original factors. Compare the resulting alphas with simulations from structural macrofinance models.\",\"key_assumptions\":[\"Realized variance is forecastable at short horizons\",\"Lagged variance only weakly forecasts returns\",\"Strategy is implementable in real time\"],\"threats_addressed\":[\"transaction costs\",\"leverage constraints\",\"overlap with cross-sectional low-risk anomalies and risk parity\",\"common volatility component\",\"rebalancing horizon\",\"business cycle risk\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"robustness\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the regression equation explicitly (managed factor on original factor, with optional controls or recession interactions) and describe how each managed factor is constructed. The visible excerpts do not describe standard errors.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'suggesting'\",\"'challenge to' rather than rejection\",\"'weakly related'\"]},\"claim_evidence\":[{\"claim\":\"Volatility-managed factors earn large positive alphas against the original factors.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Lagged volatility predicts future volatility but barely predicts average returns, so the mean-variance trade-off worsens when volatility is high.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"The managed strategy takes relatively less risk in recessions.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Alphas survive transaction costs and leverage limits.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V\"},{\"claim\":\"Structural macrofinance models rarely generate alphas as large as the historical ones.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"Alphas are proportional to the covariance between variance and the factor price of risk.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III\"},{\"claim\":\"Utility gains of about 65% for mean-variance investors.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"}],\"structural_summary\":\"The paper opens with a simple strategy and its headline result: alphas across many factors. It motivates the strategy with mean-variance logic and gives intuition from volatility sorts. It then previews a theory link from alphas to price-of-risk dynamics and the stochastic discount factor, and frames the findings as a challenge to structural models. The body presents empirical results, followed by a long section of mechanism and robustness checks. Theory and general-equilibrium sections close the paper, ending with a brief conclusion.\",\"writing_profile_keywords\":[\"trading strategy\",\"factor alphas\",\"Sharpe ratio\",\"volatility timing\",\"macrofinance challenge\",\"simulation of models\",\"self-contained robustness subsections\"],\"confidence\":0.8,\"notes\":\"The abstract is truncated, and part of the introduction (sentence 5 onward) is merged into it. Introduction paragraphs are fragmented by line breaks in the extraction, which affects paragraph indices. Statistical reporting conventions are not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Time-series alpha of volatility-scaled factor returns relative to the unmanaged factors, plus Sharpe ratio and utility gains.", "strategy": "Scale monthly factor returns by the inverse of the prior month's realized variance and regress the managed returns on the original factors. Compare the resulting alphas with simulations from structural macrofinance models.", "key_assumptions": ["Realized variance is forecastable at short horizons", "Lagged variance only weakly forecasts returns", "Strategy is implementable in real time"], "threats_addressed": ["transaction costs", "leverage constraints", "overlap with cross-sectional low-risk anomalies and risk parity", "common volatility component", "rebalancing horizon", "business cycle risk"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "tension_or_puzzle"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 8, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "robustness"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Table notes state the regression equation explicitly (managed factor on original factor, with optional controls or recession interactions) and describe how each managed factor is constructed. The visible excerpts do not describe standard errors.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'suggesting'", "'challenge to' rather than rejection", "'weakly related'"]}, "claim_evidence": [{"claim": "Volatility-managed factors earn large positive alphas against the original factors.", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Lagged volatility predicts future volatility but barely predicts average returns, so the mean-variance trade-off worsens when volatility is high.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "The managed strategy takes relatively less risk in recessions.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Alphas survive transaction costs and leverage limits.", "evidence_type": "table", "evidence_ref": "Tables IV, V"}, {"claim": "Structural macrofinance models rarely generate alphas as large as the historical ones.", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "Alphas are proportional to the covariance between variance and the factor price of risk.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III"}, {"claim": "Utility gains of about 65% for mean-variance investors.", "evidence_type": "figure", "evidence_ref": "Figure 4"}], "structural_summary": "The paper opens with a simple strategy and its headline result: alphas across many factors. It motivates the strategy with mean-variance logic and gives intuition from volatility sorts. It then previews a theory link from alphas to price-of-risk dynamics and the stochastic discount factor, and frames the findings as a challenge to structural models. The body presents empirical results, followed by a long section of mechanism and robustness checks. Theory and general-equilibrium sections close the paper, ending with a brief conclusion.", "writing_profile_keywords": ["trading strategy", "factor alphas", "Sharpe ratio", "volatility timing", "macrofinance challenge", "simulation of models", "self-contained robustness subsections"], "confidence": 0.8, "notes": "The abstract is truncated, and part of the introduction (sentence 5 onward) is merged into it. Introduction paragraphs are fragmented by line breaks in the extraction, which affects paragraph indices. Statistical reporting conventions are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.77, "usage": {"input": 5877, "output": 2016, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064622, "request_id": "req_011CggpeTfcrxHG5drcG6sTs", "cached": false, "ts": "2026-11-03T22:58:08+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/27406ebc0ae1376a90cbf92731e906e708ab2bd36e265ede7c2b912382ec6c2e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance (M&A / auctions)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How potential acquirers' characteristics (distance, asset overlap, concentration gains) and capitalization relate to the probability of acquiring a failed bank, and how local acquirer capitalization relates to FDIC resolution costs.\",\"strategy\":\"Fixed-effects logit models on failed bank × potential acquirer pairs, with failed bank and acquirer-quarter fixed effects; interactions with acquirer capitalization; OLS regressions of resolution cost on local acquirer capitalization; samples restricted to eligible and actual bidders to rule out an FDIC-eligibility explanation.\",\"key_assumptions\":[\"Fixed effects absorb inherent failed bank and acquirer characteristics.\",\"Acquisition likelihood reveals relative willingness to pay, per the auction model.\",\"Capitalization of local potential acquirers is not driven by omitted regional shocks that also drive costs.\"],\"threats_addressed\":[\"FDIC eligibility criteria as an alternative explanation\",\"regional shocks\",\"bid characteristics and measurement of resolution costs\",\"selection from focusing only on participating bidders\",\"soft information vs. other interpretations of distance\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":0},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"data\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Each table note names the estimator (logit, FE logit, OLS) and defines the dependent variable and key regressors, including units (e.g., distance in 100-mile increments, cost as % of assets, constant 2009 prices). The excerpts do not show standard error details.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests'\",\"'consistent with'\",\"'unlikely to be driven by'\",\"'partially explains'\",\"'may lead to'\",\"framework 'rationalizes' rather than proves\"]},\"claim_evidence\":[{\"claim\":\"Average FDIC loss from selling a failed bank is about 28% of assets.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I; Figure 2\"},{\"claim\":\"Failed bank sales are predominantly local; acquisition likelihood falls with distance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 3\"},{\"claim\":\"Banks with similar loan portfolios are more likely to acquire failed banks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 4\"},{\"claim\":\"Market concentration gains have an economically small but statistically significant effect on acquisition.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 5\"},{\"claim\":\"Best-capitalized local buyers acquire over 30% of failed banks versus 10% for the worst capitalized.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6; Table IX\"},{\"claim\":\"Results persist when the sample is restricted to FDIC-eligible bidders, ruling out the eligibility explanation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Poor capitalization of local potential acquirers is associated with higher FDIC resolution costs.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI; Figure 7\"},{\"claim\":\"Higher-valuation bidders bid weakly more; budget constraints create a wedge between willingness and ability to pay.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Appendix\"}],\"structural_summary\":\"The paper opens with the scale of FDIC losses in the Great Recession and notes how little is known about failed bank allocation since the S&L crisis. It then introduces an auction model with budget constraints. Next it documents three willingness-to-pay determinants (distance, asset overlap, concentration gains) and shows that capitalization limits ability to pay, which distorts allocation toward lower-valuation acquirers. An FDIC-eligibility alternative is ruled out using bidder data, and the misallocation is linked to resolution costs, with robustness checks for regional shocks and cost measurement. The paper closes with links to fire-sale and banking theories and with policy implications for capital regulation and resolution.\",\"writing_profile_keywords\":[\"policy-motivated\",\"auction framework\",\"fixed-effects logit\",\"pair-level data\",\"alternative hypothesis testing\",\"misallocation\",\"bank resolution\"],\"confidence\":0.8,\"notes\":\"Institutional background (idx 1) is labeled 'literature' by the role rule but is descriptive institutional content, so it is coded as 'other'. Sections V and VI are both results; Section VI contains robustness subsections. No equations are counted, despite a stylized model and an appendix proof.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance (M&A / auctions)", "primary_method": "reduced_form_regression", "identification": {"estimand": "How potential acquirers' characteristics (distance, asset overlap, concentration gains) and capitalization relate to the probability of acquiring a failed bank, and how local acquirer capitalization relates to FDIC resolution costs.", "strategy": "Fixed-effects logit models on failed bank × potential acquirer pairs, with failed bank and acquirer-quarter fixed effects; interactions with acquirer capitalization; OLS regressions of resolution cost on local acquirer capitalization; samples restricted to eligible and actual bidders to rule out an FDIC-eligibility explanation.", "key_assumptions": ["Fixed effects absorb inherent failed bank and acquirer characteristics.", "Acquisition likelihood reveals relative willingness to pay, per the auction model.", "Capitalization of local potential acquirers is not driven by omitted regional shocks that also drive costs."], "threats_addressed": ["FDIC eligibility criteria as an alternative explanation", "regional shocks", "bid characteristics and measurement of resolution costs", "selection from focusing only on participating bidders", "soft information vs. other interpretations of distance"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 0}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "related_literature_positioning"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "data"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "alternative_specification", "other"]}, "tables": {"convention_notes": "Each table note names the estimator (logit, FE logit, OLS) and defines the dependent variable and key regressors, including units (e.g., distance in 100-mile increments, cost as % of assets, constant 2009 prices). The excerpts do not show standard error details.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests'", "'consistent with'", "'unlikely to be driven by'", "'partially explains'", "'may lead to'", "framework 'rationalizes' rather than proves"]}, "claim_evidence": [{"claim": "Average FDIC loss from selling a failed bank is about 28% of assets.", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I; Figure 2"}, {"claim": "Failed bank sales are predominantly local; acquisition likelihood falls with distance.", "evidence_type": "table", "evidence_ref": "Table III; Figure 3"}, {"claim": "Banks with similar loan portfolios are more likely to acquire failed banks.", "evidence_type": "table", "evidence_ref": "Table VI; Figure 4"}, {"claim": "Market concentration gains have an economically small but statistically significant effect on acquisition.", "evidence_type": "table", "evidence_ref": "Table VII; Figure 5"}, {"claim": "Best-capitalized local buyers acquire over 30% of failed banks versus 10% for the worst capitalized.", "evidence_type": "figure", "evidence_ref": "Figure 6; Table IX"}, {"claim": "Results persist when the sample is restricted to FDIC-eligible bidders, ruling out the eligibility explanation.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Poor capitalization of local potential acquirers is associated with higher FDIC resolution costs.", "evidence_type": "table", "evidence_ref": "Table XI; Figure 7"}, {"claim": "Higher-valuation bidders bid weakly more; budget constraints create a wedge between willingness and ability to pay.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Appendix"}], "structural_summary": "The paper opens with the scale of FDIC losses in the Great Recession and notes how little is known about failed bank allocation since the S&L crisis. It then introduces an auction model with budget constraints. Next it documents three willingness-to-pay determinants (distance, asset overlap, concentration gains) and shows that capitalization limits ability to pay, which distorts allocation toward lower-valuation acquirers. An FDIC-eligibility alternative is ruled out using bidder data, and the misallocation is linked to resolution costs, with robustness checks for regional shocks and cost measurement. The paper closes with links to fire-sale and banking theories and with policy implications for capital regulation and resolution.", "writing_profile_keywords": ["policy-motivated", "auction framework", "fixed-effects logit", "pair-level data", "alternative hypothesis testing", "misallocation", "bank resolution"], "confidence": 0.8, "notes": "Institutional background (idx 1) is labeled 'literature' by the role rule but is descriptive institutional content, so it is coded as 'other'. Sections V and VI are both results; Section VI contains robustness subsections. No equations are counted, despite a stylized model and an appendix proof."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.95, "usage": {"input": 9752, "output": 2470, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.089202, "request_id": "req_011CggpfM1YjYEn5V5xt9y6b", "cached": false, "ts": "2026-11-03T22:58:23+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/27565031974f4a9b3c05db909c11391e43347b0c0fbba4db2e65ea2bf710b220.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"international_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Cross-sectional association between countries' trade composition (import ratio of finished vs. basic goods) and currency interest rates, excess returns and global risk exposure. Time-series association between relative productivity and real exchange rates and interest rate differentials.\",\"strategy\":\"A general equilibrium model generates predictions, which are tested with cross-sectional regressions, portfolio sorts, and GMM/Fama-MacBeth asset pricing tests. Time-series regressions relate relative productivity to real exchange rates and rates, and predictive regressions use the Baltic Dry Index.\",\"key_assumptions\":[\"Covered interest parity holds, so forward discounts approximate interest differentials\",\"Model structure: comparative advantage, convex trade costs, inelastic trade channel\",\"Import ratio proxies a country's specialization type\"],\"threats_addressed\":[\"Controls for GDP size and inflation in cross-sectional regressions\",\"Carry-sorted portfolios compared with import-ratio sorts\",\"G10 subsample versus full sample\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"conclusion\"},{\"idx\":4,\"role\":\"appendix\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes define variable construction (import ratio, relative productivity against an equal-weighted basket of trading partners) and data frequency, and annualize returns. Results are reported for the full sample and the G10 subsample. Asset pricing tests use both GMM and Fama-MacBeth.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\",\"note\":\"Model figures plot comparative statics against relative productivity. Empirical figures are scatter and time-series plots.\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'tend to' generalizations\",\"'helps rationalize' framing of the mechanism\",\"'consistent with' the data\"]},\"claim_evidence\":[{\"claim\":\"Sorting currencies on import ratios yields large spreads in interest rates and excess returns, largely subsuming the carry trade\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Table VII\"},{\"claim\":\"Import ratios explain the cross-section of forward discounts and FX returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 5\"},{\"claim\":\"Final-good producers are more exposed to global output shocks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"},{\"claim\":\"Relative productivity tracks real exchange rates and real interest rate differentials\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-IV, Figures 7-8\"},{\"claim\":\"The Baltic Dry Index (shipping costs) forecasts carry trade returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"In the model, the commodity-country currency is risky because domestic production insulates that country from global shocks\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I, Appendix A\"},{\"claim\":\"Unconditional currency risk premia are substantial\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Lustig, Roussanov, and Verdelhan (2011); Hassan and Mano (2014)\"}],\"structural_summary\":\"The introduction opens with the carry trade and the puzzle of heterogeneous currency risk, and proposes trade composition as its source. It then previews model predictions and empirical confirmations, and positions the paper within the UIP-failure literature. A long model section derives exchange rates, risk premia, import ratios, trade costs and conditional implications, ending with a summary of implications. A long empirical section tests each implication in turn: portfolios, asset pricing tests, productivity regressions, and a case study of the financial crisis. A brief conclusion restates the evidence and the mechanism, and appendices hold proofs and derivations.\",\"writing_profile_keywords\":[\"model-then-test\",\"general equilibrium\",\"prediction-by-prediction validation\",\"portfolio sorts\",\"carry trade\",\"case study\"],\"confidence\":0.8,\"notes\":\"The introduction is truncated, so there may be a roadmap or a more explicit contribution paragraph that is not visible. The type of standard errors is not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "international_finance", "subfield_secondary": "asset_pricing", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Cross-sectional association between countries' trade composition (import ratio of finished vs. basic goods) and currency interest rates, excess returns and global risk exposure. Time-series association between relative productivity and real exchange rates and interest rate differentials.", "strategy": "A general equilibrium model generates predictions, which are tested with cross-sectional regressions, portfolio sorts, and GMM/Fama-MacBeth asset pricing tests. Time-series regressions relate relative productivity to real exchange rates and rates, and predictive regressions use the Baltic Dry Index.", "key_assumptions": ["Covered interest parity holds, so forward discounts approximate interest differentials", "Model structure: comparative advantage, convex trade costs, inelastic trade channel", "Import ratio proxies a country's specialization type"], "threats_addressed": ["Controls for GDP size and inflation in cross-sectional regressions", "Carry-sorted portfolios compared with import-ratio sorts", "G10 subsample versus full sample"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "related_literature_positioning"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model", "new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "conclusion"}, {"idx": 4, "role": "appendix"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["subsample", "alternative_specification"]}, "tables": {"convention_notes": "Notes define variable construction (import ratio, relative productivity against an equal-weighted basket of trading partners) and data frequency, and annualize returns. Results are reported for the full sample and the G10 subsample. Asset pricing tests use both GMM and Fama-MacBeth.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear", "note": "Model figures plot comparative statics against relative productivity. Empirical figures are scatter and time-series plots."}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'tend to' generalizations", "'helps rationalize' framing of the mechanism", "'consistent with' the data"]}, "claim_evidence": [{"claim": "Sorting currencies on import ratios yields large spreads in interest rates and excess returns, largely subsuming the carry trade", "evidence_type": "table", "evidence_ref": "Table VI, Table VII"}, {"claim": "Import ratios explain the cross-section of forward discounts and FX returns", "evidence_type": "table", "evidence_ref": "Table II, Figure 5"}, {"claim": "Final-good producers are more exposed to global output shocks", "evidence_type": "figure", "evidence_ref": "Figure 6"}, {"claim": "Relative productivity tracks real exchange rates and real interest rate differentials", "evidence_type": "table", "evidence_ref": "Tables III-IV, Figures 7-8"}, {"claim": "The Baltic Dry Index (shipping costs) forecasts carry trade returns", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "In the model, the commodity-country currency is risky because domestic production insulates that country from global shocks", "evidence_type": "equation_or_proof", "evidence_ref": "Section I, Appendix A"}, {"claim": "Unconditional currency risk premia are substantial", "evidence_type": "prior_literature", "evidence_ref": "Lustig, Roussanov, and Verdelhan (2011); Hassan and Mano (2014)"}], "structural_summary": "The introduction opens with the carry trade and the puzzle of heterogeneous currency risk, and proposes trade composition as its source. It then previews model predictions and empirical confirmations, and positions the paper within the UIP-failure literature. A long model section derives exchange rates, risk premia, import ratios, trade costs and conditional implications, ending with a summary of implications. A long empirical section tests each implication in turn: portfolios, asset pricing tests, productivity regressions, and a case study of the financial crisis. A brief conclusion restates the evidence and the mechanism, and appendices hold proofs and derivations.", "writing_profile_keywords": ["model-then-test", "general equilibrium", "prediction-by-prediction validation", "portfolio sorts", "carry trade", "case study"], "confidence": 0.8, "notes": "The introduction is truncated, so there may be a roadmap or a more explicit contribution paragraph that is not visible. The type of standard errors is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.99, "usage": {"input": 6759, "output": 2113, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07009, "request_id": "req_011Cggpn37DmwWz9ecfjEuPU", "cached": false, "ts": "2026-11-03T22:59:51+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/27f19f86ee71c7a8c4a76df91bc799e2b62bc67a354b269fd599a55a629dd275.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"corporate_finance (political connections, labor)\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Difference in post-buyout establishment-level employment growth between targets of politically connected and nonconnected private equity firms\",\"strategy\":\"Pre/post establishment panel regressions comparing connected vs nonconnected PE targets, supplemented by propensity score matching on buyout year/state/type and establishment traits, and placebo/DDD comparisons with failed buyouts withdrawn for presumably exogenous reasons\",\"key_assumptions\":[\"Matching on observables removes relevant differences between connected and nonconnected PE targets\",\"Failed buyouts (deals cancelled by the target or a third party, or presumably positive-NPV deals) form a valid counterfactual for successful buyouts\",\"Political connection classification from keyword searches is accurate\"],\"threats_addressed\":[\"selection of targets by connected PE firms\",\"nonrandom assignment to treatment and control groups (local market conditions, buyout type, PE firm size)\",\"endogeneity of the deal withdrawal decision\",\"missed job creation at newly opened establishments\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable as an explicit log-difference formula and define indicator variables. They specify the estimator (panel OLS, conditional logit, negative binomial with fixed effects) and the matching procedure. Univariate pre/post means come first, followed by the regression tables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"presumably exogenous\",\"mitigate rather than eliminate concerns\",\"only suggestive of a causal relation\",\"appears to\"]},\"claim_evidence\":[{\"claim\":\"Establishments of politically connected PE targets have higher post-buyout employment growth than those of nonconnected PE targets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I, II\"},{\"claim\":\"The employment growth difference persists against propensity-score-matched peers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Failed buyouts show no systematic job creation; DDD estimates favor successful connected buyouts\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Employment increases are larger in election years and in more corrupt states\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"State employment changes are positively associated with incumbent reelection\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Targets of connected PE firms are more likely to receive government contracts and grants\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Results are not driven by differences in establishment openings or closures, or by using employment levels instead of growth\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"}],\"structural_summary\":\"The introduction opens with a public controversy anecdote, then links it to the recent literature on PE and employment. It previews the data, followed by a sequence of identification concerns, each paired with a design (matching, failed-buyout placebo/DDD). It next poses a 'why' question and lists alternative explanations before testing the exchange-of-favors story with three pieces of evidence. A separate literature section follows. The results are split into an employment dynamics section and a mechanism section, then come a short robustness section and a conclusion that explicitly limits causal interpretation.\",\"writing_profile_keywords\":[\"political connections\",\"private equity buyouts\",\"establishment-level employment\",\"selection concerns\",\"propensity score matching\",\"placebo failed deals\",\"exchange of favors\",\"cautious causal language\"],\"confidence\":0.8,\"notes\":\"No figures. The way standard errors are computed is not visible in the excerpts. The abstract mentions robustness to selection tests but gives no method detail. Primary method is borderline between DiD and reduced-form panel regression with matching; DiD was chosen because of the pre/post comparison across groups plus the DDD.\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "corporate_finance (political connections, labor)", "primary_method": "difference_in_differences", "identification": {"estimand": "Difference in post-buyout establishment-level employment growth between targets of politically connected and nonconnected private equity firms", "strategy": "Pre/post establishment panel regressions comparing connected vs nonconnected PE targets, supplemented by propensity score matching on buyout year/state/type and establishment traits, and placebo/DDD comparisons with failed buyouts withdrawn for presumably exogenous reasons", "key_assumptions": ["Matching on observables removes relevant differences between connected and nonconnected PE targets", "Failed buyouts (deals cancelled by the target or a third party, or presumably positive-NPV deals) form a valid counterfactual for successful buyouts", "Political connection classification from keyword searches is accurate"], "threats_addressed": ["selection of targets by connected PE firms", "nonrandom assignment to treatment and control groups (local market conditions, buyout type, PE firm size)", "endogeneity of the deal withdrawal decision", "missed job creation at newly opened establishments"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_specification", "alternative_measure"]}, "tables": {"convention_notes": "Table notes define the dependent variable as an explicit log-difference formula and define indicator variables. They specify the estimator (panel OLS, conditional logit, negative binomial with fixed effects) and the matching procedure. Univariate pre/post means come first, followed by the regression tables.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "presumably exogenous", "mitigate rather than eliminate concerns", "only suggestive of a causal relation", "appears to"]}, "claim_evidence": [{"claim": "Establishments of politically connected PE targets have higher post-buyout employment growth than those of nonconnected PE targets", "evidence_type": "table", "evidence_ref": "Tables I, II"}, {"claim": "The employment growth difference persists against propensity-score-matched peers", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Failed buyouts show no systematic job creation; DDD estimates favor successful connected buyouts", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Employment increases are larger in election years and in more corrupt states", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "State employment changes are positively associated with incumbent reelection", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Targets of connected PE firms are more likely to receive government contracts and grants", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Results are not driven by differences in establishment openings or closures, or by using employment levels instead of growth", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}], "structural_summary": "The introduction opens with a public controversy anecdote, then links it to the recent literature on PE and employment. It previews the data, followed by a sequence of identification concerns, each paired with a design (matching, failed-buyout placebo/DDD). It next poses a 'why' question and lists alternative explanations before testing the exchange-of-favors story with three pieces of evidence. A separate literature section follows. The results are split into an employment dynamics section and a mechanism section, then come a short robustness section and a conclusion that explicitly limits causal interpretation.", "writing_profile_keywords": ["political connections", "private equity buyouts", "establishment-level employment", "selection concerns", "propensity score matching", "placebo failed deals", "exchange of favors", "cautious causal language"], "confidence": 0.8, "notes": "No figures. The way standard errors are computed is not visible in the excerpts. The abstract mentions robustness to selection tests but gives no method detail. Primary method is borderline between DiD and reduced-form panel regression with matching; DiD was chosen because of the pre/post comparison across groups plus the DDD."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.36, "usage": {"input": 6083, "output": 2119, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067506, "request_id": "req_011CggpWGRxWPcC9iNnzmoAU", "cached": false, "ts": "2026-11-03T22:56:18+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2854eaa79139a46dfe374dc1308865dae423ae87c0efaa8256b21ea2182acace.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"regression_discontinuity\",\"identification\":{\"estimand\":\"Local effect of a stock's margin trading eligibility (Group 1 status) on its liquidity (effective spread, price impact), and how this effect varies with market conditions\",\"strategy\":\"Regression discontinuity around the impact-cost eligibility cutoff of 1% for margin trading in India. The sample is local to the cutoff with CCT bandwidths, and the regressions include lagged controls.\",\"key_assumptions\":[\"Stocks near the impact-cost cutoff are comparable apart from eligibility\",\"No manipulation or sorting of stocks around the cutoff\",\"Other characteristics are continuous at the threshold\"],\"threats_addressed\":[\"Common shocks driving both deleveraging and illiquidity\",\"Effects of Group 1 status other than margin trading\",\"Cannibalization of liquidity from ineligible stocks\",\"Mechanical variation in measured impact cost (addressed with placebo cutoffs)\",\"Bandwidth and functional-form sensitivity\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"placebo\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Table notes define dependent variables (Espread, Pimpact), the timing of eligibility, the local sample defined by CCT bandwidths, and how specifications differ across columns. Robustness tables refer back to specific columns of the main table.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"evidence consistent with a causal effect\",\"potential effects\",\"suggest\"]},\"claim_evidence\":[{\"claim\":\"Margin eligibility lowers effective spreads by about 2.4 basis points\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The liquidity benefit of margin eligibility reverses during severe market downturns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Margin traders follow contrarian strategies, with position changes negatively related to returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Contrarian margin trading reduces returns to reversal strategies in eligible stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Liquidity improvements are larger when margin trading is more intense\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Results are robust to alternative bandwidths, polynomials and placebo cutoffs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV-VI\"},{\"claim\":\"Liquidity measures are visibly discontinuous at the eligibility cutoff\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"}],\"structural_summary\":\"The paper opens with a question about leverage and liquidity, motivated by theory and by the identification difficulty that common shocks drive both. It previews an RDD built on India's margin eligibility rules, followed by a crisis-reversal result and mechanism evidence on contrarian margin traders. An institutional section precedes the data and RDD design, including bandwidth selection. A long results section covers the baseline, robustness, placebos, alternative interpretations, downturn interactions and mechanisms. The conclusion summarizes three findings and draws policy relevance.\",\"writing_profile_keywords\":[\"regression discontinuity\",\"margin trading\",\"liquidity spirals\",\"institutional laboratory\",\"causal framing\",\"India\"],\"confidence\":0.85,\"notes\":\"Section I (institutional setting) is labeled 'theory' by the rule but is coded here as 'other'. Standard error treatment is not visible in the excerpts. Figures 2–4 are RDD plots, but the captions do not say whether confidence bands are shown.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "asset_pricing", "primary_method": "regression_discontinuity", "identification": {"estimand": "Local effect of a stock's margin trading eligibility (Group 1 status) on its liquidity (effective spread, price impact), and how this effect varies with market conditions", "strategy": "Regression discontinuity around the impact-cost eligibility cutoff of 1% for margin trading in India. The sample is local to the cutoff with CCT bandwidths, and the regressions include lagged controls.", "key_assumptions": ["Stocks near the impact-cost cutoff are comparable apart from eligibility", "No manipulation or sorting of stocks around the cutoff", "Other characteristics are continuous at the threshold"], "threats_addressed": ["Common shocks driving both deleveraging and illiquidity", "Effects of Group 1 status other than margin trading", "Cannibalization of liquidity from ineligible stocks", "Mechanical variation in measured impact cost (addressed with placebo cutoffs)", "Bandwidth and functional-form sensitivity"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "literature_gap"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure", "placebo", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Table notes define dependent variables (Espread, Pimpact), the timing of eligibility, the local sample defined by CCT bandwidths, and how specifications differ across columns. Robustness tables refer back to specific columns of the main table.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["evidence consistent with a causal effect", "potential effects", "suggest"]}, "claim_evidence": [{"claim": "Margin eligibility lowers effective spreads by about 2.4 basis points", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The liquidity benefit of margin eligibility reverses during severe market downturns", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Margin traders follow contrarian strategies, with position changes negatively related to returns", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Contrarian margin trading reduces returns to reversal strategies in eligible stocks", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Liquidity improvements are larger when margin trading is more intense", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Results are robust to alternative bandwidths, polynomials and placebo cutoffs", "evidence_type": "table", "evidence_ref": "Tables IV-VI"}, {"claim": "Liquidity measures are visibly discontinuous at the eligibility cutoff", "evidence_type": "figure", "evidence_ref": "Figure 2"}], "structural_summary": "The paper opens with a question about leverage and liquidity, motivated by theory and by the identification difficulty that common shocks drive both. It previews an RDD built on India's margin eligibility rules, followed by a crisis-reversal result and mechanism evidence on contrarian margin traders. An institutional section precedes the data and RDD design, including bandwidth selection. A long results section covers the baseline, robustness, placebos, alternative interpretations, downturn interactions and mechanisms. The conclusion summarizes three findings and draws policy relevance.", "writing_profile_keywords": ["regression discontinuity", "margin trading", "liquidity spirals", "institutional laboratory", "causal framing", "India"], "confidence": 0.85, "notes": "Section I (institutional setting) is labeled 'theory' by the rule but is coded here as 'other'. Standard error treatment is not visible in the excerpts. Figures 2–4 are RDD plots, but the captions do not say whether confidence bands are shown."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.71, "usage": {"input": 6438, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066786, "request_id": "req_011Cggpe79LKkBbdTs8VTKky", "cached": false, "ts": "2026-11-03T22:58:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2931d45123c629803b10a2b8482a17ed8ce80e804cd9c05565644bda35c9fc81.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Differential sensitivity of local factor prices (wages, real estate) to aggregate GDP shocks by MSA local beta, and the cross-sectional relation between local beta and firms' industry-adjusted returns and conditional betas\",\"strategy\":\"Panel regressions that interact aggregate shocks with an industry-composition-weighted local beta; firm-level panel regressions and local-beta portfolio sorts within industry; supported by a calibrated production-based equilibrium model\",\"key_assumptions\":[\"No local factor mobility across markets at business-cycle frequency\",\"Firm location choice treated as given\",\"Headquarters location proxies for where the firm operates\"],\"threats_addressed\":[\"labor mobility across areas\",\"headquarters location as a noisy proxy for operations (geographically focused firms)\",\"outlier firms driving results\",\"residual correlation structure\",\"sample period dependence\",\"measurement of local beta\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":4},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"robustness_mention\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_sample_period\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes define the key variable (local beta) and refer back to its construction in an earlier table. Empirical tables are panel regressions of factor prices, betas and returns on local beta; one table reports portfolio sorts with three-factor alphas. Model tables give parameter values and the model-implied counterparts of the empirical regressions. Appendix tables list data sets and headquarters employment shares.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with the view\",\"likely to\",\"suggest\",\"explicit discussion of simplifying assumptions\"]},\"claim_evidence\":[{\"claim\":\"Wages respond more strongly to aggregate shocks in high local beta MSAs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Housing and commercial real estate returns are more procyclical in high local beta areas\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Firms in higher local beta areas have lower conditional equity betas\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Firms in higher local beta areas earn lower future returns, especially when real estate holdings are low\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII, VIII\"},{\"claim\":\"Calibrated model reproduces procyclical factor prices and lower firm risk in high-beta areas\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X, XI\"},{\"claim\":\"Labor mobility across counties is low at annual frequency\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Intro paragraph 4, footnote 2\"}],\"structural_summary\":\"The introduction moves from a stylized fact about immobile local factors to a mechanism illustrated by an example, then previews the model and results. It defends its simplifying assumptions, lists extensions and robustness checks, and positions the paper in the literature. A simplified labor-only model builds intuition before the data section constructs the local beta. The empirical section first tests factor prices, then firm risk and returns. A full three-input production model is calibrated to match the empirical patterns. The conclusion restates the findings without discussing policy.\",\"writing_profile_keywords\":[\"local beta\",\"production-based asset pricing\",\"model-then-evidence-then-full-model\",\"natural hedge mechanism\",\"panel regressions\",\"calibration\"],\"confidence\":0.8,\"notes\":\"Standard-error treatment is not visible; the introduction mentions checking various residual correlation assumptions. The model in Section IV is calibrated, not estimated.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Differential sensitivity of local factor prices (wages, real estate) to aggregate GDP shocks by MSA local beta, and the cross-sectional relation between local beta and firms' industry-adjusted returns and conditional betas", "strategy": "Panel regressions that interact aggregate shocks with an industry-composition-weighted local beta; firm-level panel regressions and local-beta portfolio sorts within industry; supported by a calibrated production-based equilibrium model", "key_assumptions": ["No local factor mobility across markets at business-cycle frequency", "Firm location choice treated as given", "Headquarters location proxies for where the firm operates"], "threats_addressed": ["labor mobility across areas", "headquarters location as a noisy proxy for operations (geographically focused firms)", "outlier firms driving results", "residual correlation structure", "sample period dependence", "measurement of local beta"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 4}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "robustness_mention"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}], "contribution_paragraph_index": 8, "contribution_types": ["new_evidence", "new_data_or_measure", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_sample_period", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Notes define the key variable (local beta) and refer back to its construction in an earlier table. Empirical tables are panel regressions of factor prices, betas and returns on local beta; one table reports portfolio sorts with three-factor alphas. Model tables give parameter values and the model-implied counterparts of the empirical regressions. Appendix tables list data sets and headquarters employment shares.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with the view", "likely to", "suggest", "explicit discussion of simplifying assumptions"]}, "claim_evidence": [{"claim": "Wages respond more strongly to aggregate shocks in high local beta MSAs", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Housing and commercial real estate returns are more procyclical in high local beta areas", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Firms in higher local beta areas have lower conditional equity betas", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Firms in higher local beta areas earn lower future returns, especially when real estate holdings are low", "evidence_type": "table", "evidence_ref": "Tables VI, VII, VIII"}, {"claim": "Calibrated model reproduces procyclical factor prices and lower firm risk in high-beta areas", "evidence_type": "table", "evidence_ref": "Tables X, XI"}, {"claim": "Labor mobility across counties is low at annual frequency", "evidence_type": "descriptive_statistic", "evidence_ref": "Intro paragraph 4, footnote 2"}], "structural_summary": "The introduction moves from a stylized fact about immobile local factors to a mechanism illustrated by an example, then previews the model and results. It defends its simplifying assumptions, lists extensions and robustness checks, and positions the paper in the literature. A simplified labor-only model builds intuition before the data section constructs the local beta. The empirical section first tests factor prices, then firm risk and returns. A full three-input production model is calibrated to match the empirical patterns. The conclusion restates the findings without discussing policy.", "writing_profile_keywords": ["local beta", "production-based asset pricing", "model-then-evidence-then-full-model", "natural hedge mechanism", "panel regressions", "calibration"], "confidence": 0.8, "notes": "Standard-error treatment is not visible; the introduction mentions checking various residual correlation assumptions. The model in Section IV is calibrated, not estimated."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.23, "usage": {"input": 8108, "output": 2021, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073646, "request_id": "req_011CggpTfUy1ngLBvHDm2Qvu", "cached": false, "ts": "2026-11-03T22:55:42+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/299bc406441467932eec9d28b244c56d8a29bfb606a3ca9458fcf52895b3918d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"limits of arbitrage / financial intermediation\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; closed-form equilibrium analysis of an infinite-horizon model with segmented markets and collateral-constrained arbitrageurs\",\"key_assumptions\":[\"markets for paired assets are segmented, with outside investors trading in only one asset\",\"a collateral-based financial constraint applies to arbitrageurs\",\"the set of active opportunity characteristics is deterministic\",\"arbitrage risk is small (near-identical payoffs) in the risky case\",\"a symmetric equilibrium in which arbitrageur wealth is the only state variable\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"No tables; results are delivered through equations and propositions.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"scope conditions (e.g., small arbitrage risk)\",\"qualifiers such as 'generally' and 'in that sense'\",\"explicit simplifying assumptions\"]},\"claim_evidence\":[{\"claim\":\"Arbitrage capital dynamics are self-correcting: depleted capital raises returns, which gradually replenishes capital.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B.2 and II.C\"},{\"claim\":\"In equilibrium arbitrageurs invest in opportunities above a cutoff return per unit of collateral, and the cutoff falls as capital rises.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B\"},{\"claim\":\"With small arbitrage risk, capital follows approximate AR(1) dynamics and risk acts mainly through the financial constraint rather than risk aversion.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.B\"},{\"claim\":\"Spreads/returns relate positively to betas with respect to arbitrage capital in the cross-section.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C.1\"},{\"claim\":\"Arbitrageur diversification causes contagion across markets but lowers wealth variance and spread volatility.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.D\"},{\"claim\":\"Empirical studies document price discrepancies and link arbitrage capital to anomaly size.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction paragraph 0, footnote 1\"}],\"structural_summary\":\"Opens with tension between frictionless arbitrage and documented anomalies, then previews a model built in stages. Riskless arbitrage is solved first as a benchmark (cutoff equilibrium, self-correcting dynamics). Small-risk arbitrage follows, with cross-sectional and shock-response implications. Capital mobility and contagion come last. Literature positioning appears inside the introduction. Conclusion summarizes the analysis.\",\"writing_profile_keywords\":[\"dynamic equilibrium model\",\"limits of arbitrage\",\"closed-form solutions\",\"benchmark-then-extension\",\"segmented markets\",\"collateral constraints\"],\"confidence\":0.85,\"notes\":\"Pure theory paper with no tables or figures. Sections II and III are analytical results within the theory, so they are labeled theory rather than results. No roadmap paragraph is seen in the introduction excerpt.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "limits of arbitrage / financial intermediation", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; closed-form equilibrium analysis of an infinite-horizon model with segmented markets and collateral-constrained arbitrageurs", "key_assumptions": ["markets for paired assets are segmented, with outside investors trading in only one asset", "a collateral-based financial constraint applies to arbitrageurs", "the set of active opportunity characteristics is deterministic", "arbitrage risk is small (near-identical payoffs) in the risky case", "a symmetric equilibrium in which arbitrageur wealth is the only state variable"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}], "contribution_paragraph_index": 5, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "No tables; results are delivered through equations and propositions.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["scope conditions (e.g., small arbitrage risk)", "qualifiers such as 'generally' and 'in that sense'", "explicit simplifying assumptions"]}, "claim_evidence": [{"claim": "Arbitrage capital dynamics are self-correcting: depleted capital raises returns, which gradually replenishes capital.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B.2 and II.C"}, {"claim": "In equilibrium arbitrageurs invest in opportunities above a cutoff return per unit of collateral, and the cutoff falls as capital rises.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B"}, {"claim": "With small arbitrage risk, capital follows approximate AR(1) dynamics and risk acts mainly through the financial constraint rather than risk aversion.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.B"}, {"claim": "Spreads/returns relate positively to betas with respect to arbitrage capital in the cross-section.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C.1"}, {"claim": "Arbitrageur diversification causes contagion across markets but lowers wealth variance and spread volatility.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.D"}, {"claim": "Empirical studies document price discrepancies and link arbitrage capital to anomaly size.", "evidence_type": "prior_literature", "evidence_ref": "Introduction paragraph 0, footnote 1"}], "structural_summary": "Opens with tension between frictionless arbitrage and documented anomalies, then previews a model built in stages. Riskless arbitrage is solved first as a benchmark (cutoff equilibrium, self-correcting dynamics). Small-risk arbitrage follows, with cross-sectional and shock-response implications. Capital mobility and contagion come last. Literature positioning appears inside the introduction. Conclusion summarizes the analysis.", "writing_profile_keywords": ["dynamic equilibrium model", "limits of arbitrage", "closed-form solutions", "benchmark-then-extension", "segmented markets", "collateral constraints"], "confidence": 0.85, "notes": "Pure theory paper with no tables or figures. Sections II and III are analytical results within the theory, so they are labeled theory rather than results. No roadmap paragraph is seen in the introduction excerpt."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.28, "usage": {"input": 5439, "output": 2009, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06273, "request_id": "req_011Cggq12oV1W83oJgBkdHSU", "cached": false, "ts": "2026-11-03T23:02:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2a9715297db6415cdec3167e8b5c80019034e68d34d40a70aade6795089c2425.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"theoretical Kyle-type model with nonfundamental speculation; corrigendum\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical correction of equilibrium characterization)\",\"strategy\":\"not applicable; analytical re-derivation of the original model's equilibrium including the omitted second-order condition\",\"key_assumptions\":[\"original Madrigal (1996) model setup and notation retained\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"replication_or_correction\"],\"roadmap_present\":true,\"literature_positioning_location\":\"none\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"appendix\"},{\"idx\":4,\"role\":\"appendix\"},{\"idx\":5,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"The tables set the original equations and the original numerical table beside corrected versions, giving each original expression next to its correction. They contain no statistical estimates.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":false},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'may not exist' conditional existence statements\",\"parameter-region qualifiers ('if information is sufficiently precise')\"]},\"claim_evidence\":[{\"claim\":\"A linear pure strategy equilibrium may fail to exist in Madrigal's model when the speculator's noise-trading information is sufficiently precise\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.A; Appendix B (Proof of Proposition 2)\"},{\"claim\":\"The original characterization omits a second-order condition\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendix A\"},{\"claim\":\"Calculation errors in the original derivations mislead implications for liquidity and early price discovery\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables CI and CII; Section II.B; Appendix C\"},{\"claim\":\"Order-flow-based speculation remains relevant, e.g., HFT and proprietary trading\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Introduction\"}],\"structural_summary\":\"This is a short corrigendum to a published Kyle-type model. A single introductory paragraph motivates current relevance, states that the original solution is partly wrong, and previews two corrections with a section roadmap. Section I restates the original setup in its original notation and gives the correct equilibrium characterization, including the second-order condition. Section II first proves that the equilibrium need not exist and then corrects calculation errors and their implications. Appendices hold the derivations and the proof, and corrected material is placed in tables and an internet appendix. There is no conclusion and essentially no literature positioning.\",\"writing_profile_keywords\":[\"corrigendum\",\"Kyle model\",\"equilibrium existence\",\"second-order condition\",\"calculation errors\",\"original notation\",\"proof in appendix\"],\"confidence\":0.88,\"notes\":\"No conclusion section is visible in the excerpts. The abstract has no explicit question or method sentence. The 'gap' role is assigned to the sentence on the omitted second-order condition, which explains the error.\"}", "json": {"genre": "replication_or_correction", "subfield": "market_microstructure", "subfield_secondary": "theoretical Kyle-type model with nonfundamental speculation; corrigendum", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical correction of equilibrium characterization)", "strategy": "not applicable; analytical re-derivation of the original model's equilibrium including the omitted second-order condition", "key_assumptions": ["original Madrigal (1996) model setup and notation retained"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["replication_or_correction"], "roadmap_present": true, "literature_positioning_location": "none"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "appendix"}, {"idx": 4, "role": "appendix"}, {"idx": 5, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "The tables set the original equations and the original numerical table beside corrected versions, giving each original expression next to its correction. They contain no statistical estimates.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": false}, "hedging": {"level": "low", "typical_devices": ["'may not exist' conditional existence statements", "parameter-region qualifiers ('if information is sufficiently precise')"]}, "claim_evidence": [{"claim": "A linear pure strategy equilibrium may fail to exist in Madrigal's model when the speculator's noise-trading information is sufficiently precise", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.A; Appendix B (Proof of Proposition 2)"}, {"claim": "The original characterization omits a second-order condition", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendix A"}, {"claim": "Calculation errors in the original derivations mislead implications for liquidity and early price discovery", "evidence_type": "table", "evidence_ref": "Tables CI and CII; Section II.B; Appendix C"}, {"claim": "Order-flow-based speculation remains relevant, e.g., HFT and proprietary trading", "evidence_type": "none_stated", "evidence_ref": "Introduction"}], "structural_summary": "This is a short corrigendum to a published Kyle-type model. A single introductory paragraph motivates current relevance, states that the original solution is partly wrong, and previews two corrections with a section roadmap. Section I restates the original setup in its original notation and gives the correct equilibrium characterization, including the second-order condition. Section II first proves that the equilibrium need not exist and then corrects calculation errors and their implications. Appendices hold the derivations and the proof, and corrected material is placed in tables and an internet appendix. There is no conclusion and essentially no literature positioning.", "writing_profile_keywords": ["corrigendum", "Kyle model", "equilibrium existence", "second-order condition", "calculation errors", "original notation", "proof in appendix"], "confidence": 0.88, "notes": "No conclusion section is visible in the excerpts. The abstract has no explicit question or method sentence. The 'gap' role is assigned to the sentence on the omitted second-order condition, which explains the error."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 12.33, "usage": {"input": 2692, "output": 1536, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.042282, "request_id": "req_011Cggpn8ywscmTMrAR9b42d", "cached": false, "ts": "2026-11-03T22:59:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2ac62d87ff3e8905a031288ca362560fded28341510bdec0fb24fcb37c5e9ef7.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"corporate_finance (real investment / real options)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-sectional relation between stock-level capacity overhang (installed minus optimal capacity) and expected stock returns, net of standard pricing factors and characteristics\",\"strategy\":\"Recursive stochastic frontier estimation of capacity overhang, followed by value-weighted portfolio sorts with factor-model alphas and Fama-MacBeth regressions with characteristic controls\",\"key_assumptions\":[\"Capacity overhang is non-negative (one-sided distributional assumption)\",\"Optimal capacity determinants and overhang determinants are correctly separated\",\"Excluded adjustment costs (fixed, convex) do not invalidate the overhang estimate\",\"Real-time recursive estimation avoids look-ahead bias\"],\"threats_addressed\":[\"Look-ahead bias, addressed by recursive estimation\",\"Results driven by micro stocks, addressed with size-subsample sorts\",\"Omitted pricing factors, addressed with CAPM, Q-theory and FF5 alphas\",\"Measurement validity, addressed by validation against investment behavior and survey capacity utilization\",\"Fixed and convex adjustment costs, discussed as a caveat\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"conclusion\"},{\"idx\":4,\"role\":\"appendix\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes open with what the table presents and define the capacity proxy used (PPE plus intangibles, PPE, or total assets) and the optimal-capacity determinants. They point readers to the appendix variable-definition tables. Return timing (month t on t-1 characteristics) is stated explicitly. Panels separate specifications.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests' / 'our evidence supports'\",\"'potentially explaining'\",\"explicit caveat paragraph on excluded adjustment costs\",\"'we try to close this gap'\",\"'unlikely to greatly distort'\"]},\"claim_evidence\":[{\"claim\":\"An extended Pindyck-type real options model can yield positive, negative, or U-shaped expected return–capacity overhang relations.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Figures 1-3; Appendix B Proposition 1\"},{\"claim\":\"The capacity overhang estimate tracks firm investment behavior and industry capacity utilization from surveys.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6; Section II.A.4\"},{\"claim\":\"Value-weighted high-minus-low capacity overhang portfolios earn about -12.5% per year, with t around -4.2.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The negative relation survives CAPM, Q-theory and five-factor adjustment, loading negatively on profitability factors.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 7\"},{\"claim\":\"The negative relation persists across size, reversibility and economic-state subsamples.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-V\"},{\"claim\":\"Fama-MacBeth regressions yield significant negative capacity overhang premia, with t-statistics below -3.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Capacity overhang helps explain momentum and profitability anomalies but not value and investment anomalies.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII-IX\"},{\"claim\":\"Convex adjustment costs are negligible relative to other adjustment costs.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Cooper and Haltiwanger (2006); Bloom (2009)\"}],\"structural_summary\":\"The introduction opens with conflicting theoretical predictions about how capacity overhang relates to expected returns and notes that empirical work is scarce. It then sketches the model mechanism, previews a novel stochastic-frontier measure and the main return results with magnitudes, and defends the measure against excluded adjustment costs. Literature positioning and contributions, both empirical and theoretical, close the introduction. A theory section extends a real options model to generate testable shapes, with proofs in appendices. One long empirical section follows: measure construction and validation, then pricing tests (univariate and double sorts, state timing, Fama-MacBeth), then horse races against anomalies. A short conclusion restates the findings as support for divestment-option models.\",\"writing_profile_keywords\":[\"real options\",\"capacity overhang\",\"stochastic frontier\",\"portfolio sorts\",\"Fama-MacBeth\",\"anomaly horse races\",\"theory-guided measurement\",\"competing predictions\"],\"confidence\":0.82,\"notes\":\"Standard error treatment is not visible in the excerpts. Robustness tests appear within the results subsections, and an Internet Appendix is referenced, so location is coded as mixed. The empirical section mixes measurement design and results; it is coded as results.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "corporate_finance (real investment / real options)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-sectional relation between stock-level capacity overhang (installed minus optimal capacity) and expected stock returns, net of standard pricing factors and characteristics", "strategy": "Recursive stochastic frontier estimation of capacity overhang, followed by value-weighted portfolio sorts with factor-model alphas and Fama-MacBeth regressions with characteristic controls", "key_assumptions": ["Capacity overhang is non-negative (one-sided distributional assumption)", "Optimal capacity determinants and overhang determinants are correctly separated", "Excluded adjustment costs (fixed, convex) do not invalidate the overhang estimate", "Real-time recursive estimation avoids look-ahead bias"], "threats_addressed": ["Look-ahead bias, addressed by recursive estimation", "Results driven by micro stocks, addressed with size-subsample sorts", "Omitted pricing factors, addressed with CAPM, Q-theory and FF5 alphas", "Measurement validity, addressed by validation against investment behavior and survey capacity utilization", "Fixed and convex adjustment costs, discussed as a caveat"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "contribution_statement"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_data_or_measure", "new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "conclusion"}, {"idx": 4, "role": "appendix"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "alternative_specification"]}, "tables": {"convention_notes": "Table notes open with what the table presents and define the capacity proxy used (PPE plus intangibles, PPE, or total assets) and the optimal-capacity determinants. They point readers to the appendix variable-definition tables. Return timing (month t on t-1 characteristics) is stated explicitly. Panels separate specifications.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests' / 'our evidence supports'", "'potentially explaining'", "explicit caveat paragraph on excluded adjustment costs", "'we try to close this gap'", "'unlikely to greatly distort'"]}, "claim_evidence": [{"claim": "An extended Pindyck-type real options model can yield positive, negative, or U-shaped expected return–capacity overhang relations.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Figures 1-3; Appendix B Proposition 1"}, {"claim": "The capacity overhang estimate tracks firm investment behavior and industry capacity utilization from surveys.", "evidence_type": "figure", "evidence_ref": "Figures 5-6; Section II.A.4"}, {"claim": "Value-weighted high-minus-low capacity overhang portfolios earn about -12.5% per year, with t around -4.2.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The negative relation survives CAPM, Q-theory and five-factor adjustment, loading negatively on profitability factors.", "evidence_type": "table", "evidence_ref": "Table II; Figure 7"}, {"claim": "The negative relation persists across size, reversibility and economic-state subsamples.", "evidence_type": "table", "evidence_ref": "Tables III-V"}, {"claim": "Fama-MacBeth regressions yield significant negative capacity overhang premia, with t-statistics below -3.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Capacity overhang helps explain momentum and profitability anomalies but not value and investment anomalies.", "evidence_type": "table", "evidence_ref": "Tables VII-IX"}, {"claim": "Convex adjustment costs are negligible relative to other adjustment costs.", "evidence_type": "prior_literature", "evidence_ref": "Cooper and Haltiwanger (2006); Bloom (2009)"}], "structural_summary": "The introduction opens with conflicting theoretical predictions about how capacity overhang relates to expected returns and notes that empirical work is scarce. It then sketches the model mechanism, previews a novel stochastic-frontier measure and the main return results with magnitudes, and defends the measure against excluded adjustment costs. Literature positioning and contributions, both empirical and theoretical, close the introduction. A theory section extends a real options model to generate testable shapes, with proofs in appendices. One long empirical section follows: measure construction and validation, then pricing tests (univariate and double sorts, state timing, Fama-MacBeth), then horse races against anomalies. A short conclusion restates the findings as support for divestment-option models.", "writing_profile_keywords": ["real options", "capacity overhang", "stochastic frontier", "portfolio sorts", "Fama-MacBeth", "anomaly horse races", "theory-guided measurement", "competing predictions"], "confidence": 0.82, "notes": "Standard error treatment is not visible in the excerpts. Robustness tests appear within the results subsections, and an Internet Appendix is referenced, so location is coded as mixed. The empirical section mixes measurement design and results; it is coded as results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.98, "usage": {"input": 8479, "output": 2508, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08487, "request_id": "req_011CggpwSa757mg6NkeWrKJB", "cached": false, "ts": "2026-11-03T23:02:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2ac63d13c067c9a6e4963e037323128c54af00d32a50f15c3f61663d475157ab.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"capital structure / dynamic trade-off theory\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model); comparative statics of optimal debt and leverage with respect to current and expected future profitability\",\"strategy\":\"analytical characterization of a stochastic dynamic model with zero-maturity debt, Poisson regime arrivals for EBIT, endogenous default cost and a borrowing constraint\",\"key_assumptions\":[\"EBIT i.i.d. across Poisson-arriving regimes (relaxed in Section VII)\",\"zero-maturity debt valued at par\",\"risk-neutral lenders with the same information as shareholders\",\"no renegotiation and no flotation costs\",\"nondecreasing density of EBIT, which ensures concavity\"],\"threats_addressed\":[\"i.i.d. assumption relaxed by allowing persistence across regimes\",\"future profitability examined via rightward translation of the distribution\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"theory\"},{\"idx\":7,\"role\":\"theory\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"A single qualitative summary table maps the sign of profitability effects on optimal debt across scenarios, noting whether the borrowing constraint binds. It contains no estimates.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"conditioning results on whether the trade-off theory is operative\",\"acknowledging that results depend on the i.i.d. assumption\",\"qualifiers such as 'potentially' and 'at least in the example'\",\"framing results as an alternative explanation\"]},\"claim_evidence\":[{\"claim\":\"When the borrowing constraint is slack, optimal debt equates the marginal interest tax shield with the marginal expected default cost.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV.A; Appendix A\"},{\"claim\":\"Under i.i.d. EBIT, optimal debt is invariant to current EBIT where the trade-off theory holds, so leverage falls as EBIT rises.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections II-IV; Figure 1\"},{\"claim\":\"The tax rate determines which of the three scenarios prevails, via explicit threshold tax rates.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section V; Figure 2\"},{\"claim\":\"With a uniform distribution, a rightward shift of the profitability distribution reduces optimal debt in the trade-off region.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section VI; Figure 3; Table I\"},{\"claim\":\"The negative profitability-leverage relation matches empirical findings.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction footnotes 3-4\"},{\"claim\":\"With persistence across regimes in a uniform example, both debt and leverage decrease in EBIT.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section VII\"}],\"structural_summary\":\"A single-author theory paper opens with a puzzle: the conventional reading of trade-off theory conflicts with the empirical negative relation between profitability and leverage. The introduction motivates the modeling choices (zero-maturity debt, Poisson EBIT regimes, endogenous borrowing limit), previews the analytical results, and gives a roadmap. The body builds the environment, derives the Bellman problem, partitions outcomes into three scenarios, and analyzes the interior scenario as the trade-off case. It then derives tax-rate thresholds, examines shifts in future profitability, and extends the model to persistent EBIT. The conclusion restates which conditions make the trade-off theory operative and lists extensions. Proofs are placed in appendices.\",\"writing_profile_keywords\":[\"analytical dynamic model\",\"trade-off theory\",\"puzzle resolution\",\"scenario-based characterization\",\"comparative statics\",\"proofs in appendix\",\"single author\",\"modeling-assumption justification\"],\"confidence\":0.88,\"notes\":\"Purely theoretical paper with no empirical estimation; empirical facts are cited only as motivation. The section on persistence across regimes is treated as a robustness-type extension within the main text.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "capital structure / dynamic trade-off theory", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model); comparative statics of optimal debt and leverage with respect to current and expected future profitability", "strategy": "analytical characterization of a stochastic dynamic model with zero-maturity debt, Poisson regime arrivals for EBIT, endogenous default cost and a borrowing constraint", "key_assumptions": ["EBIT i.i.d. across Poisson-arriving regimes (relaxed in Section VII)", "zero-maturity debt valued at par", "risk-neutral lenders with the same information as shareholders", "no renegotiation and no flotation costs", "nondecreasing density of EBIT, which ensures concavity"], "threats_addressed": ["i.i.d. assumption relaxed by allowing persistence across regimes", "future profitability examined via rightward translation of the distribution"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "theory"}, {"idx": 7, "role": "theory"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["other"]}, "tables": {"convention_notes": "A single qualitative summary table maps the sign of profitability effects on optimal debt across scenarios, noting whether the borrowing constraint binds. It contains no estimates.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["conditioning results on whether the trade-off theory is operative", "acknowledging that results depend on the i.i.d. assumption", "qualifiers such as 'potentially' and 'at least in the example'", "framing results as an alternative explanation"]}, "claim_evidence": [{"claim": "When the borrowing constraint is slack, optimal debt equates the marginal interest tax shield with the marginal expected default cost.", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV.A; Appendix A"}, {"claim": "Under i.i.d. EBIT, optimal debt is invariant to current EBIT where the trade-off theory holds, so leverage falls as EBIT rises.", "evidence_type": "equation_or_proof", "evidence_ref": "Sections II-IV; Figure 1"}, {"claim": "The tax rate determines which of the three scenarios prevails, via explicit threshold tax rates.", "evidence_type": "equation_or_proof", "evidence_ref": "Section V; Figure 2"}, {"claim": "With a uniform distribution, a rightward shift of the profitability distribution reduces optimal debt in the trade-off region.", "evidence_type": "equation_or_proof", "evidence_ref": "Section VI; Figure 3; Table I"}, {"claim": "The negative profitability-leverage relation matches empirical findings.", "evidence_type": "prior_literature", "evidence_ref": "Introduction footnotes 3-4"}, {"claim": "With persistence across regimes in a uniform example, both debt and leverage decrease in EBIT.", "evidence_type": "equation_or_proof", "evidence_ref": "Section VII"}], "structural_summary": "A single-author theory paper opens with a puzzle: the conventional reading of trade-off theory conflicts with the empirical negative relation between profitability and leverage. The introduction motivates the modeling choices (zero-maturity debt, Poisson EBIT regimes, endogenous borrowing limit), previews the analytical results, and gives a roadmap. The body builds the environment, derives the Bellman problem, partitions outcomes into three scenarios, and analyzes the interior scenario as the trade-off case. It then derives tax-rate thresholds, examines shifts in future profitability, and extends the model to persistent EBIT. The conclusion restates which conditions make the trade-off theory operative and lists extensions. Proofs are placed in appendices.", "writing_profile_keywords": ["analytical dynamic model", "trade-off theory", "puzzle resolution", "scenario-based characterization", "comparative statics", "proofs in appendix", "single author", "modeling-assumption justification"], "confidence": 0.88, "notes": "Purely theoretical paper with no empirical estimation; empirical facts are cited only as motivation. The section on persistence across regimes is treated as a robustness-type extension within the main text."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.05, "usage": {"input": 8022, "output": 2281, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078502, "request_id": "req_011CggpNrJwyPzEQPVYETQou", "cached": false, "ts": "2026-11-03T22:54:38+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2d0b9356ab798fd506afbacc97cf09f75d9ee809d88fd47d46e8ec1ba60238ce.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"asset_pricing (term structure / monetary policy communication)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model; equilibrium degree of policy-rate adjustment and welfare under discretion vs commitment)\",\"strategy\":\"not applicable; analytical rational-expectations model with private information and volatility-averse central bank, solved numerically for some parameter cases\",\"key_assumptions\":[\"Fed has private information about its long-run target rate\",\"Fed's objective includes bond market volatility (reduced form)\",\"Market forms rational expectations about Fed's adjustment\",\"Discretionary meeting-by-meeting decision making\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"The tables report numerical equilibrium characterizations across parameter grids: volatility aversion and variance ratios, adjustment coefficients with and without commitment, and the optimal appointed-banker aversion. They are not regression tables. Notes define each symbol.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'in our setting' qualifiers\",\"'can help explain' / 'can be valuable'\",\"explicit caveat paragraph disclaiming guidance on transition\",\"'behaves as if' modeling framing\"]},\"claim_evidence\":[{\"claim\":\"FOMC members articulate gradual adjustment toward a target and cite financial market stability as a rationale.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section I; Coibion and Gorodnichenko (2012) transcript quotes\"},{\"claim\":\"Attention to financial market terms in FOMC deliberations varies over time and is substantial.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"In any rational expectations equilibrium, the policy rate underadjusts relative to the first best.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B; Appendix proofs\"},{\"claim\":\"For some parameters, Pareto-ranked multiple equilibria arise from feedback between market conjecture and Fed's gradualism.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2; Section II.B.2\"},{\"claim\":\"Discretion yields welfare losses vs commitment, and a less market-sensitive appointed central banker improves welfare.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.C; Table I; Appendix proofs\"},{\"claim\":\"In dynamic extensions, private-information shocks are impounded into the policy rate gradually over time.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV–V; Section III\"}],\"structural_summary\":\"Opens with an anecdotal motivation about Fed-watching, then introduces a model built on private information and volatility aversion. The introduction sketches the signal-jamming intuition, the commitment problem, and a Rogoff-style remedy, then previews positive and normative results and gives a roadmap. A short section of narrative transcript evidence follows. A static model carries the core propositions on underadjustment, multiple equilibria, and time inconsistency, plus extensions to the objective function and to additional shocks. A dynamic section adds three-period and near-rational versions, followed by a discussion of further applications. The conclusion restates the contribution with an explicit caveat about transition policy, and proofs are in the appendix.\",\"writing_profile_keywords\":[\"theory\",\"signal-jamming\",\"time-consistency\",\"monetary policy\",\"gradualism\",\"normative implications\",\"numerical equilibrium tables\",\"narrative motivation\"],\"confidence\":0.85,\"notes\":\"Sections I and IV are labeled 'other': Section I is motivating narrative/transcript evidence rather than formal results, and Section IV is a discussion of applications. The roadmap numbering in the introduction matches the section headings.\"}", "json": {"genre": "theoretical", "subfield": "macro_finance", "subfield_secondary": "asset_pricing (term structure / monetary policy communication)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model; equilibrium degree of policy-rate adjustment and welfare under discretion vs commitment)", "strategy": "not applicable; analytical rational-expectations model with private information and volatility-averse central bank, solved numerically for some parameter cases", "key_assumptions": ["Fed has private information about its long-run target rate", "Fed's objective includes bond market volatility (reduced form)", "Market forms rational expectations about Fed's adjustment", "Discretionary meeting-by-meeting decision making"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "The tables report numerical equilibrium characterizations across parameter grids: volatility aversion and variance ratios, adjustment coefficients with and without commitment, and the optimal appointed-banker aversion. They are not regression tables. Notes define each symbol.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'in our setting' qualifiers", "'can help explain' / 'can be valuable'", "explicit caveat paragraph disclaiming guidance on transition", "'behaves as if' modeling framing"]}, "claim_evidence": [{"claim": "FOMC members articulate gradual adjustment toward a target and cite financial market stability as a rationale.", "evidence_type": "prior_literature", "evidence_ref": "Section I; Coibion and Gorodnichenko (2012) transcript quotes"}, {"claim": "Attention to financial market terms in FOMC deliberations varies over time and is substantial.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "In any rational expectations equilibrium, the policy rate underadjusts relative to the first best.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B; Appendix proofs"}, {"claim": "For some parameters, Pareto-ranked multiple equilibria arise from feedback between market conjecture and Fed's gradualism.", "evidence_type": "figure", "evidence_ref": "Figure 2; Section II.B.2"}, {"claim": "Discretion yields welfare losses vs commitment, and a less market-sensitive appointed central banker improves welfare.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.C; Table I; Appendix proofs"}, {"claim": "In dynamic extensions, private-information shocks are impounded into the policy rate gradually over time.", "evidence_type": "table", "evidence_ref": "Tables IV–V; Section III"}], "structural_summary": "Opens with an anecdotal motivation about Fed-watching, then introduces a model built on private information and volatility aversion. The introduction sketches the signal-jamming intuition, the commitment problem, and a Rogoff-style remedy, then previews positive and normative results and gives a roadmap. A short section of narrative transcript evidence follows. A static model carries the core propositions on underadjustment, multiple equilibria, and time inconsistency, plus extensions to the objective function and to additional shocks. A dynamic section adds three-period and near-rational versions, followed by a discussion of further applications. The conclusion restates the contribution with an explicit caveat about transition policy, and proofs are in the appendix.", "writing_profile_keywords": ["theory", "signal-jamming", "time-consistency", "monetary policy", "gradualism", "normative implications", "numerical equilibrium tables", "narrative motivation"], "confidence": 0.85, "notes": "Sections I and IV are labeled 'other': Section I is motivating narrative/transcript evidence rather than formal results, and Section IV is a discussion of applications. The roadmap numbering in the introduction matches the section headings."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.12, "usage": {"input": 6525, "output": 2098, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068854, "request_id": "req_011CggptoPfLDa5p1KWqJVr5", "cached": false, "ts": "2026-11-03T23:01:23+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2d0f038836fbd5ddea5aba9f28832c2a6766c0547be9fe244ce48fe407c7c9d2.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"fixed income / public finance (municipal bonds)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Share of tax-adjusted municipal bond spreads attributable to default risk versus liquidity, and the implied default risk premium.\",\"strategy\":\"Triangulates three decompositions: (1) transaction-based regressions of spreads on liquidity measures to back out a liquidity component; (2) CDS premia used to measure the default component; (3) a quasi-natural experiment on within-bond spread changes around pre-refunding events, with bond and time controls.\",\"key_assumptions\":[\"Pre-refunding removes default risk without an offsetting liquidity effect (lower bound interpretation)\",\"Tax adjustment using top statutory rates correctly converts tax-exempt yields to taxable equivalents\",\"CDS premia measure pure default compensation\",\"Swap rate is an appropriate riskless benchmark\"],\"threats_addressed\":[\"omitted issuer/bond-specific factors in cross-pool comparisons (within-bond design)\",\"sensitivity to the tax adjustment method\",\"benchmark and sample selection\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes define dependent variables (tax-adjusted spread over interpolated swap rate), explain the observation level (bond-month, issuer-month), and describe sample filters. Tables move from sample construction and summary statistics to pricing regressions, cross-method comparisons, and implied premia.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'likely due to' for mechanism attribution\",\"lower-bound framing ('at least', 'assuming no offsetting effect')\",\"ranges rather than point estimates\",\"'could' for candidate channels\"]},\"claim_evidence\":[{\"claim\":\"Default risk explains roughly three-quarters or more of the average tax-adjusted muni spread.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, VII, VIII\"},{\"claim\":\"Pre-refunding lowers within-bond spreads by at least about 41%.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Implied default risk premium is an order of magnitude larger than in corporate bonds.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X; Berndt et al. (2005)\"},{\"claim\":\"Distressed issuers such as Detroit, Chicago and Illinois carry elevated default spreads, consistent with pension underfunding.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Municipal defaults are historically extremely rare.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Moody's (2015) default rates\"}],\"structural_summary\":\"The paper opens by contrasting recent municipal distress with very low historical default rates, then notes that the literature emphasizes taxes and illiquidity. It poses the question of how much of the spread is default versus liquidity and previews the headline share and the implied premium. It positions itself against a contemporaneous paper that reaches opposing conclusions, defending its own within-bond design. A data section is followed by a methods section covering the tax adjustment and three decompositions, then results, robustness, and a conclusion. The conclusion speculates on tax-related channels and draws policy implications.\",\"writing_profile_keywords\":[\"triangulation of methods\",\"decomposition\",\"contrast with contemporaneous work\",\"quasi-natural experiment\",\"policy implications\",\"single-author first person\"],\"confidence\":0.82,\"notes\":\"Standard-error conventions are not visible in the table note excerpts. Causal language is limited to the pre-refunding experiment; the main claims are decomposition shares.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "fixed income / public finance (municipal bonds)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Share of tax-adjusted municipal bond spreads attributable to default risk versus liquidity, and the implied default risk premium.", "strategy": "Triangulates three decompositions: (1) transaction-based regressions of spreads on liquidity measures to back out a liquidity component; (2) CDS premia used to measure the default component; (3) a quasi-natural experiment on within-bond spread changes around pre-refunding events, with bond and time controls.", "key_assumptions": ["Pre-refunding removes default risk without an offsetting liquidity effect (lower bound interpretation)", "Tax adjustment using top statutory rates correctly converts tax-exempt yields to taxable equivalents", "CDS premia measure pure default compensation", "Swap rate is an appropriate riskless benchmark"], "threats_addressed": ["omitted issuer/bond-specific factors in cross-pool comparisons (within-bond design)", "sensitivity to the tax adjustment method", "benchmark and sample selection"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_measure", "subsample", "alternative_specification"]}, "tables": {"convention_notes": "Notes define dependent variables (tax-adjusted spread over interpolated swap rate), explain the observation level (bond-month, issuer-month), and describe sample filters. Tables move from sample construction and summary statistics to pricing regressions, cross-method comparisons, and implied premia.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'likely due to' for mechanism attribution", "lower-bound framing ('at least', 'assuming no offsetting effect')", "ranges rather than point estimates", "'could' for candidate channels"]}, "claim_evidence": [{"claim": "Default risk explains roughly three-quarters or more of the average tax-adjusted muni spread.", "evidence_type": "table", "evidence_ref": "Tables IV, VII, VIII"}, {"claim": "Pre-refunding lowers within-bond spreads by at least about 41%.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Implied default risk premium is an order of magnitude larger than in corporate bonds.", "evidence_type": "table", "evidence_ref": "Table X; Berndt et al. (2005)"}, {"claim": "Distressed issuers such as Detroit, Chicago and Illinois carry elevated default spreads, consistent with pension underfunding.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Municipal defaults are historically extremely rare.", "evidence_type": "descriptive_statistic", "evidence_ref": "Moody's (2015) default rates"}], "structural_summary": "The paper opens by contrasting recent municipal distress with very low historical default rates, then notes that the literature emphasizes taxes and illiquidity. It poses the question of how much of the spread is default versus liquidity and previews the headline share and the implied premium. It positions itself against a contemporaneous paper that reaches opposing conclusions, defending its own within-bond design. A data section is followed by a methods section covering the tax adjustment and three decompositions, then results, robustness, and a conclusion. The conclusion speculates on tax-related channels and draws policy implications.", "writing_profile_keywords": ["triangulation of methods", "decomposition", "contrast with contemporaneous work", "quasi-natural experiment", "policy implications", "single-author first person"], "confidence": 0.82, "notes": "Standard-error conventions are not visible in the table note excerpts. Causal language is limited to the pre-refunding experiment; the main claims are decomposition shares."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.34, "usage": {"input": 7027, "output": 2049, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069882, "request_id": "req_011Cggpf7iLfvmEhFxGrLnum", "cached": false, "ts": "2026-11-03T22:58:18+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2e48830ce6dbd2ff98c14df5c17e5620b4b1e6e237a19c949c5dd799152353e3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Ratio of the variance of news about average expected inflation over a bond's life to the variance of nominal yield shocks (inflation variance ratio)\",\"strategy\":\"Campbell-Ammer accounting decomposition. Inflation news is measured from revisions in survey forecasts (Blue Chip, SPF), extended to long horizons with a trend-cycle model. Yield innovations come from forecasting regressions or a martingale assumption. Data ratios are compared with model-implied ratios from equilibrium models.\",\"key_assumptions\":[\"Survey consensus forecasts proxy for market inflation expectations\",\"Yield innovations are well measured by OLS forecasts or a martingale assumption\",\"Long-horizon expectations follow a random walk plus AR(1) trend-cycle structure\",\"VAR with cointegration of yields and inflation and stationary real rates (Section IV)\"],\"threats_addressed\":[\"Stability across monetary regimes and subperiods\",\"Alternative methods for constructing yield innovations\",\"Alternative parameterizations of equilibrium models\",\"Statistical uncertainty in persistence of real rates\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":true,\"sentence_index\":3},\"results_note_placeholder\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"other\"},{\"idx\":8,\"role\":\"results\"},{\"idx\":9,\"role\":\"theory\"},{\"idx\":10,\"role\":\"other\"},{\"idx\":11,\"role\":\"theory\"},{\"idx\":12,\"role\":\"other\"},{\"idx\":13,\"role\":\"theory\"},{\"idx\":14,\"role\":\"results\"},{\"idx\":15,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"alternative_measure\",\"alternative_specification\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Notes describe how shocks are constructed (survey revisions, forecasting model or martingale assumption) and state units (annualized percent per quarter). Model-property tables report variance decompositions for alternative parameterizations.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"explicit statements that components cannot be determined reliably\",\"'at least without imposing restrictive assumptions'\",\"'point estimates are consistent with... but uncertainty cannot rule out'\",\"'in the neighborhood of'\"]},\"claim_evidence\":[{\"claim\":\"Inflation news variance is roughly 10-20% of quarterly yield shock variance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Table III\"},{\"claim\":\"Low inflation variance ratios hold across monetary regimes and subperiods.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Yield innovations are more than twice as volatile as inflation news.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-4\"},{\"claim\":\"Long-run risk models imply variance ratios near one because real-rate news is small.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 7-8, Table V\"},{\"claim\":\"Habit formation models fit the documented volatilities somewhat better.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 10\"},{\"claim\":\"The data cannot separate real-rate news from term premium news in yield variance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Table VI, Figure 11\"}],\"structural_summary\":\"The introduction frames a literature lacking consensus facts and poses one focused question about the inflation share of yield shocks, built on an accounting identity. It previews the low ratio and the inability to split the remainder between real rates and term premia. Section I defines the ratio and its measurement. Section II estimates it with surveys at short horizons and a trend-cycle model at long horizons, including subsamples and a revisit of earlier work. Section III compares the facts with long-run risk, New Keynesian and habit models. Section IV estimates a dynamic model and finds the remaining decomposition is not identified. The brief conclusion restates the robust fact and its use for evaluating models.\",\"writing_profile_keywords\":[\"accounting identity\",\"stylized fact\",\"model evaluation\",\"survey expectations\",\"variance decomposition\",\"candid negative result\",\"single-author voice\"],\"confidence\":0.75,\"notes\":\"Several section entries (idx 3-7, 10, 12) are figure-panel artifacts parsed as headings. Significance reporting conventions are not visible in the excerpts. The abstract-moves object contains one non-schema key, \\\"results_note_placeholder\\\" (set to absent), which should be removed if strict schema validation is required.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "other", "identification": {"estimand": "Ratio of the variance of news about average expected inflation over a bond's life to the variance of nominal yield shocks (inflation variance ratio)", "strategy": "Campbell-Ammer accounting decomposition. Inflation news is measured from revisions in survey forecasts (Blue Chip, SPF), extended to long horizons with a trend-cycle model. Yield innovations come from forecasting regressions or a martingale assumption. Data ratios are compared with model-implied ratios from equilibrium models.", "key_assumptions": ["Survey consensus forecasts proxy for market inflation expectations", "Yield innovations are well measured by OLS forecasts or a martingale assumption", "Long-horizon expectations follow a random walk plus AR(1) trend-cycle structure", "VAR with cointegration of yields and inflation and stationary real rates (Section IV)"], "threats_addressed": ["Stability across monetary regimes and subperiods", "Alternative methods for constructing yield innovations", "Alternative parameterizations of equilibrium models", "Statistical uncertainty in persistence of real rates"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": true, "sentence_index": 3}, "results_note_placeholder": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "other"}, {"idx": 8, "role": "results"}, {"idx": 9, "role": "theory"}, {"idx": 10, "role": "other"}, {"idx": 11, "role": "theory"}, {"idx": 12, "role": "other"}, {"idx": 13, "role": "theory"}, {"idx": 14, "role": "results"}, {"idx": 15, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["subsample", "alternative_measure", "alternative_specification", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Notes describe how shocks are constructed (survey revisions, forecasting model or martingale assumption) and state units (annualized percent per quarter). Model-property tables report variance decompositions for alternative parameterizations.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["explicit statements that components cannot be determined reliably", "'at least without imposing restrictive assumptions'", "'point estimates are consistent with... but uncertainty cannot rule out'", "'in the neighborhood of'"]}, "claim_evidence": [{"claim": "Inflation news variance is roughly 10-20% of quarterly yield shock variance.", "evidence_type": "table", "evidence_ref": "Table I, Table III"}, {"claim": "Low inflation variance ratios hold across monetary regimes and subperiods.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Yield innovations are more than twice as volatile as inflation news.", "evidence_type": "figure", "evidence_ref": "Figures 1-4"}, {"claim": "Long-run risk models imply variance ratios near one because real-rate news is small.", "evidence_type": "figure", "evidence_ref": "Figures 7-8, Table V"}, {"claim": "Habit formation models fit the documented volatilities somewhat better.", "evidence_type": "figure", "evidence_ref": "Figure 10"}, {"claim": "The data cannot separate real-rate news from term premium news in yield variance.", "evidence_type": "table", "evidence_ref": "Table IV, Table VI, Figure 11"}], "structural_summary": "The introduction frames a literature lacking consensus facts and poses one focused question about the inflation share of yield shocks, built on an accounting identity. It previews the low ratio and the inability to split the remainder between real rates and term premia. Section I defines the ratio and its measurement. Section II estimates it with surveys at short horizons and a trend-cycle model at long horizons, including subsamples and a revisit of earlier work. Section III compares the facts with long-run risk, New Keynesian and habit models. Section IV estimates a dynamic model and finds the remaining decomposition is not identified. The brief conclusion restates the robust fact and its use for evaluating models.", "writing_profile_keywords": ["accounting identity", "stylized fact", "model evaluation", "survey expectations", "variance decomposition", "candid negative result", "single-author voice"], "confidence": 0.75, "notes": "Several section entries (idx 3-7, 10, 12) are figure-panel artifacts parsed as headings. Significance reporting conventions are not visible in the excerpts. The abstract-moves object contains one non-schema key, \"results_note_placeholder\" (set to absent), which should be removed if strict schema validation is required."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.12, "usage": {"input": 7669, "output": 2227, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07601, "request_id": "req_011Cggq4EWGXCbkSqozjtKs9", "cached": false, "ts": "2026-11-03T23:03:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2ecb08379d79c8d07ae8ef2a950b8ce9fa0999fa753c1294e0bc0928da10bf8b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Changes in customer execution costs and in dealer capital commitment and market-quality measures across subperiods relative to the precrisis benchmark, and how these changes differ between bank-affiliated and nonbank dealers.\",\"strategy\":\"Time-series regressions on period indicators with controls, with the precrisis period as benchmark. Bank-affiliated dealers are compared with nonbank dealers to attribute changes to bank-focused regulation. Trades likely to be electronic are compared with traditional trades, and stressful days are examined separately.\",\"key_assumptions\":[\"Nonbank dealers serve as a comparison group unaffected by bank-focused regulation\",\"Control variables capture relevant changes in market conditions over time\",\"Trade characteristics proxy for the likelihood of electronic intermediation\"],\"threats_addressed\":[\"Compositional shifts in trading toward more liquid bonds\",\"Growth of electronic trading reducing search costs\",\"Competition from corporate bond ETFs and mutual funds\",\"Confounding from the TRACE transparency phase-in, handled by excluding 2002–2005\",\"Liquidity changes concentrated on stressful days\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"p_values\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables move from sample construction and summary statistics to transaction-cost estimates, then capital-commitment statistics and time-series regressions, then bank-versus-nonbank and stressful-day splits. Regressions use period indicators with the precrisis period as benchmark. Notes state the data sources, sample period, aggregation level, and dollar size cutoffs.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'likely contribute'\",\"'supports the view'\",\"'may point to'\",\"explicit disclaimer that net welfare and macroeconomic stability effects are not assessed\",\"'it will be of interest' framing of future questions\"]},\"claim_evidence\":[{\"claim\":\"Average customer execution costs are roughly unchanged over 2006–2016 apart from a temporary crisis spike.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 1\"},{\"claim\":\"Dealer capital commitment, turnover and trade size fell in the recent period relative to the precrisis period.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V; Figure 2\"},{\"claim\":\"The declines are driven by bank-affiliated dealers, while nonbank dealers increased their commitment.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"Electronic trading reduced the need for dealer capital in segments where electronic venues are most likely.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Section VI.D\"},{\"claim\":\"Bank versus nonbank differences also appear on stressful days.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"}],\"structural_summary\":\"The paper opens with policy and practitioner concern about corporate bond liquidity and notes the disagreement over whether liquidity has deteriorated. It argues that execution costs alone are insufficient and introduces capital commitment measures. A literature and regulation section precedes the data section. Results then proceed from transaction costs, to dealer-level measures, to time-series regressions with period dummies. Next comes a bank versus nonbank comparison as the main attribution device for regulation, alongside electronic and ETF alternative explanations, followed by a stressful-day analysis. The conclusion restates the findings, supports Duffie's regulatory prediction, disclaims any welfare assessment, and poses future questions.\",\"writing_profile_keywords\":[\"descriptive-to-attributional progression\",\"policy-relevant\",\"subperiod comparison\",\"bank vs nonbank contrast\",\"alternative explanations weighed\",\"measured causal hedging\"],\"confidence\":0.82,\"notes\":\"No equations are visible. Causal attribution to regulation rests on comparing bank-affiliated and nonbank dealers. Robustness checks are partly in an Internet Appendix and partly in the stressful-day section.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Changes in customer execution costs and in dealer capital commitment and market-quality measures across subperiods relative to the precrisis benchmark, and how these changes differ between bank-affiliated and nonbank dealers.", "strategy": "Time-series regressions on period indicators with controls, with the precrisis period as benchmark. Bank-affiliated dealers are compared with nonbank dealers to attribute changes to bank-focused regulation. Trades likely to be electronic are compared with traditional trades, and stressful days are examined separately.", "key_assumptions": ["Nonbank dealers serve as a comparison group unaffected by bank-focused regulation", "Control variables capture relevant changes in market conditions over time", "Trade characteristics proxy for the likelihood of electronic intermediation"], "threats_addressed": ["Compositional shifts in trading toward more liquid bonds", "Growth of electronic trading reducing search costs", "Competition from corporate bond ETFs and mutual funds", "Confounding from the TRACE transparency phase-in, handled by excluding 2002–2005", "Liquidity changes concentrated on stressful days"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "p_values", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "other"]}, "tables": {"convention_notes": "Tables move from sample construction and summary statistics to transaction-cost estimates, then capital-commitment statistics and time-series regressions, then bank-versus-nonbank and stressful-day splits. Regressions use period indicators with the precrisis period as benchmark. Notes state the data sources, sample period, aggregation level, and dollar size cutoffs.", "units_stated_in_notes": true, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'likely contribute'", "'supports the view'", "'may point to'", "explicit disclaimer that net welfare and macroeconomic stability effects are not assessed", "'it will be of interest' framing of future questions"]}, "claim_evidence": [{"claim": "Average customer execution costs are roughly unchanged over 2006–2016 apart from a temporary crisis spike.", "evidence_type": "table", "evidence_ref": "Table III; Figure 1"}, {"claim": "Dealer capital commitment, turnover and trade size fell in the recent period relative to the precrisis period.", "evidence_type": "table", "evidence_ref": "Tables IV, V; Figure 2"}, {"claim": "The declines are driven by bank-affiliated dealers, while nonbank dealers increased their commitment.", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "Electronic trading reduced the need for dealer capital in segments where electronic venues are most likely.", "evidence_type": "table", "evidence_ref": "Section VI.D"}, {"claim": "Bank versus nonbank differences also appear on stressful days.", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}], "structural_summary": "The paper opens with policy and practitioner concern about corporate bond liquidity and notes the disagreement over whether liquidity has deteriorated. It argues that execution costs alone are insufficient and introduces capital commitment measures. A literature and regulation section precedes the data section. Results then proceed from transaction costs, to dealer-level measures, to time-series regressions with period dummies. Next comes a bank versus nonbank comparison as the main attribution device for regulation, alongside electronic and ETF alternative explanations, followed by a stressful-day analysis. The conclusion restates the findings, supports Duffie's regulatory prediction, disclaims any welfare assessment, and poses future questions.", "writing_profile_keywords": ["descriptive-to-attributional progression", "policy-relevant", "subperiod comparison", "bank vs nonbank contrast", "alternative explanations weighed", "measured causal hedging"], "confidence": 0.82, "notes": "No equations are visible. Causal attribution to regulation rests on comparing bank-affiliated and nonbank dealers. Robustness checks are partly in an Internet Appendix and partly in the stressful-day section."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 22.18, "usage": {"input": 8055, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076454, "request_id": "req_011Cggpz56MnZDUbEpiDsuin", "cached": false, "ts": "2026-11-03T23:02:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2ed2f326d2baff68a61418c25f88130c95eba899ef53f2413788a3c2180d1679.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical welfare comparison of benchmark vs. no-benchmark equilibria)\",\"strategy\":\"not applicable; search-model equilibrium comparison and mechanism design\",\"key_assumptions\":[\"search frictions with slow and fast traders\",\"dealer cost is common (homogeneous) or heterogeneous\",\"parameter conditions vary by result\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":4},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"No tables. Results are stated as propositions and theorems, with proofs in the appendices. Figures illustrate numerical examples of surplus, entry probability, prices, and equilibrium regimes.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"under natural parameter assumptions\",\"conditions under which\",\"may\",\"in some cases\",\"is likely to be\",\"suggests that there may be\"]},\"claim_evidence\":[{\"claim\":\"Publishing a benchmark encourages efficient trader entry and mitigates dealer hold-up\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II, Appendix A\"},{\"claim\":\"Benchmarks improve matching efficiency and raise the market share of low-cost dealers\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III, Appendix B\"},{\"claim\":\"Dealers may introduce benchmarks when volume gains offset reduced margins\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.E\"},{\"claim\":\"The optimal revelation mechanism coincides with announcing the benchmark under stated conditions\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.F-G\"},{\"claim\":\"Benchmarks underlie trillions of dollars in loans and derivatives\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"MPG (2014); BIS (2014)\"},{\"claim\":\"Benchmarks increase surplus relative to no benchmark in a numerical example\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-3\"}],\"structural_summary\":\"Theory paper. The introduction motivates the question with the scale of benchmark-referenced OTC markets. It then previews three welfare channels (entry, matching, search costs), cases where a benchmark lowers welfare, an optimal-mechanism result, and dealers' incentives to introduce benchmarks. A brief literature comparison and a roadmap follow. An institutional section describes how benchmarks are used. The core search model with homogeneous dealer costs compares equilibria with and without a benchmark, derives the welfare comparison and dealer incentives, and characterizes the optimal mechanism. An extension adds heterogeneous dealer costs to study matching efficiency. A short discussion of manipulation precedes a conclusion with policy remarks. Proofs are in the appendices.\",\"writing_profile_keywords\":[\"search model\",\"OTC markets\",\"price transparency\",\"mechanism design\",\"welfare comparison\",\"proposition-proof structure\",\"policy relevance\"],\"confidence\":0.88,\"notes\":\"Purely theoretical; there are no tables or empirical estimates. Institutional magnitudes (trillions of dollars) are used only as motivation. One author chaired the MPG report.\"}", "json": {"genre": "theoretical", "subfield": "market_microstructure", "subfield_secondary": "financial_intermediation_banking", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical welfare comparison of benchmark vs. no-benchmark equilibria)", "strategy": "not applicable; search-model equilibrium comparison and mechanism design", "key_assumptions": ["search frictions with slow and fast traders", "dealer cost is common (homogeneous) or heterogeneous", "parameter conditions vary by result"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 4}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "policy_or_broader_implication"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "No tables. Results are stated as propositions and theorems, with proofs in the appendices. Figures illustrate numerical examples of surplus, entry probability, prices, and equilibrium regimes.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["under natural parameter assumptions", "conditions under which", "may", "in some cases", "is likely to be", "suggests that there may be"]}, "claim_evidence": [{"claim": "Publishing a benchmark encourages efficient trader entry and mitigates dealer hold-up", "evidence_type": "equation_or_proof", "evidence_ref": "Section II, Appendix A"}, {"claim": "Benchmarks improve matching efficiency and raise the market share of low-cost dealers", "evidence_type": "equation_or_proof", "evidence_ref": "Section III, Appendix B"}, {"claim": "Dealers may introduce benchmarks when volume gains offset reduced margins", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.E"}, {"claim": "The optimal revelation mechanism coincides with announcing the benchmark under stated conditions", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.F-G"}, {"claim": "Benchmarks underlie trillions of dollars in loans and derivatives", "evidence_type": "prior_literature", "evidence_ref": "MPG (2014); BIS (2014)"}, {"claim": "Benchmarks increase surplus relative to no benchmark in a numerical example", "evidence_type": "figure", "evidence_ref": "Figures 1-3"}], "structural_summary": "Theory paper. The introduction motivates the question with the scale of benchmark-referenced OTC markets. It then previews three welfare channels (entry, matching, search costs), cases where a benchmark lowers welfare, an optimal-mechanism result, and dealers' incentives to introduce benchmarks. A brief literature comparison and a roadmap follow. An institutional section describes how benchmarks are used. The core search model with homogeneous dealer costs compares equilibria with and without a benchmark, derives the welfare comparison and dealer incentives, and characterizes the optimal mechanism. An extension adds heterogeneous dealer costs to study matching efficiency. A short discussion of manipulation precedes a conclusion with policy remarks. Proofs are in the appendices.", "writing_profile_keywords": ["search model", "OTC markets", "price transparency", "mechanism design", "welfare comparison", "proposition-proof structure", "policy relevance"], "confidence": 0.88, "notes": "Purely theoretical; there are no tables or empirical estimates. Institutional magnitudes (trillions of dollars) are used only as motivation. One author chaired the MPG report."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.11, "usage": {"input": 6278, "output": 1937, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064646, "request_id": "req_011Cggph274eHbEbFRmDdQqx", "cached": false, "ts": "2026-11-03T22:58:41+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2ef52b123465f7d1de149601106d2aaf2ea333486f2c9d1d69510c7e5a24d054.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Effect of depression-era local (region-occupation) labor market deterioration on long-run risky asset investment more than a decade later, plus the part not working through income, employment or wealth\",\"strategy\":\"Cross-sectional variation in region-occupation unemployment changes during the 1991-93 Finnish depression, conditional on region and occupation fixed effects and predepression controls; secondhand experiences of neighbors and family isolate non-wealth channels; matching as an alternative estimator\",\"key_assumptions\":[\"Conditional independence: exposure to local labor market distress is unrelated to unobserved worker characteristics given fixed effects and predepression controls\",\"Secondhand experiences do not affect the individual's wealth, income or employment\",\"Depression shocks (Soviet collapse, twin crisis) were unexpected\"],\"threats_addressed\":[\"Omitted variables / selection into occupations and regions\",\"Wealth, income and employment channels as bad controls\",\"Measurement error from recall bias in survey data\",\"Sensitivity to the labor market definition\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_approach\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":13,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":14,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"new_method\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"falsification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes report coefficients with t-values and spell out the dependent variable, control sets and sample. Income and wealth are in 2005 CPI-adjusted terms. Later tables cross-reference the control sets of earlier columns. Table I pairs a falsification test with descriptive statistics.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"plausibly exogenous\",\"suggest\",\"likely\",\"unlikely to explain\",\"interpret as lower bound\",\"candidate explanations\"]},\"claim_evidence\":[{\"claim\":\"A one-SD worse labor market experience lowers stock market participation by about 2.8-3.6 pp, against a 22% base rate\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Depression-era labor conditions do not predict predepression stock market participation (falsification)\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table I\"},{\"claim\":\"Labor market conditions affect later income, unemployment and net worth\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Results hold with family fixed effects, parental controls and alternative labor market definitions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Matching adversely affected workers to less affected ones yields similar results\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Adverse experiences also reduce holdings of fixed income funds, balanced funds and derivatives\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Neighbors' and family members' adverse experiences reduce risky investment by 0.4-0.6 pp, so at least 11% of the effect is not via wealth or income\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"}],\"structural_summary\":\"The introduction frames portfolio heterogeneity as a puzzle that twin studies leave open. It then organizes the design around three identification challenges, each answered by a feature of the Finnish depression setting, and previews effect sizes with clustered t-values. Four literature strands follow, then a roadmap. Section I covers the timeline, register data and a conditional independence defense. Section II gives baseline firsthand-experience regressions followed by robustness checks: controls, labor market definitions, regional analyses, matching and alternative assets. Section III uses secondhand experiences of neighbors and family to separate preference and belief channels from wealth channels. The conclusion restates the contribution, bounds the mechanism to preferences, beliefs or trust, and suggests extensions.\",\"writing_profile_keywords\":[\"natural experiment\",\"register data\",\"identification challenges enumerated\",\"falsification test\",\"secondhand experiences\",\"lower-bound interpretation\",\"clustered t-values\",\"mechanism isolation\"],\"confidence\":0.85,\"notes\":\"Section II carries the main results; the 'other' role assigned by the rule was reinterpreted as results. Section III was labeled robustness by the rule but serves as channel analysis, so it is classified as results. Whether figures show uncertainty cannot be determined from the captions.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Effect of depression-era local (region-occupation) labor market deterioration on long-run risky asset investment more than a decade later, plus the part not working through income, employment or wealth", "strategy": "Cross-sectional variation in region-occupation unemployment changes during the 1991-93 Finnish depression, conditional on region and occupation fixed effects and predepression controls; secondhand experiences of neighbors and family isolate non-wealth channels; matching as an alternative estimator", "key_assumptions": ["Conditional independence: exposure to local labor market distress is unrelated to unobserved worker characteristics given fixed effects and predepression controls", "Secondhand experiences do not affect the individual's wealth, income or employment", "Depression shocks (Soviet collapse, twin crisis) were unexpected"], "threats_addressed": ["Omitted variables / selection into occupations and regions", "Wealth, income and employment channels as bad controls", "Measurement error from recall bias in survey data", "Sensitivity to the labor market definition"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_approach"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "policy_or_broader_implication"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "related_literature_positioning"}, {"paragraph_index": 12, "move": "related_literature_positioning"}, {"paragraph_index": 13, "move": "related_literature_positioning"}, {"paragraph_index": 14, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_identification", "new_method"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure", "falsification", "subsample", "other"]}, "tables": {"convention_notes": "Notes report coefficients with t-values and spell out the dependent variable, control sets and sample. Income and wealth are in 2005 CPI-adjusted terms. Later tables cross-reference the control sets of earlier columns. Table I pairs a falsification test with descriptive statistics.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["plausibly exogenous", "suggest", "likely", "unlikely to explain", "interpret as lower bound", "candidate explanations"]}, "claim_evidence": [{"claim": "A one-SD worse labor market experience lowers stock market participation by about 2.8-3.6 pp, against a 22% base rate", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Depression-era labor conditions do not predict predepression stock market participation (falsification)", "evidence_type": "figure", "evidence_ref": "Figure 3; Table I"}, {"claim": "Labor market conditions affect later income, unemployment and net worth", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Results hold with family fixed effects, parental controls and alternative labor market definitions", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Matching adversely affected workers to less affected ones yields similar results", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Adverse experiences also reduce holdings of fixed income funds, balanced funds and derivatives", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Neighbors' and family members' adverse experiences reduce risky investment by 0.4-0.6 pp, so at least 11% of the effect is not via wealth or income", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}], "structural_summary": "The introduction frames portfolio heterogeneity as a puzzle that twin studies leave open. It then organizes the design around three identification challenges, each answered by a feature of the Finnish depression setting, and previews effect sizes with clustered t-values. Four literature strands follow, then a roadmap. Section I covers the timeline, register data and a conditional independence defense. Section II gives baseline firsthand-experience regressions followed by robustness checks: controls, labor market definitions, regional analyses, matching and alternative assets. Section III uses secondhand experiences of neighbors and family to separate preference and belief channels from wealth channels. The conclusion restates the contribution, bounds the mechanism to preferences, beliefs or trust, and suggests extensions.", "writing_profile_keywords": ["natural experiment", "register data", "identification challenges enumerated", "falsification test", "secondhand experiences", "lower-bound interpretation", "clustered t-values", "mechanism isolation"], "confidence": 0.85, "notes": "Section II carries the main results; the 'other' role assigned by the rule was reinterpreted as results. Section III was labeled robustness by the rule but serves as channel analysis, so it is classified as results. Whether figures show uncertainty cannot be determined from the captions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.78, "usage": {"input": 7744, "output": 2385, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07947, "request_id": "req_011CggpSEW2QrqobxsXayFPw", "cached": false, "ts": "2026-11-03T22:55:25+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/327b0caf6a62f7e9717830edf8818e7babae68d31b67ee75dc638badcb08d4b3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_intermediation_banking (asset management, search frictions)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (equilibrium comparative statics of market efficiency, fees, and manager performance)\",\"strategy\":\"Analytical general equilibrium of a Grossman-Stiglitz noisy rational expectations model extended with search for asset managers and Nash-bargained fees\",\"key_assumptions\":[\"costly search for informed managers\",\"costly information acquisition by managers\",\"noise traders and noise allocators\",\"interior equilibrium with investor indifference\",\"linear asset-market equilibrium\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":7,\"contribution_types\":[\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"appendix\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"The single table is a literature summary that maps model predictions to cited studies and quotes, organized in panels. It reports no estimates.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\",\"notes_placeholder_removed\":\"\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'under certain conditions' qualifiers\",\"'consistent with' when linking predictions to evidence\",\"conditional 'if search costs have diminished' framing\"]},\"claim_evidence\":[{\"claim\":\"Lower search costs for finding managers raise active allocation, lower fees, and make prices more efficient\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II-III, Figure 3, Appendix A.6\"},{\"claim\":\"Informed managers outperform after fees and uninformed managers underperform\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 1, Section III.A\"},{\"claim\":\"Average manager performance depends on the number of noise allocators and can be zero or negative with free random search\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A, Appendix A.2\"},{\"claim\":\"Large and sophisticated investors benefit from searching, so managers with such investors outperform\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV, Appendix A.3\"},{\"claim\":\"Cross-sectional manager skill evidence and rising market efficiency are consistent with model predictions\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Intro paras 5-6, Section V, Table I\"},{\"claim\":\"Market inefficiency magnitude is approximately 6%\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C\"}],\"structural_summary\":\"The paper opens by motivating search frictions in asset management and poses questions about efficiency and who outperforms. It then lists numbered predictions and sketches the equilibrium mechanism. Next, it positions the results against puzzles in the mutual fund literature (Fama, Berk-Green) and states its contribution, followed by a roadmap. The body sets up the model, solves the equilibrium, and derives propositions and comparative statics, including a calibrated magnitude. It then extends the model to heterogeneous investors and managers and maps the predictions to existing evidence in a literature table. The conclusion is brief and restates the contribution. Proofs, extensions, and an institutional description of due diligence sit in the appendices.\",\"writing_profile_keywords\":[\"Grossman-Stiglitz extension\",\"search frictions\",\"numbered predictions\",\"tractability emphasis\",\"prediction-to-literature mapping\",\"proposition-based\"],\"confidence\":0.85,\"notes\":\"Pure theory paper with an empirical-implications section that relies on prior literature rather than new estimation. Robustness takes the form of model extensions in the appendix (AUM-based search, endogenous noise allocators, partial equilibrium). The 'figures' object contains an extra key not permitted by the schema; the intended value is uncertainty_shown='no'.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "financial_intermediation_banking (asset management, search frictions)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (equilibrium comparative statics of market efficiency, fees, and manager performance)", "strategy": "Analytical general equilibrium of a Grossman-Stiglitz noisy rational expectations model extended with search for asset managers and Nash-bargained fees", "key_assumptions": ["costly search for informed managers", "costly information acquisition by managers", "noise traders and noise allocators", "interior equilibrium with investor indifference", "linear asset-market equilibrium"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "tension_or_puzzle"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 7, "contribution_types": ["new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "appendix", "types": ["other"]}, "tables": {"convention_notes": "The single table is a literature summary that maps model predictions to cited studies and quotes, organized in panels. It reports no estimates.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no", "notes_placeholder_removed": ""}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'under certain conditions' qualifiers", "'consistent with' when linking predictions to evidence", "conditional 'if search costs have diminished' framing"]}, "claim_evidence": [{"claim": "Lower search costs for finding managers raise active allocation, lower fees, and make prices more efficient", "evidence_type": "equation_or_proof", "evidence_ref": "Section II-III, Figure 3, Appendix A.6"}, {"claim": "Informed managers outperform after fees and uninformed managers underperform", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 1, Section III.A"}, {"claim": "Average manager performance depends on the number of noise allocators and can be zero or negative with free random search", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A, Appendix A.2"}, {"claim": "Large and sophisticated investors benefit from searching, so managers with such investors outperform", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV, Appendix A.3"}, {"claim": "Cross-sectional manager skill evidence and rising market efficiency are consistent with model predictions", "evidence_type": "prior_literature", "evidence_ref": "Intro paras 5-6, Section V, Table I"}, {"claim": "Market inefficiency magnitude is approximately 6%", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C"}], "structural_summary": "The paper opens by motivating search frictions in asset management and poses questions about efficiency and who outperforms. It then lists numbered predictions and sketches the equilibrium mechanism. Next, it positions the results against puzzles in the mutual fund literature (Fama, Berk-Green) and states its contribution, followed by a roadmap. The body sets up the model, solves the equilibrium, and derives propositions and comparative statics, including a calibrated magnitude. It then extends the model to heterogeneous investors and managers and maps the predictions to existing evidence in a literature table. The conclusion is brief and restates the contribution. Proofs, extensions, and an institutional description of due diligence sit in the appendices.", "writing_profile_keywords": ["Grossman-Stiglitz extension", "search frictions", "numbered predictions", "tractability emphasis", "prediction-to-literature mapping", "proposition-based"], "confidence": 0.85, "notes": "Pure theory paper with an empirical-implications section that relies on prior literature rather than new estimation. Robustness takes the form of model extensions in the appendix (AUM-based search, endogenous noise allocators, partial equilibrium). The 'figures' object contains an extra key not permitted by the schema; the intended value is uncertainty_shown='no'."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.48, "usage": {"input": 6183, "output": 2057, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066666, "request_id": "req_011CggpzR1SNXEjmjWPEQNby", "cached": false, "ts": "2026-11-03T23:02:39+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/338e0e5b8572193f6da6b0a692302dddf9abc9f02363eaafb2b444740849361f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"macro_finance (misallocation and aggregate productivity)\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Differential change in employment, capital growth and bank debt share of higher-productivity young firms relative to lower-productivity firms after state banking deregulation\",\"strategy\":\"Triple-difference design exploiting staggered state banking deregulation (intrastate and interstate) between the 1970s and early 1990s, interacted with firm TFP and firm age groups; bank versus nonbank debt share used to separate credit supply from credit demand; falsification tests against alternative mechanisms\",\"key_assumptions\":[\"Timing of deregulation is not correlated with preexisting relative growth trends of productive firms\",\"Credit demand shocks would move bank and nonbank debt similarly, while supply shocks move the bank debt share\",\"Local demand and economic conditions do not differentially favor productive young firms\",\"The monopolistic competition misallocation model (Hsieh-Klenow) is valid for the quantification exercise\"],\"threats_addressed\":[\"pre-trends (dynamics analysis)\",\"credit demand versus supply\",\"local economic growth shocks\",\"industry-specific local shocks\",\"alternative mechanisms other than bank credit\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"falsification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Each table note opens by stating what the table links (deregulation to an outcome for higher-productivity firms) and emphasizes the differential effect for young firms. Notes cross-reference earlier specifications and equations. Some coefficients are rescaled by the average TFP gap between productivity groups.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'our analysis suggests'\",\"'we provide evidence that'\",\"'is associated with'\",\"'consistent with the view'\",\"'intuitively ... should'\"]},\"claim_evidence\":[{\"claim\":\"Deregulation raises relative employment growth of higher-productivity young firms but not old firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Relative capital growth of productive young firms also rises after deregulation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Growth effects occur right after deregulation and show no preexisting trends\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Bank debt share rises for high-productivity young firms, indicating a shift in bank credit supply composition\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table VI, Figure 2\"},{\"claim\":\"Credit demand and local economic shocks do not explain the bank debt and labor results\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"},{\"claim\":\"Marginal revenue product gaps between high- and low-productivity young firms converge after deregulation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Implied industry productivity gains equal 1.1%-2.1% of value added, with labor misallocation central\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X; Section V.C; Appendix C\"}],\"structural_summary\":\"The introduction moves from the broad finance-growth link to a gap on mechanisms and the overlooked role of labor. It then motivates why young firms should matter and previews a triple-difference design around staggered state banking deregulation. Results and a three-part contribution statement follow, positioned against the literature. The body describes the institutional background and Census microdata, then documents reallocation of labor and capital, including dynamics. It next uses bank versus nonbank debt shares to argue for a credit supply mechanism and runs falsification tests against alternative mechanisms. A Hsieh-Klenow misallocation framework is then used to test convergence of marginal products and to quantify industry productivity gains, with derivations and parameter assumptions in appendices. The conclusion restates the findings.\",\"writing_profile_keywords\":[\"staggered deregulation\",\"triple difference\",\"Census microdata\",\"misallocation\",\"mechanism tests\",\"quantification\",\"young firms\",\"labor reallocation\"],\"confidence\":0.8,\"notes\":\"Panel headings (idx 5, 6) are parsing artifacts. Section 6 contains the alternative-mechanisms subsection, so it is labeled robustness. The statistical significance convention and standard error type are not visible in the excerpts. Section 1 is an institutional background section.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "macro_finance (misallocation and aggregate productivity)", "primary_method": "difference_in_differences", "identification": {"estimand": "Differential change in employment, capital growth and bank debt share of higher-productivity young firms relative to lower-productivity firms after state banking deregulation", "strategy": "Triple-difference design exploiting staggered state banking deregulation (intrastate and interstate) between the 1970s and early 1990s, interacted with firm TFP and firm age groups; bank versus nonbank debt share used to separate credit supply from credit demand; falsification tests against alternative mechanisms", "key_assumptions": ["Timing of deregulation is not correlated with preexisting relative growth trends of productive firms", "Credit demand shocks would move bank and nonbank debt similarly, while supply shocks move the bank debt share", "Local demand and economic conditions do not differentially favor productive young firms", "The monopolistic competition misallocation model (Hsieh-Klenow) is valid for the quantification exercise"], "threats_addressed": ["pre-trends (dynamics analysis)", "credit demand versus supply", "local economic growth shocks", "industry-specific local shocks", "alternative mechanisms other than bank credit"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["falsification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Each table note opens by stating what the table links (deregulation to an outcome for higher-productivity firms) and emphasizes the differential effect for young firms. Notes cross-reference earlier specifications and equations. Some coefficients are rescaled by the average TFP gap between productivity groups.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'our analysis suggests'", "'we provide evidence that'", "'is associated with'", "'consistent with the view'", "'intuitively ... should'"]}, "claim_evidence": [{"claim": "Deregulation raises relative employment growth of higher-productivity young firms but not old firms", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Relative capital growth of productive young firms also rises after deregulation", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Growth effects occur right after deregulation and show no preexisting trends", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Bank debt share rises for high-productivity young firms, indicating a shift in bank credit supply composition", "evidence_type": "table", "evidence_ref": "Table V, Table VI, Figure 2"}, {"claim": "Credit demand and local economic shocks do not explain the bank debt and labor results", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}, {"claim": "Marginal revenue product gaps between high- and low-productivity young firms converge after deregulation", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Implied industry productivity gains equal 1.1%-2.1% of value added, with labor misallocation central", "evidence_type": "table", "evidence_ref": "Table X; Section V.C; Appendix C"}], "structural_summary": "The introduction moves from the broad finance-growth link to a gap on mechanisms and the overlooked role of labor. It then motivates why young firms should matter and previews a triple-difference design around staggered state banking deregulation. Results and a three-part contribution statement follow, positioned against the literature. The body describes the institutional background and Census microdata, then documents reallocation of labor and capital, including dynamics. It next uses bank versus nonbank debt shares to argue for a credit supply mechanism and runs falsification tests against alternative mechanisms. A Hsieh-Klenow misallocation framework is then used to test convergence of marginal products and to quantify industry productivity gains, with derivations and parameter assumptions in appendices. The conclusion restates the findings.", "writing_profile_keywords": ["staggered deregulation", "triple difference", "Census microdata", "misallocation", "mechanism tests", "quantification", "young firms", "labor reallocation"], "confidence": 0.8, "notes": "Panel headings (idx 5, 6) are parsing artifacts. Section 6 contains the alternative-mechanisms subsection, so it is labeled robustness. The statistical significance convention and standard error type are not visible in the excerpts. Section 1 is an institutional background section."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.84, "usage": {"input": 7600, "output": 2388, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078954, "request_id": "req_011Cggq9KmAJMWbHJ6ftMwtH", "cached": false, "ts": "2026-11-03T23:04:42+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/35a082abb20dc96de5483bff59de8242cd9ecfd5396cca14ef9e17b712a4b99e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"asset management / securitization (CDO collateral managers)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association of a manager's specialization (CDO Share) with deal and collateral performance; association of past deal performance with future deal assignments; effect of CDO involvement on mutual fund flows during the crisis\",\"strategy\":\"Cross-sectional regressions with deal type × vintage fixed effects comparing similar deals closed at the same time. Poisson regressions of future deal counts on past deal characteristics. Security-level comparisons of collateral underwritten by the CDO's arranger versus other collateral, conditional on asset class × rating cells. Fund-flow regressions with objective × time fixed effects. Alternative explanations are tested.\",\"key_assumptions\":[\"CDO Share proxies for franchise-value and reputational concerns\",\"Deals within the same type and vintage are comparable after controls\",\"Observationally similar securities are comparable apart from arranger underwriting\"],\"threats_addressed\":[\"Differences in direct incentives (fees, equity stakes)\",\"Employee skill and experience\",\"Manager optimism about housing\",\"Deal characteristics under manager discretion\",\"Magnetar-sponsored deals excluded\",\"Selection in securities with price data\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"other\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes give the full regression equation with fixed-effect structure (type × vintage, asset class × rating × time), define variables, and state the sample and data window (e.g., IRR from cash flows through December 2015). Note openings often state the table's takeaway ('This table shows that...'). Appendix tables cover deal type mapping, alternative performance measures, and price-data selection.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' framing of results\",\"'suggests' / 'potentially understate'\",\"explicit acknowledgement that time-series race-to-bottom explanations cannot be ruled out\",\"restricting claims to cross-sectional differences\"]},\"claim_evidence\":[{\"claim\":\"Managers whose past CDOs performed poorly received more subsequent deal assignments\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 1\"},{\"claim\":\"Moving from 0% to 100% CDO Share is associated with 6.6% lower deal IRR\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 2\"},{\"claim\":\"Specialists' underperformance is not explained by direct incentives, skill, or optimism\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI\"},{\"claim\":\"Arranger-underwritten collateral held by specialized managers performed worse than similar securities\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII; Figure 4\"},{\"claim\":\"Institutional investors withdrew from mutual funds of families that managed ABS CDOs during the crisis\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"},{\"claim\":\"Specialist underperformance is robust to alternative performance measures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table A.III\"}],\"structural_summary\":\"The introduction motivates the paper with securitization's role in the crisis and notes that collateral managers have been neglected. It then states a popular 'front men' narrative and derives four testable predictions from it, previews tests and results (including one magnitude), and openly bounds the claims to cross-sectional evidence. It closes by positioning against the securitization-conflicts and demand literatures and giving a roadmap. The body has three sections: institutional background, data built from merged sources, and a long results section. The results section tests the predictions in order, with an explicit 'alternative explanations' subsection. The conclusion restates the contribution, flags limitations, and draws implications for asset-manager policy.\",\"writing_profile_keywords\":[\"prediction-driven narrative\",\"quid pro quo hypothesis\",\"anecdotal motivation\",\"alternative explanations ruled out\",\"fixed-effects regressions\",\"consistent-with framing\",\"financial crisis\",\"single-author first person\"],\"confidence\":0.8,\"notes\":\"Standard-error type and significance-reporting conventions are not visible in the excerpts. The institutional background section is classed as 'other' rather than literature. Hypotheses come from a verbal narrative, not a formal model.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "asset management / securitization (CDO collateral managers)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association of a manager's specialization (CDO Share) with deal and collateral performance; association of past deal performance with future deal assignments; effect of CDO involvement on mutual fund flows during the crisis", "strategy": "Cross-sectional regressions with deal type × vintage fixed effects comparing similar deals closed at the same time. Poisson regressions of future deal counts on past deal characteristics. Security-level comparisons of collateral underwritten by the CDO's arranger versus other collateral, conditional on asset class × rating cells. Fund-flow regressions with objective × time fixed effects. Alternative explanations are tested.", "key_assumptions": ["CDO Share proxies for franchise-value and reputational concerns", "Deals within the same type and vintage are comparable after controls", "Observationally similar securities are comparable apart from arranger underwriting"], "threats_addressed": ["Differences in direct incentives (fees, equity stakes)", "Employee skill and experience", "Manager optimism about housing", "Deal characteristics under manager discretion", "Magnetar-sponsored deals excluded", "Selection in securities with price data"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 5}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "other"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Table notes give the full regression equation with fixed-effect structure (type × vintage, asset class × rating × time), define variables, and state the sample and data window (e.g., IRR from cash flows through December 2015). Note openings often state the table's takeaway ('This table shows that...'). Appendix tables cover deal type mapping, alternative performance measures, and price-data selection.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' framing of results", "'suggests' / 'potentially understate'", "explicit acknowledgement that time-series race-to-bottom explanations cannot be ruled out", "restricting claims to cross-sectional differences"]}, "claim_evidence": [{"claim": "Managers whose past CDOs performed poorly received more subsequent deal assignments", "evidence_type": "table", "evidence_ref": "Table III; Figure 1"}, {"claim": "Moving from 0% to 100% CDO Share is associated with 6.6% lower deal IRR", "evidence_type": "table", "evidence_ref": "Table IV; Figure 2"}, {"claim": "Specialists' underperformance is not explained by direct incentives, skill, or optimism", "evidence_type": "table", "evidence_ref": "Tables V, VI"}, {"claim": "Arranger-underwritten collateral held by specialized managers performed worse than similar securities", "evidence_type": "table", "evidence_ref": "Table VIII; Figure 4"}, {"claim": "Institutional investors withdrew from mutual funds of families that managed ABS CDOs during the crisis", "evidence_type": "table", "evidence_ref": "Table XII"}, {"claim": "Specialist underperformance is robust to alternative performance measures", "evidence_type": "table", "evidence_ref": "Table A.III"}], "structural_summary": "The introduction motivates the paper with securitization's role in the crisis and notes that collateral managers have been neglected. It then states a popular 'front men' narrative and derives four testable predictions from it, previews tests and results (including one magnitude), and openly bounds the claims to cross-sectional evidence. It closes by positioning against the securitization-conflicts and demand literatures and giving a roadmap. The body has three sections: institutional background, data built from merged sources, and a long results section. The results section tests the predictions in order, with an explicit 'alternative explanations' subsection. The conclusion restates the contribution, flags limitations, and draws implications for asset-manager policy.", "writing_profile_keywords": ["prediction-driven narrative", "quid pro quo hypothesis", "anecdotal motivation", "alternative explanations ruled out", "fixed-effects regressions", "consistent-with framing", "financial crisis", "single-author first person"], "confidence": 0.8, "notes": "Standard-error type and significance-reporting conventions are not visible in the excerpts. The institutional background section is classed as 'other' rather than literature. Hypotheses come from a verbal narrative, not a formal model."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.28, "usage": {"input": 9141, "output": 2343, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084218, "request_id": "req_011CggpgQ3daNZ4LAGW4zepz", "cached": false, "ts": "2026-11-03T22:58:38+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/36006ac1366745a5882d383986a6868311c1ee20dd95c876668765a0f9831720.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance; political economy of government credit in China\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of CDB government credit (industrial SOE loans and infrastructure loans) on private firms' and SOEs' assets, employment, sales and efficiency, in the same industry and in downstream industries\",\"strategy\":\"2SLS using the predicted timing of municipal party-secretary turnover cycles, interacted with each city's predetermined focal SOE industry, as instruments for province-industry CDB loan flows; OLS baseline; input-output matrix used to identify downstream industries\",\"key_assumptions\":[\"Predicted political turnover timing is exogenous to local economic conditions\",\"Exclusion: turnover cycles affect firms only through CDB credit, not through other channels such as other-bank borrowing, land sales, fiscal transfers or tax enforcement\",\"Focal industry is predetermined and stable over time\",\"Effects of SOE loans are separable from those of infrastructure loans\"],\"threats_addressed\":[\"Turnover timing correlated with local conditions (hazard regressions)\",\"Alternative political channels (other banks, land sales, transfers, tax enforcement)\",\"Confounding by infrastructure loans\",\"Coincidence with national political cycles (off-cycle cities)\",\"Unobserved investment-opportunity shocks over the turnover cycle\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":4},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_identification\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"falsification\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes give unit of observation, sample coverage (cities, provinces, industries, years), estimator (OLS, 2SLS, Cox hazard, Probit) and instrument definitions. An appendix table defines variables and gives loan units (100 million RMB). Appendix tables report first-stage results and an off-national-cycle subsample.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggestive evidence' for the political career-concern channel\",\"'alleviate the concern' when defending the exclusion restriction\",\"'could explain' when linking results to mixed prior findings\",\"'not likely to be confounded'\"]},\"claim_evidence\":[{\"claim\":\"CDB SOE industrial loans crowd out private firms in the same industry, reducing their assets, employment and sales\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Table V\"},{\"claim\":\"CDB SOE loans to upstream focal industries raise assets and sales of downstream private firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"CDB loans to cities rise early in a city secretary's term, giving a zig-zag borrowing pattern\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Figure 3, Table AIII\"},{\"claim\":\"Other channels of politician influence are uncorrelated with the turnover cycles\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Infrastructure loans crowd in private firms, while industrial SOE loans have the opposite effect\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX, Figure A2\"},{\"claim\":\"Turnover timing is not predicted by local economic conditions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Turnover-cycle effects are stronger in cities with better access to CDB credit\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"}],\"structural_summary\":\"The introduction frames a crowd-out versus crowd-in debate and attributes mixed prior evidence to aggregate data. It previews a supply-chain decomposition built on proprietary bank data, with an instrument from political turnover timing, and reports magnitudes for first-stage and 2SLS effects. It then defends the exclusion restriction, positions the paper in the political business cycle literature and gives a roadmap. A background section on the bank and a data section are followed by one long results section. That section moves from politicians' career concerns, to the first stage, to same-industry and downstream effects, to alternative channels, infrastructure versus SOE loans, reduced form and overall effects. A one-paragraph conclusion restates the findings and draws broader implications. Further results are in an internet appendix.\",\"writing_profile_keywords\":[\"proprietary data\",\"political turnover instrument\",\"2SLS\",\"supply-chain spillovers\",\"crowding out versus crowding in\",\"China\",\"state-owned enterprises\",\"exclusion restriction defense\"],\"confidence\":0.8,\"notes\":\"Section I is institutional background, so it is coded 'other' rather than literature. The excerpts do not show whether results are reported with stars or standard errors, or how standard errors are computed. Robustness checks (off-cycle cities, other channels, heterogeneity by CDB access) appear within the results section and the appendix tables. The conclusion is truncated, so policy discussion is inferred only from its broader framing of government credit's costs and benefits.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance; political economy of government credit in China", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of CDB government credit (industrial SOE loans and infrastructure loans) on private firms' and SOEs' assets, employment, sales and efficiency, in the same industry and in downstream industries", "strategy": "2SLS using the predicted timing of municipal party-secretary turnover cycles, interacted with each city's predetermined focal SOE industry, as instruments for province-industry CDB loan flows; OLS baseline; input-output matrix used to identify downstream industries", "key_assumptions": ["Predicted political turnover timing is exogenous to local economic conditions", "Exclusion: turnover cycles affect firms only through CDB credit, not through other channels such as other-bank borrowing, land sales, fiscal transfers or tax enforcement", "Focal industry is predetermined and stable over time", "Effects of SOE loans are separable from those of infrastructure loans"], "threats_addressed": ["Turnover timing correlated with local conditions (hazard regressions)", "Alternative political channels (other banks, land sales, transfers, tax enforcement)", "Confounding by infrastructure loans", "Coincidence with national political cycles (off-cycle cities)", "Unobserved investment-opportunity shocks over the turnover cycle"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 4}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 2, "contribution_types": ["new_evidence", "new_data_or_measure", "new_identification", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "falsification", "alternative_specification"]}, "tables": {"convention_notes": "Notes give unit of observation, sample coverage (cities, provinces, industries, years), estimator (OLS, 2SLS, Cox hazard, Probit) and instrument definitions. An appendix table defines variables and gives loan units (100 million RMB). Appendix tables report first-stage results and an off-national-cycle subsample.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggestive evidence' for the political career-concern channel", "'alleviate the concern' when defending the exclusion restriction", "'could explain' when linking results to mixed prior findings", "'not likely to be confounded'"]}, "claim_evidence": [{"claim": "CDB SOE industrial loans crowd out private firms in the same industry, reducing their assets, employment and sales", "evidence_type": "table", "evidence_ref": "Table IV, Table V"}, {"claim": "CDB SOE loans to upstream focal industries raise assets and sales of downstream private firms", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "CDB loans to cities rise early in a city secretary's term, giving a zig-zag borrowing pattern", "evidence_type": "table", "evidence_ref": "Table III, Figure 3, Table AIII"}, {"claim": "Other channels of politician influence are uncorrelated with the turnover cycles", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Infrastructure loans crowd in private firms, while industrial SOE loans have the opposite effect", "evidence_type": "table", "evidence_ref": "Table IX, Figure A2"}, {"claim": "Turnover timing is not predicted by local economic conditions", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Turnover-cycle effects are stronger in cities with better access to CDB credit", "evidence_type": "table", "evidence_ref": "Table VIII"}], "structural_summary": "The introduction frames a crowd-out versus crowd-in debate and attributes mixed prior evidence to aggregate data. It previews a supply-chain decomposition built on proprietary bank data, with an instrument from political turnover timing, and reports magnitudes for first-stage and 2SLS effects. It then defends the exclusion restriction, positions the paper in the political business cycle literature and gives a roadmap. A background section on the bank and a data section are followed by one long results section. That section moves from politicians' career concerns, to the first stage, to same-industry and downstream effects, to alternative channels, infrastructure versus SOE loans, reduced form and overall effects. A one-paragraph conclusion restates the findings and draws broader implications. Further results are in an internet appendix.", "writing_profile_keywords": ["proprietary data", "political turnover instrument", "2SLS", "supply-chain spillovers", "crowding out versus crowding in", "China", "state-owned enterprises", "exclusion restriction defense"], "confidence": 0.8, "notes": "Section I is institutional background, so it is coded 'other' rather than literature. The excerpts do not show whether results are reported with stars or standard errors, or how standard errors are computed. Robustness checks (off-cycle cities, other channels, heterogeneity by CDB access) appear within the results section and the appendix tables. The conclusion is truncated, so policy discussion is inferred only from its broader framing of government credit's costs and benefits."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.54, "usage": {"input": 7666, "output": 2450, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080458, "request_id": "req_011CggpoHp3AX4XaoHuxAaUo", "cached": false, "ts": "2026-11-03T23:00:11+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/36fc61e29d7e4edf9e3003a1f3c5103056deae144898354b4439fe22222d4cfc.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"international_finance\",\"subfield_secondary\":\"asset_pricing; financial_intermediation_banking (limits to arbitrage, banking regulation)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Size and persistence of CIP deviations (cross-currency basis), and the effect of bank balance-sheet constraints, proxied by quarter-end reporting, on the basis\",\"strategy\":\"Measure the basis with credit-risk-free instruments (GC repo, KfW bonds) to establish arbitrage net of transaction costs. Then run regressions with quarter-end indicator variables for contracts that cross the quarter-end balance-sheet date, compare the term structure of the basis, use IOER-based arbitrage as a balance-sheet cost proxy, and run an event study of ECB monetary policy announcements.\",\"key_assumptions\":[\"Repo and KfW instruments carry no credit-risk differences across currencies\",\"Contracts that cross the quarter-end differ from other contracts only through their appearance on banks' balance sheets on reporting dates\",\"Intraday yield changes around ECB announcements capture interest rate shocks\"],\"threats_addressed\":[\"credit risk differences across Libor panels\",\"indicative nature of Libor quotes\",\"transaction costs (bid-ask)\",\"term-structure confounds in quarter-end effects\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"literature_gap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Notes define variables and sample periods and report means in basis points with standard deviations in parentheses. Regression tables define the quarter-end indicator variables. The KfW basis is defined by formula in the notes.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"hypothesize\",\"point to\",\"appear\",\"consistent with\",\"can be rationalized by\",\"beyond the scope\"]},\"claim_evidence\":[{\"claim\":\"Libor-based CIP has failed persistently for G10 currencies since 2008\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Figures 2, 4\"},{\"claim\":\"Repo-based basis implies arbitrage free of credit risk, net of transaction costs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"KfW bond basis is nonzero for EUR, CHF and JPY, giving long-maturity arbitrage of about 10-20 bps\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"CIP deviations widen for contracts crossing quarter-ends, implying regulation affects prices\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V; Figure 5\"},{\"claim\":\"Balance-sheet cost proxies (IOER arbitrage) account for about two-thirds of CIP deviations\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 6\"},{\"claim\":\"Basis comoves with other near-risk-free spreads such as the KfW-bund basis and the Libor tenor basis\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"Basis correlates with nominal interest rates in the cross section and in response to ECB shocks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 8, 9\"}],\"structural_summary\":\"The paper opens with market size and the textbook status of CIP, then frames postcrisis CIP violations as a puzzle for no-arbitrage and limits-of-arbitrage models. The introduction rules out credit risk and transaction costs, establishes arbitrage with repo and KfW instruments, and proposes a two-factor hypothesis: intermediary constraints combined with international funding and investment imbalances. It previews four stylized facts and positions the paper against the crisis-era CIP literature and intermediary asset pricing theory. Body sections proceed in order: definitions, arbitrage evidence, a verbal hypothesis section, and tests of the basis characteristics. The conclusion restates the four facts, points to welfare and monetary-transmission implications, and invites future research.\",\"writing_profile_keywords\":[\"puzzle framing\",\"stylized facts\",\"arbitrage documentation\",\"verbal hypothesis then tests\",\"quarter-end regulatory test\",\"cash-flow diagrams\",\"bps magnitudes\"],\"confidence\":0.8,\"notes\":\"Section III develops verbal hypotheses rather than a formal model and is labeled theory. Standard error type is not visible in the excerpts. The abstract ends with a fragment of the introduction's first sentence, which was ignored for sentence indexing (4 real sentences).\"}", "json": {"genre": "empirical", "subfield": "international_finance", "subfield_secondary": "asset_pricing; financial_intermediation_banking (limits to arbitrage, banking regulation)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Size and persistence of CIP deviations (cross-currency basis), and the effect of bank balance-sheet constraints, proxied by quarter-end reporting, on the basis", "strategy": "Measure the basis with credit-risk-free instruments (GC repo, KfW bonds) to establish arbitrage net of transaction costs. Then run regressions with quarter-end indicator variables for contracts that cross the quarter-end balance-sheet date, compare the term structure of the basis, use IOER-based arbitrage as a balance-sheet cost proxy, and run an event study of ECB monetary policy announcements.", "key_assumptions": ["Repo and KfW instruments carry no credit-risk differences across currencies", "Contracts that cross the quarter-end differ from other contracts only through their appearance on banks' balance sheets on reporting dates", "Intraday yield changes around ECB announcements capture interest rate shocks"], "threats_addressed": ["credit risk differences across Libor panels", "indicative nature of Libor quotes", "transaction costs (bid-ask)", "term-structure confounds in quarter-end effects"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "literature_gap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Notes define variables and sample periods and report means in basis points with standard deviations in parentheses. Regression tables define the quarter-end indicator variables. The KfW basis is defined by formula in the notes.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["hypothesize", "point to", "appear", "consistent with", "can be rationalized by", "beyond the scope"]}, "claim_evidence": [{"claim": "Libor-based CIP has failed persistently for G10 currencies since 2008", "evidence_type": "table", "evidence_ref": "Table I; Figures 2, 4"}, {"claim": "Repo-based basis implies arbitrage free of credit risk, net of transaction costs", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "KfW bond basis is nonzero for EUR, CHF and JPY, giving long-maturity arbitrage of about 10-20 bps", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "CIP deviations widen for contracts crossing quarter-ends, implying regulation affects prices", "evidence_type": "table", "evidence_ref": "Tables IV, V; Figure 5"}, {"claim": "Balance-sheet cost proxies (IOER arbitrage) account for about two-thirds of CIP deviations", "evidence_type": "table", "evidence_ref": "Table VI; Figure 6"}, {"claim": "Basis comoves with other near-risk-free spreads such as the KfW-bund basis and the Libor tenor basis", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "Basis correlates with nominal interest rates in the cross section and in response to ECB shocks", "evidence_type": "figure", "evidence_ref": "Figures 8, 9"}], "structural_summary": "The paper opens with market size and the textbook status of CIP, then frames postcrisis CIP violations as a puzzle for no-arbitrage and limits-of-arbitrage models. The introduction rules out credit risk and transaction costs, establishes arbitrage with repo and KfW instruments, and proposes a two-factor hypothesis: intermediary constraints combined with international funding and investment imbalances. It previews four stylized facts and positions the paper against the crisis-era CIP literature and intermediary asset pricing theory. Body sections proceed in order: definitions, arbitrage evidence, a verbal hypothesis section, and tests of the basis characteristics. The conclusion restates the four facts, points to welfare and monetary-transmission implications, and invites future research.", "writing_profile_keywords": ["puzzle framing", "stylized facts", "arbitrage documentation", "verbal hypothesis then tests", "quarter-end regulatory test", "cash-flow diagrams", "bps magnitudes"], "confidence": 0.8, "notes": "Section III develops verbal hypotheses rather than a formal model and is labeled theory. Standard error type is not visible in the excerpts. The abstract ends with a fragment of the introduction's first sentence, which was ignored for sentence indexing (4 real sentences)."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.73, "usage": {"input": 8256, "output": 2356, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080938, "request_id": "req_011CggpxLktKSEqCHawdPPA2", "cached": false, "ts": "2026-11-03T23:02:13+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3805d6b5d9706c0378663a49bf01dc0cb28fa500fe0d6394be94b3ac0a7c1dd2.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"real estate / mortgage default\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Distribution across borrowers of walk-away (threshold) negative equity / CLTV levels at which ruthless default occurs\",\"strategy\":\"Two steps: a discrete-time logit hazard model of default on flexible CLTV dummies, exploiting ZIP-level variation in the timing and size of house price declines among zero-down-payment borrowers; then maximum likelihood estimation of a gamma distribution of default costs, with defaults probabilistically classified as liquidity-driven or ruthless.\",\"key_assumptions\":[\"Default arises only from unwillingness to pay (an equity response) or inability to pay (liquidity)\",\"The effect of equity on default can be separated from liquidity shocks once loan-age fixed effects, quarter fixed effects and county economic conditions are controlled\",\"ZIP-code HPI accurately measures individual home equity\",\"Borrowers walk away when the financial benefits exceed the costs\"],\"threats_addressed\":[\"liquidity/income shocks confounding default timing\",\"endogenous down payment choice (all borrowers start with CLTV of 100)\",\"lender recourse\",\"measurement error in house values\",\"non-representativeness of the nonprime sample (external validity via LPS sample)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":8,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_method\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes describe the estimating equation, the number of observations and the fixed effects that are included but not shown. The logit table reports coefficients, standard errors and odds ratios. Walk-away distributions are reported as CLTV percentiles (25th, median, 75th) with N, and the baseline row is repeated across the mechanism, robustness and broader-sample tables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggests/appears\",\"probably\",\"we argue\",\"may be\",\"it is difficult to pin down\"]},\"claim_evidence\":[{\"claim\":\"Median borrower does not walk away until CLTV reaches about 174 (equity of -74%)\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 7\"},{\"claim\":\"CLTV alone does not drive default until it exceeds roughly 120\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 6\"},{\"claim\":\"Frictionless option model predicts certain default at about -20% equity\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Kau, Keenan, and Kim (1994)\"},{\"claim\":\"Lender recourse is not the main driver of high walk-away thresholds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"High thresholds also hold in a broader sample including GSE and portfolio loans\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"House prices fell sharply in the four studied states\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-2\"}],\"structural_summary\":\"The introduction moves from the crisis facts to the ruthless-default debate, then previews the headline magnitude against model predictions. It defends a two-step method that separates liquidity-driven from ruthless defaults and closes with policy implications on moral hazard. A short conceptual section defines ruthless default, followed by the data section. The empirical section estimates a hazard model and then a maximum likelihood distribution of walk-away thresholds. A long discussion section weighs financial versus behavioral mechanisms, identification, external validity and robustness. The conclusion restates the findings and leaves the behavioral drivers to future research.\",\"writing_profile_keywords\":[\"walk-away threshold\",\"negative equity\",\"hazard model\",\"maximum likelihood\",\"behavioral explanation\",\"policy implication\",\"mechanism discussion\"],\"confidence\":0.8,\"notes\":\"Primary method could also be classed as reduced_form_regression, since step 1 is a logit hazard; it was labelled structural because step 2 estimates a parametric default-cost distribution by maximum likelihood. The Discussion section's role_rule says conclusion, but its content is mechanisms and robustness, so it was assigned robustness. The way statistical significance is reported is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "real estate / mortgage default", "primary_method": "structural_estimation", "identification": {"estimand": "Distribution across borrowers of walk-away (threshold) negative equity / CLTV levels at which ruthless default occurs", "strategy": "Two steps: a discrete-time logit hazard model of default on flexible CLTV dummies, exploiting ZIP-level variation in the timing and size of house price declines among zero-down-payment borrowers; then maximum likelihood estimation of a gamma distribution of default costs, with defaults probabilistically classified as liquidity-driven or ruthless.", "key_assumptions": ["Default arises only from unwillingness to pay (an equity response) or inability to pay (liquidity)", "The effect of equity on default can be separated from liquidity shocks once loan-age fixed effects, quarter fixed effects and county economic conditions are controlled", "ZIP-code HPI accurately measures individual home equity", "Borrowers walk away when the financial benefits exceed the costs"], "threats_addressed": ["liquidity/income shocks confounding default timing", "endogenous down payment choice (all borrowers start with CLTV of 100)", "lender recourse", "measurement error in house values", "non-representativeness of the nonprime sample (external validity via LPS sample)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 8, "move": "policy_or_broader_implication"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_method"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Notes describe the estimating equation, the number of observations and the fixed effects that are included but not shown. The logit table reports coefficients, standard errors and odds ratios. Walk-away distributions are reported as CLTV percentiles (25th, median, 75th) with N, and the baseline row is repeated across the mechanism, robustness and broader-sample tables.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggests/appears", "probably", "we argue", "may be", "it is difficult to pin down"]}, "claim_evidence": [{"claim": "Median borrower does not walk away until CLTV reaches about 174 (equity of -74%)", "evidence_type": "table", "evidence_ref": "Table III; Figure 7"}, {"claim": "CLTV alone does not drive default until it exceeds roughly 120", "evidence_type": "table", "evidence_ref": "Table II; Figure 6"}, {"claim": "Frictionless option model predicts certain default at about -20% equity", "evidence_type": "prior_literature", "evidence_ref": "Kau, Keenan, and Kim (1994)"}, {"claim": "Lender recourse is not the main driver of high walk-away thresholds", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "High thresholds also hold in a broader sample including GSE and portfolio loans", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "House prices fell sharply in the four studied states", "evidence_type": "figure", "evidence_ref": "Figures 1-2"}], "structural_summary": "The introduction moves from the crisis facts to the ruthless-default debate, then previews the headline magnitude against model predictions. It defends a two-step method that separates liquidity-driven from ruthless defaults and closes with policy implications on moral hazard. A short conceptual section defines ruthless default, followed by the data section. The empirical section estimates a hazard model and then a maximum likelihood distribution of walk-away thresholds. A long discussion section weighs financial versus behavioral mechanisms, identification, external validity and robustness. The conclusion restates the findings and leaves the behavioral drivers to future research.", "writing_profile_keywords": ["walk-away threshold", "negative equity", "hazard model", "maximum likelihood", "behavioral explanation", "policy implication", "mechanism discussion"], "confidence": 0.8, "notes": "Primary method could also be classed as reduced_form_regression, since step 1 is a logit hazard; it was labelled structural because step 2 estimates a parametric default-cost distribution by maximum likelihood. The Discussion section's role_rule says conclusion, but its content is mechanisms and robustness, so it was assigned robustness. The way statistical significance is reported is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.09, "usage": {"input": 7965, "output": 2234, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077334, "request_id": "req_011CggpjbeyRwnQVvhd3o8eJ", "cached": false, "ts": "2026-11-03T22:59:19+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/391304afeba68f9701a99c579179912ab8baa4c9f9ba1d9d23d97a283811872c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"fixed_income_mbs_liquidity\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of TBA eligibility and of parallel TBA trading volume on SP trading costs; dealer hedge ratios of SP inventory with TBA trades\",\"strategy\":\"Trading-cost regressions on consecutive trades; regression discontinuity at the 1.05 LTV eligibility threshold; propensity score matching on the estimated probability of eligibility; predictable dollar-roll volume before settlement used as exogenous variation in TBA volume; dealer-level inventory regressions\",\"key_assumptions\":[\"Trading costs vary smoothly in LTV except for the eligibility threshold\",\"Observable pool characteristics capture selection into eligibility\",\"Predicted dollar roll volume is exogenous to SP liquidity\"],\"threats_addressed\":[\"Characteristics associated with TBA eligibility, rather than eligibility itself, driving liquidity\",\"Endogeneity of TBA trading volume\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Notes specify the regression equation, sample construction and variable definitions; some tables report coefficients with z-statistics and marginal effects.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggest\",\"appears to\",\"may\",\"consistent with\",\"one (but not necessarily the only) way\"]},\"claim_evidence\":[{\"claim\":\"TBA-eligible SPs have significantly lower trading costs than ineligible SPs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Trading costs jump when the LTV eligibility threshold is crossed, supporting an effect of eligibility itself\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 3\"},{\"claim\":\"Eligibility lowers costs after matching on estimated eligibility probability\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV-V\"},{\"claim\":\"SP trading costs fall on days of predictably high TBA dollar roll volume\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 4\"},{\"claim\":\"Dealers hedge SP inventory changes with offsetting TBA trades, less so for ineligible SPs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"TBA trades hedge price changes of eligible SPs better than those of ineligible SPs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Prearranged trades are more common for hard-to-hedge SPs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The paper opens with the puzzle of high liquidity in a market of unique securities and proposes three channels through which TBA trading creates liquidity. It positions itself against prior trading-cost studies and previews the hedging result and a policy implication for bond markets. After institutional background and data sections, the results proceed in steps: trading-cost benchmarks, then the causal effect of eligibility using RD and matching, then spillovers using predictable dollar-roll volume, then evidence on dealer hedging and prearranged trades. The conclusion restates findings and proposes forward markets for municipal and corporate bonds.\",\"writing_profile_keywords\":[\"institutional detail\",\"OTC liquidity\",\"TRACE data\",\"regression discontinuity\",\"propensity score matching\",\"dealer hedging\",\"policy extension\"],\"confidence\":0.8,\"notes\":\"Robustness checks are not visible in the excerpts. Standard-error treatment is not described in the excerpts. The roadmap's description of Sections IV and V differs slightly from the section headings.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "fixed_income_mbs_liquidity", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of TBA eligibility and of parallel TBA trading volume on SP trading costs; dealer hedge ratios of SP inventory with TBA trades", "strategy": "Trading-cost regressions on consecutive trades; regression discontinuity at the 1.05 LTV eligibility threshold; propensity score matching on the estimated probability of eligibility; predictable dollar-roll volume before settlement used as exogenous variation in TBA volume; dealer-level inventory regressions", "key_assumptions": ["Trading costs vary smoothly in LTV except for the eligibility threshold", "Observable pool characteristics capture selection into eligibility", "Predicted dollar roll volume is exogenous to SP liquidity"], "threats_addressed": ["Characteristics associated with TBA eligibility, rather than eligibility itself, driving liquidity", "Endogeneity of TBA trading volume"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 2, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Notes specify the regression equation, sample construction and variable definitions; some tables report coefficients with z-statistics and marginal effects.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggest", "appears to", "may", "consistent with", "one (but not necessarily the only) way"]}, "claim_evidence": [{"claim": "TBA-eligible SPs have significantly lower trading costs than ineligible SPs", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Trading costs jump when the LTV eligibility threshold is crossed, supporting an effect of eligibility itself", "evidence_type": "table", "evidence_ref": "Table III; Figure 3"}, {"claim": "Eligibility lowers costs after matching on estimated eligibility probability", "evidence_type": "table", "evidence_ref": "Tables IV-V"}, {"claim": "SP trading costs fall on days of predictably high TBA dollar roll volume", "evidence_type": "table", "evidence_ref": "Table VI; Figure 4"}, {"claim": "Dealers hedge SP inventory changes with offsetting TBA trades, less so for ineligible SPs", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "TBA trades hedge price changes of eligible SPs better than those of ineligible SPs", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Prearranged trades are more common for hard-to-hedge SPs", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The paper opens with the puzzle of high liquidity in a market of unique securities and proposes three channels through which TBA trading creates liquidity. It positions itself against prior trading-cost studies and previews the hedging result and a policy implication for bond markets. After institutional background and data sections, the results proceed in steps: trading-cost benchmarks, then the causal effect of eligibility using RD and matching, then spillovers using predictable dollar-roll volume, then evidence on dealer hedging and prearranged trades. The conclusion restates findings and proposes forward markets for municipal and corporate bonds.", "writing_profile_keywords": ["institutional detail", "OTC liquidity", "TRACE data", "regression discontinuity", "propensity score matching", "dealer hedging", "policy extension"], "confidence": 0.8, "notes": "Robustness checks are not visible in the excerpts. Standard-error treatment is not described in the excerpts. The roadmap's description of Sections IV and V differs slightly from the section headings."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.33, "usage": {"input": 8070, "output": 2001, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073094, "request_id": "req_011CggpaGDvrU3R5jfFgrJGt", "cached": false, "ts": "2026-11-03T22:57:11+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/39c4a0cadfd55619b3d575999c04db8991a4ccf19b2b62595a2aa7cc5233ebe1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Out-of-sample predictability (R2) of dividend growth and stock index returns, and the incremental contribution of learning about parameters relative to full information\",\"strategy\":\"A latent variable dividend model is estimated with the Kalman filter and embedded in a long-run risks equilibrium model. Out-of-sample forecasts use recursive (real-time) parameter estimates to represent learning, and these are compared against full-information forecasts using bootstrap-simulated p-values.\",\"key_assumptions\":[\"Investors behave as if they learn about dividend dynamics using the authors' model\",\"Model-implied expected returns proxy true expected returns\",\"A better asset pricing model forecasts returns better (Proposition 1)\"],\"threats_addressed\":[\"Gap between in-sample and out-of-sample predictability\",\"Benchmarking against alternative dividend models\",\"Sensitivity to the preference specification (CRRA versus Epstein-Zin)\",\"Sampling frequency\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":0},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\",\"new_method\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state the sample (nonoverlapping annual data since 1946). Coefficients are shown with standard errors in parentheses, either Newey-West or bootstrap-simulated. Out-of-sample R2s are reported with bootstrap-simulated p-values and compared against benchmark models. Appendix tables repeat the analyses at other sampling frequencies.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'support the view' framing\",\"'behave as if' assumption\",\"'suggests'\",\"'can contribute'\"]},\"claim_evidence\":[{\"claim\":\"The dividend model predicts roughly 40% of annual dividend growth variation out-of-sample and beats alternative dividend models\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Parameter uncertainty is high and resolves slowly; parameter estimates fluctuate over time\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 2\"},{\"claim\":\"Stock yields under learning explain 18.7% of return variation versus 13.0% under full information\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The long-run risks model with learning predicts 25-27% of return variation out-of-sample, with learning contributing about half\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; Figures 7-8\"},{\"claim\":\"A better model forecasts returns better\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 1; Appendix C\"}],\"structural_summary\":\"The introduction motivates the paper with the equity premium and the debate over in-sample versus out-of-sample return predictability, then previews the dividend model, the learning evidence and the return-forecasting results with magnitudes. The body moves through four steps: the dividend model and its estimation; parameter uncertainty and learning; stock yields under learning; and an embedded long-run risks model, followed by subperiod and preference checks. The conclusion restates the findings. Appendices cover Kalman filter estimation, the bootstrap, a proof, model derivations and GMM estimation.\",\"writing_profile_keywords\":[\"out-of-sample R2\",\"parameter learning\",\"long-run risks\",\"Kalman filter\",\"bootstrap p-values\",\"model horse race\"],\"confidence\":0.8,\"notes\":\"Section headings labeled '0,' are parsing artifacts; their content belongs to the dividend model section and to Appendix A respectively. No explicit roadmap or standalone contribution paragraph is visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "structural_estimation", "identification": {"estimand": "Out-of-sample predictability (R2) of dividend growth and stock index returns, and the incremental contribution of learning about parameters relative to full information", "strategy": "A latent variable dividend model is estimated with the Kalman filter and embedded in a long-run risks equilibrium model. Out-of-sample forecasts use recursive (real-time) parameter estimates to represent learning, and these are compared against full-information forecasts using bootstrap-simulated p-values.", "key_assumptions": ["Investors behave as if they learn about dividend dynamics using the authors' model", "Model-implied expected returns proxy true expected returns", "A better asset pricing model forecasts returns better (Proposition 1)"], "threats_addressed": ["Gap between in-sample and out-of-sample predictability", "Benchmarking against alternative dividend models", "Sensitivity to the preference specification (CRRA versus Epstein-Zin)", "Sampling frequency"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 0}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}], "contribution_paragraph_index": -1, "contribution_types": ["new_theory_or_model", "new_evidence", "new_method"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Notes state the sample (nonoverlapping annual data since 1946). Coefficients are shown with standard errors in parentheses, either Newey-West or bootstrap-simulated. Out-of-sample R2s are reported with bootstrap-simulated p-values and compared against benchmark models. Appendix tables repeat the analyses at other sampling frequencies.", "units_stated_in_notes": false, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'support the view' framing", "'behave as if' assumption", "'suggests'", "'can contribute'"]}, "claim_evidence": [{"claim": "The dividend model predicts roughly 40% of annual dividend growth variation out-of-sample and beats alternative dividend models", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Parameter uncertainty is high and resolves slowly; parameter estimates fluctuate over time", "evidence_type": "table", "evidence_ref": "Table VI; Figure 2"}, {"claim": "Stock yields under learning explain 18.7% of return variation versus 13.0% under full information", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The long-run risks model with learning predicts 25-27% of return variation out-of-sample, with learning contributing about half", "evidence_type": "table", "evidence_ref": "Table IX; Figures 7-8"}, {"claim": "A better model forecasts returns better", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 1; Appendix C"}], "structural_summary": "The introduction motivates the paper with the equity premium and the debate over in-sample versus out-of-sample return predictability, then previews the dividend model, the learning evidence and the return-forecasting results with magnitudes. The body moves through four steps: the dividend model and its estimation; parameter uncertainty and learning; stock yields under learning; and an embedded long-run risks model, followed by subperiod and preference checks. The conclusion restates the findings. Appendices cover Kalman filter estimation, the bootstrap, a proof, model derivations and GMM estimation.", "writing_profile_keywords": ["out-of-sample R2", "parameter learning", "long-run risks", "Kalman filter", "bootstrap p-values", "model horse race"], "confidence": 0.8, "notes": "Section headings labeled '0,' are parsing artifacts; their content belongs to the dividend model section and to Appendix A respectively. No explicit roadmap or standalone contribution paragraph is visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.04, "usage": {"input": 9507, "output": 2029, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079402, "request_id": "req_011CggqCRXGhZ1DhUmaMhmCE", "cached": false, "ts": "2026-11-03T23:05:22+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/39e75956c9d05070cbfb1c8c8b982077aea761cdf6b07254f63ae685a9c27cab.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance (security design, securitization regulation)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; analytical characterization of first-best, second-best optimal mechanism and market equilibria\",\"key_assumptions\":[\"risk-neutral agents, three periods\",\"manager more impatient than investors (gains from trade)\",\"screening effort and asset quality are manager's private information\",\"lack of commitment in market equilibria\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"literature_gap\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"appendix\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"No tables; results are stated as propositions and lemmas, with proofs in appendices and a numerical illustration in a figure.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'can feature' / 'may' modal verbs\",\"acknowledging the model is stylized\",\"'consistent with' / 'in line with' when linking to evidence\",\"scope limits on where policy implications apply\"],\"typical_devices_note\":null},\"claim_evidence\":[{\"claim\":\"Costly cash flow retention is essential to implement positive screening under the optimal mechanism\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Appendix D\"},{\"claim\":\"Market allocations can have too much or too little screening relative to second-best\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section III; Figure 3\"},{\"claim\":\"Multiple equilibria arise when gains from trade are large\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A; Appendix E\"},{\"claim\":\"Optimal mechanism decentralized via differential retention rules and transfers across securities\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"Retention correlates with better CMBS performance, consistent with model\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Ashcraft, Gooriah, and Kermani (2014)\"},{\"claim\":\"Securitization associated with declining lending standards and originator private information\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Keys et al. (2010), Mian and Sufi (2009), others\"}],\"structural_summary\":\"Theory paper. The introduction motivates the question with securitization facts and empirical literature on two frictions, sketches the model, and previews second-best and equilibrium results. It links predictions to crisis-era patterns, discusses policy and commitment, and closes with literature positioning; there is no roadmap. The body runs model, then optimal mechanism, then market equilibria (including multiplicity), then efficiency and policy with a securitization application. The conclusion restates findings and regulatory implications. Appendices hold the microfoundation, proofs, and alternative information structures.\",\"writing_profile_keywords\":[\"mechanism design\",\"adverse selection\",\"moral hazard\",\"securitization\",\"retention\",\"policy decentralization\",\"multiple equilibria\",\"first-person singular\"],\"confidence\":0.85,\"notes\":\"The role_rule labels for Section II (robustness) and Sections III–IV (other) were overridden to theory. Appendix F's alternative information structures serve as robustness/understanding exercises.\"}", "json": {"genre": "theoretical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance (security design, securitization regulation)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; analytical characterization of first-best, second-best optimal mechanism and market equilibria", "key_assumptions": ["risk-neutral agents, three periods", "manager more impatient than investors (gains from trade)", "screening effort and asset quality are manager's private information", "lack of commitment in market equilibria"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "policy_or_broader_implication"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "literature_gap"}], "contribution_paragraph_index": 8, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "appendix", "types": ["other"]}, "tables": {"convention_notes": "No tables; results are stated as propositions and lemmas, with proofs in appendices and a numerical illustration in a figure.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'can feature' / 'may' modal verbs", "acknowledging the model is stylized", "'consistent with' / 'in line with' when linking to evidence", "scope limits on where policy implications apply"], "typical_devices_note": null}, "claim_evidence": [{"claim": "Costly cash flow retention is essential to implement positive screening under the optimal mechanism", "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Appendix D"}, {"claim": "Market allocations can have too much or too little screening relative to second-best", "evidence_type": "figure", "evidence_ref": "Section III; Figure 3"}, {"claim": "Multiple equilibria arise when gains from trade are large", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A; Appendix E"}, {"claim": "Optimal mechanism decentralized via differential retention rules and transfers across securities", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV"}, {"claim": "Retention correlates with better CMBS performance, consistent with model", "evidence_type": "prior_literature", "evidence_ref": "Ashcraft, Gooriah, and Kermani (2014)"}, {"claim": "Securitization associated with declining lending standards and originator private information", "evidence_type": "prior_literature", "evidence_ref": "Keys et al. (2010), Mian and Sufi (2009), others"}], "structural_summary": "Theory paper. The introduction motivates the question with securitization facts and empirical literature on two frictions, sketches the model, and previews second-best and equilibrium results. It links predictions to crisis-era patterns, discusses policy and commitment, and closes with literature positioning; there is no roadmap. The body runs model, then optimal mechanism, then market equilibria (including multiplicity), then efficiency and policy with a securitization application. The conclusion restates findings and regulatory implications. Appendices hold the microfoundation, proofs, and alternative information structures.", "writing_profile_keywords": ["mechanism design", "adverse selection", "moral hazard", "securitization", "retention", "policy decentralization", "multiple equilibria", "first-person singular"], "confidence": 0.85, "notes": "The role_rule labels for Section II (robustness) and Sections III–IV (other) were overridden to theory. Appendix F's alternative information structures serve as robustness/understanding exercises."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.1, "usage": {"input": 7283, "output": 1962, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069166, "request_id": "req_011CggpgpN7qPop4HAoqdN13", "cached": false, "ts": "2026-11-03T22:58:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3b1a72f736281efa4fdc86c2ad1760beb84c5c5cae0c8ceef5c9f9ad3ac48667.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"fixed income / credit risk\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Shares of cross-sectional and time-series variance of credit spreads attributable to long-run expected excess returns versus long-run expected credit loss\",\"strategy\":\"Campbell-Shiller-style log-linearized present-value identity combined with a panel VAR of credit spreads, excess returns, default probability, and ratings; expectations are taken from VAR forecasts\",\"key_assumptions\":[\"Log-linear approximation of bond excess returns is accurate\",\"VAR adequately captures conditional expectations\",\"Recovery rate on coupons equals recovery rate on principal\",\"Price data are complete upon default\"],\"threats_addressed\":[\"Missing default prices, filled from Moody's DRS\",\"Sensitivity to the priority order across overlapping databases (Appendix B)\",\"Infrequent default, handled with a large panel\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"appendix\",\"types\":[\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes state the monthly sample period 1973-2011 and define the VAR-implied long-run coefficient formula. Panels separate long-run credit-loss and excess-return coefficients and report volatility ratios. Summary statistics are shown both raw and cross-sectionally demeaned, with variables in percent.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'nearly as much' / 'about half' approximations\",\"'can shed light' / 'can be useful' for implications\",\"future research deferrals\"]},\"claim_evidence\":[{\"claim\":\"Long-run expected credit loss volatility is about two-thirds of credit spread volatility in the cross-section.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Expected returns account for about half of cross-sectional credit spread variation, nearly as much as expected credit loss.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Risk premiums dominate time-series variation of the aggregate market portfolio's credit spread, as credit loss diversifies away.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The joint decomposition links bond credit spreads to stock book-to-market components.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Long-run expected credit loss and returns vary with credit spreads differently across ratings.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Results are not driven by the priority order among overlapping databases.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Appendix B comparison of monthly returns\"}],\"structural_summary\":\"Opens by asking what drives credit spread variation and notes that structural models disagree on the components. Adapts the Campbell-Shiller present-value decomposition to bonds, with credit spread in place of the dividend-price ratio and credit loss in place of dividend growth. Builds a large merged bond panel and previews the volatility-ratio results, then positions the paper against structural, reduced-form, and price-ratio literatures. Section I derives the identity. Section II covers data, VAR estimation, and main results. Section III extends to a joint stock-bond decomposition, and Section IV turns to the aggregate portfolio, where risk premiums dominate. The conclusion restates findings and points to investment and illiquidity extensions. Appendices give the derivation and data details.\",\"writing_profile_keywords\":[\"variance decomposition\",\"present-value identity\",\"VAR\",\"model-free\",\"credit spreads\",\"cross-section vs aggregate\",\"concise\"],\"confidence\":0.8,\"notes\":\"The primary method is coded 'other' because the approach is a VAR-based present-value variance decomposition. Robustness is visible only through the database priority-order check in Appendix B. Significance reporting conventions are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "fixed income / credit risk", "primary_method": "other", "identification": {"estimand": "Shares of cross-sectional and time-series variance of credit spreads attributable to long-run expected excess returns versus long-run expected credit loss", "strategy": "Campbell-Shiller-style log-linearized present-value identity combined with a panel VAR of credit spreads, excess returns, default probability, and ratings; expectations are taken from VAR forecasts", "key_assumptions": ["Log-linear approximation of bond excess returns is accurate", "VAR adequately captures conditional expectations", "Recovery rate on coupons equals recovery rate on principal", "Price data are complete upon default"], "threats_addressed": ["Missing default prices, filled from Moody's DRS", "Sensitivity to the priority order across overlapping databases (Appendix B)", "Infrequent default, handled with a large panel"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 2, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "appendix", "types": ["alternative_specification"]}, "tables": {"convention_notes": "Notes state the monthly sample period 1973-2011 and define the VAR-implied long-run coefficient formula. Panels separate long-run credit-loss and excess-return coefficients and report volatility ratios. Summary statistics are shown both raw and cross-sectionally demeaned, with variables in percent.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'nearly as much' / 'about half' approximations", "'can shed light' / 'can be useful' for implications", "future research deferrals"]}, "claim_evidence": [{"claim": "Long-run expected credit loss volatility is about two-thirds of credit spread volatility in the cross-section.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Expected returns account for about half of cross-sectional credit spread variation, nearly as much as expected credit loss.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Risk premiums dominate time-series variation of the aggregate market portfolio's credit spread, as credit loss diversifies away.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The joint decomposition links bond credit spreads to stock book-to-market components.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Long-run expected credit loss and returns vary with credit spreads differently across ratings.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Results are not driven by the priority order among overlapping databases.", "evidence_type": "table", "evidence_ref": "Appendix B comparison of monthly returns"}], "structural_summary": "Opens by asking what drives credit spread variation and notes that structural models disagree on the components. Adapts the Campbell-Shiller present-value decomposition to bonds, with credit spread in place of the dividend-price ratio and credit loss in place of dividend growth. Builds a large merged bond panel and previews the volatility-ratio results, then positions the paper against structural, reduced-form, and price-ratio literatures. Section I derives the identity. Section II covers data, VAR estimation, and main results. Section III extends to a joint stock-bond decomposition, and Section IV turns to the aggregate portfolio, where risk premiums dominate. The conclusion restates findings and points to investment and illiquidity extensions. Appendices give the derivation and data details.", "writing_profile_keywords": ["variance decomposition", "present-value identity", "VAR", "model-free", "credit spreads", "cross-section vs aggregate", "concise"], "confidence": 0.8, "notes": "The primary method is coded 'other' because the approach is a VAR-based present-value variance decomposition. Robustness is visible only through the database priority-order check in Appendix B. Significance reporting conventions are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.13, "usage": {"input": 5930, "output": 2034, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065194, "request_id": "req_011Cggph8QKmdoupfMwz4MyG", "cached": false, "ts": "2026-11-03T22:58:44+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3c274e8ab145f499cb1251d8a4d1d9206b3d84219d620d040fc2e678af702234.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance (labor search frictions, production-based asset pricing)\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Cross-sectional return premium associated with firm stock-return loadings on changes in aggregate labor market tightness, interpreted as compensation for exposure to matching efficiency shocks\",\"strategy\":\"Rolling firm-level regressions estimate loadings on log changes in labor market tightness, controlling for the market. These feed portfolio sorts and Fama-MacBeth regressions with factor-model alphas. A calibrated partial-equilibrium search model is then used to interpret the premium.\",\"key_assumptions\":[\"Loadings on labor market tightness proxy for exposure to priced matching efficiency shocks\",\"Exogenous pricing kernel with both productivity and matching efficiency shocks priced\",\"Positive matching efficiency shocks raise discount rates\",\"Approximate aggregation via equilibrium forecasting rules\"],\"threats_addressed\":[\"Explanation by standard factor models (four-factor alphas)\",\"Overlap with known characteristics such as size, value, momentum, hiring rate, investment and asset growth\",\"Alternative portfolio formation timing\",\"Alternative calibrations shutting down priced shocks\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define variables and give units, e.g., returns and alphas in percent per month. They report t-statistics in rows below the estimates and state the sample period. Model-versus-data comparison tables place empirical and simulated moments side by side, and calibration parameters are tabulated with symbols.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'proxy for' wording linking loadings to shocks\",\"agnostic stance on source of matching efficiency variation\",\"'consistent with' model predictions\",\"'suggest' in concluding claims\"],\"claim_evidence\":[{\"claim\":\"Firms with low tightness loadings outperform high-loading firms by about 6% per year.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 2\"},{\"claim\":\"The return spread survives four-factor adjustment and alternative portfolio constructions.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Tightness loadings predict returns after controlling for firm characteristics including hiring and investment rates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Calibrated model reproduces aggregate and firm-level labor market moments.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Model replicates the negative relation between tightness loadings and expected returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; Table X\"},{\"claim\":\"Expected excess returns follow a two-factor structure in the market and labor market tightness.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.G; Appendix B\"},{\"claim\":\"Firms sorted by loadings differ in cyclicality of labor characteristics, as the model predicts.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"}],\"structural_summary\":\"The paper opens with labor-flow facts and states its claim up front. The introduction previews the empirical measure and the search model, sketches the cash-flow versus discount-rate mechanism, and positions the work against three literatures before giving a roadmap. The body first documents the anomaly with portfolio sorts, robustness checks and Fama-MacBeth regressions. It then builds a partial-equilibrium search model with heterogeneous firms, calibrates it to match moments and return spreads, and closes by testing the mechanism's micro predictions in firm data. The conclusion restates the findings and calls for further research.\",\"writing_profile_keywords\":[\"empirics-first then model\",\"production-based asset pricing\",\"search and matching\",\"calibration\",\"portfolio sorts\",\"Fama-MacBeth\",\"mechanism inspection\",\"model-data comparison tables\"],\"confidence\":0.85,\"notes\":\"Section idx 2 has a heading-parsing artifact ('(b)'); it continues the empirical results with portfolio sorts and robustness. Standard-error type is not visible in the excerpts. Robustness appears within the results, in a model robustness subsection, and in the Internet Appendix.\"}", "json": null, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.86, "usage": {"input": 8938, "output": 2405, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084646, "request_id": "req_011Cggpi1Y7G1FaQpkW64PC5", "cached": false, "ts": "2026-11-03T22:58:59+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3e64809cbee0355608db222ccf008d046d22a1372570424d03c4df29b8a89216.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"methodological_econometric\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Posterior probabilities of competing linear factor pricing models, and Bayes factors for zero-alpha restrictions on test assets and excluded factors\",\"strategy\":\"Bayesian model comparison: priors on alphas are scaled by a multiple of the market Sharpe ratio, closed-form Bayes factors are derived from the GRS F-statistic, and probabilities are aggregated over all factor subsets, including categorical factor versions\",\"key_assumptions\":[\"Returns are i.i.d. multivariate normal conditional on the factors\",\"Prior on alphas is normal with scale tied to a plausible maximum Sharpe ratio multiple\",\"Uniform prior over models, or over factor versions in the categorical case\",\"Market factor included by default, a requirement relaxed later in the paper\"],\"threats_addressed\":[\"Prior sensitivity, checked across multiple Sharpe ratio multiples\",\"Overfitting from multiple versions of the same factor concept\",\"Models being only approximately true\",\"Out-of-sample performance of the selected models\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_method\",\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Posterior model probabilities are reported in percent across several prior Sharpe multiples. Factor alphas are annualized with t-statistics in parentheses. Out-of-sample Sharpe ratios are compared across models.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\",\"figures_note\":\"Figures plot posterior probabilities over recursive samples and as functions of the prior multiple; no confidence bands are evident.\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"Acknowledging that all models are literally false and testing approximate models\",\"Noting that maximum model probabilities stay below 50%\",\"Presenting prior sensitivity across multiple Sharpe multiples\",\"Listing limitations of the i.i.d. normal specification for future work\"]},\"claim_evidence\":[{\"claim\":\"Bayesian test of zero alphas can be computed in closed form from the GRS F-statistic\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendices A-B\"},{\"claim\":\"Among FF3 subsets, {Mkt HML} gets about 60% posterior probability and FF3 about 39%\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Introduction paragraph 2\"},{\"claim\":\"Six-factor model {Mkt IA ROE SMB HMLm UMD} has the highest posterior probability over 1972-2015\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-II; Figures 1-2\"},{\"claim\":\"HMLm is not redundant relative to the other factors in the best model\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table IV\"},{\"claim\":\"UMD becomes redundant when HMLm is excluded from the best model\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"Top-ranked models achieve higher in-sample Sharpe ratios than CAPM, FF3, FF5 and HXZ\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Absolute tests of the best model are shown on size/momentum and book-to-market/investment portfolios\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6\"}],\"structural_summary\":\"The paper opens by motivating the need for a joint model-comparison tool, then builds the method step by step. It first derives an absolute Bayesian alpha test, then relative tests between models, then simultaneous comparison of all factor subsets, and finally a categorical-factor extension. It applies the method to 10 prominent factors with prior-sensitivity analysis and redundancy checks. Descriptive out-of-sample Sharpe ratio evidence and absolute tests of the best model follow. The conclusion restates the method, positions it against related Bayesian work, and lists extensions.\",\"writing_profile_keywords\":[\"Bayesian model comparison\",\"closed-form derivation\",\"factor model horse race\",\"prior sensitivity\",\"method-then-application\",\"appendix derivations\"],\"confidence\":0.8,\"notes\":\"Rule-based section roles labeled the empirical sections as theory; they were reassigned to results/robustness. The primary method is a Bayesian econometric procedure, coded as 'other'. Section VI is framed as descriptive evidence on out-of-sample performance.\"}", "json": {"genre": "methodological_econometric", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "other", "identification": {"estimand": "Posterior probabilities of competing linear factor pricing models, and Bayes factors for zero-alpha restrictions on test assets and excluded factors", "strategy": "Bayesian model comparison: priors on alphas are scaled by a multiple of the market Sharpe ratio, closed-form Bayes factors are derived from the GRS F-statistic, and probabilities are aggregated over all factor subsets, including categorical factor versions", "key_assumptions": ["Returns are i.i.d. multivariate normal conditional on the factors", "Prior on alphas is normal with scale tied to a plausible maximum Sharpe ratio multiple", "Uniform prior over models, or over factor versions in the categorical case", "Market factor included by default, a requirement relaxed later in the paper"], "threats_addressed": ["Prior sensitivity, checked across multiple Sharpe ratio multiples", "Overfitting from multiple versions of the same factor concept", "Models being only approximately true", "Out-of-sample performance of the selected models"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}], "contribution_paragraph_index": 0, "contribution_types": ["new_method", "new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Posterior model probabilities are reported in percent across several prior Sharpe multiples. Factor alphas are annualized with t-statistics in parentheses. Out-of-sample Sharpe ratios are compared across models.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no", "figures_note": "Figures plot posterior probabilities over recursive samples and as functions of the prior multiple; no confidence bands are evident."}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["Acknowledging that all models are literally false and testing approximate models", "Noting that maximum model probabilities stay below 50%", "Presenting prior sensitivity across multiple Sharpe multiples", "Listing limitations of the i.i.d. normal specification for future work"]}, "claim_evidence": [{"claim": "Bayesian test of zero alphas can be computed in closed form from the GRS F-statistic", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendices A-B"}, {"claim": "Among FF3 subsets, {Mkt HML} gets about 60% posterior probability and FF3 about 39%", "evidence_type": "none_stated", "evidence_ref": "Introduction paragraph 2"}, {"claim": "Six-factor model {Mkt IA ROE SMB HMLm UMD} has the highest posterior probability over 1972-2015", "evidence_type": "table", "evidence_ref": "Tables I-II; Figures 1-2"}, {"claim": "HMLm is not redundant relative to the other factors in the best model", "evidence_type": "figure", "evidence_ref": "Figure 3; Table IV"}, {"claim": "UMD becomes redundant when HMLm is excluded from the best model", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "Top-ranked models achieve higher in-sample Sharpe ratios than CAPM, FF3, FF5 and HXZ", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Absolute tests of the best model are shown on size/momentum and book-to-market/investment portfolios", "evidence_type": "figure", "evidence_ref": "Figures 5-6"}], "structural_summary": "The paper opens by motivating the need for a joint model-comparison tool, then builds the method step by step. It first derives an absolute Bayesian alpha test, then relative tests between models, then simultaneous comparison of all factor subsets, and finally a categorical-factor extension. It applies the method to 10 prominent factors with prior-sensitivity analysis and redundancy checks. Descriptive out-of-sample Sharpe ratio evidence and absolute tests of the best model follow. The conclusion restates the method, positions it against related Bayesian work, and lists extensions.", "writing_profile_keywords": ["Bayesian model comparison", "closed-form derivation", "factor model horse race", "prior sensitivity", "method-then-application", "appendix derivations"], "confidence": 0.8, "notes": "Rule-based section roles labeled the empirical sections as theory; they were reassigned to results/robustness. The primary method is a Bayesian econometric procedure, coded as 'other'. Section VI is framed as descriptive evidence on out-of-sample performance."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.0, "usage": {"input": 8447, "output": 2312, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080822, "request_id": "req_011CggpsD3emDcfqvmpDHv9r", "cached": false, "ts": "2026-11-03T23:01:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3fd3f626f36bea4e9e417df114ee8e28a9afe2330b92c8e4bbe3e0222ba92706.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Elasticity of net (and gross) employment creation in the nontradable sector, by firm age category, to local income growth at the commuting-zone level\",\"strategy\":\"Bartik-type shift-share instrument: national manufacturing changes interacted with each commuting zone's manufacturing composition instrument local income growth, which shifts local nontradable investment opportunities\",\"key_assumptions\":[\"Local manufacturing income changes driven by national shocks are unrelated to the region's own performance (exclusion)\",\"The nontradable sector responds only to local demand, so the effect operates through local income\"],\"threats_addressed\":[\"Organizational arrangements specific to the nontradable sector such as franchising (construction sector check)\",\"Housing collateral and real estate value channels\",\"Local banking and deposit channels\",\"Short-lived jobs reflecting entrant overreaction\",\"Alternative shock measure (import penetration)\",\"Alternative geographic units, state-by-year fixed effects, alternative measurement periods\",\"Firm size confounding firm age\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes give data source (QWI/LEHD, BDS), sector NAICS codes, unit of observation (CZ-year-firm age), dependent variable definition and horizon, and say controls are the same as earlier tables. Regressions are reported separately by firm age category, and descriptive tables come first.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggesting\",\"consistent with\",\"beyond the scope of any single paper\",\"one specific channel framing\",\"presumably\"]},\"claim_evidence\":[{\"claim\":\"New firms account for most net employment creation in response to local income shocks in the nontradable sector\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table IV\"},{\"claim\":\"Existing young and old firms do not differ significantly in their elasticity to income shocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Positive shocks raise both gross job creation and destruction among existing firms, increasing churn\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Results hold in the construction sector, where franchising is unimportant\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Results are not driven by house price appreciation or collateral channels\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Startup jobs created by shocks are not especially short-lived\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Local bank access does not explain startup responsiveness\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X, Table XI\"},{\"claim\":\"Net job creation among older firms is concentrated in large firms, so the channel operates through age rather than size\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"}],\"structural_summary\":\"The introduction motivates the paper with the role of startups in job creation and the unclear mechanism behind it, then argues for comparing firms by age rather than size. It previews a Bartik-instrumented local demand design and states the main findings: startups drive net creation, existing firms of different ages do not differ, gross flows reveal churn, and age matters rather than size. Robustness and the literature follow, then a roadmap. The data and methodology section precedes a long results section with many subsections on short- and medium-term effects, gross flows, the construction sector, asymmetry, housing collateral and job durability. Separate sections then cover access to capital and size versus age. The conclusion restates the thought experiment, links the findings to the spawning literature and leaves the deeper reasons open.\",\"writing_profile_keywords\":[\"Bartik instrument\",\"firm age\",\"job creation\",\"nontradable sector\",\"commuting zones\",\"churn\",\"mechanism exploration\",\"q-theory framing\"],\"confidence\":0.8,\"notes\":\"The abstract text includes a stray first sentence of the introduction. The excerpts do not show standard error treatment or significance notation.\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "macro_finance", "primary_method": "instrumental_variables", "identification": {"estimand": "Elasticity of net (and gross) employment creation in the nontradable sector, by firm age category, to local income growth at the commuting-zone level", "strategy": "Bartik-type shift-share instrument: national manufacturing changes interacted with each commuting zone's manufacturing composition instrument local income growth, which shifts local nontradable investment opportunities", "key_assumptions": ["Local manufacturing income changes driven by national shocks are unrelated to the region's own performance (exclusion)", "The nontradable sector responds only to local demand, so the effect operates through local income"], "threats_addressed": ["Organizational arrangements specific to the nontradable sector such as franchising (construction sector check)", "Housing collateral and real estate value channels", "Local banking and deposit channels", "Short-lived jobs reflecting entrant overreaction", "Alternative shock measure (import penetration)", "Alternative geographic units, state-by-year fixed effects, alternative measurement periods", "Firm size confounding firm age"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": true, "sentence_index": 0}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Notes give data source (QWI/LEHD, BDS), sector NAICS codes, unit of observation (CZ-year-firm age), dependent variable definition and horizon, and say controls are the same as earlier tables. Regressions are reported separately by firm age category, and descriptive tables come first.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggesting", "consistent with", "beyond the scope of any single paper", "one specific channel framing", "presumably"]}, "claim_evidence": [{"claim": "New firms account for most net employment creation in response to local income shocks in the nontradable sector", "evidence_type": "table", "evidence_ref": "Table III, Table IV"}, {"claim": "Existing young and old firms do not differ significantly in their elasticity to income shocks", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Positive shocks raise both gross job creation and destruction among existing firms, increasing churn", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Results hold in the construction sector, where franchising is unimportant", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Results are not driven by house price appreciation or collateral channels", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Startup jobs created by shocks are not especially short-lived", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Local bank access does not explain startup responsiveness", "evidence_type": "table", "evidence_ref": "Table X, Table XI"}, {"claim": "Net job creation among older firms is concentrated in large firms, so the channel operates through age rather than size", "evidence_type": "table", "evidence_ref": "Table XII"}], "structural_summary": "The introduction motivates the paper with the role of startups in job creation and the unclear mechanism behind it, then argues for comparing firms by age rather than size. It previews a Bartik-instrumented local demand design and states the main findings: startups drive net creation, existing firms of different ages do not differ, gross flows reveal churn, and age matters rather than size. Robustness and the literature follow, then a roadmap. The data and methodology section precedes a long results section with many subsections on short- and medium-term effects, gross flows, the construction sector, asymmetry, housing collateral and job durability. Separate sections then cover access to capital and size versus age. The conclusion restates the thought experiment, links the findings to the spawning literature and leaves the deeper reasons open.", "writing_profile_keywords": ["Bartik instrument", "firm age", "job creation", "nontradable sector", "commuting zones", "churn", "mechanism exploration", "q-theory framing"], "confidence": 0.8, "notes": "The abstract text includes a stray first sentence of the introduction. The excerpts do not show standard error treatment or significance notation."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.63, "usage": {"input": 7343, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072966, "request_id": "req_011CggpcETKLpNq2RhMDyQ4L", "cached": false, "ts": "2026-11-03T22:57:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/436975d688e36a4927227083acbcc05fdd7dc111cc4d6dd4a0540f6c5afa3be8.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Marginal propensity to borrow out of housing collateral: the change in equity extraction per dollar increase in collateralized borrowing capacity, holding lifetime wealth fixed.\",\"strategy\":\"Difference-in-differences that compares owners of price-controlled MPDU units with owners of market-rate units in the same subdivision, before and after the fully anticipated expiration of resale price controls. Staggered expiration dates allow flexible controls for calendar-year and locality effects. The borrowing response is scaled by the estimated increase in transaction prices (pledgeable collateral) to obtain the marginal propensity.\",\"key_assumptions\":[\"Parallel trends: absent expiration, borrowing and prices of MPDU and market-rate owners in the same subdivision would have evolved similarly\",\"Expiration is fully anticipated at purchase, so it does not change expected lifetime wealth\",\"Expiration changes only collateralized borrowing capacity, not the market value of the home\"],\"threats_addressed\":[\"Wealth effects confounding collateral effects in house-price-based designs\",\"Pre-trends (graphical evidence and dynamic event-time estimates)\",\"Spurious timing (placebo expiration tests)\",\"Compositional differences between treated and control units (matching estimates, within-ownership-spell comparisons)\",\"Specification sensitivity\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"placebo\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Each regression table note states the estimation level (transaction, property-year, or loan), the dependent variable and its units, and that each column is a separate DiD regression with the treatment dummy reported. The summary-statistics table reports means with standard deviations in parentheses. The standard-error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggest\",\"potentially important role\",\"at least some fraction\",\"likely used\",\"acknowledging data limitations\"]},\"claim_evidence\":[{\"claim\":\"Expiration of price controls raises transaction prices (pledgeable collateral) of MPDU units.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 5\"},{\"claim\":\"Annual probability of equity extraction rises by about 4 percentage points for previously controlled owners after expiration.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 6\"},{\"claim\":\"MPB out of housing collateral is roughly $0.04–$0.13 and is correlated with initial leverage.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 7\"},{\"claim\":\"Home improvement permit applications rise 0.6–1 percentage points (60–100% of baseline) after expiration.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Placebo expiration dates produce no comparable borrowing response.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"},{\"claim\":\"MPDU and control units follow parallel trends before expiration.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"}],\"structural_summary\":\"The introduction moves from macro motivation (household debt cycles) to a gap: the lack of causal evidence on collateral constraints separate from wealth effects. It then poses three questions and introduces a quasi-experiment based on anticipated expiration of inclusionary-zoning price controls, with an upfront identification argument and a description of a newly assembled data set. Before the empirical design, the paper presents institutional background and a stylized model showing why the setting separates collateral from wealth effects. Results begin with graphical parallel-trends checks, then report DiD estimates for prices, extensive-margin and total borrowing, and combine them into an MPB. Heterogeneity analyses by leverage and income follow, with robustness (placebo, matching, alternative specifications) placed inside the results section. A separate section uses permits and foreclosures to assess real uses of extracted funds. A short conclusion restates the design and the main magnitudes.\",\"writing_profile_keywords\":[\"quasi-experiment\",\"difference-in-differences\",\"collateral constraints\",\"marginal propensity to borrow\",\"institutional detail\",\"stylized model motivating identification\",\"graphical evidence first\",\"heterogeneity analysis\",\"single-author first person\"],\"confidence\":0.85,\"notes\":\"Rule-based roles labeled the Empirical Framework section as theory; it is reassigned here to empirical_design. Section V is reassigned to results. The standard-error type and significance notation are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "macro_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Marginal propensity to borrow out of housing collateral: the change in equity extraction per dollar increase in collateralized borrowing capacity, holding lifetime wealth fixed.", "strategy": "Difference-in-differences that compares owners of price-controlled MPDU units with owners of market-rate units in the same subdivision, before and after the fully anticipated expiration of resale price controls. Staggered expiration dates allow flexible controls for calendar-year and locality effects. The borrowing response is scaled by the estimated increase in transaction prices (pledgeable collateral) to obtain the marginal propensity.", "key_assumptions": ["Parallel trends: absent expiration, borrowing and prices of MPDU and market-rate owners in the same subdivision would have evolved similarly", "Expiration is fully anticipated at purchase, so it does not change expected lifetime wealth", "Expiration changes only collateralized borrowing capacity, not the market value of the home"], "threats_addressed": ["Wealth effects confounding collateral effects in house-price-based designs", "Pre-trends (graphical evidence and dynamic event-time estimates)", "Spurious timing (placebo expiration tests)", "Compositional differences between treated and control units (matching estimates, within-ownership-spell comparisons)", "Specification sensitivity"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_identification", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["placebo", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Each regression table note states the estimation level (transaction, property-year, or loan), the dependent variable and its units, and that each column is a separate DiD regression with the treatment dummy reported. The summary-statistics table reports means with standard deviations in parentheses. The standard-error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggest", "potentially important role", "at least some fraction", "likely used", "acknowledging data limitations"]}, "claim_evidence": [{"claim": "Expiration of price controls raises transaction prices (pledgeable collateral) of MPDU units.", "evidence_type": "table", "evidence_ref": "Table II; Figure 5"}, {"claim": "Annual probability of equity extraction rises by about 4 percentage points for previously controlled owners after expiration.", "evidence_type": "table", "evidence_ref": "Table III; Figure 6"}, {"claim": "MPB out of housing collateral is roughly $0.04–$0.13 and is correlated with initial leverage.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 7"}, {"claim": "Home improvement permit applications rise 0.6–1 percentage points (60–100% of baseline) after expiration.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Placebo expiration dates produce no comparable borrowing response.", "evidence_type": "figure", "evidence_ref": "Figure 9"}, {"claim": "MPDU and control units follow parallel trends before expiration.", "evidence_type": "figure", "evidence_ref": "Figure 4"}], "structural_summary": "The introduction moves from macro motivation (household debt cycles) to a gap: the lack of causal evidence on collateral constraints separate from wealth effects. It then poses three questions and introduces a quasi-experiment based on anticipated expiration of inclusionary-zoning price controls, with an upfront identification argument and a description of a newly assembled data set. Before the empirical design, the paper presents institutional background and a stylized model showing why the setting separates collateral from wealth effects. Results begin with graphical parallel-trends checks, then report DiD estimates for prices, extensive-margin and total borrowing, and combine them into an MPB. Heterogeneity analyses by leverage and income follow, with robustness (placebo, matching, alternative specifications) placed inside the results section. A separate section uses permits and foreclosures to assess real uses of extracted funds. A short conclusion restates the design and the main magnitudes.", "writing_profile_keywords": ["quasi-experiment", "difference-in-differences", "collateral constraints", "marginal propensity to borrow", "institutional detail", "stylized model motivating identification", "graphical evidence first", "heterogeneity analysis", "single-author first person"], "confidence": 0.85, "notes": "Rule-based roles labeled the Empirical Framework section as theory; it is reassigned here to empirical_design. Section V is reassigned to results. The standard-error type and significance notation are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.86, "usage": {"input": 7759, "output": 2274, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07731, "request_id": "req_011Cggpqjn3o2jc1TEjkgPfH", "cached": false, "ts": "2026-11-03T23:00:44+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4499d259c9d3016f28ffc4d8b519489d5f773161841d89a88ca6bc62ae631ea8.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"household_finance / mutual funds\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Sensitivity of aggregate monthly flows into active value and growth funds to option-implied investor preferences for downside protection and upside potential.\",\"strategy\":\"Time-series regressions of category-level active fund flows on pricing-kernel slopes and Prelec weighting parameters estimated from S&P 500 option prices, with sentiment, return and flow controls. Cross-sectional splits by fund tail properties, Active Share, and retirement versus retail funds support the hypothesized channel.\",\"key_assumptions\":[\"Option-implied pricing kernel reflects risk preferences of mutual fund investors via no-arbitrage\",\"Preference proxies are not merely proxies for investor sentiment\",\"Prelec functional form, or the moneyness range chosen for the PK slopes, captures the relevant tail preferences\"],\"threats_addressed\":[\"investor sentiment confound (Baker-Wurgler control, low correlations)\",\"business cycle / recessions (NBER indicator)\",\"market volatility (VIX control)\",\"benchmark choice for active vs passive comparison\",\"differences in security characteristics held by active vs passive funds\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":8,\"move\":\"robustness_mention\"},{\"paragraph_index\":9,\"move\":\"identification_defense\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"identification_defense\"},{\"paragraph_index\":12,\"move\":\"preview_of_results\"},{\"paragraph_index\":13,\"move\":\"preview_of_results\"},{\"paragraph_index\":14,\"move\":\"preview_of_results\"},{\"paragraph_index\":15,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":16,\"move\":\"contribution_statement\"},{\"paragraph_index\":17,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":18,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":16,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes describe sample period, bootstrap design (100 funds, 400 repetitions), variable construction and controls. Summary statistics state units (millions USD, %). Regression tables are described as aggregate flow regressions with listed controls. Tercile sorts are defined in the notes. Standard error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'can help explain' / 'might help explain'\",\"'appear to' / 'appear consistent with'\",\"'to the extent that'\",\"explicit caveat that results need not imply investor segmentation\",\"'suggest' for implications\"],\"typical_devices_note\":\"\"},\"claim_evidence\":[{\"claim\":\"Active value funds hedge downside risk better than passive benchmarks (lower co-skewness, downside beta, jump-risk hedging).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-IV\"},{\"claim\":\"Active growth funds show higher skewness and positive idiosyncratic skewness relative to benchmarks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-III\"},{\"claim\":\"In an RDEU portfolio model, demand for active value (growth) funds rises with downside-protection (upside-seeking) preference.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-3\"},{\"claim\":\"Investors would pay reasonable fees for access to active fund portfolios under the model.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"Option-implied downside (upside) preferences explain flows into active large value (growth) funds, with magnitudes comparable to past performance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V-VI\"},{\"claim\":\"Flow sensitivity to preferences is stronger for funds with better tail properties and higher Active Share.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-XI\"},{\"claim\":\"Retirement value funds are more sensitive to downside preferences; retail growth funds more sensitive to upside preferences.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"}],\"structural_summary\":\"Opens with the puzzle of large active-fund assets despite mediocre performance. First documents distributional differences between active and passive funds using bootstrapped returns and several benchmarks. Then builds a rank-dependent expected utility portfolio model showing these differences matter to tail-sensitive investors. Next constructs option-implied preference proxies and runs time-series flow regressions with controls. Cross-sectional splits sharpen the channel. Closes by restating the findings and linking them to debates on managerial skill and non-EU preferences.\",\"writing_profile_keywords\":[\"puzzle-driven\",\"model-then-test\",\"bootstrap comparisons\",\"option-implied preferences\",\"cross-sectional validation\",\"alternative proxies\",\"literature complement\"],\"confidence\":0.8,\"notes\":\"Mixed paper: a numerical theory section sits inside the main text, followed by empirical tests. Standard error conventions and significance notation are not visible in the excerpts. The Internet Appendix holds robustness checks.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "household_finance / mutual funds", "primary_method": "reduced_form_regression", "identification": {"estimand": "Sensitivity of aggregate monthly flows into active value and growth funds to option-implied investor preferences for downside protection and upside potential.", "strategy": "Time-series regressions of category-level active fund flows on pricing-kernel slopes and Prelec weighting parameters estimated from S&P 500 option prices, with sentiment, return and flow controls. Cross-sectional splits by fund tail properties, Active Share, and retirement versus retail funds support the hypothesized channel.", "key_assumptions": ["Option-implied pricing kernel reflects risk preferences of mutual fund investors via no-arbitrage", "Preference proxies are not merely proxies for investor sentiment", "Prelec functional form, or the moneyness range chosen for the PK slopes, captures the relevant tail preferences"], "threats_addressed": ["investor sentiment confound (Baker-Wurgler control, low correlations)", "business cycle / recessions (NBER indicator)", "market volatility (VIX control)", "benchmark choice for active vs passive comparison", "differences in security characteristics held by active vs passive funds"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": true, "sentence_index": 0}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 8, "move": "robustness_mention"}, {"paragraph_index": 9, "move": "identification_defense"}, {"paragraph_index": 10, "move": "preview_of_results"}, {"paragraph_index": 11, "move": "identification_defense"}, {"paragraph_index": 12, "move": "preview_of_results"}, {"paragraph_index": 13, "move": "preview_of_results"}, {"paragraph_index": 14, "move": "preview_of_results"}, {"paragraph_index": 15, "move": "related_literature_positioning"}, {"paragraph_index": 16, "move": "contribution_statement"}, {"paragraph_index": 17, "move": "policy_or_broader_implication"}, {"paragraph_index": 18, "move": "roadmap"}], "contribution_paragraph_index": 16, "contribution_types": ["new_evidence", "new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "subsample", "alternative_specification"]}, "tables": {"convention_notes": "Notes describe sample period, bootstrap design (100 funds, 400 repetitions), variable construction and controls. Summary statistics state units (millions USD, %). Regression tables are described as aggregate flow regressions with listed controls. Tercile sorts are defined in the notes. Standard error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'can help explain' / 'might help explain'", "'appear to' / 'appear consistent with'", "'to the extent that'", "explicit caveat that results need not imply investor segmentation", "'suggest' for implications"], "typical_devices_note": ""}, "claim_evidence": [{"claim": "Active value funds hedge downside risk better than passive benchmarks (lower co-skewness, downside beta, jump-risk hedging).", "evidence_type": "table", "evidence_ref": "Tables II-IV"}, {"claim": "Active growth funds show higher skewness and positive idiosyncratic skewness relative to benchmarks.", "evidence_type": "table", "evidence_ref": "Tables II-III"}, {"claim": "In an RDEU portfolio model, demand for active value (growth) funds rises with downside-protection (upside-seeking) preference.", "evidence_type": "figure", "evidence_ref": "Figures 2-3"}, {"claim": "Investors would pay reasonable fees for access to active fund portfolios under the model.", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "Option-implied downside (upside) preferences explain flows into active large value (growth) funds, with magnitudes comparable to past performance.", "evidence_type": "table", "evidence_ref": "Tables V-VI"}, {"claim": "Flow sensitivity to preferences is stronger for funds with better tail properties and higher Active Share.", "evidence_type": "table", "evidence_ref": "Tables VIII-XI"}, {"claim": "Retirement value funds are more sensitive to downside preferences; retail growth funds more sensitive to upside preferences.", "evidence_type": "table", "evidence_ref": "Table XII"}], "structural_summary": "Opens with the puzzle of large active-fund assets despite mediocre performance. First documents distributional differences between active and passive funds using bootstrapped returns and several benchmarks. Then builds a rank-dependent expected utility portfolio model showing these differences matter to tail-sensitive investors. Next constructs option-implied preference proxies and runs time-series flow regressions with controls. Cross-sectional splits sharpen the channel. Closes by restating the findings and linking them to debates on managerial skill and non-EU preferences.", "writing_profile_keywords": ["puzzle-driven", "model-then-test", "bootstrap comparisons", "option-implied preferences", "cross-sectional validation", "alternative proxies", "literature complement"], "confidence": 0.8, "notes": "Mixed paper: a numerical theory section sits inside the main text, followed by empirical tests. Standard error conventions and significance notation are not visible in the excerpts. The Internet Appendix holds robustness checks."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.59, "usage": {"input": 11201, "output": 2497, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.095538, "request_id": "req_011CggqFZsej3JzRp4mSAUVi", "cached": false, "ts": "2026-11-03T23:06:07+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/449f407999a6a4ce5ec43bbf42efda2cfe53558d735ade566282c7906bacc6c1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure; financial_econometrics\",\"primary_method\":\"machine_learning_textual\",\"identification\":{\"estimand\":\"Increase in out-of-sample adjusted R2 (and forecast-implied Sharpe ratio) from adding LASSO cross-stock lagged-return forecasts to benchmark one-minute-ahead return forecasts\",\"strategy\":\"Rolling 30-minute-window LASSO estimation with cross-validated penalty on all NYSE lagged returns, out-of-sample evaluation against OLS benchmarks, Giacomini-White significance test; event study of selection rate around Ravenpack news\",\"key_assumptions\":[\"Gains only arise if the cross-section contains a sparse set of signals (fewer than about 30)\",\"Out-of-sample evaluation guards against in-sample overfitting\"],\"threats_addressed\":[\"in-sample overfitting from large candidate set\",\"gains coming from shrinkage versus selection (Appendix C)\",\"no-signal or dense-signal data-generating processes (simulation placebo)\",\"trading costs eroding Sharpe ratios\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":6,\"move\":\"other\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"contribution_statement\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"other\"}],\"contribution_paragraph_index\":10,\"contribution_types\":[\"new_evidence\",\"new_method\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes give sample (250 randomly chosen NYSE stocks per day, 2005-2012), define fit as percentage-point increase in adjusted R2, and report standard errors in parentheses clustered by stock-day or by year. Table VII states units as selections per minute.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests' framing\",\"explicit caveat that LASSO complements rather than replaces intuition\",\"'tends to identify'\",\"conditional statements about when gains occur\"]},\"claim_evidence\":[{\"claim\":\"Adding LASSO forecasts to benchmark models raises out-of-sample adjusted R2 by at least about 1.2 percentage points\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Table II\"},{\"claim\":\"Forecast-implied trading strategy earns an annualized Sharpe ratio of about 1.8 net of trading costs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"LASSO predictors are unexpected: the cross-validated penalty averages 2.5% per month\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"Selected predictors are short-lived and sparse\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5; Figure 6; Table VI\"},{\"claim\":\"LASSO more often selects lagged returns of stocks with news flashes about fundamentals\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 7\"},{\"claim\":\"LASSO raises out-of-sample fit only when signals are sparse, not when there are none or many\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Appendix A; Figures A1-A5\"}],\"structural_summary\":\"The introduction motivates the problem by contrasting intuition-based predictor identification with fast, complex markets, using a concrete stock example. It previews the method and the out-of-sample results, including magnitudes. It then defends the out-of-sample framing and bounds the claim by presenting LASSO as a complement to intuition. Literature positioning follows inside the introduction, ending with notation. The body explains the LASSO, then reports out-of-sample fit and Sharpe ratios. It next characterizes the selected predictors as unexpected, short-lived and sparse, and links them to news via an event study. A one-paragraph conclusion restates the findings. Appendices cover simulations with placebo data-generating processes, the significance test, and a decomposition of the gains.\",\"writing_profile_keywords\":[\"high-frequency return predictability\",\"LASSO\",\"out-of-sample forecasting\",\"sparsity\",\"conversational tone\",\"concrete examples\",\"news event study\"],\"confidence\":0.8,\"notes\":\"Primary method classified as machine learning (LASSO), although the evaluation relies on regressions. The introduction is partly truncated, so a roadmap may exist but is not visible; it uses inline subsection labels such as 'Out-of-Sample Performance' and 'Predictor Analysis' instead.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure; financial_econometrics", "primary_method": "machine_learning_textual", "identification": {"estimand": "Increase in out-of-sample adjusted R2 (and forecast-implied Sharpe ratio) from adding LASSO cross-stock lagged-return forecasts to benchmark one-minute-ahead return forecasts", "strategy": "Rolling 30-minute-window LASSO estimation with cross-validated penalty on all NYSE lagged returns, out-of-sample evaluation against OLS benchmarks, Giacomini-White significance test; event study of selection rate around Ravenpack news", "key_assumptions": ["Gains only arise if the cross-section contains a sparse set of signals (fewer than about 30)", "Out-of-sample evaluation guards against in-sample overfitting"], "threats_addressed": ["in-sample overfitting from large candidate set", "gains coming from shrinkage versus selection (Appendix C)", "no-signal or dense-signal data-generating processes (simulation placebo)", "trading costs eroding Sharpe ratios"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 6, "move": "other"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "contribution_statement"}, {"paragraph_index": 11, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "other"}], "contribution_paragraph_index": 10, "contribution_types": ["new_evidence", "new_method"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Notes give sample (250 randomly chosen NYSE stocks per day, 2005-2012), define fit as percentage-point increase in adjusted R2, and report standard errors in parentheses clustered by stock-day or by year. Table VII states units as selections per minute.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests' framing", "explicit caveat that LASSO complements rather than replaces intuition", "'tends to identify'", "conditional statements about when gains occur"]}, "claim_evidence": [{"claim": "Adding LASSO forecasts to benchmark models raises out-of-sample adjusted R2 by at least about 1.2 percentage points", "evidence_type": "table", "evidence_ref": "Table I; Table II"}, {"claim": "Forecast-implied trading strategy earns an annualized Sharpe ratio of about 1.8 net of trading costs", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "LASSO predictors are unexpected: the cross-validated penalty averages 2.5% per month", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "Selected predictors are short-lived and sparse", "evidence_type": "figure", "evidence_ref": "Figure 5; Figure 6; Table VI"}, {"claim": "LASSO more often selects lagged returns of stocks with news flashes about fundamentals", "evidence_type": "table", "evidence_ref": "Table VII; Figure 7"}, {"claim": "LASSO raises out-of-sample fit only when signals are sparse, not when there are none or many", "evidence_type": "figure", "evidence_ref": "Appendix A; Figures A1-A5"}], "structural_summary": "The introduction motivates the problem by contrasting intuition-based predictor identification with fast, complex markets, using a concrete stock example. It previews the method and the out-of-sample results, including magnitudes. It then defends the out-of-sample framing and bounds the claim by presenting LASSO as a complement to intuition. Literature positioning follows inside the introduction, ending with notation. The body explains the LASSO, then reports out-of-sample fit and Sharpe ratios. It next characterizes the selected predictors as unexpected, short-lived and sparse, and links them to news via an event study. A one-paragraph conclusion restates the findings. Appendices cover simulations with placebo data-generating processes, the significance test, and a decomposition of the gains.", "writing_profile_keywords": ["high-frequency return predictability", "LASSO", "out-of-sample forecasting", "sparsity", "conversational tone", "concrete examples", "news event study"], "confidence": 0.8, "notes": "Primary method classified as machine learning (LASSO), although the evaluation relies on regressions. The introduction is partly truncated, so a roadmap may exist but is not visible; it uses inline subsection labels such as 'Out-of-Sample Performance' and 'Predictor Analysis' instead."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.44, "usage": {"input": 8820, "output": 2304, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082154, "request_id": "req_011CggqD1ZPsxXcVhjHqJ5YE", "cached": false, "ts": "2026-11-03T23:05:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/45b8805e39601d8e791639f18bb37d2764747b025124b270d1682df9ee3fac7d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"methodological_econometric\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Latent quarterly time series of average realized PE returns, split into traded-factor exposures (betas) and a time-varying PE premium (alpha); secondarily, the association of buyout returns with cross-market yield spreads.\",\"strategy\":\"Bayesian MCMC / likelihood estimation of a latent return index. One moment condition per fund equates the present value of distributions to the present value of contributions, which overidentifies the system. The method is validated by Monte Carlo simulations and followed by time-series regressions of the buyout index on macro variables.\",\"key_assumptions\":[\"PV of LP distributions equals PV of contributions when discounted at the latent index return\",\"Fund-level error structure as specified in the model\",\"More funds than time periods, giving overidentification\",\"Priors chosen for the Bayesian estimation\"],\"threats_addressed\":[\"Sparse or unbalanced data\",\"Violations of the model's error structure\",\"Unmodeled fee structure\",\"Realistic return dynamics, checked with listed-stock pseudo-funds\",\"Identifying problem zones such as high idiosyncratic volatility\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_method\",\"new_data_or_measure\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Simulation tables report mean, volatility, correlation and MSE of estimated vs. true series. Bayesian factor-model tables report posterior means and standard deviations by fund-type panel. Comparison tables report annualized statistics against industry indices. The macro regression table gives Newey-West t-statistics in italics beneath coefficients.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with the conjecture'\",\"'to a first approximation'\",\"'suggests' / 'may be beneficial'\",\"explicit discussion of when the method degrades\"]},\"claim_evidence\":[{\"claim\":\"The estimation method recovers true return series well and is robust to many assumption violations, but degrades under some conditions.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Figures 1-3\"},{\"claim\":\"PE returns are only partly spanned by traded factors; PE resembles a levered investment in small and mid-cap equities.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"The cash-flow-based indices differ from industry appraisal-based indices in their return properties.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 4\"},{\"claim\":\"Buyout returns are higher when the equity-versus-debt yield spread is larger, supporting market segmentation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Cycles of PE sub-classes are not highly correlated, suggesting gains from diversification within PE.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"}],\"structural_summary\":\"The paper opens with the data problem of illiquid PE: no transaction-based performance measures. It then presents a Bayesian MCMC method that extracts a latent quarterly return index from LP cash flows. A long methodology section derives the model and stress-tests it with several simulation designs, including model-consistent errors, listed-stock pseudo-funds, problem zones and fees. A short data section on Preqin follows. Results cover factor exposures and the PE premium, a comparison with industry indices, and a test of the market segmentation hypothesis. The conclusion restates findings and notes diversification and regulatory implications.\",\"writing_profile_keywords\":[\"methodology paper\",\"latent index estimation\",\"Bayesian MCMC\",\"simulation validation\",\"private equity\",\"market segmentation test\",\"cash-flow data\"],\"confidence\":0.8,\"notes\":\"The introduction labels a 'primary' and a 'third' contribution; the second is likely in the truncated paragraph 1. The standard error description is mixed because tables report posterior SDs and Newey-West t-stats. Robustness checks appear both in Section I simulations and in the Internet Appendix.\"}", "json": {"genre": "methodological_econometric", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "asset_pricing", "primary_method": "structural_estimation", "identification": {"estimand": "Latent quarterly time series of average realized PE returns, split into traded-factor exposures (betas) and a time-varying PE premium (alpha); secondarily, the association of buyout returns with cross-market yield spreads.", "strategy": "Bayesian MCMC / likelihood estimation of a latent return index. One moment condition per fund equates the present value of distributions to the present value of contributions, which overidentifies the system. The method is validated by Monte Carlo simulations and followed by time-series regressions of the buyout index on macro variables.", "key_assumptions": ["PV of LP distributions equals PV of contributions when discounted at the latent index return", "Fund-level error structure as specified in the model", "More funds than time periods, giving overidentification", "Priors chosen for the Bayesian estimation"], "threats_addressed": ["Sparse or unbalanced data", "Violations of the model's error structure", "Unmodeled fee structure", "Realistic return dynamics, checked with listed-stock pseudo-funds", "Identifying problem zones such as high idiosyncratic volatility"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_method", "new_data_or_measure", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters", "alternative_specification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Simulation tables report mean, volatility, correlation and MSE of estimated vs. true series. Bayesian factor-model tables report posterior means and standard deviations by fund-type panel. Comparison tables report annualized statistics against industry indices. The macro regression table gives Newey-West t-statistics in italics beneath coefficients.", "units_stated_in_notes": true, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with the conjecture'", "'to a first approximation'", "'suggests' / 'may be beneficial'", "explicit discussion of when the method degrades"]}, "claim_evidence": [{"claim": "The estimation method recovers true return series well and is robust to many assumption violations, but degrades under some conditions.", "evidence_type": "table", "evidence_ref": "Table I; Figures 1-3"}, {"claim": "PE returns are only partly spanned by traded factors; PE resembles a levered investment in small and mid-cap equities.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "The cash-flow-based indices differ from industry appraisal-based indices in their return properties.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 4"}, {"claim": "Buyout returns are higher when the equity-versus-debt yield spread is larger, supporting market segmentation.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Cycles of PE sub-classes are not highly correlated, suggesting gains from diversification within PE.", "evidence_type": "figure", "evidence_ref": "Figure 4"}], "structural_summary": "The paper opens with the data problem of illiquid PE: no transaction-based performance measures. It then presents a Bayesian MCMC method that extracts a latent quarterly return index from LP cash flows. A long methodology section derives the model and stress-tests it with several simulation designs, including model-consistent errors, listed-stock pseudo-funds, problem zones and fees. A short data section on Preqin follows. Results cover factor exposures and the PE premium, a comparison with industry indices, and a test of the market segmentation hypothesis. The conclusion restates findings and notes diversification and regulatory implications.", "writing_profile_keywords": ["methodology paper", "latent index estimation", "Bayesian MCMC", "simulation validation", "private equity", "market segmentation test", "cash-flow data"], "confidence": 0.8, "notes": "The introduction labels a 'primary' and a 'third' contribution; the second is likely in the truncated paragraph 1. The standard error description is mixed because tables report posterior SDs and Newey-West t-stats. Robustness checks appear both in Section I simulations and in the Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.2, "usage": {"input": 4574, "output": 2084, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06077, "request_id": "req_011Cggq13955aR1X3ngVFmCv", "cached": false, "ts": "2026-11-03T23:02:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/466048fb6368fbae0427d80a3e781a24f911ddcd695a480cf08b68afb46908e6.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between predicted dynamic short-selling risk (forecast variance of future loan fees) and future returns, price efficiency, and short-selling activity on mispricings\",\"strategy\":\"Two-stage approach: forecast loan-fee variance with an OLS panel model, then use the fitted ShortRisk in double-sorted portfolios (five-factor alphas), Fama-MacBeth regressions, and panel regressions with fixed effects, including interactions with option time to expiration\",\"key_assumptions\":[\"Fitted ShortRisk captures ex ante uncertainty about future loan fees and availability\",\"Controls for current loan fees and supply separate dynamic risk from static constraints\",\"Factor models adequately adjust for risk in the return tests\"],\"threats_addressed\":[\"Confounding with static short-sale constraints such as current loan fees and supply\",\"Interaction with the short interest return-predictability anomaly\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":5,\"move\":\"motivation\"},{\"paragraph_index\":5,\"move\":\"literature_gap\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes state the estimating equation in full, define the variables, give the sample period and data sources, and say whether portfolios are equal- or value-weighted. Portfolio results are reported as monthly percentages, with raw returns and five-factor alphas in separate tables. Fama-MacBeth and panel regressions report fixed effects in the equation.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'is associated with' wording\",\"'suggests'\",\"'may explain some of'\",\"conditional framing ('if short-selling risk is truly a limit to arbitrage')\"]},\"claim_evidence\":[{\"claim\":\"Long-short portfolios sorted on short-selling risk earn five-factor alphas of about 75 bps per month\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Higher short-selling risk predicts lower future returns in the cross-section, controlling for short interest\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table V, Figure 1\"},{\"claim\":\"A one-SD increase in ShortRisk is associated with a 6.8% increase in price delay, net of current loan conditions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Arbitrageurs short less on put-call disparity when short-selling risk is high and option maturity is long\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Loan fees and supply vary with past 20-day returns, linking noise trader risk to short-selling risk\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII, Figures 2-3\"}],\"structural_summary\":\"Opens with a practitioner epigraph and frames the static-cost view of short-selling constraints in prior work. It then reframes these constraints as dynamic risk, using a two-stock thought experiment. The introduction previews three tests in turn: returns, price efficiency (with a magnitude), and short selling on mispricings by holding horizon. A background section reviews the literature and formalizes hypotheses, and a data section builds a forecast-based risk measure. A single long results section covers arbitrageur effects and the link to noise trader risk. The conclusion restates the findings and links them to existing puzzles.\",\"writing_profile_keywords\":[\"epigraph opening\",\"static-vs-dynamic reframing\",\"thought experiment\",\"portfolio sorts\",\"Fama-MacBeth\",\"limits to arbitrage\",\"put-call parity test\",\"hypothesis development section\"],\"confidence\":0.78,\"notes\":\"The introduction epigraph is split across paragraphs 0-4. Robustness checks and standard error details are not visible in the excerpts. The abstract uses causal wording ('affects'), while the conclusion mostly uses associational language.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between predicted dynamic short-selling risk (forecast variance of future loan fees) and future returns, price efficiency, and short-selling activity on mispricings", "strategy": "Two-stage approach: forecast loan-fee variance with an OLS panel model, then use the fitted ShortRisk in double-sorted portfolios (five-factor alphas), Fama-MacBeth regressions, and panel regressions with fixed effects, including interactions with option time to expiration", "key_assumptions": ["Fitted ShortRisk captures ex ante uncertainty about future loan fees and availability", "Controls for current loan fees and supply separate dynamic risk from static constraints", "Factor models adequately adjust for risk in the return tests"], "threats_addressed": ["Confounding with static short-sale constraints such as current loan fees and supply", "Interaction with the short interest return-predictability anomaly"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 5, "move": "motivation"}, {"paragraph_index": 5, "move": "literature_gap"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "preview_of_results"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_data_or_measure", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes state the estimating equation in full, define the variables, give the sample period and data sources, and say whether portfolios are equal- or value-weighted. Portfolio results are reported as monthly percentages, with raw returns and five-factor alphas in separate tables. Fama-MacBeth and panel regressions report fixed effects in the equation.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'is associated with' wording", "'suggests'", "'may explain some of'", "conditional framing ('if short-selling risk is truly a limit to arbitrage')"]}, "claim_evidence": [{"claim": "Long-short portfolios sorted on short-selling risk earn five-factor alphas of about 75 bps per month", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Higher short-selling risk predicts lower future returns in the cross-section, controlling for short interest", "evidence_type": "table", "evidence_ref": "Table III, Table V, Figure 1"}, {"claim": "A one-SD increase in ShortRisk is associated with a 6.8% increase in price delay, net of current loan conditions", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Arbitrageurs short less on put-call disparity when short-selling risk is high and option maturity is long", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Loan fees and supply vary with past 20-day returns, linking noise trader risk to short-selling risk", "evidence_type": "table", "evidence_ref": "Table VIII, Figures 2-3"}], "structural_summary": "Opens with a practitioner epigraph and frames the static-cost view of short-selling constraints in prior work. It then reframes these constraints as dynamic risk, using a two-stock thought experiment. The introduction previews three tests in turn: returns, price efficiency (with a magnitude), and short selling on mispricings by holding horizon. A background section reviews the literature and formalizes hypotheses, and a data section builds a forecast-based risk measure. A single long results section covers arbitrageur effects and the link to noise trader risk. The conclusion restates the findings and links them to existing puzzles.", "writing_profile_keywords": ["epigraph opening", "static-vs-dynamic reframing", "thought experiment", "portfolio sorts", "Fama-MacBeth", "limits to arbitrage", "put-call parity test", "hypothesis development section"], "confidence": 0.78, "notes": "The introduction epigraph is split across paragraphs 0-4. Robustness checks and standard error details are not visible in the excerpts. The abstract uses causal wording ('affects'), while the conclusion mostly uses associational language."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 22.63, "usage": {"input": 4934, "output": 2500, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07053, "request_id": "req_011CggpsEr3sggcL996niFU6", "cached": false, "ts": "2026-11-03T23:01:07+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/469c06408ebb74919053bfe6cc5d25c76d4ce1b1613ad7dffcdd7ab3c577a116.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_intermediation_banking; small business credit / government loan guarantee evaluation\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Average effect on the treated of SBA loan amount (per $1 million) on firm employment growth, measured as post-loan minus pre-loan average employment, for growing small firms\",\"strategy\":\"Longitudinal matching (exact matching plus kernel propensity scores on employment history) combined with treatment-control group fixed-effects regressions. IVs built from the lagged local presence or share of branches of Preferred Lender Program banks, with PLP activity defined outside the local market.\",\"key_assumptions\":[\"Bank entry into a county is not based on potential growth of SBA borrowers there\",\"A bank's choice to become a PLP lender is not based on local SBA-recipient growth potential\",\"For the branch-share IV, PLP banks do not gain local market share based on SBA-recipient growth potential\",\"Small businesses rely on local lenders\",\"For matching/OLS: no correlated time-varying unobservables\"],\"threats_addressed\":[\"Selection into loan receipt (positive or negative)\",\"Pre-loan growth trends of recipients\",\"Correlation of SBA availability with local financial development and growth\",\"Firm exit/survivorship\",\"Displacement of competing firms (nontradables)\",\"Time-varying idiosyncratic demand, productivity or cost shocks\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":5,\"move\":\"motivation\"},{\"paragraph_index\":5,\"move\":\"literature_gap\"},{\"paragraph_index\":5,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":8,\"move\":\"identification_defense\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"other\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_identification\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable as the change in average employment, t+1..t+3 versus t-3..t-1. They state that loan amount is in millions of 2010 dollars, list the county-level controls, and distinguish OLS from IV and scaled from unscaled specifications. Summary-statistics tables come first, then OLS, IV, alternative-IV and heterogeneity tables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"explicit caveats paragraph on non-random allocation and external validity\",\"'arguably exogenous' framing of instruments\",\"'if these estimates are correct' conditional\",\"lower-bound interpretation\",\"'preliminary investigation' and 'rough estimates' qualifiers\",\"'suggesting' when linking results to credit constraints\"]},\"claim_evidence\":[{\"claim\":\"Each $1 million of SBA loans raises recipient employment by about 3–3.5 jobs over three post-loan years\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, VI, VII\"},{\"claim\":\"Effects are larger for younger and larger firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"No systematic difference between 7(a) and 504 programs or over the business cycle\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX, X\"},{\"claim\":\"Effects are smaller where local conventional credit is more plentiful\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Displacement of competing firms is around 10% of the total effect in nontradables\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"},{\"claim\":\"Results are similar when exited firms are assigned zero employment\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XIII\"},{\"claim\":\"Taxpayer cost per job is roughly $21,000–$25,000, low relative to other government programs\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section V.B; Neumark (2013) comparison in conclusion\"},{\"claim\":\"PLP lenders account for a large share of SBA loans, especially larger loans\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Only 3.2% of SBA loans occur in counties with no PLP branch in the prior year\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Introduction\"}],\"structural_summary\":\"The paper is a policy-evaluation empirical study. The introduction opens with a historical epigraph and the size of the SBA program, then notes that the employment effect is theoretically ambiguous. It previews linked administrative-census data and the matching method, and motivates instruments from lender institutional features while stating the identifying assumptions. Literature positioning follows, then an explicit caveats paragraph on external validity. The body describes the program institutions and the data, presents matching/OLS results, then the IV approach, and finally heterogeneity, displacement, exit and cost-per-job extensions. The conclusion restates the estimates, bounds them as a lower bound, and draws a cost-effectiveness policy implication.\",\"writing_profile_keywords\":[\"program evaluation\",\"linked administrative data\",\"matching estimator\",\"instrumental variables\",\"explicit caveats\",\"cost per job\",\"heterogeneous effects\",\"policy relevance\"],\"confidence\":0.8,\"notes\":\"The introduction-paragraph indices are fragmented: paragraphs 0–4 are pieces of the epigraph, and paragraph 5 contains the opening motivation, theoretical ambiguity and empirical difficulty. Section IV is labeled 'data' by the role rule but is actually IV methods and results. Inference reporting conventions (stars or SEs) are not visible in the excerpts. Robustness checks (alternative IVs, exit zeros, controls) appear within the results and extensions sections, and an internet appendix exists.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking; small business credit / government loan guarantee evaluation", "primary_method": "instrumental_variables", "identification": {"estimand": "Average effect on the treated of SBA loan amount (per $1 million) on firm employment growth, measured as post-loan minus pre-loan average employment, for growing small firms", "strategy": "Longitudinal matching (exact matching plus kernel propensity scores on employment history) combined with treatment-control group fixed-effects regressions. IVs built from the lagged local presence or share of branches of Preferred Lender Program banks, with PLP activity defined outside the local market.", "key_assumptions": ["Bank entry into a county is not based on potential growth of SBA borrowers there", "A bank's choice to become a PLP lender is not based on local SBA-recipient growth potential", "For the branch-share IV, PLP banks do not gain local market share based on SBA-recipient growth potential", "Small businesses rely on local lenders", "For matching/OLS: no correlated time-varying unobservables"], "threats_addressed": ["Selection into loan receipt (positive or negative)", "Pre-loan growth trends of recipients", "Correlation of SBA availability with local financial development and growth", "Firm exit/survivorship", "Displacement of competing firms (nontradables)", "Time-varying idiosyncratic demand, productivity or cost shocks"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 5, "move": "motivation"}, {"paragraph_index": 5, "move": "literature_gap"}, {"paragraph_index": 5, "move": "tension_or_puzzle"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 8, "move": "identification_defense"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "other"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_data_or_measure", "new_identification"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Table notes define the dependent variable as the change in average employment, t+1..t+3 versus t-3..t-1. They state that loan amount is in millions of 2010 dollars, list the county-level controls, and distinguish OLS from IV and scaled from unscaled specifications. Summary-statistics tables come first, then OLS, IV, alternative-IV and heterogeneity tables.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["explicit caveats paragraph on non-random allocation and external validity", "'arguably exogenous' framing of instruments", "'if these estimates are correct' conditional", "lower-bound interpretation", "'preliminary investigation' and 'rough estimates' qualifiers", "'suggesting' when linking results to credit constraints"]}, "claim_evidence": [{"claim": "Each $1 million of SBA loans raises recipient employment by about 3–3.5 jobs over three post-loan years", "evidence_type": "table", "evidence_ref": "Tables IV, VI, VII"}, {"claim": "Effects are larger for younger and larger firms", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "No systematic difference between 7(a) and 504 programs or over the business cycle", "evidence_type": "table", "evidence_ref": "Tables IX, X"}, {"claim": "Effects are smaller where local conventional credit is more plentiful", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Displacement of competing firms is around 10% of the total effect in nontradables", "evidence_type": "table", "evidence_ref": "Table XII"}, {"claim": "Results are similar when exited firms are assigned zero employment", "evidence_type": "table", "evidence_ref": "Table XIII"}, {"claim": "Taxpayer cost per job is roughly $21,000–$25,000, low relative to other government programs", "evidence_type": "prior_literature", "evidence_ref": "Section V.B; Neumark (2013) comparison in conclusion"}, {"claim": "PLP lenders account for a large share of SBA loans, especially larger loans", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Only 3.2% of SBA loans occur in counties with no PLP branch in the prior year", "evidence_type": "descriptive_statistic", "evidence_ref": "Introduction"}], "structural_summary": "The paper is a policy-evaluation empirical study. The introduction opens with a historical epigraph and the size of the SBA program, then notes that the employment effect is theoretically ambiguous. It previews linked administrative-census data and the matching method, and motivates instruments from lender institutional features while stating the identifying assumptions. Literature positioning follows, then an explicit caveats paragraph on external validity. The body describes the program institutions and the data, presents matching/OLS results, then the IV approach, and finally heterogeneity, displacement, exit and cost-per-job extensions. The conclusion restates the estimates, bounds them as a lower bound, and draws a cost-effectiveness policy implication.", "writing_profile_keywords": ["program evaluation", "linked administrative data", "matching estimator", "instrumental variables", "explicit caveats", "cost per job", "heterogeneous effects", "policy relevance"], "confidence": 0.8, "notes": "The introduction-paragraph indices are fragmented: paragraphs 0–4 are pieces of the epigraph, and paragraph 5 contains the opening motivation, theoretical ambiguity and empirical difficulty. Section IV is labeled 'data' by the role rule but is actually IV methods and results. Inference reporting conventions (stars or SEs) are not visible in the excerpts. Robustness checks (alternative IVs, exit zeros, controls) appear within the results and extensions sections, and an internet appendix exists."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.52, "usage": {"input": 9001, "output": 2564, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088078, "request_id": "req_011CggpZDKXBkaGYQ6qHe2x2", "cached": false, "ts": "2026-11-03T22:57:01+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/46c8de1094f213281ef592eaa709b90a556509f4530eaba3217f210c07b8490d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"value premium and cash-flow duration\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Differences in future cash-flow (dividend/earnings) growth rates between low and high book-to-market portfolios\",\"strategy\":\"Descriptive portfolio sorts (buy-and-hold and rebalanced) tracking average real dividends and earnings, with adjustments for survivorship and look-back bias; Fama-MacBeth firm-level regressions; valuation-model and accounting decompositions\",\"key_assumptions\":[\"Averages across formation years are representative of expected growth\",\"Reinvesting delisting proceeds properly corrects survivorship bias\",\"Terminal growth perpetuity assumption beyond year 20\"],\"threats_addressed\":[\"survivorship bias\",\"look-back bias\",\"rebalancing effect\",\"small-stock driven results (value vs equal weighting)\",\"negative earnings\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"other\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"resolution_of_puzzle\"===\"x\"?\"other\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"alternative_sample_period\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Long table notes repeat the portfolio construction (June sorts, NYSE breakpoints, value-weighted quintiles, $100 initial investment, July–June dividend aggregation, CPI deflation). Regression tables report Fama-MacBeth estimates with Newey-West t-statistics in parentheses and state winsorization.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'do not have substantially higher' qualifier\",\"'suggest' / 'unlikely to explain'\",\"'help explain'\",\"'in some settings ... appear to'\"]},\"claim_evidence\":[{\"claim\":\"In buy-and-hold portfolios, growth stocks do not have substantially faster dividend growth than value stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Figure 1, Table IV\"},{\"claim\":\"In annually rebalanced portfolios, value stocks' dividends grow faster than growth stocks'\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table VI\"},{\"claim\":\"Back-of-the-envelope calculations from Fama-French ROE imply initially lower earnings growth for growth stocks, explained by efficiency growth\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"After correcting for survivorship bias, book-to-market no longer predicts lower future firm dividend growth\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII–IX\"},{\"claim\":\"Rebalancing raises value stocks' growth rates and lowers growth stocks' relative to buy-and-hold\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"Valuation models that account for higher value returns imply similar growth for growth and value stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"}],\"structural_summary\":\"The paper opens by challenging a widely held belief: that growth stocks have faster cash-flow growth. It ties this belief to duration-based explanations of the value premium. Portfolio-level dividend evidence is presented first, for buy-and-hold and rebalanced portfolios and for long-run growth. A dedicated section then explains why the belief persists, attributing it to efficiency growth, survivorship bias and look-back bias. An analytical section derives the rebalancing effect, followed by additional tests. The conclusion is a brief restatement, and the appendices give delta-method standard errors and a survivorship-adjustment procedure.\",\"writing_profile_keywords\":[\"conventional-wisdom challenge\",\"single-author first-person\",\"descriptive portfolio evidence\",\"bias reconciliation\",\"duration-based value premium\",\"question-form title\"],\"confidence\":0.75,\"notes\":\"The intro paragraph containing Gordon's formula is fragmented across two entries (paragraphs 1–2) by parsing. Significance language is not clearly visible apart from the Newey-West t-statistics in the regression tables. One introduction move entry is malformed: paragraph 3 should read 'mechanism_or_theory_sketch' (it previews how efficiency growth, survivorship and look-back biases reconcile the results).\"}", "json": null, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.59, "usage": {"input": 7763, "output": 2158, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075006, "request_id": "req_011CggpjJVqsfs4uEEvQbLBR", "cached": false, "ts": "2026-11-03T22:59:16+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/471a4f9935dd471327c89504592ec0b66c247ab73d335f7db72174cf8bf2ce4f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"market_microstructure (experimental asset markets)\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Association between trader cognitive skills (Raven fluid intelligence, CRT cognitive reflection, ToM) and per-period trading earnings, and the interaction of CRT with ToM, in a controlled experimental asset market\",\"strategy\":\"Laboratory asset markets in which private signals are randomized across traders so that skill is not confounded with information. Earnings are regressed on skill scores in linear panel regressions with subject random effects, controlling for individual and market characteristics.\",\"key_assumptions\":[\"Random allocation of private signals makes high-skill traders no better informed on average\",\"Raven, CRT and ToM scores validly measure the targeted skills\",\"The individual controls (financial literacy, personality, risk attitudes) capture relevant confounding traits\"],\"threats_addressed\":[\"Confounding of skill with insider/private information, addressed by randomizing signals\",\"Omitted individual characteristics, addressed by a large set of individual-level controls\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":4},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes are descriptive. They state the model (linear panel regressions with subject random effects) and list the controls, and they open with a one-sentence statement of the result. Robust standard errors clustered at the session level are reported in parentheses. A design summary table describes the treatments.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\",\"figures\":null},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"conjecture framing (we expect, we conjecture)\",\"scoping results to 'our market environment'\",\"noting that other market settings may differ\",\"citing an independent replication as support\"]},\"claim_evidence\":[{\"claim\":\"Fluid intelligence, cognitive reflection and ToM scores are positively related to trader earnings\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 1\"},{\"claim\":\"Cognitive reflection and ToM skills matter more when a trader's private hint is imprecise\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Skills matter more when skill heterogeneity among traders in a market is high\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"High-skill traders submit a larger share of potentially profitable orders\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"ToM is key to assessing the informational content of trades\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Bruguier, Quartz, and Bossaerts (2010)\"},{\"claim\":\"The main findings are confirmed in a later independent study\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Hefti, Heinke, and Schneider (2016)\"}],\"structural_summary\":\"The introduction motivates the question by challenging the rationality assumption with evidence of cognitive heterogeneity. It notes that prior work studies trader skills in isolation and proposes a unified laboratory market with randomized private signals. It then sketches the mechanisms for each skill and for the CRT–ToM complementarity. A separate literature section is organized by skill, followed by the design section. A dedicated conjectures section derives five hypotheses, which the results section tests one by one using random-effects panel regressions. The conclusion restates the findings, bounds their scope to information-driven markets, and cites an independent replication.\",\"writing_profile_keywords\":[\"laboratory experiment\",\"conjecture-driven\",\"cognitive skills\",\"theory of mind\",\"random effects panel\",\"behavioral finance\"],\"confidence\":0.8,\"notes\":\"Robustness checks are not visible in the excerpts; an Internet Appendix exists, but its contents are unknown. The conjectures section was classified as theory, although it is informal reasoning rather than a model. Significance reporting conventions (stars, t-statistics) are not visible.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "market_microstructure (experimental asset markets)", "primary_method": "survey_or_experiment", "identification": {"estimand": "Association between trader cognitive skills (Raven fluid intelligence, CRT cognitive reflection, ToM) and per-period trading earnings, and the interaction of CRT with ToM, in a controlled experimental asset market", "strategy": "Laboratory asset markets in which private signals are randomized across traders so that skill is not confounded with information. Earnings are regressed on skill scores in linear panel regressions with subject random effects, controlling for individual and market characteristics.", "key_assumptions": ["Random allocation of private signals makes high-skill traders no better informed on average", "Raven, CRT and ToM scores validly measure the targeted skills", "The individual controls (financial literacy, personality, risk attitudes) capture relevant confounding traits"], "threats_addressed": ["Confounding of skill with insider/private information, addressed by randomizing signals", "Omitted individual characteristics, addressed by a large set of individual-level controls"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 4}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 2, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 2, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes are descriptive. They state the model (linear panel regressions with subject random effects) and list the controls, and they open with a one-sentence statement of the result. Robust standard errors clustered at the session level are reported in parentheses. A design summary table describes the treatments.", "units_stated_in_notes": true, "standard_errors_description": "clustered", "figures": null}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["conjecture framing (we expect, we conjecture)", "scoping results to 'our market environment'", "noting that other market settings may differ", "citing an independent replication as support"]}, "claim_evidence": [{"claim": "Fluid intelligence, cognitive reflection and ToM scores are positively related to trader earnings", "evidence_type": "table", "evidence_ref": "Table II; Figure 1"}, {"claim": "Cognitive reflection and ToM skills matter more when a trader's private hint is imprecise", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Skills matter more when skill heterogeneity among traders in a market is high", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "High-skill traders submit a larger share of potentially profitable orders", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "ToM is key to assessing the informational content of trades", "evidence_type": "prior_literature", "evidence_ref": "Bruguier, Quartz, and Bossaerts (2010)"}, {"claim": "The main findings are confirmed in a later independent study", "evidence_type": "prior_literature", "evidence_ref": "Hefti, Heinke, and Schneider (2016)"}], "structural_summary": "The introduction motivates the question by challenging the rationality assumption with evidence of cognitive heterogeneity. It notes that prior work studies trader skills in isolation and proposes a unified laboratory market with randomized private signals. It then sketches the mechanisms for each skill and for the CRT–ToM complementarity. A separate literature section is organized by skill, followed by the design section. A dedicated conjectures section derives five hypotheses, which the results section tests one by one using random-effects panel regressions. The conclusion restates the findings, bounds their scope to information-driven markets, and cites an independent replication.", "writing_profile_keywords": ["laboratory experiment", "conjecture-driven", "cognitive skills", "theory of mind", "random effects panel", "behavioral finance"], "confidence": 0.8, "notes": "Robustness checks are not visible in the excerpts; an Internet Appendix exists, but its contents are unknown. The conjectures section was classified as theory, although it is informal reasoning rather than a model. Significance reporting conventions (stars, t-statistics) are not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.37, "usage": {"input": 5071, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061318, "request_id": "req_011Cggpuj5yY8TLUtu2MwPiY", "cached": false, "ts": "2026-11-03T23:01:36+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/48aa94471af6f752f948cdf51239c83b1aaccc0db9e317c80b9bf6fba8314e20.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance (production-based asset pricing, production networks)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Risk prices of innovations in network sparsity and concentration in the cross-section of stock returns; return spreads of network-beta-sorted portfolios\",\"strategy\":\"General equilibrium multisector model derives the network factors in closed form; factors are computed from input-output/Compustat data; stocks are sorted on factor betas; Fama-MacBeth estimation of risk prices; time-series regressions of dividend growth on the factors\",\"key_assumptions\":[\"Cobb-Douglas production with an exogenous, stochastically evolving network\",\"Perfect competition and a representative household\",\"Concentration and sparsity are sufficient statistics for aggregate risk\",\"Factor betas estimated from returns proxy for true exposures\"],\"threats_addressed\":[\"Spreads explained by CAPM or Fama-French three-factor model\",\"Sensitivity to the choice of test assets (seven sets)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":4},\"magnitude\":{\"present\":true,\"sentence_index\":4},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"roadmap\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\",\"new_data_or_measure\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state what is reported (statistics, excess returns, alphas, risk prices, coefficients with t-statistics, R2) and describe the panel and column structure. Market capitalization units (inflation-adjusted billions of 2013 dollars) are given, and test-asset sets are enumerated by column.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'consistent with the model predictions'\",\"'supportive evidence'\",\"'can be interpreted as'\",\"'to the best of my knowledge'\"]},\"claim_evidence\":[{\"claim\":\"Sparsity-beta sorted portfolios earn a high-minus-low spread of about 4.6% per year\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Concentration-beta sorted portfolios earn a spread of about -3.2% per year, not explained by CAPM or FF3\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Sparsity and concentration are priced across multiple sets of test assets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Sparsity innovations relate to higher and concentration innovations to lower aggregate dividend growth\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Concentration and sparsity are sufficient statistics for aggregate risk and determine consumption dynamics\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.C-D\"},{\"claim\":\"Distinct networks can share the same factor values, illustrating how the factors differ\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-4\"}],\"structural_summary\":\"The paper opens with a concrete motivation (technological change reshaping input-output links) and states a first-mover contribution. It then sketches the economic mechanisms behind each factor and previews the empirical tests and magnitudes, followed by a related-literature subsection inside the introduction. A long closed-form theory section derives the two network factors and illustrates them with stylized network figures. A compact evidence section tests the signed risk-price predictions through beta-sorted portfolios, Fama-MacBeth estimation and macro regressions. The conclusion restates the findings and lists extensions: endogenous network formation and credit frictions.\",\"writing_profile_keywords\":[\"production-based asset pricing\",\"theory-first then test\",\"closed-form model\",\"first-person singular\",\"signed prediction testing\",\"portfolio sorts\",\"Fama-MacBeth\"],\"confidence\":0.8,\"notes\":\"Single-author model-plus-test paper. The theory section dominates by equation count (23 of 24). Robustness appears as a subsection within the evidence section; its details are not visible. Standard error type is not stated in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance (production-based asset pricing, production networks)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Risk prices of innovations in network sparsity and concentration in the cross-section of stock returns; return spreads of network-beta-sorted portfolios", "strategy": "General equilibrium multisector model derives the network factors in closed form; factors are computed from input-output/Compustat data; stocks are sorted on factor betas; Fama-MacBeth estimation of risk prices; time-series regressions of dividend growth on the factors", "key_assumptions": ["Cobb-Douglas production with an exogenous, stochastically evolving network", "Perfect competition and a representative household", "Concentration and sparsity are sufficient statistics for aggregate risk", "Factor betas estimated from returns proxy for true exposures"], "threats_addressed": ["Spreads explained by CAPM or Fama-French three-factor model", "Sensitivity to the choice of test assets (seven sets)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 4}, "magnitude": {"present": true, "sentence_index": 4}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "roadmap"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model", "new_data_or_measure", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "other"]}, "tables": {"convention_notes": "Notes state what is reported (statistics, excess returns, alphas, risk prices, coefficients with t-statistics, R2) and describe the panel and column structure. Market capitalization units (inflation-adjusted billions of 2013 dollars) are given, and test-asset sets are enumerated by column.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'consistent with the model predictions'", "'supportive evidence'", "'can be interpreted as'", "'to the best of my knowledge'"]}, "claim_evidence": [{"claim": "Sparsity-beta sorted portfolios earn a high-minus-low spread of about 4.6% per year", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Concentration-beta sorted portfolios earn a spread of about -3.2% per year, not explained by CAPM or FF3", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Sparsity and concentration are priced across multiple sets of test assets", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Sparsity innovations relate to higher and concentration innovations to lower aggregate dividend growth", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Concentration and sparsity are sufficient statistics for aggregate risk and determine consumption dynamics", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.C-D"}, {"claim": "Distinct networks can share the same factor values, illustrating how the factors differ", "evidence_type": "figure", "evidence_ref": "Figures 2-4"}], "structural_summary": "The paper opens with a concrete motivation (technological change reshaping input-output links) and states a first-mover contribution. It then sketches the economic mechanisms behind each factor and previews the empirical tests and magnitudes, followed by a related-literature subsection inside the introduction. A long closed-form theory section derives the two network factors and illustrates them with stylized network figures. A compact evidence section tests the signed risk-price predictions through beta-sorted portfolios, Fama-MacBeth estimation and macro regressions. The conclusion restates the findings and lists extensions: endogenous network formation and credit frictions.", "writing_profile_keywords": ["production-based asset pricing", "theory-first then test", "closed-form model", "first-person singular", "signed prediction testing", "portfolio sorts", "Fama-MacBeth"], "confidence": 0.8, "notes": "Single-author model-plus-test paper. The theory section dominates by equation count (23 of 24). Robustness appears as a subsection within the evidence section; its details are not visible. Standard error type is not stated in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.55, "usage": {"input": 6159, "output": 2065, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06673, "request_id": "req_011Cggq1dsNFCpXhUZj8U3gC", "cached": false, "ts": "2026-11-03T23:02:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4d59b6136497e94ba7bb0ee520742fd86af2f0712b475227fdc47d8f2e27ffb5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure / macro_finance (liquidity dynamics, adverse selection, social learning)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theory paper)\",\"strategy\":\"not applicable; analytical characterization of equilibrium plus numerical simulations of belief dynamics\",\"key_assumptions\":[\"Akerlof-style lemons market with sellers informed and buyers relying on public signals\",\"the joint distribution of signals and qualities (parameter mu) is unknown and learned from public history\",\"trading generates observations to learn from (Assumption 2), microfounded in Appendix B\",\"short-lived risk-neutral agents\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"appendix\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"No tables. Figures are numerical illustrations whose captions list the exact parameter values used in each example or simulation.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"openly calling the learning-from-trade link an assumption\",\"mentioning microfoundations for that assumption\",\"'can arise' / 'may be slow' modal phrasing\",\"presenting the model as a minor modification of a standard setup\"]},\"claim_evidence\":[{\"claim\":\"An illiquidity trap can arise: excessive asymmetric information halts trade, which stops learning and perpetuates illiquidity.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.B; Appendix A proofs; Figure 2\"},{\"claim\":\"After unlikely observations push the economy into the illiquid region, liquidity can be slow to recover.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5 impulse response; Figure 3\"},{\"claim\":\"When structure mu changes over time, liquidity is higher in stable periods and remains fragile even in the long run.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C; Figure 4\"},{\"claim\":\"Positive information flow from trading (Assumption 2) can be microfounded in several ways, some of which produce a trap.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Appendix B\"},{\"claim\":\"More informativeness need not raise seller value when a regularity condition fails.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"The paper is a short theory article.\\n\\n- **Introduction:** It states the aim, defines liquidity as the share of potential gains from trade that are realized, and sketches a lemons model in which buyers learn how to interpret public signals from past history. It then previews the main results and the extension in which the economy's structure changes. A contribution paragraph frames trading-generated expertise as the novelty. Three literature-positioning paragraphs follow, then a roadmap.\\n- **Body:** Model setup, then static equilibrium characterization, then a long dynamics section covering short-run and long-run outcomes and a changing mu, illustrated with numerical simulations.\\n- **Conclusion:** One paragraph restating the implications.\\n- **Appendices:** Proofs, plus microfoundations for the key assumption.\",\"writing_profile_keywords\":[\"pure theory\",\"Akerlof lemons\",\"social learning\",\"illiquidity trap\",\"propositions and proofs\",\"numerical illustration\",\"short conclusion\",\"literature strands enumerated\"],\"confidence\":0.85,\"notes\":\"No empirical work, so empirical fields are set to not applicable or none. Figure 5 reports simulation quantiles, which counts as partial uncertainty display; the other figures are deterministic examples. The robustness category 'other' refers to the alternative microfoundations in Appendix B. The section headed 'Dynamics' (idx 3) has rule-based role 'other' but contains theoretical results, so it is assigned 'theory'.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure / macro_finance (liquidity dynamics, adverse selection, social learning)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theory paper)", "strategy": "not applicable; analytical characterization of equilibrium plus numerical simulations of belief dynamics", "key_assumptions": ["Akerlof-style lemons market with sellers informed and buyers relying on public signals", "the joint distribution of signals and qualities (parameter mu) is unknown and learned from public history", "trading generates observations to learn from (Assumption 2), microfounded in Appendix B", "short-lived risk-neutral agents"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "appendix", "types": ["other"]}, "tables": {"convention_notes": "No tables. Figures are numerical illustrations whose captions list the exact parameter values used in each example or simulation.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["openly calling the learning-from-trade link an assumption", "mentioning microfoundations for that assumption", "'can arise' / 'may be slow' modal phrasing", "presenting the model as a minor modification of a standard setup"]}, "claim_evidence": [{"claim": "An illiquidity trap can arise: excessive asymmetric information halts trade, which stops learning and perpetuates illiquidity.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.B; Appendix A proofs; Figure 2"}, {"claim": "After unlikely observations push the economy into the illiquid region, liquidity can be slow to recover.", "evidence_type": "figure", "evidence_ref": "Figure 5 impulse response; Figure 3"}, {"claim": "When structure mu changes over time, liquidity is higher in stable periods and remains fragile even in the long run.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C; Figure 4"}, {"claim": "Positive information flow from trading (Assumption 2) can be microfounded in several ways, some of which produce a trap.", "evidence_type": "equation_or_proof", "evidence_ref": "Appendix B"}, {"claim": "More informativeness need not raise seller value when a regularity condition fails.", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "The paper is a short theory article.\n\n- **Introduction:** It states the aim, defines liquidity as the share of potential gains from trade that are realized, and sketches a lemons model in which buyers learn how to interpret public signals from past history. It then previews the main results and the extension in which the economy's structure changes. A contribution paragraph frames trading-generated expertise as the novelty. Three literature-positioning paragraphs follow, then a roadmap.\n- **Body:** Model setup, then static equilibrium characterization, then a long dynamics section covering short-run and long-run outcomes and a changing mu, illustrated with numerical simulations.\n- **Conclusion:** One paragraph restating the implications.\n- **Appendices:** Proofs, plus microfoundations for the key assumption.", "writing_profile_keywords": ["pure theory", "Akerlof lemons", "social learning", "illiquidity trap", "propositions and proofs", "numerical illustration", "short conclusion", "literature strands enumerated"], "confidence": 0.85, "notes": "No empirical work, so empirical fields are set to not applicable or none. Figure 5 reports simulation quantiles, which counts as partial uncertainty display; the other figures are deterministic examples. The robustness category 'other' refers to the alternative microfoundations in Appendix B. The section headed 'Dynamics' (idx 3) has rule-based role 'other' but contains theoretical results, so it is assigned 'theory'."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.54, "usage": {"input": 4680, "output": 2057, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.060654, "request_id": "req_011CggpqvzEnvVgrfFzbnuD9", "cached": false, "ts": "2026-11-03T23:00:45+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4d6992a34ecf27cb2dd16f7d4bd5776a49f56ce78fe67ed5b805c16b0ec7325d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"corporate_finance (theory of the firm, internal capital allocation in mutual fund firms)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Change in manager value added (excess return over benchmark times AUM) following a firm-initiated capital reallocation (promotion/demotion), bounded from below\",\"strategy\":\"Panel regressions of manager value added on firm capital reallocation dummies, with counterfactual adjustments (e.g., excluding poorly performing funds, filling in zeros for fired managers) to produce a lower bound; external-hire comparisons and probit predictability tests support the information channel\",\"key_assumptions\":[\"Vanguard index fund benchmarks correctly measure the investor's alternative investment opportunity\",\"Counterfactual adjustments yield a conservative lower bound\",\"Null hypothesis that reallocation decisions add no value is the appropriate test\"],\"threats_addressed\":[\"Endogeneity of manager moves in firm/manager fixed-effects attribution models\",\"Survivorship from managers dropping out after demotion\",\"Predictability of reallocations from public information\",\"Behavioral explanations such as investor inattention (acknowledged, not ruled out)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"other\"},{\"paragraph_index\":3,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"data\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"robustness\"},{\"idx\":9,\"role\":\"conclusion\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"sensitivity_to_parameters\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes name the dependent variable and point to its defining equation number; units are given for value added ($ millions/month). Columns are described as alternative regression specifications. Later tables state that they repeat earlier specifications with a modified sample or variable.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'at least' / lower-bound framing\",\"'consistent with' the hypothesis\",\"explicit acknowledgment that behavioral explanations cannot be ruled out\",\"limiting generalization to the mutual fund industry\"],\"claim_evidence\":[],\"structural_summary\":\"\",\"writing_profile_keywords\":[],\"confidence\":0,\"notes\":\"\"},\"claim_evidence\":[{\"claim\":\"Firm reallocation of capital to managers raises manager value added by at least about $474K per month\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III\"},{\"claim\":\"The firm accounts for at least 30% of total manager value added under conservative counterfactuals\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4, Figure 5\"},{\"claim\":\"Investors reward firms after reallocation decisions with additional capital\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Table V\"},{\"claim\":\"External hires with AUM changes do not produce detectable value-added changes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Public information such as past performance has weak power to predict firm reallocation decisions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Table VIII\"},{\"claim\":\"Under decreasing returns to scale, value added rather than gross alpha measures skill\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Section II model\"},{\"claim\":\"Firm-driven AUM changes would take investors years to replicate through flows\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"}],\"structural_summary\":\"The paper opens with an assertive contribution statement and the headline magnitude. It then sketches an information-asymmetry mechanism with a numbered list of supporting findings, adds a paragraph acknowledging a behavioral alternative it cannot rule out, and closes the introduction with a roadmap. A short theory section frames a Coasian benchmark against asymmetric information. Data definitions argue for value added over gross alpha. The main results test a null of no value added and bound the firm's contribution from below under counterfactuals. Follow-up sections show investor responses and the information source of firm skill, and a section reports gross alpha results at referees' request. The conclusion ties the findings to the theory of the firm and limits generalization to other industries.\",\"writing_profile_keywords\":[\"theory of the firm\",\"lower bound\",\"value added\",\"information asymmetry\",\"mutual fund firms\",\"counterfactuals\",\"assertive opening\"],\"confidence\":0.8,\"notes\":\"No statistical significance conventions or standard error descriptions are visible in the excerpts. Section VIII (gross alpha) is treated as robustness. The causal reading relies on the bounding approach rather than a quasi-experiment.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "corporate_finance (theory of the firm, internal capital allocation in mutual fund firms)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Change in manager value added (excess return over benchmark times AUM) following a firm-initiated capital reallocation (promotion/demotion), bounded from below", "strategy": "Panel regressions of manager value added on firm capital reallocation dummies, with counterfactual adjustments (e.g., excluding poorly performing funds, filling in zeros for fired managers) to produce a lower bound; external-hire comparisons and probit predictability tests support the information channel", "key_assumptions": ["Vanguard index fund benchmarks correctly measure the investor's alternative investment opportunity", "Counterfactual adjustments yield a conservative lower bound", "Null hypothesis that reallocation decisions add no value is the appropriate test"], "threats_addressed": ["Endogeneity of manager moves in firm/manager fixed-effects attribution models", "Survivorship from managers dropping out after demotion", "Predictability of reallocations from public information", "Behavioral explanations such as investor inattention (acknowledged, not ruled out)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "other"}, {"paragraph_index": 3, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "data"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "robustness"}, {"idx": 9, "role": "conclusion"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "sensitivity_to_parameters", "alternative_specification"]}, "tables": {"convention_notes": "Table notes name the dependent variable and point to its defining equation number; units are given for value added ($ millions/month). Columns are described as alternative regression specifications. Later tables state that they repeat earlier specifications with a modified sample or variable.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'at least' / lower-bound framing", "'consistent with' the hypothesis", "explicit acknowledgment that behavioral explanations cannot be ruled out", "limiting generalization to the mutual fund industry"], "claim_evidence": [], "structural_summary": "", "writing_profile_keywords": [], "confidence": 0, "notes": ""}, "claim_evidence": [{"claim": "Firm reallocation of capital to managers raises manager value added by at least about $474K per month", "evidence_type": "table", "evidence_ref": "Table II, Table III"}, {"claim": "The firm accounts for at least 30% of total manager value added under conservative counterfactuals", "evidence_type": "figure", "evidence_ref": "Figure 4, Figure 5"}, {"claim": "Investors reward firms after reallocation decisions with additional capital", "evidence_type": "table", "evidence_ref": "Table IV, Table V"}, {"claim": "External hires with AUM changes do not produce detectable value-added changes", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Public information such as past performance has weak power to predict firm reallocation decisions", "evidence_type": "table", "evidence_ref": "Table VII, Table VIII"}, {"claim": "Under decreasing returns to scale, value added rather than gross alpha measures skill", "evidence_type": "figure", "evidence_ref": "Figure 1; Section II model"}, {"claim": "Firm-driven AUM changes would take investors years to replicate through flows", "evidence_type": "figure", "evidence_ref": "Figure 6"}], "structural_summary": "The paper opens with an assertive contribution statement and the headline magnitude. It then sketches an information-asymmetry mechanism with a numbered list of supporting findings, adds a paragraph acknowledging a behavioral alternative it cannot rule out, and closes the introduction with a roadmap. A short theory section frames a Coasian benchmark against asymmetric information. Data definitions argue for value added over gross alpha. The main results test a null of no value added and bound the firm's contribution from below under counterfactuals. Follow-up sections show investor responses and the information source of firm skill, and a section reports gross alpha results at referees' request. The conclusion ties the findings to the theory of the firm and limits generalization to other industries.", "writing_profile_keywords": ["theory of the firm", "lower bound", "value added", "information asymmetry", "mutual fund firms", "counterfactuals", "assertive opening"], "confidence": 0.8, "notes": "No statistical significance conventions or standard error descriptions are visible in the excerpts. Section VIII (gross alpha) is treated as robustness. The causal reading relies on the bounding approach rather than a quasi-experiment."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.35, "usage": {"input": 6854, "output": 2242, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07305, "request_id": "req_011CggpkgFa27VHncb9jkV9D", "cached": false, "ts": "2026-11-03T22:59:34+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4d9a97cf0c9f923308ec4265b5b9b472b9a89d20971eecb99fefd7af347a3fcc.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"corporate_finance (managerial compensation / agency in asset management)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-sectional associations between the likelihood of compensation contract features (performance pay, AUM pay, profit pay, deferred pay, evaluation period) and proxies for agency conflicts, alternative monitoring, and manager characteristics; also associations of contract features with future performance and fees.\",\"strategy\":\"Logistic and OLS regressions on hand-collected SAI disclosure data with advisor-, fund-, and manager-level controls. Supplemented by bootstrap simulations of randomly assigned contracts and by change regressions around switches of advisory firm.\",\"key_assumptions\":[\"Proxies (clientele dispersion, financial affiliations, ownership) capture the intensity of agency conflicts\",\"Disclosed SAI compensation descriptions accurately reflect actual contracts\",\"Contract choice reflects an optimal contracting equilibrium\"],\"threats_addressed\":[\"Spurious correlation from contract clustering within families, addressed with randomly assigned contract simulations\",\"Time-invariant unobservables, addressed with changes in contracts around advisory firm changes\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"tension_or_puzzle\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"placebo\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes describe the regression type (logit with coefficients and marginal effects at means; OLS) and the sample. Variables are deferred to the Appendix, whose definitions include units and sources. Summary tables report mean, SD and percentiles. The bootstrap table reports simulated t-statistic distributions.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'broadly consistent with' an optimal contracting equilibrium\",\"'as proxied by'\",\"'arguably'\",\"'we speculate'\",\"'little evidence'\",\"'mixed evidence'\"]},\"claim_evidence\":[{\"claim\":\"Most portfolio managers receive explicit performance-linked bonuses, often large relative to base salary.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Performance-based pay and deferred pay are more likely when agency conflicts are more intense.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Investor sophistication, flow discipline and outsourcing dismissal threat substitute for explicit incentives; managerial ownership does not.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Determinant results are not an artifact of chance, per randomly assigned contract simulations.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Contract changes after advisory firm switches track changes in determinants.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Compensation arrangements are not associated with differences in future fund performance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Performance pay goes with higher advisory fees and lower distribution fees, leaving total fees unchanged.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"}],\"structural_summary\":\"The paper motivates the topic with the economic importance of mutual funds and the lack of data on manager-level incentives. It then introduces a new data source created by an SEC disclosure rule and describes pay patterns. Three hypotheses (agency intensity, substitute monitoring, manager characteristics) are previewed with findings. The body runs: institutional background; data and construction of compensation variables; a large determinants section with hypothesis development, logit results and robustness (simulation, advisory firm changes); evaluation-period determinants; and links to performance and fees. The conclusion restates findings within an optimal-contracting interpretation.\",\"writing_profile_keywords\":[\"hand-collected data\",\"first systematic evidence\",\"hypothesis-testing structure\",\"optimal contracting interpretation\",\"agency theory\",\"descriptive-then-determinants\",\"null result on performance\"],\"confidence\":0.8,\"notes\":\"No roadmap paragraph is visible in the excerpts. The way statistical significance is reported is not visible in the table notes. Section I is institutional background rather than a literature review, so it is labeled 'other'. Robustness is placed inside Section III.D.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "corporate_finance (managerial compensation / agency in asset management)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-sectional associations between the likelihood of compensation contract features (performance pay, AUM pay, profit pay, deferred pay, evaluation period) and proxies for agency conflicts, alternative monitoring, and manager characteristics; also associations of contract features with future performance and fees.", "strategy": "Logistic and OLS regressions on hand-collected SAI disclosure data with advisor-, fund-, and manager-level controls. Supplemented by bootstrap simulations of randomly assigned contracts and by change regressions around switches of advisory firm.", "key_assumptions": ["Proxies (clientele dispersion, financial affiliations, ownership) capture the intensity of agency conflicts", "Disclosed SAI compensation descriptions accurately reflect actual contracts", "Contract choice reflects an optimal contracting equilibrium"], "threats_addressed": ["Spurious correlation from contract clustering within families, addressed with randomly assigned contract simulations", "Time-invariant unobservables, addressed with changes in contracts around advisory firm changes"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "other"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "tension_or_puzzle"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["placebo", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes describe the regression type (logit with coefficients and marginal effects at means; OLS) and the sample. Variables are deferred to the Appendix, whose definitions include units and sources. Summary tables report mean, SD and percentiles. The bootstrap table reports simulated t-statistic distributions.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'broadly consistent with' an optimal contracting equilibrium", "'as proxied by'", "'arguably'", "'we speculate'", "'little evidence'", "'mixed evidence'"]}, "claim_evidence": [{"claim": "Most portfolio managers receive explicit performance-linked bonuses, often large relative to base salary.", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Performance-based pay and deferred pay are more likely when agency conflicts are more intense.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Investor sophistication, flow discipline and outsourcing dismissal threat substitute for explicit incentives; managerial ownership does not.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Determinant results are not an artifact of chance, per randomly assigned contract simulations.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Contract changes after advisory firm switches track changes in determinants.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Compensation arrangements are not associated with differences in future fund performance.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Performance pay goes with higher advisory fees and lower distribution fees, leaving total fees unchanged.", "evidence_type": "table", "evidence_ref": "Table VIII"}], "structural_summary": "The paper motivates the topic with the economic importance of mutual funds and the lack of data on manager-level incentives. It then introduces a new data source created by an SEC disclosure rule and describes pay patterns. Three hypotheses (agency intensity, substitute monitoring, manager characteristics) are previewed with findings. The body runs: institutional background; data and construction of compensation variables; a large determinants section with hypothesis development, logit results and robustness (simulation, advisory firm changes); evaluation-period determinants; and links to performance and fees. The conclusion restates findings within an optimal-contracting interpretation.", "writing_profile_keywords": ["hand-collected data", "first systematic evidence", "hypothesis-testing structure", "optimal contracting interpretation", "agency theory", "descriptive-then-determinants", "null result on performance"], "confidence": 0.8, "notes": "No roadmap paragraph is visible in the excerpts. The way statistical significance is reported is not visible in the table notes. Section I is institutional background rather than a literature review, so it is labeled 'other'. Robustness is placed inside Section III.D."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.72, "usage": {"input": 7914, "output": 2258, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07761, "request_id": "req_011CggqFyJK4gmDMYqXZQF5D", "cached": false, "ts": "2026-11-03T23:06:11+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4e650c20bf4f14df88172d81773935f49bc255f2cfb1b93c66d0938e740ba439.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in average delta-hedged and unhedged option returns between nontrading and trading periods, plus the gap between realized and implied variance during market closure.\",\"strategy\":\"Calendar-based comparison of portfolio returns (nontrading minus trading means) with Newey-West t-statistics. Risk-adjusted performance measures (alphas, Sharpe ratios, certainty equivalents, Kadan-Liu indexes) and Fama-MacBeth cross-sectional regressions are used to rule out risk-based explanations.\",\"key_assumptions\":[\"Risk adjustment measures adequately capture option return risk, including higher moments\",\"Bid-ask midpoint quotes reflect tradable prices\",\"Nontrading timing is exogenous to option-specific information\"],\"threats_addressed\":[\"Differential risk across trading and nontrading periods\",\"Increased aversion to unbounded downside risk over weekends\",\"Portfolio construction and weighting choices\",\"Subsample and year-by-year stability\",\"Transaction costs\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":9,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":7,\"contribution_types\":[\"new_evidence\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Notes define each portfolio construction, report Newey-West t-statistics in parentheses, and present differences as nontrading mean minus trading mean. Volume is given in 100-share contracts and spreads in percent. Fama-MacBeth regressions are used for the cross-sectional analysis.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"yes\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"our evidence suggests\",\"we believe\",\"likely\",\"may\",\"possible that\"]},\"claim_evidence\":[{\"claim\":\"Delta-hedged option returns are strongly negative over nontrading periods and near zero on trading days.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III\"},{\"claim\":\"The effect holds across puts and calls, maturities, moneyness, weighting schemes and samples.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table IV\"},{\"claim\":\"The nontrading effect is negative in every sample year.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Implied volatilities embed sizable variance during market closure, while realized variance over closure is small.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Risk-adjusted returns from writing options remain much higher over nontrading periods despite somewhat higher risk.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Cross-sectional risk patterns do not explain nontrading returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The nontrading effect is related to volume and open interest patterns.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"}],\"structural_summary\":\"The paper documents an anomaly and then tests explanations for it. The introduction motivates with institutional reasons why risk and return may differ when markets are closed. It then previews pervasive nontrading return effects and a gap between implied and realized variance. Three explanations are laid out (risk, risk aversion, mispricing), and the evidence is previewed as favoring mispricing. The introduction closes with implications for research, trading and industry practice, and a single-paper literature positioning. After the data section, a results section establishes the effect and its robustness. A second results section rules out risk-based explanations. The conclusion quantifies the dollar transfer from buyers to writers and recommends trading-time conventions for the VIX and for data providers.\",\"writing_profile_keywords\":[\"anomaly documentation\",\"option returns\",\"weekend effect\",\"mispricing interpretation\",\"risk adjustment\",\"industry practice implications\",\"calendar vs trading time\"],\"confidence\":0.85,\"notes\":\"The abstract gives no method or magnitude sentence. Robustness appears within a results subsection, and an Internet Appendix is mentioned. Sections II and III are both results-oriented and are labeled results.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in average delta-hedged and unhedged option returns between nontrading and trading periods, plus the gap between realized and implied variance during market closure.", "strategy": "Calendar-based comparison of portfolio returns (nontrading minus trading means) with Newey-West t-statistics. Risk-adjusted performance measures (alphas, Sharpe ratios, certainty equivalents, Kadan-Liu indexes) and Fama-MacBeth cross-sectional regressions are used to rule out risk-based explanations.", "key_assumptions": ["Risk adjustment measures adequately capture option return risk, including higher moments", "Bid-ask midpoint quotes reflect tradable prices", "Nontrading timing is exogenous to option-specific information"], "threats_addressed": ["Differential risk across trading and nontrading periods", "Increased aversion to unbounded downside risk over weekends", "Portfolio construction and weighting choices", "Subsample and year-by-year stability", "Transaction costs"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "policy_or_broader_implication"}, {"paragraph_index": 9, "move": "policy_or_broader_implication"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "roadmap"}], "contribution_paragraph_index": 7, "contribution_types": ["new_evidence", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "alternative_measure", "alternative_sample_period"]}, "tables": {"convention_notes": "Notes define each portfolio construction, report Newey-West t-statistics in parentheses, and present differences as nontrading mean minus trading mean. Volume is given in 100-share contracts and spreads in percent. Fama-MacBeth regressions are used for the cross-sectional analysis.", "units_stated_in_notes": true, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "yes"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["our evidence suggests", "we believe", "likely", "may", "possible that"]}, "claim_evidence": [{"claim": "Delta-hedged option returns are strongly negative over nontrading periods and near zero on trading days.", "evidence_type": "table", "evidence_ref": "Table II, Table III"}, {"claim": "The effect holds across puts and calls, maturities, moneyness, weighting schemes and samples.", "evidence_type": "table", "evidence_ref": "Table III, Table IV"}, {"claim": "The nontrading effect is negative in every sample year.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Implied volatilities embed sizable variance during market closure, while realized variance over closure is small.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Risk-adjusted returns from writing options remain much higher over nontrading periods despite somewhat higher risk.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Cross-sectional risk patterns do not explain nontrading returns.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The nontrading effect is related to volume and open interest patterns.", "evidence_type": "figure", "evidence_ref": "Figure 2"}], "structural_summary": "The paper documents an anomaly and then tests explanations for it. The introduction motivates with institutional reasons why risk and return may differ when markets are closed. It then previews pervasive nontrading return effects and a gap between implied and realized variance. Three explanations are laid out (risk, risk aversion, mispricing), and the evidence is previewed as favoring mispricing. The introduction closes with implications for research, trading and industry practice, and a single-paper literature positioning. After the data section, a results section establishes the effect and its robustness. A second results section rules out risk-based explanations. The conclusion quantifies the dollar transfer from buyers to writers and recommends trading-time conventions for the VIX and for data providers.", "writing_profile_keywords": ["anomaly documentation", "option returns", "weekend effect", "mispricing interpretation", "risk adjustment", "industry practice implications", "calendar vs trading time"], "confidence": 0.85, "notes": "The abstract gives no method or magnitude sentence. Robustness appears within a results subsection, and an Internet Appendix is mentioned. Sections II and III are both results-oriented and are labeled results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.73, "usage": {"input": 7800, "output": 2164, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075274, "request_id": "req_011CggptcbXCJjzwwWnSpLeS", "cached": false, "ts": "2026-11-03T23:01:21+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/52300d61a9177b256b0aeb37eecdde7a2ae40f2a7ed0c3f7907985b1930b8752.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of the tightness of investors' leverage constraints (Regulation T initial margin requirement) on the slope and intercept of the security market line\",\"strategy\":\"Time-series regressions of monthly cross-sectional SML slope and intercept, estimated from beta-sorted portfolios, on the lagged Fed-set initial margin requirement, exploiting 22 regulatory changes from 1934 to 1974, with controls for market return, other risk factors and the cost of leverage\",\"key_assumptions\":[\"Margin requirement changes are not driven by factors that independently affect the price of beta risk\",\"The margin requirement is a direct and binding measure of maximum leverage, not an endogenous price\",\"Ex ante portfolio betas adequately measure systematic risk\"],\"threats_addressed\":[\"Endogeneity of margin changes to market conditions (determinants of Regulation T changes analysed)\",\"Confounding by the cost of leverage (borrowing-lending rate spread control)\",\"Omitted risk factors\",\"Sensitivity to test assets, portfolio construction and estimation method\",\"Short-sales constraints interacting with disagreement\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Each table note opens by stating what is regressed on what, then defines the variables and their construction, such as SML coefficients from monthly cross-sectional regressions of beta-sorted portfolios. Standard-error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' theory phrasing\",\"'at least in part'\",\"explicit acknowledgment that constraints cannot fully explain the flat SML\",\"listing alternative explanatory factors\"]},\"claim_evidence\":[{\"claim\":\"Tighter margin requirements are associated with a flatter SML slope and a higher intercept.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The result holds when controlling for additional risk factors and using alternative test assets.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"},{\"claim\":\"The borrowing-lending rate spread does not explain the SML slope or weaken the margin effect.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Increases in margin requirements reduce margin credit.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table III\"},{\"claim\":\"The SML is flatter in subsamples with higher initial margin requirements.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Figure 4\"},{\"claim\":\"Under a common margin constraint, the SML is flatter than in the unconstrained CAPM.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I eq. (1); Appendix derivation\"},{\"claim\":\"The TED spread is a poor proxy because it is endogenous and has the wrong-signed relation in prior work.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Cohen, Polk, Vuolteenaho (2005); Frazzini and Pedersen (2014)\"}],\"structural_summary\":\"The paper opens with the flat-SML anomaly and the leverage-constraint theory that explains it. It then argues that existing proxies such as the TED spread are flawed and proposes the Fed's margin requirement as a direct measure. The theory section summarizes a constrained-CAPM model, with the derivation in the appendix. A long institutional section documents the regulation, what drives its changes, and its effects on margin credit. A two-step SML estimation, following Cohen, Polk and Vuolteenaho, feeds time-series regressions on the margin level. Results are followed by robustness checks, a cost-of-leverage control and a short-sales/disagreement analysis. The conclusion restates the finding but concedes that constraints alone cannot explain negative slopes, and it lists other possible factors.\",\"writing_profile_keywords\":[\"regulatory variation\",\"constrained CAPM\",\"betting-against-beta\",\"security market line\",\"historical data\",\"test of theory\",\"Fama-MacBeth style two-step\"],\"confidence\":0.82,\"notes\":\"Paragraphs that apparently contain the first and second contributions are truncated or missing, so the introduction moves are partial. The sample runs to 1975. The standard-error method is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of the tightness of investors' leverage constraints (Regulation T initial margin requirement) on the slope and intercept of the security market line", "strategy": "Time-series regressions of monthly cross-sectional SML slope and intercept, estimated from beta-sorted portfolios, on the lagged Fed-set initial margin requirement, exploiting 22 regulatory changes from 1934 to 1974, with controls for market return, other risk factors and the cost of leverage", "key_assumptions": ["Margin requirement changes are not driven by factors that independently affect the price of beta risk", "The margin requirement is a direct and binding measure of maximum leverage, not an endogenous price", "Ex ante portfolio betas adequately measure systematic risk"], "threats_addressed": ["Endogeneity of margin changes to market conditions (determinants of Regulation T changes analysed)", "Confounding by the cost of leverage (borrowing-lending rate spread control)", "Omitted risk factors", "Sensitivity to test assets, portfolio construction and estimation method", "Short-sales constraints interacting with disagreement"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Each table note opens by stating what is regressed on what, then defines the variables and their construction, such as SML coefficients from monthly cross-sectional regressions of beta-sorted portfolios. Standard-error treatment is not visible in the excerpts.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' theory phrasing", "'at least in part'", "explicit acknowledgment that constraints cannot fully explain the flat SML", "listing alternative explanatory factors"]}, "claim_evidence": [{"claim": "Tighter margin requirements are associated with a flatter SML slope and a higher intercept.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The result holds when controlling for additional risk factors and using alternative test assets.", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}, {"claim": "The borrowing-lending rate spread does not explain the SML slope or weaken the margin effect.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Increases in margin requirements reduce margin credit.", "evidence_type": "figure", "evidence_ref": "Figure 3; Table III"}, {"claim": "The SML is flatter in subsamples with higher initial margin requirements.", "evidence_type": "figure", "evidence_ref": "Figure 1, Figure 4"}, {"claim": "Under a common margin constraint, the SML is flatter than in the unconstrained CAPM.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I eq. (1); Appendix derivation"}, {"claim": "The TED spread is a poor proxy because it is endogenous and has the wrong-signed relation in prior work.", "evidence_type": "prior_literature", "evidence_ref": "Cohen, Polk, Vuolteenaho (2005); Frazzini and Pedersen (2014)"}], "structural_summary": "The paper opens with the flat-SML anomaly and the leverage-constraint theory that explains it. It then argues that existing proxies such as the TED spread are flawed and proposes the Fed's margin requirement as a direct measure. The theory section summarizes a constrained-CAPM model, with the derivation in the appendix. A long institutional section documents the regulation, what drives its changes, and its effects on margin credit. A two-step SML estimation, following Cohen, Polk and Vuolteenaho, feeds time-series regressions on the margin level. Results are followed by robustness checks, a cost-of-leverage control and a short-sales/disagreement analysis. The conclusion restates the finding but concedes that constraints alone cannot explain negative slopes, and it lists other possible factors.", "writing_profile_keywords": ["regulatory variation", "constrained CAPM", "betting-against-beta", "security market line", "historical data", "test of theory", "Fama-MacBeth style two-step"], "confidence": 0.82, "notes": "Paragraphs that apparently contain the first and second contributions are truncated or missing, so the introduction moves are partial. The sample runs to 1975. The standard-error method is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.89, "usage": {"input": 6992, "output": 2286, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074482, "request_id": "req_011Cggpw1bwUb5qw4g8FPHvq", "cached": false, "ts": "2026-11-03T23:01:54+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/580b3f76810d041a434a6172e92c5a01ce123770c612789fc07d6efdd31e6d27.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"methodological_econometric\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"computational methods / numerical solution of long-run risk models (financial_econometrics)\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Not applicable. The target quantity is the approximation error of Campbell-Shiller log-linearized solutions relative to global projection solutions, measured in model moments and predictions.\",\"strategy\":\"Numerical comparison under calibrated or estimated parameters from published studies. Log-linearized and global (projection/spline) solutions are compared for equity premium, price-dividend volatility, predictability regressions and yield curves.\",\"key_assumptions\":[\"The global projection solution is accurate enough to serve as the benchmark, supported by Euler-error and moment accuracy checks in the appendix.\",\"Parameter values taken from the original studies are representative of the literature.\"],\"threats_addressed\":[\"Accuracy of the benchmark solution, via Euler errors and moment convergence.\",\"Dependence of errors on parameter choices, via sensitivity plots over γ, ψ and persistence.\",\"Special case where the SDF is linear (unitary EIS), with closed forms in the appendix.\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"replication_or_correction\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"},{\"idx\":13,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables set annualized moments from the log-linear solution beside the global solution, with relative errors in percent. Parameter values are listed in the notes with page references to the original studies. Euler-error tables report MAE and RMSE. No standard errors are shown because the outputs are numerical.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"modal 'can' for error magnitudes\",\"conditioning on high persistence or parameter regions\",\"noting that lower-persistence models have small errors\"]},\"claim_evidence\":[{\"claim\":\"Log-linearization overstates the equity premium by about 100 bps and return predictability by roughly a factor of two in the Bansal-Kiku-Yaron calibration.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table VI\"},{\"claim\":\"Approximation errors in moments can exceed 70% at the high-persistence (95% quantile) estimates of Schorfheide-Song-Yaron.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Appendix table of annualized moments and errors for Schorfheide, Song, and Yaron (2018); Figure E3\"},{\"claim\":\"Nonlinearities stem from highly persistent state processes and their interaction with stochastic volatility.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1, 2, 6, 7; Table I\"},{\"claim\":\"Errors grow with persistence and risk aversion across parameter choices.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 4, 5\"},{\"claim\":\"Log-linearization can invert the slope of the nominal yield curve in bond models.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 8, 9\"},{\"claim\":\"The projection benchmark is highly accurate, with Euler errors below 4e-9.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Appendix B\"}],\"structural_summary\":\"The paper opens from a landmark model and the standard approximation technique, then asks whether neglected higher-order terms matter. It previews large errors with concrete magnitudes across six published models, closing the introduction with a usage recommendation and a roadmap. A model-framework section is followed by an analysis isolating the sources of nonlinearity (persistence, volatility interaction, the unitary-EIS special case). A model-by-model comparison of moments, predictability and yield curves comes next. The conclusion restates the findings with bounds, noting small errors for low-persistence models. Appendices document solution methods, benchmark accuracy, closed forms and extra figures.\",\"writing_profile_keywords\":[\"numerical accuracy\",\"log-linearization critique\",\"global projection methods\",\"long-run risk\",\"replication of published calibrations\",\"relative error reporting\",\"parameter sensitivity\"],\"confidence\":0.8,\"notes\":\"Genre could alternatively be read as theoretical or computational; I classed it as methodological because the core contribution is assessing a solution technique. Sections II and III are tagged theory by the rule, but their content is numerical results comparison. No empirical data or statistical inference is used.\"}", "json": {"genre": "methodological_econometric", "subfield": "asset_pricing", "subfield_secondary": "computational methods / numerical solution of long-run risk models (financial_econometrics)", "primary_method": "calibration_simulation", "identification": {"estimand": "Not applicable. The target quantity is the approximation error of Campbell-Shiller log-linearized solutions relative to global projection solutions, measured in model moments and predictions.", "strategy": "Numerical comparison under calibrated or estimated parameters from published studies. Log-linearized and global (projection/spline) solutions are compared for equity premium, price-dividend volatility, predictability regressions and yield curves.", "key_assumptions": ["The global projection solution is accurate enough to serve as the benchmark, supported by Euler-error and moment accuracy checks in the appendix.", "Parameter values taken from the original studies are representative of the literature."], "threats_addressed": ["Accuracy of the benchmark solution, via Euler errors and moment convergence.", "Dependence of errors on parameter choices, via sensitivity plots over γ, ψ and persistence.", "Special case where the SDF is linear (unitary EIS), with closed forms in the appendix."], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "replication_or_correction"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}, {"idx": 13, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters", "alternative_specification", "other"]}, "tables": {"convention_notes": "Tables set annualized moments from the log-linear solution beside the global solution, with relative errors in percent. Parameter values are listed in the notes with page references to the original studies. Euler-error tables report MAE and RMSE. No standard errors are shown because the outputs are numerical.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["modal 'can' for error magnitudes", "conditioning on high persistence or parameter regions", "noting that lower-persistence models have small errors"]}, "claim_evidence": [{"claim": "Log-linearization overstates the equity premium by about 100 bps and return predictability by roughly a factor of two in the Bansal-Kiku-Yaron calibration.", "evidence_type": "table", "evidence_ref": "Table V, Table VI"}, {"claim": "Approximation errors in moments can exceed 70% at the high-persistence (95% quantile) estimates of Schorfheide-Song-Yaron.", "evidence_type": "table", "evidence_ref": "Appendix table of annualized moments and errors for Schorfheide, Song, and Yaron (2018); Figure E3"}, {"claim": "Nonlinearities stem from highly persistent state processes and their interaction with stochastic volatility.", "evidence_type": "figure", "evidence_ref": "Figures 1, 2, 6, 7; Table I"}, {"claim": "Errors grow with persistence and risk aversion across parameter choices.", "evidence_type": "figure", "evidence_ref": "Figures 4, 5"}, {"claim": "Log-linearization can invert the slope of the nominal yield curve in bond models.", "evidence_type": "figure", "evidence_ref": "Figures 8, 9"}, {"claim": "The projection benchmark is highly accurate, with Euler errors below 4e-9.", "evidence_type": "table", "evidence_ref": "Table II; Appendix B"}], "structural_summary": "The paper opens from a landmark model and the standard approximation technique, then asks whether neglected higher-order terms matter. It previews large errors with concrete magnitudes across six published models, closing the introduction with a usage recommendation and a roadmap. A model-framework section is followed by an analysis isolating the sources of nonlinearity (persistence, volatility interaction, the unitary-EIS special case). A model-by-model comparison of moments, predictability and yield curves comes next. The conclusion restates the findings with bounds, noting small errors for low-persistence models. Appendices document solution methods, benchmark accuracy, closed forms and extra figures.", "writing_profile_keywords": ["numerical accuracy", "log-linearization critique", "global projection methods", "long-run risk", "replication of published calibrations", "relative error reporting", "parameter sensitivity"], "confidence": 0.8, "notes": "Genre could alternatively be read as theoretical or computational; I classed it as methodological because the core contribution is assessing a solution technique. Sections II and III are tagged theory by the rule, but their content is numerical results comparison. No empirical data or statistical inference is used."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 27.11, "usage": {"input": 8895, "output": 3213, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.100634, "request_id": "req_011CggptuewogAJ5vrMxMAZo", "cached": false, "ts": "2026-11-03T23:01:35+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/5d32141da98f4fd3c0f4cde643e0084d36d59122f6583ef1d3260996099da9a3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"dynamic contracting / executive compensation and CEO turnover\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Associations between a firm's growth prospects (initial Q, ratio of Q) and the probability of CEO turnover and the front-loading of CEO pay\",\"strategy\":\"Probit regressions of turnover and panel regressions of log pay on tenure-year interacted with initial Q, run on ExecuComp/Compustat/CRSP 1992-2014, used to test model predictions; no quasi-experimental design visible\",\"key_assumptions\":[\"Average Q measures growth prospects\",\"The model's comparative statics map onto cross-firm variation in observed Q\"],\"threats_addressed\":[\"Acknowledges other theories may explain the same patterns\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_theory_or_model\",\"new_method\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes name the merged data set and sample period, define each variable, and identify the estimator (probit marginal effects, compensation regressions). The visible excerpts do not describe standard errors.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\",\"figures\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'broadly supports' / 'consistent with' for empirical findings\",\"'may' for model outcomes\",\"explicit acknowledgment that other theories may fit the evidence\",\"'we believe' for extensions\"]},\"claim_evidence\":[{\"claim\":\"The risk of growth-induced dismissal raises managers' effective discount rate and therefore agency costs, which leads to front-loaded pay.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections I-III\"},{\"claim\":\"Low-growth firms optimally forgo growth opportunities after good performance, which amounts to job protection.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.D, Figure 2\"},{\"claim\":\"Severance pay is suboptimal, but incoming managers may receive a signing bonus.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections II-III\"},{\"claim\":\"Firms with better growth prospects have higher CEO turnover.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-III, Figure 5\"},{\"claim\":\"Firms with better growth prospects use more front-loaded CEO compensation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Figure 6\"},{\"claim\":\"High-growth firms have shorter tenures and shorter compensation duration in model simulations.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 3-4\"},{\"claim\":\"Managers matter for firm performance, especially in growing environments.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Bertrand and Schoar (2003); Bennedsen et al. (2012)\"}],\"structural_summary\":\"The paper is theory-led and adds a short empirical test. The introduction motivates a tension between replacing managers to seize growth and the cost of incentive provision. It then previews the mechanism and the main contract results (two firm types, drift adjustment, no severance pay), positions the work against the CEO-match and dynamic-contracting literatures, lists technical contributions, and closes with a roadmap. The body builds a two-period model with derived empirical implications, then a continuous-time model with a state-space representation of contracts. Next it characterizes the optimal contract with numerical illustrations, and finally reports probit and compensation regressions on ExecuComp data. The conclusion restates the insight, suggests extensions, and hedges on the empirical channel.\",\"writing_profile_keywords\":[\"dynamic moral hazard\",\"continuous-time contracting\",\"CEO turnover\",\"front-loaded compensation\",\"model-then-evidence\",\"consistent-with empirics\"],\"confidence\":0.85,\"notes\":\"The primary contribution is theoretical. The empirical section is supportive and its claims are associational. Standard errors and robustness checks are not visible in the excerpts, and an Internet Appendix is mentioned.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "corporate_finance", "subfield_secondary": "dynamic contracting / executive compensation and CEO turnover", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Associations between a firm's growth prospects (initial Q, ratio of Q) and the probability of CEO turnover and the front-loading of CEO pay", "strategy": "Probit regressions of turnover and panel regressions of log pay on tenure-year interacted with initial Q, run on ExecuComp/Compustat/CRSP 1992-2014, used to test model predictions; no quasi-experimental design visible", "key_assumptions": ["Average Q measures growth prospects", "The model's comparative statics map onto cross-firm variation in observed Q"], "threats_addressed": ["Acknowledges other theories may explain the same patterns"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_theory_or_model", "new_method", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes name the merged data set and sample period, define each variable, and identify the estimator (probit marginal effects, compensation regressions). The visible excerpts do not describe standard errors.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no", "figures": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'broadly supports' / 'consistent with' for empirical findings", "'may' for model outcomes", "explicit acknowledgment that other theories may fit the evidence", "'we believe' for extensions"]}, "claim_evidence": [{"claim": "The risk of growth-induced dismissal raises managers' effective discount rate and therefore agency costs, which leads to front-loaded pay.", "evidence_type": "equation_or_proof", "evidence_ref": "Sections I-III"}, {"claim": "Low-growth firms optimally forgo growth opportunities after good performance, which amounts to job protection.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.D, Figure 2"}, {"claim": "Severance pay is suboptimal, but incoming managers may receive a signing bonus.", "evidence_type": "equation_or_proof", "evidence_ref": "Sections II-III"}, {"claim": "Firms with better growth prospects have higher CEO turnover.", "evidence_type": "table", "evidence_ref": "Tables II-III, Figure 5"}, {"claim": "Firms with better growth prospects use more front-loaded CEO compensation.", "evidence_type": "table", "evidence_ref": "Table IV, Figure 6"}, {"claim": "High-growth firms have shorter tenures and shorter compensation duration in model simulations.", "evidence_type": "figure", "evidence_ref": "Figures 3-4"}, {"claim": "Managers matter for firm performance, especially in growing environments.", "evidence_type": "prior_literature", "evidence_ref": "Bertrand and Schoar (2003); Bennedsen et al. (2012)"}], "structural_summary": "The paper is theory-led and adds a short empirical test. The introduction motivates a tension between replacing managers to seize growth and the cost of incentive provision. It then previews the mechanism and the main contract results (two firm types, drift adjustment, no severance pay), positions the work against the CEO-match and dynamic-contracting literatures, lists technical contributions, and closes with a roadmap. The body builds a two-period model with derived empirical implications, then a continuous-time model with a state-space representation of contracts. Next it characterizes the optimal contract with numerical illustrations, and finally reports probit and compensation regressions on ExecuComp data. The conclusion restates the insight, suggests extensions, and hedges on the empirical channel.", "writing_profile_keywords": ["dynamic moral hazard", "continuous-time contracting", "CEO turnover", "front-loaded compensation", "model-then-evidence", "consistent-with empirics"], "confidence": 0.85, "notes": "The primary contribution is theoretical. The empirical section is supportive and its claims are associational. Standard errors and robustness checks are not visible in the excerpts, and an Internet Appendix is mentioned."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.7, "usage": {"input": 7726, "output": 2176, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075218, "request_id": "req_011CggppYqhVccuPdcDQBtVp", "cached": false, "ts": "2026-11-03T23:00:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/5ed40cbf05976698685318b11ae9b53e62e099eb8f9c5b0a6a929a8a03bbfd79.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Multiperiod exposures of interest-rate-sorted currency baskets to identified macroeconomic shocks, and the horizon-dependent prices of those exposures (shock exposure and shock price elasticities).\",\"strategy\":\"A structural VAR with stochastic variance for U.S. consumption growth, inflation, the nominal rate and basket exchange-rate growth, with cross-equation restrictions from recursive preferences. Structural shocks are identified through zero restrictions on the impact matrix. The model appears to be estimated with Bayesian methods, given that credible intervals are reported.\",\"key_assumptions\":[\"A representative U.S. agent with recursive preferences prices currency returns.\",\"The zero restrictions on the impact matrix correctly identify the structural shocks.\",\"Macro and exchange-rate dynamics follow a VAR with a single common stochastic variance factor.\",\"The foreign pricing kernel need not be modeled, so market completeness is not assumed.\"],\"threats_addressed\":[\"Comparison of unrestricted and restricted exposure elasticities.\",\"Statistical insignificance of risk premia at long horizons is acknowledged.\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"literature_gap\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":9,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"confidence_intervals\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"appendix\",\"types\":[\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes state model equations and identifying restrictions, data frequency and sample period. Uncertainty is reported as 95% credible intervals in parentheses rather than stars.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"yes\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests' and 'qualitatively similar'\",\"explicitly noting statistical insignificance at long horizons\",\"deferring the economic origins of the shock to future research\",\"describing the study as complementary to existing work\"]},\"claim_evidence\":[{\"claim\":\"The term structure of average carry log returns slopes slightly downward, from about 1.13% to 0.80% per quarter.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Higher yield currencies have larger multiperiod exposures to interest rate and inflation shocks.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-3; Table III\"},{\"claim\":\"Prices of the interest rate, inflation and consumption growth risks are positive, with the interest rate risk priced highest.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 4\"},{\"claim\":\"The interest rate shock affects consumption with a lag and behaves like Bansal-Yaron long-run risk.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Three-month currency risk premia decompose into contributions from the separate macro shocks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The variance shock is the sole driver of time variation in risk premia.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure D1\"}],\"structural_summary\":\"Opens with a new stylized fact (a downward-sloping carry return term structure) and asks which macro risks it compensates. The introduction previews findings shock by shock, then positions the paper against term-structure and long-run risk work and lists facts future models must match. Section I sorts currencies into baskets and estimates a reduced-form VAR with exposure elasticities. Section II builds a recursive-preference pricing kernel and describes estimation. Section III reports exposure and price elasticities and links the interest rate shock to long-run risk. A short conclusion restates the findings, and appendices cover data, model solution and parameter estimates.\",\"writing_profile_keywords\":[\"first-person singular\",\"shock elasticities\",\"term structure\",\"structural VAR\",\"recursive preferences\",\"credible intervals\",\"stylized-fact motivation\",\"complementarity framing\"],\"confidence\":0.8,\"notes\":\"Section I mixes reduced-form evidence with model setup and is labeled results. The 'credible intervals' suggest Bayesian estimation. Robustness is limited to the unrestricted-versus-restricted comparison in Figure C1.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "international_finance", "primary_method": "structural_estimation", "identification": {"estimand": "Multiperiod exposures of interest-rate-sorted currency baskets to identified macroeconomic shocks, and the horizon-dependent prices of those exposures (shock exposure and shock price elasticities).", "strategy": "A structural VAR with stochastic variance for U.S. consumption growth, inflation, the nominal rate and basket exchange-rate growth, with cross-equation restrictions from recursive preferences. Structural shocks are identified through zero restrictions on the impact matrix. The model appears to be estimated with Bayesian methods, given that credible intervals are reported.", "key_assumptions": ["A representative U.S. agent with recursive preferences prices currency returns.", "The zero restrictions on the impact matrix correctly identify the structural shocks.", "Macro and exchange-rate dynamics follow a VAR with a single common stochastic variance factor.", "The foreign pricing kernel need not be modeled, so market completeness is not assumed."], "threats_addressed": ["Comparison of unrestricted and restricted exposure elasticities.", "Statistical insignificance of risk premia at long horizons is acknowledged."], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "literature_gap"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "roadmap"}], "contribution_paragraph_index": 9, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "confidence_intervals", "magnitude_statements_in_intro": true}, "robustness": {"location": "appendix", "types": ["alternative_specification"]}, "tables": {"convention_notes": "Notes state model equations and identifying restrictions, data frequency and sample period. Uncertainty is reported as 95% credible intervals in parentheses rather than stars.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "yes"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests' and 'qualitatively similar'", "explicitly noting statistical insignificance at long horizons", "deferring the economic origins of the shock to future research", "describing the study as complementary to existing work"]}, "claim_evidence": [{"claim": "The term structure of average carry log returns slopes slightly downward, from about 1.13% to 0.80% per quarter.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Higher yield currencies have larger multiperiod exposures to interest rate and inflation shocks.", "evidence_type": "figure", "evidence_ref": "Figures 2-3; Table III"}, {"claim": "Prices of the interest rate, inflation and consumption growth risks are positive, with the interest rate risk priced highest.", "evidence_type": "table", "evidence_ref": "Table VI; Figure 4"}, {"claim": "The interest rate shock affects consumption with a lag and behaves like Bansal-Yaron long-run risk.", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Three-month currency risk premia decompose into contributions from the separate macro shocks.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The variance shock is the sole driver of time variation in risk premia.", "evidence_type": "figure", "evidence_ref": "Figure D1"}], "structural_summary": "Opens with a new stylized fact (a downward-sloping carry return term structure) and asks which macro risks it compensates. The introduction previews findings shock by shock, then positions the paper against term-structure and long-run risk work and lists facts future models must match. Section I sorts currencies into baskets and estimates a reduced-form VAR with exposure elasticities. Section II builds a recursive-preference pricing kernel and describes estimation. Section III reports exposure and price elasticities and links the interest rate shock to long-run risk. A short conclusion restates the findings, and appendices cover data, model solution and parameter estimates.", "writing_profile_keywords": ["first-person singular", "shock elasticities", "term structure", "structural VAR", "recursive preferences", "credible intervals", "stylized-fact motivation", "complementarity framing"], "confidence": 0.8, "notes": "Section I mixes reduced-form evidence with model setup and is labeled results. The 'credible intervals' suggest Bayesian estimation. Robustness is limited to the unrestricted-versus-restricted comparison in Figure C1."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.34, "usage": {"input": 8419, "output": 2343, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08133, "request_id": "req_011CggpdcTRn5zCxq5bG1H3D", "cached": false, "ts": "2026-11-03T22:58:00+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/61f586c8a687efe4fee98bebfe95e190e46201bb0653111cf819f341cf15119c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure; financial fragility / limits to arbitrage\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theory): comparative statics of equilibrium prices, asset quality, and welfare with respect to the amount of arbitrage capital\",\"strategy\":\"Noisy rational expectations equilibrium with an endogenous lemons problem; arbitrage capital is varied in comparative statics\",\"key_assumptions\":[\"two-period economy with a risk-free asset in elastic supply\",\"informed participants (seller, arbitrageurs) are liquidity- or margin-constrained\",\"uninformed hedgers with random exposures generate noise\",\"risk-neutral competitive uninformed investors set prices\",\"seller chooses between selling the marketable asset and liquidating a nonmarketable asset\"],\"threats_addressed\":[\"exogenous margin constraints relaxed via an endogenous-margin extension\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"No tables. Results are shown through numerical comparative-statics figures, with the x-axis scaled as a percentage of baseline arbitrage capital.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'plausible mechanism' framing\",\"modal verbs (can, may)\",\"'shed light on' phrasing\"]},\"claim_evidence\":[{\"claim\":\"When arbitrage capital is ample, prices are informative and there is no price dislocation.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV.A; Appendix B proofs\"},{\"claim\":\"Scarce arbitrage capital makes prices less informative and creates an endogenous lemons problem, causing fire sales and market freezes.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections III-IV; Figures 3-4\"},{\"claim\":\"Reducing arbitrage capital lowers the expected price of the high-quality asset and the average quality of traded assets.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-7\"},{\"claim\":\"Welfare declines as arbitrage capital falls, across parameter values.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 8, Section IV.B\"},{\"claim\":\"The results are robust to endogenizing margin constraints.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section V, Figure 9\"}],\"structural_summary\":\"The paper opens with a puzzle: why do fire sales occur when well-capitalized nonspecialists exist? It contrasts this with existing explanations based on specialist valuations and limited arbitrage. It then sketches a two-period model combining informed-but-constrained and uninformed-but-unconstrained agents, and previews the capital-multiplier mechanism. A separate literature review follows. The model and its equilibrium are set out formally, and the main results cover fire sales, market freezes, and welfare. An extension endogenizes margin constraints as a robustness check. A discussion section covers fragility, policy, and the 2007-08 crisis, and the conclusion restates the mechanism. Proofs are in the appendix.\",\"writing_profile_keywords\":[\"puzzle-driven\",\"noisy REE\",\"lemons problem\",\"limits to arbitrage\",\"comparative statics figures\",\"policy discussion\",\"crisis interpretation\"],\"confidence\":0.85,\"notes\":\"Pure theory paper. The figures are numerical illustrations, not empirical results. The Discussion section (idx 6) is coded 'other' rather than 'conclusion'; its content is policy implications and an interpretation of the 2007-08 crisis.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure; financial fragility / limits to arbitrage", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theory): comparative statics of equilibrium prices, asset quality, and welfare with respect to the amount of arbitrage capital", "strategy": "Noisy rational expectations equilibrium with an endogenous lemons problem; arbitrage capital is varied in comparative statics", "key_assumptions": ["two-period economy with a risk-free asset in elastic supply", "informed participants (seller, arbitrageurs) are liquidity- or margin-constrained", "uninformed hedgers with random exposures generate noise", "risk-neutral competitive uninformed investors set prices", "seller chooses between selling the marketable asset and liquidating a nonmarketable asset"], "threats_addressed": ["exogenous margin constraints relaxed via an endogenous-margin extension"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}], "contribution_paragraph_index": 2, "contribution_types": ["new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "No tables. Results are shown through numerical comparative-statics figures, with the x-axis scaled as a percentage of baseline arbitrage capital.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'plausible mechanism' framing", "modal verbs (can, may)", "'shed light on' phrasing"]}, "claim_evidence": [{"claim": "When arbitrage capital is ample, prices are informative and there is no price dislocation.", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV.A; Appendix B proofs"}, {"claim": "Scarce arbitrage capital makes prices less informative and creates an endogenous lemons problem, causing fire sales and market freezes.", "evidence_type": "equation_or_proof", "evidence_ref": "Sections III-IV; Figures 3-4"}, {"claim": "Reducing arbitrage capital lowers the expected price of the high-quality asset and the average quality of traded assets.", "evidence_type": "figure", "evidence_ref": "Figures 5-7"}, {"claim": "Welfare declines as arbitrage capital falls, across parameter values.", "evidence_type": "figure", "evidence_ref": "Figure 8, Section IV.B"}, {"claim": "The results are robust to endogenizing margin constraints.", "evidence_type": "figure", "evidence_ref": "Section V, Figure 9"}], "structural_summary": "The paper opens with a puzzle: why do fire sales occur when well-capitalized nonspecialists exist? It contrasts this with existing explanations based on specialist valuations and limited arbitrage. It then sketches a two-period model combining informed-but-constrained and uninformed-but-unconstrained agents, and previews the capital-multiplier mechanism. A separate literature review follows. The model and its equilibrium are set out formally, and the main results cover fire sales, market freezes, and welfare. An extension endogenizes margin constraints as a robustness check. A discussion section covers fragility, policy, and the 2007-08 crisis, and the conclusion restates the mechanism. Proofs are in the appendix.", "writing_profile_keywords": ["puzzle-driven", "noisy REE", "lemons problem", "limits to arbitrage", "comparative statics figures", "policy discussion", "crisis interpretation"], "confidence": 0.85, "notes": "Pure theory paper. The figures are numerical illustrations, not empirical results. The Discussion section (idx 6) is coded 'other' rather than 'conclusion'; its content is policy implications and an interpretation of the 2007-08 crisis."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.56, "usage": {"input": 5670, "output": 1936, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062194, "request_id": "req_011CggpoHhLdfXjkQEptAiWg", "cached": false, "ts": "2026-11-03T23:00:07+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/62fb6cad517007f71f64d2578aaae8b557476a9f41e25095045b05892277067b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"international_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"The share of time-series variation in bilateral exchange rates explained by systematic currency factors (carry and dollar), plus the risk premia on portfolios of currencies sorted by dollar beta.\",\"strategy\":\"Country-level time-series factor regressions of exchange rate changes on carry, conditional carry and dollar factors, with the dependent currency excluded from the factors. These are combined with portfolio sorts on dollar betas and GMM/Fama-MacBeth asset pricing tests.\",\"key_assumptions\":[\"Portfolios average out country-specific shocks (law of large numbers).\",\"High-minus-low factors are immune to U.S.-specific shocks.\",\"Markets are complete, so exchange rate changes equal differences in log SDFs (used in the model interpretation).\"],\"threats_addressed\":[\"Mechanical correlation, handled by excluding the dependent currency from the factors.\",\"U.S.-specific shocks, handled with the global dollar component and other base currencies.\",\"Data frequency, checked with daily, quarterly and annual data.\",\"Sample composition, checked with developed versus emerging countries.\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_sample_period\",\"subsample\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes write out the full country-level regression equation and define each variable. Tables are organised by country group and data frequency. One table reports portfolio summary statistics alongside GMM and Fama-MacBeth asset pricing results. Cross-country second-stage regressions relate shares of systematic exchange rate variation to comovement measures.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"appear/appears\",\"suggest\",\"tentative interpretation\",\"open question\",\"could be interpreted\"]},\"claim_evidence\":[{\"claim\":\"Carry and dollar factors together explain a large share of monthly bilateral exchange rate changes in developed countries.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Sorting currencies on dollar beta produces a cross section of excess returns that is priced by a dollar slope factor.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 1\"},{\"claim\":\"Dollar-neutral global factors explain 18% to 83% of exchange rate variation and also describe yen- and pound-based rates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V and VI\"},{\"claim\":\"At the daily frequency, the dollar factor explains much more variation than the carry factor.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Cross-country shares of systematic exchange rate variation are related to the comovement of capital flows.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Table VII; Figure 2\"},{\"claim\":\"SDFs must contain at least two global shocks to generate two independent cross sections of currency risk premia.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Internet Appendix\"}],\"structural_summary\":\"The introduction contrasts purely statistical principal components with interpretable risk factors. It then previews the factor regressions and their R2 ranges, names the dollar-neutral regressions as the key result, and treats the economic source of the shocks as an open question. A reduced-form SDF model is sketched, followed by in-intro literature positioning. The body moves from the model framework to time-series factor regressions across frequencies and samples, then to dollar-beta portfolio pricing. A section of implications and additional tests covers global shocks, other base currencies, conditional carry and links to macro comovement. A short conclusion draws implications for practitioners and researchers.\",\"writing_profile_keywords\":[\"factor regressions\",\"R2 decomposition\",\"currency portfolios\",\"reduced-form SDF model\",\"first-person singular\",\"dollar factor\",\"global shocks\"],\"confidence\":0.8,\"notes\":\"Single-author paper. Section IV's first paragraph announces 'three novel facts' but then lists four. Section II is classified as results. Significance conventions and standard errors are not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "international_finance", "subfield_secondary": "asset_pricing", "primary_method": "reduced_form_regression", "identification": {"estimand": "The share of time-series variation in bilateral exchange rates explained by systematic currency factors (carry and dollar), plus the risk premia on portfolios of currencies sorted by dollar beta.", "strategy": "Country-level time-series factor regressions of exchange rate changes on carry, conditional carry and dollar factors, with the dependent currency excluded from the factors. These are combined with portfolio sorts on dollar betas and GMM/Fama-MacBeth asset pricing tests.", "key_assumptions": ["Portfolios average out country-specific shocks (law of large numbers).", "High-minus-low factors are immune to U.S.-specific shocks.", "Markets are complete, so exchange rate changes equal differences in log SDFs (used in the model interpretation)."], "threats_addressed": ["Mechanical correlation, handled by excluding the dependent currency from the factors.", "U.S.-specific shocks, handled with the global dollar component and other base currencies.", "Data frequency, checked with daily, quarterly and annual data.", "Sample composition, checked with developed versus emerging countries."], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}], "contribution_paragraph_index": 2, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_sample_period", "subsample", "alternative_measure", "other"]}, "tables": {"convention_notes": "Table notes write out the full country-level regression equation and define each variable. Tables are organised by country group and data frequency. One table reports portfolio summary statistics alongside GMM and Fama-MacBeth asset pricing results. Cross-country second-stage regressions relate shares of systematic exchange rate variation to comovement measures.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["appear/appears", "suggest", "tentative interpretation", "open question", "could be interpreted"]}, "claim_evidence": [{"claim": "Carry and dollar factors together explain a large share of monthly bilateral exchange rate changes in developed countries.", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Sorting currencies on dollar beta produces a cross section of excess returns that is priced by a dollar slope factor.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 1"}, {"claim": "Dollar-neutral global factors explain 18% to 83% of exchange rate variation and also describe yen- and pound-based rates.", "evidence_type": "table", "evidence_ref": "Tables V and VI"}, {"claim": "At the daily frequency, the dollar factor explains much more variation than the carry factor.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Cross-country shares of systematic exchange rate variation are related to the comovement of capital flows.", "evidence_type": "figure", "evidence_ref": "Table VII; Figure 2"}, {"claim": "SDFs must contain at least two global shocks to generate two independent cross sections of currency risk premia.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Internet Appendix"}], "structural_summary": "The introduction contrasts purely statistical principal components with interpretable risk factors. It then previews the factor regressions and their R2 ranges, names the dollar-neutral regressions as the key result, and treats the economic source of the shocks as an open question. A reduced-form SDF model is sketched, followed by in-intro literature positioning. The body moves from the model framework to time-series factor regressions across frequencies and samples, then to dollar-beta portfolio pricing. A section of implications and additional tests covers global shocks, other base currencies, conditional carry and links to macro comovement. A short conclusion draws implications for practitioners and researchers.", "writing_profile_keywords": ["factor regressions", "R2 decomposition", "currency portfolios", "reduced-form SDF model", "first-person singular", "dollar factor", "global shocks"], "confidence": 0.8, "notes": "Single-author paper. Section IV's first paragraph announces 'three novel facts' but then lists four. Section II is classified as results. Significance conventions and standard errors are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.53, "usage": {"input": 6777, "output": 2179, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071482, "request_id": "req_011CggppSbA8GCouf3nWJZXN", "cached": false, "ts": "2026-11-03T23:00:24+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/64039def73ed5d504824a573cd2581c61f6f2786dc7428ad1c6f748c042a8ddd.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"not applicable (theoretical model; equilibrium effects of uncertainty shocks on shadow banking, asset prices and investment)\",\"strategy\":\"Analytical static model to microfound fragile liquidity, followed by a calibrated dynamic continuous-time model solved with projection methods; impulse responses and counterfactual comparisons (with vs. without shadow banking, with and without policy interventions)\",\"key_assumptions\":[\"Investors face liquidity events and value liquid securities\",\"Producing liquid securities requires scarce collateral\",\"Shadow money is liquid in quiet times but becomes illiquid when uncertainty is high\",\"Net worth frictions are abstracted from in the baseline\"],\"threats_addressed\":[\"Omission of net worth frictions (an extension with such a friction is placed in the Internet Appendix)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"motivation\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":8,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"robustness_mention\"}],\"contribution_paragraph_index\":9,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"internet_appendix\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"Only two tables appear: a glossary of model variables and a list of benchmark parameter values. Neither reports regressions. Notes describe the functional forms chosen for the parameterization.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\",\"notes\":\"\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'can be interpreted as'\",\"'consistent with' when linking model dynamics to stylized facts\",\"'sheds light on'\"]},\"claim_evidence\":[{\"claim\":\"When uncertainty is low, shadow banking expands liquidity, lowers discount rates and raises asset prices and growth.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 3-5\"},{\"claim\":\"An uncertainty shock collapses shadow money, shrinks liquidity, raises discount rates and lowers investment.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"},{\"claim\":\"Crashes followed by collateral mining and the collateral decelerator produce slow recoveries.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"LSAP raises asset prices both at announcement and ex ante, by supplying collateral when it is scarce.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 8\"},{\"claim\":\"Operation Twist lowers aggregate collateral and is counterproductive even though long-term bond yields fall.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"},{\"claim\":\"Fragile liquidity creates a trade-off between the size and the stability of liquidity supply.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendix A (Proposition 1)\"},{\"claim\":\"The pre-crisis credit expansion was concentrated among riskier borrowers, consistent with the model's boom dynamics.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Mian and Sufi (2009); Krishnamurthy and Muir (2015); Ranciere et al. (2008)\"}],\"structural_summary\":\"The paper opens with a narrative of boom, bust and slow recovery and frames shadow banking as fragile liquidity transformation. The introduction then walks through the model's cycle phase by phase (boom, shock, collapse, slow recovery, flight to quality), linking each phase to stylized facts. It previews policy results and ends by positioning the paper against net-worth-based macrofinance models. The body moves from a static microfoundation of the core trade-off, to a dynamic continuous-time model, to calibrated numerical results presented through figures and impulse responses, and then to policy counterfactuals. A short conclusion restates the mechanism. Proofs, the numerical method and parameter details are placed in appendices.\",\"writing_profile_keywords\":[\"macrofinance theory\",\"narrative mechanism walkthrough\",\"static-to-dynamic model progression\",\"calibrated impulse responses\",\"policy counterfactuals\",\"figure-driven results\",\"stylized facts as motivation\"],\"confidence\":0.85,\"notes\":\"This is a pure theory and calibration paper with no empirical estimation, so fields such as statistical significance and standard errors are not applicable. The only robustness exercise visible is the net-worth friction extension in the Internet Appendix. The Policy Interventions section is coded 'other' because it is a counterfactual policy-analysis section, not a robustness check.\"}", "json": {"genre": "theoretical", "subfield": "macro_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "calibration_simulation", "identification": {"estimand": "not applicable (theoretical model; equilibrium effects of uncertainty shocks on shadow banking, asset prices and investment)", "strategy": "Analytical static model to microfound fragile liquidity, followed by a calibrated dynamic continuous-time model solved with projection methods; impulse responses and counterfactual comparisons (with vs. without shadow banking, with and without policy interventions)", "key_assumptions": ["Investors face liquidity events and value liquid securities", "Producing liquid securities requires scarce collateral", "Shadow money is liquid in quiet times but becomes illiquid when uncertainty is high", "Net worth frictions are abstracted from in the baseline"], "threats_addressed": ["Omission of net worth frictions (an extension with such a friction is placed in the Internet Appendix)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "motivation"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "policy_or_broader_implication"}, {"paragraph_index": 8, "move": "policy_or_broader_implication"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "robustness_mention"}], "contribution_paragraph_index": 9, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "internet_appendix", "types": ["other"]}, "tables": {"convention_notes": "Only two tables appear: a glossary of model variables and a list of benchmark parameter values. Neither reports regressions. Notes describe the functional forms chosen for the parameterization.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no", "notes": ""}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'can be interpreted as'", "'consistent with' when linking model dynamics to stylized facts", "'sheds light on'"]}, "claim_evidence": [{"claim": "When uncertainty is low, shadow banking expands liquidity, lowers discount rates and raises asset prices and growth.", "evidence_type": "figure", "evidence_ref": "Figures 3-5"}, {"claim": "An uncertainty shock collapses shadow money, shrinks liquidity, raises discount rates and lowers investment.", "evidence_type": "figure", "evidence_ref": "Figure 6"}, {"claim": "Crashes followed by collateral mining and the collateral decelerator produce slow recoveries.", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "LSAP raises asset prices both at announcement and ex ante, by supplying collateral when it is scarce.", "evidence_type": "figure", "evidence_ref": "Figure 8"}, {"claim": "Operation Twist lowers aggregate collateral and is counterproductive even though long-term bond yields fall.", "evidence_type": "figure", "evidence_ref": "Figure 9"}, {"claim": "Fragile liquidity creates a trade-off between the size and the stability of liquidity supply.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendix A (Proposition 1)"}, {"claim": "The pre-crisis credit expansion was concentrated among riskier borrowers, consistent with the model's boom dynamics.", "evidence_type": "prior_literature", "evidence_ref": "Mian and Sufi (2009); Krishnamurthy and Muir (2015); Ranciere et al. (2008)"}], "structural_summary": "The paper opens with a narrative of boom, bust and slow recovery and frames shadow banking as fragile liquidity transformation. The introduction then walks through the model's cycle phase by phase (boom, shock, collapse, slow recovery, flight to quality), linking each phase to stylized facts. It previews policy results and ends by positioning the paper against net-worth-based macrofinance models. The body moves from a static microfoundation of the core trade-off, to a dynamic continuous-time model, to calibrated numerical results presented through figures and impulse responses, and then to policy counterfactuals. A short conclusion restates the mechanism. Proofs, the numerical method and parameter details are placed in appendices.", "writing_profile_keywords": ["macrofinance theory", "narrative mechanism walkthrough", "static-to-dynamic model progression", "calibrated impulse responses", "policy counterfactuals", "figure-driven results", "stylized facts as motivation"], "confidence": 0.85, "notes": "This is a pure theory and calibration paper with no empirical estimation, so fields such as statistical significance and standard errors are not applicable. The only robustness exercise visible is the net-worth friction extension in the Internet Appendix. The Policy Interventions section is coded 'other' because it is a counterfactual policy-analysis section, not a robustness check."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.06, "usage": {"input": 8517, "output": 2384, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082542, "request_id": "req_011CggpjQVG8CmPPRnu6HP6T", "cached": false, "ts": "2026-11-03T22:59:18+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/65bbba998d9417f425d090d4cac61dfb184a8afedda834e213a9a651dae062bc.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in anomaly (long-minus-short) daily returns on firm-specific information days (earnings announcements, Dow Jones news days) versus noninformation days.\",\"strategy\":\"Panel regressions of daily stock returns on an aggregate anomaly measure interacted with information-day dummies, with time fixed effects. Competing explanations (dynamic risk, biased expectations, data mining) are distinguished through dynamic beta controls, analyst forecast errors, controls for contemporaneous monthly returns, and out-of-sample anomalies.\",\"key_assumptions\":[\"Time fixed effects absorb daily variation in risk premia\",\"The aggregate anomaly factor adequately captures time-varying beta exposure\",\"Out-of-sample anomalies are unlikely to be products of data mining\",\"Analyst forecast errors are not subject to the joint-hypothesis problem\"],\"threats_addressed\":[\"time-varying risk premia\",\"time-varying betas\",\"day-of-the-week effects\",\"macroeconomic news days\",\"reverse causality\",\"high-volatility days\",\"data mining / spurious anomalies\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":true,\"sentence_index\":0},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"identification_defense\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"contribution_statement\"},{\"paragraph_index\":11,\"move\":\"contribution_statement\"},{\"paragraph_index\":12,\"move\":\"contribution_statement\"},{\"paragraph_index\":13,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":10,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"subsample\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the regression specification, the unit of observation (firm-day), and that the dependent variable is scaled by 100. They describe sample filters (e.g., a $5 price screen) and how anomaly portfolios are built from quintile sorts. Control-variable coefficients are left unreported. In the visible excerpts, the notes do not describe standard errors.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"most consistent with\",\"could be explained by\",\"it is unlikely that\",\"at least partly\",\"likely reflect\"]},\"claim_evidence\":[{\"claim\":\"Anomaly returns are about 50% higher on news days and six times higher on earnings announcement days.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Information-day effects appear on both the long and short sides of anomalies.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The effect holds across the four anomaly categories.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The effects are not explained by day-of-week effects or macroeconomic news days.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Dynamic betas rise on earnings days but do not explain the earnings-day anomaly effect.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Analyst forecasts are too optimistic for anomaly-shorts and too pessimistic for anomaly-longs.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Out-of-sample anomalies still show information-day effects, making data mining unlikely as the explanation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX, Table X\"},{\"claim\":\"Anomaly returns are concentrated within the window around earnings announcements.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"The paper opens with a long-standing puzzle and three competing explanations for it: risk, mispricing, and data mining. It previews tests that discriminate among them and situates its contributions inside the introduction. The body first establishes a stylized fact, information-day anomaly returns, and runs falsification checks on it. Separate sections then test each explanation in turn. The conclusion favors biased expectations while keeping the dynamic-risk alternative open.\",\"writing_profile_keywords\":[\"horse-race of explanations\",\"stylized facts first\",\"anomaly portfolios\",\"information days\",\"analyst forecast errors\",\"data mining tests\",\"hedged mispricing conclusion\"],\"confidence\":0.85,\"notes\":\"The Internet Appendix contains a simple representative-agent model. Statistical significance conventions are not visible in the table-note excerpts.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in anomaly (long-minus-short) daily returns on firm-specific information days (earnings announcements, Dow Jones news days) versus noninformation days.", "strategy": "Panel regressions of daily stock returns on an aggregate anomaly measure interacted with information-day dummies, with time fixed effects. Competing explanations (dynamic risk, biased expectations, data mining) are distinguished through dynamic beta controls, analyst forecast errors, controls for contemporaneous monthly returns, and out-of-sample anomalies.", "key_assumptions": ["Time fixed effects absorb daily variation in risk premia", "The aggregate anomaly factor adequately captures time-varying beta exposure", "Out-of-sample anomalies are unlikely to be products of data mining", "Analyst forecast errors are not subject to the joint-hypothesis problem"], "threats_addressed": ["time-varying risk premia", "time-varying betas", "day-of-the-week effects", "macroeconomic news days", "reverse causality", "high-volatility days", "data mining / spurious anomalies"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": true, "sentence_index": 0}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "identification_defense"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "contribution_statement"}, {"paragraph_index": 11, "move": "contribution_statement"}, {"paragraph_index": 12, "move": "contribution_statement"}, {"paragraph_index": 13, "move": "roadmap"}], "contribution_paragraph_index": 10, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "subsample", "falsification", "other"]}, "tables": {"convention_notes": "Table notes state the regression specification, the unit of observation (firm-day), and that the dependent variable is scaled by 100. They describe sample filters (e.g., a $5 price screen) and how anomaly portfolios are built from quintile sorts. Control-variable coefficients are left unreported. In the visible excerpts, the notes do not describe standard errors.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["most consistent with", "could be explained by", "it is unlikely that", "at least partly", "likely reflect"]}, "claim_evidence": [{"claim": "Anomaly returns are about 50% higher on news days and six times higher on earnings announcement days.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Information-day effects appear on both the long and short sides of anomalies.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The effect holds across the four anomaly categories.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "The effects are not explained by day-of-week effects or macroeconomic news days.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Dynamic betas rise on earnings days but do not explain the earnings-day anomaly effect.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Analyst forecasts are too optimistic for anomaly-shorts and too pessimistic for anomaly-longs.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Out-of-sample anomalies still show information-day effects, making data mining unlikely as the explanation.", "evidence_type": "table", "evidence_ref": "Table IX, Table X"}, {"claim": "Anomaly returns are concentrated within the window around earnings announcements.", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "The paper opens with a long-standing puzzle and three competing explanations for it: risk, mispricing, and data mining. It previews tests that discriminate among them and situates its contributions inside the introduction. The body first establishes a stylized fact, information-day anomaly returns, and runs falsification checks on it. Separate sections then test each explanation in turn. The conclusion favors biased expectations while keeping the dynamic-risk alternative open.", "writing_profile_keywords": ["horse-race of explanations", "stylized facts first", "anomaly portfolios", "information days", "analyst forecast errors", "data mining tests", "hedged mispricing conclusion"], "confidence": 0.85, "notes": "The Internet Appendix contains a simple representative-agent model. Statistical significance conventions are not visible in the table-note excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.53, "usage": {"input": 8184, "output": 2331, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08015, "request_id": "req_011Cggq2wa1WmUbXuGjYDRrA", "cached": false, "ts": "2026-11-03T23:03:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/666773bb31a6d893186453ada85a9af560ecc96a7315afb19f041902a5b33be4.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"public finance / taxation; retail industry\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Change in household tax-exclusive Amazon spending after states require Amazon to collect sales tax, converted into price elasticities; also substitution to competing retailers\",\"strategy\":\"Staggered difference-in-differences across 19 adopting states versus non-adopting states using household-month transaction data with fixed effects; event-time plot of coefficients; tax-rate intensity specification\",\"key_assumptions\":[\"Parallel trends between adopting and non-adopting states absent the tax\",\"Adoption not driven by state economic shocks that also reduce consumption\",\"Household residence correctly identified\"],\"threats_addressed\":[\"Non-random state adoption due to economic weakness (GDP growth and household income checks)\",\"Pre-trends (event-time coefficients)\",\"Sample filter on 2011 Amazon spending\",\"Location measurement error\",\"Income effects on other spending\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"appendix\",\"types\":[\"alternative_measure\",\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes open by stating the table's purpose, unit of observation (household-month), dependent variable and treatment indicator definitions. Appendix tables are labeled as replications of main-table columns. Dollar amounts are described as tax-exclusive or tax-inclusive. Standard-error treatment is not visible in the truncated notes.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'some evidence' for substitution\",\"explicit acknowledgment of external-validity limits of the sample\",\"'consistent with' and 'suggests' for mechanism\"]},\"claim_evidence\":[{\"claim\":\"Households in adopting states cut tax-exclusive Amazon spending by about 9.4%, implying an elasticity near -1.2\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"A one-point increase in sales tax lowers monthly Amazon spending by $0.54, implying an elasticity of about -1.4\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV / Table D1\"},{\"claim\":\"Low-income households reduce Amazon spending more than high-income households\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Purchases of at least $250 fall about 29%, implying an elasticity near -3.9\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Elasticity rises with transaction size\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table VII\"},{\"claim\":\"Adopting states did not differ in GDP growth or household income around adoption\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Some substitution toward Newegg but not toward Best Buy\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Spending decline persists after implementation without clear pre-trends\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"}],\"structural_summary\":\"The paper motivates the question with the growth of online retail and its sales-tax advantage, then notes the absence of systematic study of state Amazon tax laws. It presents a novel account-aggregator transaction dataset and acknowledges that the sample may not represent the general population. Section III defends against non-random adoption. Section IV gives difference-in-differences estimates, then cross-sectional splits by income and spending, large-purchase effects, a size gradient in elasticity, and substitution toward competitors. The appendices hold robustness checks: alternative residence identification, removal of the sample filter, tax-inclusive outcomes, alternative elasticity calculation, purchase probability, and income effects. The conclusion briefly restates the magnitudes.\",\"writing_profile_keywords\":[\"policy natural experiment\",\"transaction-level data\",\"elasticity translation\",\"staggered DiD\",\"heterogeneity by purchase size\",\"appendix robustness\"],\"confidence\":0.8,\"notes\":\"Section III, which tests adopter versus non-adopter differences, is classified as empirical_design. The excerpts do not reveal the standard-error clustering or whether stars are used. The intro and conclusion give different start dates for the sample period (2011 vs 2012).\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "public finance / taxation; retail industry", "primary_method": "difference_in_differences", "identification": {"estimand": "Change in household tax-exclusive Amazon spending after states require Amazon to collect sales tax, converted into price elasticities; also substitution to competing retailers", "strategy": "Staggered difference-in-differences across 19 adopting states versus non-adopting states using household-month transaction data with fixed effects; event-time plot of coefficients; tax-rate intensity specification", "key_assumptions": ["Parallel trends between adopting and non-adopting states absent the tax", "Adoption not driven by state economic shocks that also reduce consumption", "Household residence correctly identified"], "threats_addressed": ["Non-random state adoption due to economic weakness (GDP growth and household income checks)", "Pre-trends (event-time coefficients)", "Sample filter on 2011 Amazon spending", "Location measurement error", "Income effects on other spending"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "appendix", "types": ["alternative_measure", "subsample", "alternative_specification"]}, "tables": {"convention_notes": "Notes open by stating the table's purpose, unit of observation (household-month), dependent variable and treatment indicator definitions. Appendix tables are labeled as replications of main-table columns. Dollar amounts are described as tax-exclusive or tax-inclusive. Standard-error treatment is not visible in the truncated notes.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'some evidence' for substitution", "explicit acknowledgment of external-validity limits of the sample", "'consistent with' and 'suggests' for mechanism"]}, "claim_evidence": [{"claim": "Households in adopting states cut tax-exclusive Amazon spending by about 9.4%, implying an elasticity near -1.2", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "A one-point increase in sales tax lowers monthly Amazon spending by $0.54, implying an elasticity of about -1.4", "evidence_type": "table", "evidence_ref": "Table IV / Table D1"}, {"claim": "Low-income households reduce Amazon spending more than high-income households", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Purchases of at least $250 fall about 29%, implying an elasticity near -3.9", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Elasticity rises with transaction size", "evidence_type": "figure", "evidence_ref": "Figure 3; Table VII"}, {"claim": "Adopting states did not differ in GDP growth or household income around adoption", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Some substitution toward Newegg but not toward Best Buy", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Spending decline persists after implementation without clear pre-trends", "evidence_type": "figure", "evidence_ref": "Figure 4"}], "structural_summary": "The paper motivates the question with the growth of online retail and its sales-tax advantage, then notes the absence of systematic study of state Amazon tax laws. It presents a novel account-aggregator transaction dataset and acknowledges that the sample may not represent the general population. Section III defends against non-random adoption. Section IV gives difference-in-differences estimates, then cross-sectional splits by income and spending, large-purchase effects, a size gradient in elasticity, and substitution toward competitors. The appendices hold robustness checks: alternative residence identification, removal of the sample filter, tax-inclusive outcomes, alternative elasticity calculation, purchase probability, and income effects. The conclusion briefly restates the magnitudes.", "writing_profile_keywords": ["policy natural experiment", "transaction-level data", "elasticity translation", "staggered DiD", "heterogeneity by purchase size", "appendix robustness"], "confidence": 0.8, "notes": "Section III, which tests adopter versus non-adopter differences, is classified as empirical_design. The excerpts do not reveal the standard-error clustering or whether stars are used. The intro and conclusion give different start dates for the sample period (2011 vs 2012)."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.96, "usage": {"input": 8748, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080826, "request_id": "req_011Cggq1iF3NHiLxmnSX3QLq", "cached": false, "ts": "2026-11-03T23:02:58+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/66c6bbacc5f66e6f4cb33ad83597fdaa1613c45f9a30f38af24e6753a519090f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Whether a low-dimensional principal-component SDF prices anomaly and size-B/M portfolios, and whether models with sentiment investors imply the same factor structure\",\"strategy\":\"No-near-arbitrage bound on SDF variance combined with the covariance structure of returns; PC factor pricing tests; pseudo out-of-sample split-sample exercise; equilibrium model with arbitrageurs and sentiment investors\",\"key_assumptions\":[\"Near-arbitrage opportunities (extremely high Sharpe ratios) are absent\",\"Some arbitrageurs are present\",\"Sentiment-investor position sizes and leverage are bounded\",\"IID/normal dividend structure in the static model\"],\"threats_addressed\":[\"In-sample overfitting of higher-order PCs, addressed by the out-of-sample split\",\"Unbounded sentiment positions as in Daniel, Hirshleifer, and Subrahmanyam (2001)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":9,\"move\":\"preview_of_approach\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"preview_of_approach\"},{\"paragraph_index\":12,\"move\":\"preview_of_results\"},{\"paragraph_index\":13,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":14,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":15,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":15,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_sample_period\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Notes state the sample period and data source (Novy-Marx and Velikov anomalies). Returns, alphas and squared Sharpe ratios are annualized. Table IV reports Newey-West t-statistics with four lags.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'should' predictions conditional on assumptions\",\"'do not appear to' qualifiers\",\"explicit conditionality ('as long as', 'only if')\",\"acknowledging limits of structural tests\"]},\"claim_evidence\":[{\"claim\":\"Absence of near-arbitrage plus dominant covariance factors implies the SDF is approximately a function of a few PCs\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Appendix A\"},{\"claim\":\"A few PC factors price anomaly portfolios about as well as popular reduced-form factor models\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Higher-order PCs raise in-sample Sharpe ratios but not out-of-sample ones\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4, Figure 5\"},{\"claim\":\"In a model where sentiment drives all expected-return variation, the first few PCs still explain the cross section\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Figures 6-7\"},{\"claim\":\"Analyst forecast bias, as a sentiment proxy, lines up with the high-eigenvalue PC portfolios\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-IV\"},{\"claim\":\"A dynamic sentiment model yields an ICAPM with good and bad beta\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV; Appendix C\"}],\"structural_summary\":\"The paper opens by questioning the conventional reading of reduced-form factor models and quotes the Cochrane challenge to set up the debate. It then argues conceptually that no-near-arbitrage combined with a dominant covariance structure forces a low-dimensional SDF. Empirical support follows from PC factor pricing and split-sample out-of-sample tests. An equilibrium model with arbitrageurs and sentiment investors then shows observational equivalence with rational pricing, and a dynamic extension delivers an ICAPM. The conclusion calls for structural models of beliefs and preferences.\",\"writing_profile_keywords\":[\"argumentative\",\"conceptual critique\",\"observational equivalence\",\"principal components\",\"no-near-arbitrage\",\"sentiment model\",\"figures-heavy\",\"model plus evidence\"],\"confidence\":0.8,\"notes\":\"The introduction paragraph list splits a block quote into fragments (paragraphs 2-7), so the paragraph indices are shifted accordingly. Statistical inference is barely visible in the excerpts; only Table IV's notes mention Newey-West t-statistics.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Whether a low-dimensional principal-component SDF prices anomaly and size-B/M portfolios, and whether models with sentiment investors imply the same factor structure", "strategy": "No-near-arbitrage bound on SDF variance combined with the covariance structure of returns; PC factor pricing tests; pseudo out-of-sample split-sample exercise; equilibrium model with arbitrageurs and sentiment investors", "key_assumptions": ["Near-arbitrage opportunities (extremely high Sharpe ratios) are absent", "Some arbitrageurs are present", "Sentiment-investor position sizes and leverage are bounded", "IID/normal dividend structure in the static model"], "threats_addressed": ["In-sample overfitting of higher-order PCs, addressed by the out-of-sample split", "Unbounded sentiment positions as in Daniel, Hirshleifer, and Subrahmanyam (2001)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 9, "move": "preview_of_approach"}, {"paragraph_index": 10, "move": "preview_of_results"}, {"paragraph_index": 11, "move": "preview_of_results"}, {"paragraph_index": 11, "move": "preview_of_approach"}, {"paragraph_index": 12, "move": "preview_of_results"}, {"paragraph_index": 13, "move": "policy_or_broader_implication"}, {"paragraph_index": 14, "move": "related_literature_positioning"}, {"paragraph_index": 15, "move": "related_literature_positioning"}, {"paragraph_index": 15, "move": "roadmap"}], "contribution_paragraph_index": 8, "contribution_types": ["new_theory_or_model", "new_evidence", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_sample_period", "alternative_measure"]}, "tables": {"convention_notes": "Notes state the sample period and data source (Novy-Marx and Velikov anomalies). Returns, alphas and squared Sharpe ratios are annualized. Table IV reports Newey-West t-statistics with four lags.", "units_stated_in_notes": true, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'should' predictions conditional on assumptions", "'do not appear to' qualifiers", "explicit conditionality ('as long as', 'only if')", "acknowledging limits of structural tests"]}, "claim_evidence": [{"claim": "Absence of near-arbitrage plus dominant covariance factors implies the SDF is approximately a function of a few PCs", "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Appendix A"}, {"claim": "A few PC factors price anomaly portfolios about as well as popular reduced-form factor models", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Higher-order PCs raise in-sample Sharpe ratios but not out-of-sample ones", "evidence_type": "figure", "evidence_ref": "Figure 4, Figure 5"}, {"claim": "In a model where sentiment drives all expected-return variation, the first few PCs still explain the cross section", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Figures 6-7"}, {"claim": "Analyst forecast bias, as a sentiment proxy, lines up with the high-eigenvalue PC portfolios", "evidence_type": "table", "evidence_ref": "Tables III-IV"}, {"claim": "A dynamic sentiment model yields an ICAPM with good and bad beta", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV; Appendix C"}], "structural_summary": "The paper opens by questioning the conventional reading of reduced-form factor models and quotes the Cochrane challenge to set up the debate. It then argues conceptually that no-near-arbitrage combined with a dominant covariance structure forces a low-dimensional SDF. Empirical support follows from PC factor pricing and split-sample out-of-sample tests. An equilibrium model with arbitrageurs and sentiment investors then shows observational equivalence with rational pricing, and a dynamic extension delivers an ICAPM. The conclusion calls for structural models of beliefs and preferences.", "writing_profile_keywords": ["argumentative", "conceptual critique", "observational equivalence", "principal components", "no-near-arbitrage", "sentiment model", "figures-heavy", "model plus evidence"], "confidence": 0.8, "notes": "The introduction paragraph list splits a block quote into fragments (paragraphs 2-7), so the paragraph indices are shifted accordingly. Statistical inference is barely visible in the excerpts; only Table IV's notes mention Newey-West t-statistics."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.56, "usage": {"input": 8690, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080594, "request_id": "req_011Cggpv4srUfcHvY4htXQxh", "cached": false, "ts": "2026-11-03T23:01:42+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/68573fb6287686843e7ef5441b110b3ba2ecb45f4855c727763d828684bbac10.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"international_finance\",\"subfield_secondary\":\"asset_pricing; macro_finance (monetary policy announcements)\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Difference in average currency excess returns (vs. USD) on scheduled FOMC announcement days relative to nonannouncement days, and how this difference varies with interest rate differentials, monetary policy uncertainty, and the sign of the policy shock\",\"strategy\":\"High-frequency event-study comparison of announcement and nonannouncement days. Time-series regressions use an announcement dummy interacted with uncertainty proxies and a monetary policy indicator, with returns split into pre- and postannouncement windows. Results are interpreted through a constrained-financier model.\",\"key_assumptions\":[\"Scheduled FOMC dates are predetermined, so announcement-day timing is exogenous to currency returns\",\"The uncertainty proxies capture market uncertainty about monetary policy\",\"The high-frequency monetary policy indicator measures the policy surprise\"],\"threats_addressed\":[\"Unscheduled meetings excluded\",\"Outliers handled with a truncated sample\",\"Inference checked with bootstrapped standard errors and random sampling\",\"Transaction costs\",\"Other macroeconomic announcements (payroll, CPI, PPI)\",\"Other central bank announcements (Bank of Japan)\",\"Alternative measures of uncertainty\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"falsification\",\"other\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes define panels (full sample, nonannouncement days, announcement days), the return windows in ET clock times, portfolio labels (pf1–pf3, dol), and regressors such as the announcement dummy and the standardized uncertainty measure (TIV). Summary statistics are reported in daily basis points alongside Sharpe ratios. Regression tables are time-series regressions of portfolio excess returns.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\",\"convention_notes_placeholder_removed\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'we interpret' framing of results through the model\",\"'consistent with / in line with our model'\",\"'associated with' wording for the uncertainty link\",\"'potentially' in the motivation\"]},\"claim_evidence\":[{\"claim\":\"Short-USD strategy earns larger excess returns on FOMC announcement days than on other days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Table II, Figure 1\"},{\"claim\":\"Announcement-day returns are larger for high interest rate currency portfolios (14.47 vs 1.73 bps)\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III, Figure 3\"},{\"claim\":\"Higher monetary policy uncertainty is associated with higher preannouncement returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table IX\"},{\"claim\":\"Postannouncement returns are higher after monetary easing than after tightening\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 4\"},{\"claim\":\"Conditioning the strategy on the policy stance and interest rate differential raises announcement-day returns from 10.77 to 20.54 bps\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Volatility does not rise proportionally on announcement days, so Sharpe ratios are higher\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2, Table I\"},{\"claim\":\"In the model, monetary policy uncertainty raises currency risk premia, more so for high interest rate currencies\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I, Appendix A\"}],\"structural_summary\":\"The introduction motivates the paper with the importance of FOMC announcements and the size and leverage of currency markets. It then previews a sequence of results with magnitudes: announcement-day excess returns, the interest-differential cross-section, the uncertainty link, the pre/post decomposition, and improved trading strategies. A contribution paragraph contrasts the paper with equity-market announcement work, followed by literature positioning and a roadmap. A Gabaix–Maggiori-style constrained-financier model comes first and generates predictions. Data and empirical sections follow: summary statistics, time-series regressions with announcement dummies interacted with uncertainty and policy-shock measures, trading strategies, other central banks, and an in-text robustness subsection. A brief conclusion restates findings through the model, with proofs in the appendix.\",\"writing_profile_keywords\":[\"announcement-day returns\",\"high-frequency data\",\"model-guided empirics\",\"risk premium interpretation\",\"currency portfolios\",\"trading strategy\",\"robustness battery\"],\"confidence\":0.85,\"notes\":\"Sections idx 4–6 are parsing artifacts (currency labels taken as headings); idx 6 actually holds the time-series analysis, trading strategy, other-central-bank, and robustness subsections. The Internet Appendix holds further robustness checks. The standard error type is not visible in the excerpts. In the figures field, the 'convention_notes_placeholder_removed' key is not in the schema and should be dropped.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "international_finance", "subfield_secondary": "asset_pricing; macro_finance (monetary policy announcements)", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Difference in average currency excess returns (vs. USD) on scheduled FOMC announcement days relative to nonannouncement days, and how this difference varies with interest rate differentials, monetary policy uncertainty, and the sign of the policy shock", "strategy": "High-frequency event-study comparison of announcement and nonannouncement days. Time-series regressions use an announcement dummy interacted with uncertainty proxies and a monetary policy indicator, with returns split into pre- and postannouncement windows. Results are interpreted through a constrained-financier model.", "key_assumptions": ["Scheduled FOMC dates are predetermined, so announcement-day timing is exogenous to currency returns", "The uncertainty proxies capture market uncertainty about monetary policy", "The high-frequency monetary policy indicator measures the policy surprise"], "threats_addressed": ["Unscheduled meetings excluded", "Outliers handled with a truncated sample", "Inference checked with bootstrapped standard errors and random sampling", "Transaction costs", "Other macroeconomic announcements (payroll, CPI, PPI)", "Other central bank announcements (Bank of Japan)", "Alternative measures of uncertainty"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "falsification", "other", "alternative_specification"]}, "tables": {"convention_notes": "Table notes define panels (full sample, nonannouncement days, announcement days), the return windows in ET clock times, portfolio labels (pf1–pf3, dol), and regressors such as the announcement dummy and the standardized uncertainty measure (TIV). Summary statistics are reported in daily basis points alongside Sharpe ratios. Regression tables are time-series regressions of portfolio excess returns.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed", "convention_notes_placeholder_removed": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'we interpret' framing of results through the model", "'consistent with / in line with our model'", "'associated with' wording for the uncertainty link", "'potentially' in the motivation"]}, "claim_evidence": [{"claim": "Short-USD strategy earns larger excess returns on FOMC announcement days than on other days", "evidence_type": "table", "evidence_ref": "Table I, Table II, Figure 1"}, {"claim": "Announcement-day returns are larger for high interest rate currency portfolios (14.47 vs 1.73 bps)", "evidence_type": "table", "evidence_ref": "Table II, Table III, Figure 3"}, {"claim": "Higher monetary policy uncertainty is associated with higher preannouncement returns", "evidence_type": "table", "evidence_ref": "Table V, Table IX"}, {"claim": "Postannouncement returns are higher after monetary easing than after tightening", "evidence_type": "table", "evidence_ref": "Table VI, Figure 4"}, {"claim": "Conditioning the strategy on the policy stance and interest rate differential raises announcement-day returns from 10.77 to 20.54 bps", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Volatility does not rise proportionally on announcement days, so Sharpe ratios are higher", "evidence_type": "figure", "evidence_ref": "Figure 2, Table I"}, {"claim": "In the model, monetary policy uncertainty raises currency risk premia, more so for high interest rate currencies", "evidence_type": "equation_or_proof", "evidence_ref": "Section I, Appendix A"}], "structural_summary": "The introduction motivates the paper with the importance of FOMC announcements and the size and leverage of currency markets. It then previews a sequence of results with magnitudes: announcement-day excess returns, the interest-differential cross-section, the uncertainty link, the pre/post decomposition, and improved trading strategies. A contribution paragraph contrasts the paper with equity-market announcement work, followed by literature positioning and a roadmap. A Gabaix–Maggiori-style constrained-financier model comes first and generates predictions. Data and empirical sections follow: summary statistics, time-series regressions with announcement dummies interacted with uncertainty and policy-shock measures, trading strategies, other central banks, and an in-text robustness subsection. A brief conclusion restates findings through the model, with proofs in the appendix.", "writing_profile_keywords": ["announcement-day returns", "high-frequency data", "model-guided empirics", "risk premium interpretation", "currency portfolios", "trading strategy", "robustness battery"], "confidence": 0.85, "notes": "Sections idx 4–6 are parsing artifacts (currency labels taken as headings); idx 6 actually holds the time-series analysis, trading strategy, other-central-bank, and robustness subsections. The Internet Appendix holds further robustness checks. The standard error type is not visible in the excerpts. In the figures field, the 'convention_notes_placeholder_removed' key is not in the schema and should be dropped."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 22.64, "usage": {"input": 8219, "output": 2639, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08645, "request_id": "req_011CggpaZC8nFHh4jFxnj4se", "cached": false, "ts": "2026-11-03T22:57:21+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/68d6bae21147f0d83f71c9bca92374523d2d8af9bddc95f1aaa39b1c0f359551.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"macro_finance (monetary policy transmission)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of the short-term policy rate (federal funds rate) on the ex ante risk rating of newly issued bank loans, and how that effect varies with bank capitalization\",\"strategy\":\"Loan-level panel regressions of internal loan risk ratings on the federal funds rate interacted with bank capital, with loan, bank, regional and macro controls. Endogeneity is addressed by using new loans, Taylor rule residuals, regional business-cycle asynchrony, and small-bank and state exclusions.\",\"key_assumptions\":[\"Policy rate is plausibly exogenous to individual bank lending risk at loan origination\",\"Internal risk ratings comparably measure ex ante loan risk across banks\",\"Controls capture demand-side and economic-condition confounders\"],\"threats_addressed\":[\"Reverse causality of monetary policy responding to bank risk-taking\",\"Changes in economic conditions\",\"Securitization and loan sales\",\"Loans under preexisting commitment\",\"Large-bank influence on policy\",\"Liquidity versus capital as the relevant bank characteristic\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"alternative_measure\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the sample period (1997Q2–2011Q4), define the dependent variable as the STBL internal risk rating on a 1–5 scale, list the data sources for the controls, and describe sample restrictions column by column.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'associated with' wording alongside causal claims\",\"'consistent with' a theoretical channel\",\"scope qualifier limiting results to the new-loan margin\",\"noting difficulty of quantifying the channel versus establishing it\"]},\"claim_evidence\":[{\"claim\":\"Loan risk ratings rise when the federal funds rate falls; a one-SD rate decrease raises ratings by about 0.11\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The negative rate–risk relationship is weaker for poorly capitalized banks, consistent with risk-shifting\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Average loan risk rating co-moves negatively with the target federal funds rate over time\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Results are robust to Taylor rule residuals and to controls for economic conditions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Theory gives ambiguous predictions depending on maturity mismatch and leverage\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Appendix A\"},{\"claim\":\"Capital, not liquidity, drives the heterogeneity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The introduction motivates the topic with post-crisis policy debate and lays out competing theoretical channels (portfolio, risk-shifting, search for yield), concluding that the sign is an empirical question. It previews confidential loan-level data, main results with magnitudes, and an endogeneity defense, then positions the paper against prior risk measures and Jimenez et al. The body follows a short methodology section, a data section, and a long results section whose subsections are mostly robustness checks. The conclusion restates magnitudes and contrasts the findings with Spain, and a simple theoretical model sits in an appendix.\",\"writing_profile_keywords\":[\"loan-level confidential data\",\"interaction with bank capital\",\"risk-taking channel\",\"competing theoretical predictions\",\"extensive robustness subsections\",\"policy-relevant motivation\"],\"confidence\":0.8,\"notes\":\"Standard-error and significance conventions are not visible in the table-note excerpts. The abstract mentions regional business-cycle synchrony and financial distress, which correspond to the Table X analyses.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "macro_finance (monetary policy transmission)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of the short-term policy rate (federal funds rate) on the ex ante risk rating of newly issued bank loans, and how that effect varies with bank capitalization", "strategy": "Loan-level panel regressions of internal loan risk ratings on the federal funds rate interacted with bank capital, with loan, bank, regional and macro controls. Endogeneity is addressed by using new loans, Taylor rule residuals, regional business-cycle asynchrony, and small-bank and state exclusions.", "key_assumptions": ["Policy rate is plausibly exogenous to individual bank lending risk at loan origination", "Internal risk ratings comparably measure ex ante loan risk across banks", "Controls capture demand-side and economic-condition confounders"], "threats_addressed": ["Reverse causality of monetary policy responding to bank risk-taking", "Changes in economic conditions", "Securitization and loan sales", "Loans under preexisting commitment", "Large-bank influence on policy", "Liquidity versus capital as the relevant bank characteristic"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["subsample", "alternative_measure", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes state the sample period (1997Q2–2011Q4), define the dependent variable as the STBL internal risk rating on a 1–5 scale, list the data sources for the controls, and describe sample restrictions column by column.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'associated with' wording alongside causal claims", "'consistent with' a theoretical channel", "scope qualifier limiting results to the new-loan margin", "noting difficulty of quantifying the channel versus establishing it"]}, "claim_evidence": [{"claim": "Loan risk ratings rise when the federal funds rate falls; a one-SD rate decrease raises ratings by about 0.11", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The negative rate–risk relationship is weaker for poorly capitalized banks, consistent with risk-shifting", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Average loan risk rating co-moves negatively with the target federal funds rate over time", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Results are robust to Taylor rule residuals and to controls for economic conditions", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Theory gives ambiguous predictions depending on maturity mismatch and leverage", "evidence_type": "equation_or_proof", "evidence_ref": "Appendix A"}, {"claim": "Capital, not liquidity, drives the heterogeneity", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The introduction motivates the topic with post-crisis policy debate and lays out competing theoretical channels (portfolio, risk-shifting, search for yield), concluding that the sign is an empirical question. It previews confidential loan-level data, main results with magnitudes, and an endogeneity defense, then positions the paper against prior risk measures and Jimenez et al. The body follows a short methodology section, a data section, and a long results section whose subsections are mostly robustness checks. The conclusion restates magnitudes and contrasts the findings with Spain, and a simple theoretical model sits in an appendix.", "writing_profile_keywords": ["loan-level confidential data", "interaction with bank capital", "risk-taking channel", "competing theoretical predictions", "extensive robustness subsections", "policy-relevant motivation"], "confidence": 0.8, "notes": "Standard-error and significance conventions are not visible in the table-note excerpts. The abstract mentions regional business-cycle synchrony and financial distress, which correspond to the Table X analyses."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.16, "usage": {"input": 7754, "output": 2111, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07403, "request_id": "req_011CggpWUprHSiSyMixMJ3cu", "cached": false, "ts": "2026-11-03T22:56:21+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/68e686251a536e60fc5256267b033c0f879ca8fab6d203a8a04392209d1bb6a2.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"international_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of capital account openness on within-sector dispersion in the marginal revenue product of capital, differentially by sector external financial dependence, and implied aggregate TFP gains\",\"strategy\":\"Cross-country variation in the timing of liberalization (Chinn-Ito index, lagged), interacted with Rajan-Zingales sectoral external financial dependence, with country-year, country-sector and sector-year fixed effects. Aggregate country-year panel with country and year fixed effects. A misallocation model maps estimates into TFP.\",\"key_assumptions\":[\"Liberalization timing is driven largely by EU accession rather than sector-specific shocks\",\"Rajan-Zingales US-based financial dependence captures the technological need for external finance in these countries\",\"Dispersion in MRPK reflects misallocation from financial frictions\"],\"threats_addressed\":[\"Concurrent institutional changes triggered by liberalization (alternative channels)\",\"Measurement of openness (alternative de jure and de facto measures)\",\"Political connections and corruption\",\"Pre-trends and dynamics (event-study dynamic effects)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Each table note opens by writing out the full regression equation, with its fixed-effect structure and variable definitions. The aggregation table explains how its computation follows from a model equation. Excerpts are truncated, so standard-error and significance conventions are not visible.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'is associated with' for aggregate results\",\"'evidence consistent with the argument'\",\"'suggestive evidence'\",\"'we argue'\"]},\"claim_evidence\":[{\"claim\":\"Liberalization raises capital inflows and private credit relative to GDP, narrows interest spreads, and is associated with higher aggregate TFP\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Liberalization reduces MRPK dispersion more in sectors more dependent on external finance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Results survive controls for alternative institutional channels\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Young firms in financially dependent sectors expand most after liberalization\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V-VI\"},{\"claim\":\"Decline in dispersion stems from the age-explained component of MRPK\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Results are robust to alternative openness measures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Improved capital allocation raises aggregate TFP by 10-16%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; equation (15)\"},{\"claim\":\"Dynamic effects of liberalization on MRPK dispersion\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"}],\"structural_summary\":\"The paper motivates the question with cross-country growth evidence. It first presents aggregate panel evidence linking liberalization to financial development and TFP. It then builds a heterogeneous-firm misallocation model with bank-dependent firms. Firm-level tests follow, using a difference-in-differences design that interacts openness with sectoral financial dependence. Mechanism tests address alternative channels and firm age. Robustness checks cover corruption and alternative openness measures. The model then aggregates the reduced-form estimates into TFP gains, and the conclusion restates the contributions.\",\"writing_profile_keywords\":[\"cross-country firm-level panel\",\"misallocation\",\"Rajan-Zingales interaction\",\"model-guided empirics\",\"aggregation to TFP\",\"mechanism tests\",\"first-to claim\"],\"confidence\":0.8,\"notes\":\"The model section appears after the aggregate evidence rather than before it. Standard-error conventions are not visible in the excerpts. The introduction is truncated, and no roadmap is visible in it.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "international_finance", "subfield_secondary": "macro_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of capital account openness on within-sector dispersion in the marginal revenue product of capital, differentially by sector external financial dependence, and implied aggregate TFP gains", "strategy": "Cross-country variation in the timing of liberalization (Chinn-Ito index, lagged), interacted with Rajan-Zingales sectoral external financial dependence, with country-year, country-sector and sector-year fixed effects. Aggregate country-year panel with country and year fixed effects. A misallocation model maps estimates into TFP.", "key_assumptions": ["Liberalization timing is driven largely by EU accession rather than sector-specific shocks", "Rajan-Zingales US-based financial dependence captures the technological need for external finance in these countries", "Dispersion in MRPK reflects misallocation from financial frictions"], "threats_addressed": ["Concurrent institutional changes triggered by liberalization (alternative channels)", "Measurement of openness (alternative de jure and de facto measures)", "Political connections and corruption", "Pre-trends and dynamics (event-study dynamic effects)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "related_literature_positioning"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Each table note opens by writing out the full regression equation, with its fixed-effect structure and variable definitions. The aggregation table explains how its computation follows from a model equation. Excerpts are truncated, so standard-error and significance conventions are not visible.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'is associated with' for aggregate results", "'evidence consistent with the argument'", "'suggestive evidence'", "'we argue'"]}, "claim_evidence": [{"claim": "Liberalization raises capital inflows and private credit relative to GDP, narrows interest spreads, and is associated with higher aggregate TFP", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Liberalization reduces MRPK dispersion more in sectors more dependent on external finance", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Results survive controls for alternative institutional channels", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Young firms in financially dependent sectors expand most after liberalization", "evidence_type": "table", "evidence_ref": "Tables V-VI"}, {"claim": "Decline in dispersion stems from the age-explained component of MRPK", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Results are robust to alternative openness measures", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Improved capital allocation raises aggregate TFP by 10-16%", "evidence_type": "table", "evidence_ref": "Table IX; equation (15)"}, {"claim": "Dynamic effects of liberalization on MRPK dispersion", "evidence_type": "figure", "evidence_ref": "Figure 3"}], "structural_summary": "The paper motivates the question with cross-country growth evidence. It first presents aggregate panel evidence linking liberalization to financial development and TFP. It then builds a heterogeneous-firm misallocation model with bank-dependent firms. Firm-level tests follow, using a difference-in-differences design that interacts openness with sectoral financial dependence. Mechanism tests address alternative channels and firm age. Robustness checks cover corruption and alternative openness measures. The model then aggregates the reduced-form estimates into TFP gains, and the conclusion restates the contributions.", "writing_profile_keywords": ["cross-country firm-level panel", "misallocation", "Rajan-Zingales interaction", "model-guided empirics", "aggregation to TFP", "mechanism tests", "first-to claim"], "confidence": 0.8, "notes": "The model section appears after the aggregate evidence rather than before it. Standard-error conventions are not visible in the excerpts. The introduction is truncated, and no roadmap is visible in it."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.91, "usage": {"input": 7084, "output": 2132, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07177, "request_id": "req_011Cggpfr2oUQttk2HfTG2E9", "cached": false, "ts": "2026-11-03T22:58:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/6b4c77f8b3ffbdceef7b6b71e1fb6cff33a9479bfee127ccc881cc1f8c89f2cf.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"hedge funds / investment management\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between a manager's sensation-seeking trait, proxied by the attributes of the car they buy, and fund risk, performance, operational risk and trading behavior.\",\"strategy\":\"Hand-collected vehicle purchase records are matched to hedge fund managers. The paper uses portfolio sorts and multivariate OLS, logit and Cox regressions with controls. Reverse causality is checked by regressing risk measured before the car purchase on the car attributes.\",\"key_assumptions\":[\"Car attributes (sports car, horsepower, torque, minivan, safety) reveal an innate and durable sensation-seeking trait.\",\"The trait is stable over time, so it explains risk both before and after the car purchase.\",\"Controls absorb other fund-level drivers of risk and performance.\"],\"threats_addressed\":[\"Reverse causality from buying the car to risk tolerance\",\"Alternative explanations such as wealth effects (implied by the section on alternative explanations)\",\"Name-matching errors (ambiguous matches are dropped)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":10,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"subsample\",\"falsification\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variables (RISK, IDIORISK, SHARPE, ALPHA, FLOW) at length, along with the data sources and the pro- and antisensation attributes. Tables move from summary statistics to sorts, then multivariate regressions, then subsamples and robustness in a combined table. Standard error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"we argue\",\"more often than not\",\"consistent with\",\"in line with the view\",\"suggest\"]},\"claim_evidence\":[{\"claim\":\"Sports car owners run funds with about 1.8 pp higher annualized return volatility\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III\"},{\"claim\":\"Owners of prosensation cars deliver lower Sharpe ratios, information ratios and alphas\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Performance car owners show higher operational risk: more terminations, Form ADV violations and higher ω-Scores\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Sensation seekers trade more actively and unconventionally and prefer lottery-like stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Sensation-seeking fund-of-funds managers take more risk and load more on sensation-seeking hedge funds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"},{\"claim\":\"Flows to sensation-seeking funds are more sensitive to past performance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Car attributes also explain fund risk before the purchase, arguing against reverse causality\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X Panel E\"}],\"structural_summary\":\"The paper opens with a practitioner quote and frames personality-trait research as concentrated on retail investors. It introduces a novel proxy for sensation seeking based on car purchases, previews striking results with magnitudes, and then positions itself against several literatures in turn. Data construction is followed by results on financial risk, operational risk and trading behavior. A separate section turns to the investor side: fund-of-funds demand, flows, incentive alignment and perceptions. A section on alternative explanations and robustness addresses reverse causality. The conclusion restates the findings and frames them as validating allocator advice.\",\"writing_profile_keywords\":[\"novel hand-collected data\",\"personality proxy\",\"vivid framing\",\"magnitudes in introduction\",\"multi-literature positioning\",\"investor-side extension\",\"confident causal tone\"],\"confidence\":0.8,\"notes\":\"The introduction's paragraphs 0–2 are split from a quote; the first substantive paragraph is index 2. Statistical significance conventions and standard errors are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "hedge funds / investment management", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between a manager's sensation-seeking trait, proxied by the attributes of the car they buy, and fund risk, performance, operational risk and trading behavior.", "strategy": "Hand-collected vehicle purchase records are matched to hedge fund managers. The paper uses portfolio sorts and multivariate OLS, logit and Cox regressions with controls. Reverse causality is checked by regressing risk measured before the car purchase on the car attributes.", "key_assumptions": ["Car attributes (sports car, horsepower, torque, minivan, safety) reveal an innate and durable sensation-seeking trait.", "The trait is stable over time, so it explains risk both before and after the car purchase.", "Controls absorb other fund-level drivers of risk and performance."], "threats_addressed": ["Reverse causality from buying the car to risk tolerance", "Alternative explanations such as wealth effects (implied by the section on alternative explanations)", "Name-matching errors (ambiguous matches are dropped)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 2, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "contribution_statement"}], "contribution_paragraph_index": 10, "contribution_types": ["new_evidence", "new_data_or_measure", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["subsample", "falsification", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes define the dependent variables (RISK, IDIORISK, SHARPE, ALPHA, FLOW) at length, along with the data sources and the pro- and antisensation attributes. Tables move from summary statistics to sorts, then multivariate regressions, then subsamples and robustness in a combined table. Standard error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["we argue", "more often than not", "consistent with", "in line with the view", "suggest"]}, "claim_evidence": [{"claim": "Sports car owners run funds with about 1.8 pp higher annualized return volatility", "evidence_type": "table", "evidence_ref": "Tables II, III"}, {"claim": "Owners of prosensation cars deliver lower Sharpe ratios, information ratios and alphas", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Performance car owners show higher operational risk: more terminations, Form ADV violations and higher ω-Scores", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Sensation seekers trade more actively and unconventionally and prefer lottery-like stocks", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Sensation-seeking fund-of-funds managers take more risk and load more on sensation-seeking hedge funds", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}, {"claim": "Flows to sensation-seeking funds are more sensitive to past performance", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Car attributes also explain fund risk before the purchase, arguing against reverse causality", "evidence_type": "table", "evidence_ref": "Table X Panel E"}], "structural_summary": "The paper opens with a practitioner quote and frames personality-trait research as concentrated on retail investors. It introduces a novel proxy for sensation seeking based on car purchases, previews striking results with magnitudes, and then positions itself against several literatures in turn. Data construction is followed by results on financial risk, operational risk and trading behavior. A separate section turns to the investor side: fund-of-funds demand, flows, incentive alignment and perceptions. A section on alternative explanations and robustness addresses reverse causality. The conclusion restates the findings and frames them as validating allocator advice.", "writing_profile_keywords": ["novel hand-collected data", "personality proxy", "vivid framing", "magnitudes in introduction", "multi-literature positioning", "investor-side extension", "confident causal tone"], "confidence": 0.8, "notes": "The introduction's paragraphs 0–2 are split from a quote; the first substantive paragraph is index 2. Statistical significance conventions and standard errors are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.9, "usage": {"input": 7557, "output": 2191, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074842, "request_id": "req_011Cggq9dktfbhb85ekoojAD", "cached": false, "ts": "2026-11-03T23:04:44+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/6c16d6e4b905758791709941f97f8c9c578b541e63f152ff29bbee0ab842c3d9.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of a lender's retained share of outstanding mortgages in a local area on the probability of foreclosing on 90+ day delinquent loans, and of zip-code concentration of outstanding mortgages on foreclosure rates and house price changes\",\"strategy\":\"Loan-level linear probability models with lender and location fixed effects, estimated conditional on serious delinquency; 2SLS using mergers of large nonfailing banks as an instrument for retained share and concentration; zip-code cross-sectional regressions with county fixed effects; heterogeneity across states with different foreclosure procedures\",\"key_assumptions\":[\"Mergers of large nonfailing banks are unrelated to local market characteristics (exclusion/exogeneity)\",\"Conditioning on 90+ day delinquency removes lenders' prior informational advantage\",\"Zip codes are the relevant area over which foreclosure externalities operate\"],\"threats_addressed\":[\"Lender informational advantage in census tracts where it retains more mortgages\",\"Unobservable zip code or lender characteristics\",\"Alternative mechanisms such as lender organizational capabilities to foreclose or renegotiate\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Each table note names the estimator (OLS or 2SLS), the dependent variable and its window (2007–2010), and the variable of interest measured over 2004–2006. Notes refer back to Table I for variable definitions and state which fixed effects are included (e.g., county fixed effects). First-stage results are reported in separate columns.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\",\"figures_note_placeholder_removed\":null},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'evidence consistent with' phrasing\",\"'arguably' qualifiers\",\"framing hypotheses as conjectures\",\"'may' and 'suggest' in implications\"]},\"claim_evidence\":[{\"claim\":\"Lenders retaining a larger share of local outstanding mortgages are less likely to foreclose on seriously delinquent loans\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The foreclosure result holds and is quantitatively stronger when retained share is instrumented with bank mergers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"The retained-share effect is weaker where other lenders retain more, and differs in desirable versus distressed areas\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Lenders with higher retained shares are more likely to modify delinquent mortgages\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Zip codes with higher Top4 concentration have lower foreclosure rates\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"A one-SD increase in concentration is associated with about 3% smaller house price declines\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Lenders' foreclosure decisions are strategic substitutes and respond to own retained share\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.B\"}],\"structural_summary\":\"The paper opens with the fire-sale literature and notes that lenders' ex post incentives have received little attention. It states its conjecture and gives reasons why the housing crisis is a good testing ground. A simple model then derives testable implications about lender concentration, foreclosures and prices. Loan-level LPM and IV tests follow, along with cross-sectional heterogeneity and mechanism tests (modifications, alternative channels). The paper then moves to zip-code-level aggregate tests of foreclosure rates and house prices using the merger instrument. It closes with policy implications for consolidation and bad banks, and calls for future research on other collateralized markets.\",\"writing_profile_keywords\":[\"fire-sale externalities\",\"model-motivated empirics\",\"instrumental variables with bank mergers\",\"loan-level and aggregate tests\",\"policy implications\",\"cautious causal language\"],\"confidence\":0.8,\"notes\":\"Standard-error type and significance-reporting conventions are not visible in the excerpts. The introduction has no visible roadmap paragraph. The literature is positioned within the introduction, and no separate literature section is seen.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "household_finance", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of a lender's retained share of outstanding mortgages in a local area on the probability of foreclosing on 90+ day delinquent loans, and of zip-code concentration of outstanding mortgages on foreclosure rates and house price changes", "strategy": "Loan-level linear probability models with lender and location fixed effects, estimated conditional on serious delinquency; 2SLS using mergers of large nonfailing banks as an instrument for retained share and concentration; zip-code cross-sectional regressions with county fixed effects; heterogeneity across states with different foreclosure procedures", "key_assumptions": ["Mergers of large nonfailing banks are unrelated to local market characteristics (exclusion/exogeneity)", "Conditioning on 90+ day delinquency removes lenders' prior informational advantage", "Zip codes are the relevant area over which foreclosure externalities operate"], "threats_addressed": ["Lender informational advantage in census tracts where it retains more mortgages", "Unobservable zip code or lender characteristics", "Alternative mechanisms such as lender organizational capabilities to foreclose or renegotiate"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 1, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Each table note names the estimator (OLS or 2SLS), the dependent variable and its window (2007–2010), and the variable of interest measured over 2004–2006. Notes refer back to Table I for variable definitions and state which fixed effects are included (e.g., county fixed effects). First-stage results are reported in separate columns.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no", "figures_note_placeholder_removed": null}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'evidence consistent with' phrasing", "'arguably' qualifiers", "framing hypotheses as conjectures", "'may' and 'suggest' in implications"]}, "claim_evidence": [{"claim": "Lenders retaining a larger share of local outstanding mortgages are less likely to foreclose on seriously delinquent loans", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The foreclosure result holds and is quantitatively stronger when retained share is instrumented with bank mergers", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "The retained-share effect is weaker where other lenders retain more, and differs in desirable versus distressed areas", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Lenders with higher retained shares are more likely to modify delinquent mortgages", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Zip codes with higher Top4 concentration have lower foreclosure rates", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "A one-SD increase in concentration is associated with about 3% smaller house price declines", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Lenders' foreclosure decisions are strategic substitutes and respond to own retained share", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.B"}], "structural_summary": "The paper opens with the fire-sale literature and notes that lenders' ex post incentives have received little attention. It states its conjecture and gives reasons why the housing crisis is a good testing ground. A simple model then derives testable implications about lender concentration, foreclosures and prices. Loan-level LPM and IV tests follow, along with cross-sectional heterogeneity and mechanism tests (modifications, alternative channels). The paper then moves to zip-code-level aggregate tests of foreclosure rates and house prices using the merger instrument. It closes with policy implications for consolidation and bad banks, and calls for future research on other collateralized markets.", "writing_profile_keywords": ["fire-sale externalities", "model-motivated empirics", "instrumental variables with bank mergers", "loan-level and aggregate tests", "policy implications", "cautious causal language"], "confidence": 0.8, "notes": "Standard-error type and significance-reporting conventions are not visible in the excerpts. The introduction has no visible roadmap paragraph. The literature is positioned within the introduction, and no separate literature section is seen."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.07, "usage": {"input": 6067, "output": 2216, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069382, "request_id": "req_011CggpZDf7Ywsc1Kb3AhxNU", "cached": false, "ts": "2026-11-03T22:56:59+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/6c3e6bdd674ba37c6af180ae0b666d6247dddeac6623dd32c00bab519f0feb18.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Conditional minimum variance projection of the pricing kernel on tradeable realized moments of market returns, together with the implied physical moments of returns\",\"strategy\":\"Max-min Sharpe ratio optimization over Hansen-Jagannathan-type kernel projections that price tradeable realized-moment payoffs. Admissible projections are constrained by sign restrictions on risk premia, which yield bounds on physical moments. Inputs are S&P 500 option prices.\",\"key_assumptions\":[\"Arbitrage-free option markets\",\"Sign restrictions on covariances of returns with the pricing kernel (NCC conditions)\",\"Sign restrictions on risk premia of variance and higher-moment payoffs (NDP conditions)\",\"Optional SR constraint: the variance Sharpe ratio is at least the market Sharpe ratio\"],\"threats_addressed\":[\"Non-identifiability of physical beliefs without restrictions (the model-based recovery critique)\",\"Empirical validity of the moment-bound assumptions, tested with nonparametric and model-based tests\",\"Incomplete option markets with bounded moneyness, handled by corridor payoffs\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_method\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"theory\"},{\"idx\":8,\"role\":\"theory\"},{\"idx\":9,\"role\":\"results\"},{\"idx\":10,\"role\":\"conclusion\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"},{\"idx\":13,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"falsification\"]},\"tables\":{\"convention_notes\":\"There are only two tables. Predictive regressions report coefficients with standard errors in square brackets. Out-of-sample R2s are defined relative to a sample-mean forecast, following Campbell-Thompson. Most results are conveyed through time-series figures.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'(almost) model-free' qualifier\",\"'mild assumptions' framing\",\"'often/usually U-shaped'\",\"'can be' and 'potentially useful'\"]},\"claim_evidence\":[{\"claim\":\"Recovered first and second moments predict future returns and realized second moments, both in-sample and out-of-sample\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I and II\"},{\"claim\":\"Recovered equity premia are strongly countercyclical, ranging from under 1% in calm periods to about 22% (monthly horizon) during the 2008 crisis\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2 and 4\"},{\"claim\":\"MVPs are usually U-shaped at horizons above one quarter; short-horizon MVPs can be downward-sloping unless the SR constraint is imposed\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 6-9\"},{\"claim\":\"Optimal portfolios that short the MVP have countercyclical Sharpe ratios that improve on the market Sharpe ratio\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 10\"},{\"claim\":\"The sign assumptions underlying the moment bounds are empirically supported\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 11-13\"},{\"claim\":\"Model-based recovery cannot identify physical beliefs without path-independence assumptions\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Intro para 1 (Hansen-Scheinkman and the recovery literature)\"}],\"structural_summary\":\"The paper opens from the Breeden-Litzenberger benchmark and the limits of model-based recovery, then reframes the question as recovering a minimum variance kernel projection under mild sign restrictions. Theory sections build intuition in a Black-Scholes setting and then generalize to HJ projections and moment bounds. A long empirical section on S&P 500 options covers recovered moments, predictability, risk premia, kernel shape, trading strategies, and tests of the assumptions. Proofs, implementation details, and affine-model illustrations are placed in appendices.\",\"writing_profile_keywords\":[\"option-implied recovery\",\"pricing kernel projection\",\"minimum variance\",\"model-free bounds\",\"figure-heavy empirics\",\"theory-then-application\"],\"confidence\":0.72,\"notes\":\"Several section entries (idx 3-7) are equation fragments created by parsing errors and belong to Section II. Standard error type and significance conventions are not visible in the excerpts. The primary method is a custom optimization-based recovery, so it is coded as 'other'.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "other", "identification": {"estimand": "Conditional minimum variance projection of the pricing kernel on tradeable realized moments of market returns, together with the implied physical moments of returns", "strategy": "Max-min Sharpe ratio optimization over Hansen-Jagannathan-type kernel projections that price tradeable realized-moment payoffs. Admissible projections are constrained by sign restrictions on risk premia, which yield bounds on physical moments. Inputs are S&P 500 option prices.", "key_assumptions": ["Arbitrage-free option markets", "Sign restrictions on covariances of returns with the pricing kernel (NCC conditions)", "Sign restrictions on risk premia of variance and higher-moment payoffs (NDP conditions)", "Optional SR constraint: the variance Sharpe ratio is at least the market Sharpe ratio"], "threats_addressed": ["Non-identifiability of physical beliefs without restrictions (the model-based recovery critique)", "Empirical validity of the moment-bound assumptions, tested with nonparametric and model-based tests", "Incomplete option markets with bounded moneyness, handled by corridor payoffs"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_method", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "theory"}, {"idx": 8, "role": "theory"}, {"idx": 9, "role": "results"}, {"idx": 10, "role": "conclusion"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}, {"idx": 13, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "falsification"]}, "tables": {"convention_notes": "There are only two tables. Predictive regressions report coefficients with standard errors in square brackets. Out-of-sample R2s are defined relative to a sample-mean forecast, following Campbell-Thompson. Most results are conveyed through time-series figures.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'(almost) model-free' qualifier", "'mild assumptions' framing", "'often/usually U-shaped'", "'can be' and 'potentially useful'"]}, "claim_evidence": [{"claim": "Recovered first and second moments predict future returns and realized second moments, both in-sample and out-of-sample", "evidence_type": "table", "evidence_ref": "Tables I and II"}, {"claim": "Recovered equity premia are strongly countercyclical, ranging from under 1% in calm periods to about 22% (monthly horizon) during the 2008 crisis", "evidence_type": "figure", "evidence_ref": "Figures 2 and 4"}, {"claim": "MVPs are usually U-shaped at horizons above one quarter; short-horizon MVPs can be downward-sloping unless the SR constraint is imposed", "evidence_type": "figure", "evidence_ref": "Figures 6-9"}, {"claim": "Optimal portfolios that short the MVP have countercyclical Sharpe ratios that improve on the market Sharpe ratio", "evidence_type": "figure", "evidence_ref": "Figure 10"}, {"claim": "The sign assumptions underlying the moment bounds are empirically supported", "evidence_type": "figure", "evidence_ref": "Figures 11-13"}, {"claim": "Model-based recovery cannot identify physical beliefs without path-independence assumptions", "evidence_type": "prior_literature", "evidence_ref": "Intro para 1 (Hansen-Scheinkman and the recovery literature)"}], "structural_summary": "The paper opens from the Breeden-Litzenberger benchmark and the limits of model-based recovery, then reframes the question as recovering a minimum variance kernel projection under mild sign restrictions. Theory sections build intuition in a Black-Scholes setting and then generalize to HJ projections and moment bounds. A long empirical section on S&P 500 options covers recovered moments, predictability, risk premia, kernel shape, trading strategies, and tests of the assumptions. Proofs, implementation details, and affine-model illustrations are placed in appendices.", "writing_profile_keywords": ["option-implied recovery", "pricing kernel projection", "minimum variance", "model-free bounds", "figure-heavy empirics", "theory-then-application"], "confidence": 0.72, "notes": "Several section entries (idx 3-7) are equation fragments created by parsing errors and belong to Section II. Standard error type and significance conventions are not visible in the excerpts. The primary method is a custom optimization-based recovery, so it is coded as 'other'."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.67, "usage": {"input": 8140, "output": 2318, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079714, "request_id": "req_011CggqCAZMoNQNqy4vSvwLL", "cached": false, "ts": "2026-11-03T23:05:20+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/6f086478c29ee5b996c862f7898193f143fbe6fea089a9457b97c5ac6543aff9.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"credit derivatives / rare disasters\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Model-implied CDX and CDX tranche spreads (levels and time series) compared with observed spreads\",\"strategy\":\"Calibrate an equilibrium rare-disaster model to consumption data and the equity premium. Extract latent state variables from index option implied volatilities. Compare predicted tranche spreads with data and run comparative statics.\",\"key_assumptions\":[\"Representative agent with complete markets in an endowment economy\",\"Time-varying disaster probability governed by square-root processes\",\"Firm cash flows load on aggregate disasters, sector shocks and idiosyncratic shocks\",\"Preference and consumption parameters taken from prior studies without reference to CDX data\"],\"threats_addressed\":[\"Model is not fit directly to CDX data\",\"Disaster risk is linked tightly to the equity premium, so a low-equity-premium recalibration fails\",\"Removing idiosyncratic risk or sector shocks changes the fit\",\"Physical default probabilities are matched to Moody's data\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Tables report calibrated parameters and data-versus-model moments. Simulated small-sample percentiles (5th, 50th, 95th) are shown alongside population moments. Tranche spreads are given in basis points per year, with the equity tranche reported as an upfront payment. Pre-crisis and crisis subsamples are reported separately.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"framing as 'can explain' or 'is possible to account for'\",\"'at least some of the pricing behavior'\",\"explicit acknowledgment of model failures as challenges for future work\"]},\"claim_evidence\":[{\"claim\":\"Calibrated disaster model matches average five- and three-year CDX tranche spreads before and during the crisis\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI, VII\"},{\"claim\":\"Model reproduces the time series of tranche spreads using option-implied state variables\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5, 6\"},{\"claim\":\"Lower equity premium calibration produces senior tranche spreads far below the data\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"},{\"claim\":\"Model matches physical default probabilities across horizons\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Model cannot reconcile low implied-volatility autocorrelation with a persistent price-dividend ratio\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 10, Table IV\"},{\"claim\":\"Model predicts counterfactually low long-term real bond yields\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 11\"},{\"claim\":\"Model matches equity premium and bill rate moments\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"}],\"structural_summary\":\"The paper motivates the question with dramatic crisis-era senior tranche pricing and a gap: few quantitative models link structured products to fundamentals. It then positions itself against no-arbitrage models, whose limitation is incomplete markets, and turns to an equilibrium model. Next it sketches the mechanism tying the equity premium to tranche prices. A theory section presents a continuous-time disaster model, followed by data, calibration (including option-based extraction of state variables), and an evaluation of averages, time series and comparative statics. It then examines implications for options, the price-dividend ratio and bonds, honestly reporting where the model fails. The conclusion restates the fit and lists remaining puzzles.\",\"writing_profile_keywords\":[\"rare disasters\",\"equilibrium calibration\",\"model-versus-data comparison\",\"credit derivatives\",\"comparative statics\",\"acknowledged model limitations\",\"2008 crisis\"],\"confidence\":0.82,\"notes\":\"Calibration and evaluation sections are labeled 'other' in the source and were mapped here to empirical_design and results. No regression-based inference appears. Uncertainty is reported via simulated percentiles in Table II. The introduction excerpt is truncated, so a roadmap may exist but is not visible.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "credit derivatives / rare disasters", "primary_method": "calibration_simulation", "identification": {"estimand": "Model-implied CDX and CDX tranche spreads (levels and time series) compared with observed spreads", "strategy": "Calibrate an equilibrium rare-disaster model to consumption data and the equity premium. Extract latent state variables from index option implied volatilities. Compare predicted tranche spreads with data and run comparative statics.", "key_assumptions": ["Representative agent with complete markets in an endowment economy", "Time-varying disaster probability governed by square-root processes", "Firm cash flows load on aggregate disasters, sector shocks and idiosyncratic shocks", "Preference and consumption parameters taken from prior studies without reference to CDX data"], "threats_addressed": ["Model is not fit directly to CDX data", "Disaster risk is linked tightly to the equity premium, so a low-equity-premium recalibration fails", "Removing idiosyncratic risk or sector shocks changes the fit", "Physical default probabilities are matched to Moody's data"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 4, "contribution_types": ["new_theory_or_model", "new_evidence", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["sensitivity_to_parameters", "alternative_specification"]}, "tables": {"convention_notes": "Tables report calibrated parameters and data-versus-model moments. Simulated small-sample percentiles (5th, 50th, 95th) are shown alongside population moments. Tranche spreads are given in basis points per year, with the equity tranche reported as an upfront payment. Pre-crisis and crisis subsamples are reported separately.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["framing as 'can explain' or 'is possible to account for'", "'at least some of the pricing behavior'", "explicit acknowledgment of model failures as challenges for future work"]}, "claim_evidence": [{"claim": "Calibrated disaster model matches average five- and three-year CDX tranche spreads before and during the crisis", "evidence_type": "table", "evidence_ref": "Tables V, VI, VII"}, {"claim": "Model reproduces the time series of tranche spreads using option-implied state variables", "evidence_type": "figure", "evidence_ref": "Figures 5, 6"}, {"claim": "Lower equity premium calibration produces senior tranche spreads far below the data", "evidence_type": "figure", "evidence_ref": "Figure 9"}, {"claim": "Model matches physical default probabilities across horizons", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Model cannot reconcile low implied-volatility autocorrelation with a persistent price-dividend ratio", "evidence_type": "figure", "evidence_ref": "Figure 10, Table IV"}, {"claim": "Model predicts counterfactually low long-term real bond yields", "evidence_type": "figure", "evidence_ref": "Figure 11"}, {"claim": "Model matches equity premium and bill rate moments", "evidence_type": "table", "evidence_ref": "Table II"}], "structural_summary": "The paper motivates the question with dramatic crisis-era senior tranche pricing and a gap: few quantitative models link structured products to fundamentals. It then positions itself against no-arbitrage models, whose limitation is incomplete markets, and turns to an equilibrium model. Next it sketches the mechanism tying the equity premium to tranche prices. A theory section presents a continuous-time disaster model, followed by data, calibration (including option-based extraction of state variables), and an evaluation of averages, time series and comparative statics. It then examines implications for options, the price-dividend ratio and bonds, honestly reporting where the model fails. The conclusion restates the fit and lists remaining puzzles.", "writing_profile_keywords": ["rare disasters", "equilibrium calibration", "model-versus-data comparison", "credit derivatives", "comparative statics", "acknowledged model limitations", "2008 crisis"], "confidence": 0.82, "notes": "Calibration and evaluation sections are labeled 'other' in the source and were mapped here to empirical_design and results. No regression-based inference appears. Uncertainty is reported via simulated percentiles in Table II. The introduction excerpt is truncated, so a roadmap may exist but is not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.34, "usage": {"input": 6793, "output": 2217, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072306, "request_id": "req_011Cggq5yonQUcEbrsPLyUME", "cached": false, "ts": "2026-11-03T23:03:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/700cd6eede478c4dcb495bcf44e8246b7ba9a3e783316432ff0f92f05548bfc3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"experimental finance; bond markets / credit risk\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"How far experimental IPO and secondary-market bond prices deviate from the equilibrium fundamental value, and whether that deviation shrinks across repeated rounds\",\"strategy\":\"Controlled laboratory experiments with between-subject treatments (decreasing vs. increasing fundamental value path in Experiment I; structural-form model in Experiment II). Independent groups are compared with model-derived equilibrium benchmarks using nonparametric Wilcoxon signed-rank tests.\",\"key_assumptions\":[\"Groups do not interact, so group-level observations are statistically independent\",\"The equilibrium computed from the models is the correct benchmark\",\"Laboratory control isolates the feedback effect from confounds present in field markets\"],\"threats_addressed\":[\"Field confounds such as the unknown link between default risk and IPO price, auction rules, trader set and information, handled through laboratory control\",\"Learning versus initial mispricing, separated by comparing Round 1 with Round 4\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":2},\"gap\":{\"present\":true,\"sentence_index\":3},\"method\":{\"present\":true,\"sentence_index\":4},\"results\":{\"present\":true,\"sentence_index\":6},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"p_values\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Tables report round-by-round means and medians of IPO prices and of the relative absolute deviation (RAD) and relative deviation (RD) from equilibrium, with equilibrium values given alongside for comparison. Companion tables report p-values from two-sided Wilcoxon signed-rank tests, with each null hypothesis stated in the notes.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\",\"notes_placeholder_removed\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'it is likely that' when extrapolating from the lab to the field\",\"conditioning results on subject experience\",\"'a priori unclear' framing of the question\",\"explicit acknowledgment that field tests lack control\"]},\"claim_evidence\":[{\"claim\":\"Subjects learn to set IPO prices close to equilibrium after a few rounds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-II, Figure 8\"},{\"claim\":\"Deviations of market prices from fundamentals shrink with experience and bubbles disappear\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-IV, Figure 7\"},{\"claim\":\"No prior experiment studies bonds whose default probability depends on the IPO price\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Intro paragraphs 1-3; surveys by Bossaerts, Noussair & Tucker, Palan\"},{\"claim\":\"Equilibrium fundamental values and default probabilities follow from the reduced-form and structural models\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Figures 1-2, 5-6\"}],\"structural_summary\":\"The introduction motivates the topic through the feedback from IPO prices to default risk and claims the first experimental study of such assets. It then positions the paper against experimental asset-market work, the contingent-convertible feedback experiment, and research on uniform-price multiunit auctions, without a visible roadmap. Next come two lab-implementable models, reduced-form and structural, with numerically computed equilibria. The two experiments with their treatments are described, followed by results comparing IPO and market prices with equilibrium using nonparametric tests on independent groups. The conclusion restates the convergence finding, suggests implications for experimental methodology, and cautiously extrapolates to field markets.\",\"writing_profile_keywords\":[\"laboratory experiment\",\"feedback effects\",\"equilibrium benchmark\",\"learning across rounds\",\"nonparametric tests\",\"first-study claim\",\"editor-suggested extension\"],\"confidence\":0.8,\"notes\":\"The second experiment and structural model were added at the editors' suggestion, as acknowledged in the text. No explicit robustness section is visible. The figures object includes an extra field that should not be there; it is ignored here.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "experimental finance; bond markets / credit risk", "primary_method": "survey_or_experiment", "identification": {"estimand": "How far experimental IPO and secondary-market bond prices deviate from the equilibrium fundamental value, and whether that deviation shrinks across repeated rounds", "strategy": "Controlled laboratory experiments with between-subject treatments (decreasing vs. increasing fundamental value path in Experiment I; structural-form model in Experiment II). Independent groups are compared with model-derived equilibrium benchmarks using nonparametric Wilcoxon signed-rank tests.", "key_assumptions": ["Groups do not interact, so group-level observations are statistically independent", "The equilibrium computed from the models is the correct benchmark", "Laboratory control isolates the feedback effect from confounds present in field markets"], "threats_addressed": ["Field confounds such as the unknown link between default risk and IPO price, auction rules, trader set and information, handled through laboratory control", "Learning versus initial mispricing, separated by comparing Round 1 with Round 4"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 2}, "gap": {"present": true, "sentence_index": 3}, "method": {"present": true, "sentence_index": 4}, "results": {"present": true, "sentence_index": 6}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "related_literature_positioning"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "p_values", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Tables report round-by-round means and medians of IPO prices and of the relative absolute deviation (RAD) and relative deviation (RD) from equilibrium, with equilibrium values given alongside for comparison. Companion tables report p-values from two-sided Wilcoxon signed-rank tests, with each null hypothesis stated in the notes.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no", "notes_placeholder_removed": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'it is likely that' when extrapolating from the lab to the field", "conditioning results on subject experience", "'a priori unclear' framing of the question", "explicit acknowledgment that field tests lack control"]}, "claim_evidence": [{"claim": "Subjects learn to set IPO prices close to equilibrium after a few rounds", "evidence_type": "table", "evidence_ref": "Tables I-II, Figure 8"}, {"claim": "Deviations of market prices from fundamentals shrink with experience and bubbles disappear", "evidence_type": "table", "evidence_ref": "Tables III-IV, Figure 7"}, {"claim": "No prior experiment studies bonds whose default probability depends on the IPO price", "evidence_type": "prior_literature", "evidence_ref": "Intro paragraphs 1-3; surveys by Bossaerts, Noussair & Tucker, Palan"}, {"claim": "Equilibrium fundamental values and default probabilities follow from the reduced-form and structural models", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Figures 1-2, 5-6"}], "structural_summary": "The introduction motivates the topic through the feedback from IPO prices to default risk and claims the first experimental study of such assets. It then positions the paper against experimental asset-market work, the contingent-convertible feedback experiment, and research on uniform-price multiunit auctions, without a visible roadmap. Next come two lab-implementable models, reduced-form and structural, with numerically computed equilibria. The two experiments with their treatments are described, followed by results comparing IPO and market prices with equilibrium using nonparametric tests on independent groups. The conclusion restates the convergence finding, suggests implications for experimental methodology, and cautiously extrapolates to field markets.", "writing_profile_keywords": ["laboratory experiment", "feedback effects", "equilibrium benchmark", "learning across rounds", "nonparametric tests", "first-study claim", "editor-suggested extension"], "confidence": 0.8, "notes": "The second experiment and structural model were added at the editors' suggestion, as acknowledged in the text. No explicit robustness section is visible. The figures object includes an extra field that should not be there; it is ignored here."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.1, "usage": {"input": 6803, "output": 1938, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066766, "request_id": "req_011Cggq2Af3QB3oknBMjycR1", "cached": false, "ts": "2026-11-03T23:03:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/717c143d09808825d1abd93d9666ae7578fadcbcb4797fdb9520c8fd7244b203.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Age profiles of stock market participation, conditional risky share and financial wealth, plus structural preference and cost parameters (risk aversion, participation cost, crash probability)\",\"strategy\":\"Heckman selection model to handle endogenous participation, cohort effects separated with identification restrictions (Deaton-Paxson zero-sum time effects, GDP-trend proxies), and moment-matching structural estimation of an extended life-cycle model\",\"key_assumptions\":[\"Exclusion and selection restrictions in the Heckman model\",\"Time/age/cohort identification restrictions such as zero-sum time effects\",\"Administrative tax records are free of measurement error with little tax evasion\",\"Model specification: CRRA preferences, labor income process, crash probability\"],\"threats_addressed\":[\"endogeneity of participation\",\"separation of age, time and cohort effects\",\"measurement error and underreporting\",\"wealth-mortality correlation\",\"attrition\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"literature_gap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"internet_appendix\",\"types\":[\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes define variables and samples and state that monetary values are in 1995 USD. The parameterization table lists each value with its source. Some tables report t-values in parentheses. The structural estimation table describes the targeted moments.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"we argue that this is likely\",\"seem to depart\",\"suggests\",\"consistent with model predictions\"]},\"claim_evidence\":[{\"claim\":\"Households reduce the conditional risky share by just under 1 pp per year from about age 45 until retirement.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5; Table III\"},{\"claim\":\"Participation rises to about 60% by age 45 and households exit the stock market after retirement.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1, 2, 5, 6\"},{\"claim\":\"Tax-registry data minimize measurement error and tax evasion is unlikely.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section I\"},{\"claim\":\"Best-fitting parameters combine high risk aversion, a small participation cost and a crash probability in line with Norwegian history.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figures 11-12; Appendix C\"},{\"claim\":\"Existing calibrated models predict rebalancing toward safer portfolios with age but not stock market exit.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Merton 1971; Cocco, Gomes, and Maenhout 2005\"}],\"structural_summary\":\"The paper opens by claiming new stylized facts and a model that fits them. It then sets out the life-cycle theory prediction and its mismatch with micro data, and presents the Norwegian registry data as the remedy. Results are previewed with magnitudes. Descriptive cohort evidence comes next, followed by an econometric treatment of selection and cohort effects. An extended life-cycle model with a participation cost and crash risk follows, then policy functions, simulations and structural estimation. The conclusion argues the data are consistent with model predictions.\",\"writing_profile_keywords\":[\"stylized facts\",\"administrative data\",\"life-cycle model\",\"structural estimation\",\"Heckman selection\",\"cohort effects\",\"double adjustment\"],\"confidence\":0.8,\"notes\":\"The introduction and conclusion are truncated in the excerpts. No robustness section is visible; the internet appendix is referenced for estimation details, so the robustness fields are tentative. Significance conventions are not clearly stated beyond t-values in Table A.I.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "asset_pricing", "primary_method": "structural_estimation", "identification": {"estimand": "Age profiles of stock market participation, conditional risky share and financial wealth, plus structural preference and cost parameters (risk aversion, participation cost, crash probability)", "strategy": "Heckman selection model to handle endogenous participation, cohort effects separated with identification restrictions (Deaton-Paxson zero-sum time effects, GDP-trend proxies), and moment-matching structural estimation of an extended life-cycle model", "key_assumptions": ["Exclusion and selection restrictions in the Heckman model", "Time/age/cohort identification restrictions such as zero-sum time effects", "Administrative tax records are free of measurement error with little tax evasion", "Model specification: CRRA preferences, labor income process, crash probability"], "threats_addressed": ["endogeneity of participation", "separation of age, time and cohort effects", "measurement error and underreporting", "wealth-mortality correlation", "attrition"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "literature_gap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_data_or_measure", "new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "internet_appendix", "types": ["alternative_specification"]}, "tables": {"convention_notes": "Notes define variables and samples and state that monetary values are in 1995 USD. The parameterization table lists each value with its source. Some tables report t-values in parentheses. The structural estimation table describes the targeted moments.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["we argue that this is likely", "seem to depart", "suggests", "consistent with model predictions"]}, "claim_evidence": [{"claim": "Households reduce the conditional risky share by just under 1 pp per year from about age 45 until retirement.", "evidence_type": "figure", "evidence_ref": "Figure 5; Table III"}, {"claim": "Participation rises to about 60% by age 45 and households exit the stock market after retirement.", "evidence_type": "figure", "evidence_ref": "Figures 1, 2, 5, 6"}, {"claim": "Tax-registry data minimize measurement error and tax evasion is unlikely.", "evidence_type": "none_stated", "evidence_ref": "Section I"}, {"claim": "Best-fitting parameters combine high risk aversion, a small participation cost and a crash probability in line with Norwegian history.", "evidence_type": "table", "evidence_ref": "Table V; Figures 11-12; Appendix C"}, {"claim": "Existing calibrated models predict rebalancing toward safer portfolios with age but not stock market exit.", "evidence_type": "prior_literature", "evidence_ref": "Merton 1971; Cocco, Gomes, and Maenhout 2005"}], "structural_summary": "The paper opens by claiming new stylized facts and a model that fits them. It then sets out the life-cycle theory prediction and its mismatch with micro data, and presents the Norwegian registry data as the remedy. Results are previewed with magnitudes. Descriptive cohort evidence comes next, followed by an econometric treatment of selection and cohort effects. An extended life-cycle model with a participation cost and crash risk follows, then policy functions, simulations and structural estimation. The conclusion argues the data are consistent with model predictions.", "writing_profile_keywords": ["stylized facts", "administrative data", "life-cycle model", "structural estimation", "Heckman selection", "cohort effects", "double adjustment"], "confidence": 0.8, "notes": "The introduction and conclusion are truncated in the excerpts. No robustness section is visible; the internet appendix is referenced for estimation details, so the robustness fields are tentative. Significance conventions are not clearly stated beyond t-values in Table A.I."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.17, "usage": {"input": 7618, "output": 2022, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071706, "request_id": "req_011CggpXYsCnTrsijZKtwQJX", "cached": false, "ts": "2026-11-03T22:56:35+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/73e80a718367c106243d0c33ec94c0360d0187638afc2b4fe339a0152cb46153.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"regression_discontinuity\",\"identification\":{\"estimand\":\"Average 'pure' effect of an implied financial covenant violation on board composition (log number of independent directors), holding financial performance constant, for firms with restrictive loan covenants\",\"strategy\":\"Discontinuity design around covenant thresholds: panel regressions (firm FE, first differences, OLS) of board composition on a violation dummy, controlling for polynomials in the distance to the covenant threshold, estimated within a bandwidth sample, following Chava and Roberts / Roberts and Sufi\",\"key_assumptions\":[\"Conditional on distance to the covenant threshold and controls, crossing the threshold is as good as random\",\"Financial performance and other confounders vary smoothly around the covenant threshold\",\"Covenant violations implied from contract and accounting data are measured accurately enough\"],\"threats_addressed\":[\"Confounding by deteriorating financial performance\",\"Choice of polynomial order and bandwidth\",\"Pre-existing differences across violation and non-violation firms (discontinuity-based exogeneity tests)\",\"Mismeasurement of implied covenant violations (SEC-DealScan matched sample)\",\"Spurious effects at false thresholds (placebo test)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":true,\"sentence_index\":0},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"other\"},{\"paragraph_index\":4,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"placebo\",\"alternative_measure\",\"sensitivity_to_parameters\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Table notes open by naming the estimator(s) (firm FE, first differences, OLS) and the dependent variable, then define the covenant violation dummy over four covenant types. A summary statistics table and a violation vs. nonviolation comparison within the bandwidth come first. Later tables cover polynomial/bandwidth sensitivity, a placebo test, an alternative matched sample, connected directors, director characteristics and firm policies. Standard error treatment is not visible in the truncated notes.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' interpretations of policy changes\",\"'suggests' / 'appear to'\",\"explicit acknowledgment that evidence does not settle whether creditors intervene directly\",\"alternative explanation offered (management-driven hiring)\"]},\"claim_evidence\":[{\"claim\":\"Independent directors increase by roughly a quarter after covenant violations\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table V\"},{\"claim\":\"Board independence rises visibly in raw data in the years after a violation\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"New directors do not replace incumbents, so board size rises; nonindependent directors are not reduced\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Post-violation directors are mostly connected to firms borrowing from the same banks, and connected directors drive the effect\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII, Table IX\"},{\"claim\":\"Firms that reform boards issue more equity, invest more, cut payout and risk, and shift CEO pay from cash toward equity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"The effect is absent at placebo thresholds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Creditors' governance influence is subtle and exercised behind the scenes\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Baird and Rasmussen (2006); Nini, Smith, and Sufi (2012)\"}],\"structural_summary\":\"The abstract is results-led, opening with a headline magnitude, then adding facts on director links and policy consequences, and closing with a broad conclusion. The introduction situates the paper in the creditor control rights literature and states the main finding early. It then poses follow-up questions answered in sequence (who the directors are, what they do). It acknowledges the limits on attributing changes to explicit creditor intervention and discusses why creditors might care about boards, with anecdotal support, before listing contributions. The body runs data, then a discontinuity-based methodology, then results starting from a raw-data graph. Results proceed through baseline, bandwidth/polynomial sensitivity, exogeneity tests, mismeasurement checks and robustness, followed by a mechanisms-and-consequences section. A short conclusion restates the findings with 'consistent with' framing.\",\"writing_profile_keywords\":[\"regression discontinuity\",\"covenant violations\",\"board independence\",\"creditor control rights\",\"question-driven introduction\",\"raw-data figure first\",\"mechanism follow-up\",\"acknowledged limits\"],\"confidence\":0.8,\"notes\":\"Section IV is labeled robustness by the rule but functions as mechanisms and consequences results. Figure 1 shows 95% confidence intervals; for Figure 2 (nonparametric fit) this is unclear, hence 'mixed'. An Internet Appendix exists, but which robustness checks sit there is not visible. No roadmap appears in the visible intro excerpts.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "regression_discontinuity", "identification": {"estimand": "Average 'pure' effect of an implied financial covenant violation on board composition (log number of independent directors), holding financial performance constant, for firms with restrictive loan covenants", "strategy": "Discontinuity design around covenant thresholds: panel regressions (firm FE, first differences, OLS) of board composition on a violation dummy, controlling for polynomials in the distance to the covenant threshold, estimated within a bandwidth sample, following Chava and Roberts / Roberts and Sufi", "key_assumptions": ["Conditional on distance to the covenant threshold and controls, crossing the threshold is as good as random", "Financial performance and other confounders vary smoothly around the covenant threshold", "Covenant violations implied from contract and accounting data are measured accurately enough"], "threats_addressed": ["Confounding by deteriorating financial performance", "Choice of polynomial order and bandwidth", "Pre-existing differences across violation and non-violation firms (discontinuity-based exogeneity tests)", "Mismeasurement of implied covenant violations (SEC-DealScan matched sample)", "Spurious effects at false thresholds (placebo test)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": true, "sentence_index": 0}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "other"}, {"paragraph_index": 4, "move": "tension_or_puzzle"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "placebo", "alternative_measure", "sensitivity_to_parameters", "falsification"]}, "tables": {"convention_notes": "Table notes open by naming the estimator(s) (firm FE, first differences, OLS) and the dependent variable, then define the covenant violation dummy over four covenant types. A summary statistics table and a violation vs. nonviolation comparison within the bandwidth come first. Later tables cover polynomial/bandwidth sensitivity, a placebo test, an alternative matched sample, connected directors, director characteristics and firm policies. Standard error treatment is not visible in the truncated notes.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' interpretations of policy changes", "'suggests' / 'appear to'", "explicit acknowledgment that evidence does not settle whether creditors intervene directly", "alternative explanation offered (management-driven hiring)"]}, "claim_evidence": [{"claim": "Independent directors increase by roughly a quarter after covenant violations", "evidence_type": "table", "evidence_ref": "Table III, Table V"}, {"claim": "Board independence rises visibly in raw data in the years after a violation", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "New directors do not replace incumbents, so board size rises; nonindependent directors are not reduced", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Post-violation directors are mostly connected to firms borrowing from the same banks, and connected directors drive the effect", "evidence_type": "table", "evidence_ref": "Table VIII, Table IX"}, {"claim": "Firms that reform boards issue more equity, invest more, cut payout and risk, and shift CEO pay from cash toward equity", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "The effect is absent at placebo thresholds", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Creditors' governance influence is subtle and exercised behind the scenes", "evidence_type": "prior_literature", "evidence_ref": "Baird and Rasmussen (2006); Nini, Smith, and Sufi (2012)"}], "structural_summary": "The abstract is results-led, opening with a headline magnitude, then adding facts on director links and policy consequences, and closing with a broad conclusion. The introduction situates the paper in the creditor control rights literature and states the main finding early. It then poses follow-up questions answered in sequence (who the directors are, what they do). It acknowledges the limits on attributing changes to explicit creditor intervention and discusses why creditors might care about boards, with anecdotal support, before listing contributions. The body runs data, then a discontinuity-based methodology, then results starting from a raw-data graph. Results proceed through baseline, bandwidth/polynomial sensitivity, exogeneity tests, mismeasurement checks and robustness, followed by a mechanisms-and-consequences section. A short conclusion restates the findings with 'consistent with' framing.", "writing_profile_keywords": ["regression discontinuity", "covenant violations", "board independence", "creditor control rights", "question-driven introduction", "raw-data figure first", "mechanism follow-up", "acknowledged limits"], "confidence": 0.8, "notes": "Section IV is labeled robustness by the rule but functions as mechanisms and consequences results. Figure 1 shows 95% confidence intervals; for Figure 2 (nonparametric fit) this is unclear, hence 'mixed'. An Internet Appendix exists, but which robustness checks sit there is not visible. No roadmap appears in the visible intro excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 24.38, "usage": {"input": 6092, "output": 2478, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074722, "request_id": "req_011Cggq6P2HdxqAktsPBKu7J", "cached": false, "ts": "2026-11-03T23:04:08+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/73fdb4bc2200559df23a5e9f188d43b390bc22dfc72c3e714dea4027fe95a7ce.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"OTC markets / fixed income (municipal bonds), network analysis\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between a dealer's network centrality and investor execution quality: round-trip markups, intermediation chain length, prearranged versus principal trading, and inventory duration.\",\"strategy\":\"Panel regressions of markups and trade outcomes on a composite centrality index (first principal component of network measures), with state and month fixed effects and rich dealer/trade/bond controls. Within-dealer variation comes from dealer fixed effects and state-specific centrality based on tax-segmented state markets. Additional tests address alternative channels: adverse selection, bond riskiness and inventory.\",\"key_assumptions\":[\"Conditional on controls, centrality is not a proxy for dealer size or unobserved dealer quality.\",\"Central dealers do not systematically intermediate a priori riskier or more illiquid bonds.\",\"State tax segmentation creates local submarkets in which state-specific centrality varies independently of aggregate dealer attributes.\"],\"threats_addressed\":[\"dealer size / scale effects\",\"bond riskiness and illiquidity selection\",\"inventory risk channel\",\"asymmetric information / adverse selection\",\"time-invariant dealer heterogeneity (dealer fixed effects)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"other\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Long descriptive table notes state the estimator (panel OLS with state and month fixed effects, Probit, Tobit), point to Appendix D for control definitions, define the centrality index as a first principal component, and report markups in percent. Tables progress from univariate tier sorts to multivariate regressions and then subsample tests.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' theory predictions\",\"'suggests that'\",\"framing of results as tests of hypotheses derived from models\"]},\"claim_evidence\":[{\"claim\":\"The interdealer network has a stable core-periphery structure with a small set of hubs and thousands of peripheral dealers.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Figure 2, Table II\"},{\"claim\":\"Central dealers charge investors higher round-trip markups than peripheral dealers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Table V\"},{\"claim\":\"Central dealers sell more directly to end customers and are associated with shorter intermediation chains.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Table VII\"},{\"claim\":\"Central dealers prearrange fewer trades and unwind inventory faster, implying more immediacy.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII, Table XIII\"},{\"claim\":\"Investors turn more to central dealers in stress times.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"The adverse-selection component of markups is lower for central dealers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"State-specific centrality affects markups more than aggregate centrality and adds explanatory power beyond dealer fixed effects.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XIV\"}],\"structural_summary\":\"The introduction motivates the role of dealers in OTC markets and notes how little empirical work exists for lack of data. It previews the regulatory data, the main results, a state-segmentation identification refinement, tests ruling out alternative explanations, and the contribution relative to the literature. A separate section then derives hypotheses from network models. The paper next describes the data and the round-trip chain construction. Successive results sections cover network structure, the centrality premium in markups, search efficiency, and execution speed and stress-time trading. A section on alternative theories (adverse selection, riskiness, inventory, local information, subsamples) follows. The conclusion restates the cost–speed trade-off and the systemic role of central dealers.\",\"writing_profile_keywords\":[\"hypothesis-driven empirics\",\"theory-motivated tests\",\"network centrality\",\"regulatory audit trail data\",\"alternative explanations section\",\"OTC market microstructure\",\"descriptive-then-multivariate tables\"],\"confidence\":0.8,\"notes\":\"Section VII is labelled 'theory' by the role rule but functions as robustness/alternative explanations, so it was assigned robustness. Standard error treatment and significance notation are not visible in the excerpts. No roadmap is visible in the excerpted introduction paragraphs.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "OTC markets / fixed income (municipal bonds), network analysis", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between a dealer's network centrality and investor execution quality: round-trip markups, intermediation chain length, prearranged versus principal trading, and inventory duration.", "strategy": "Panel regressions of markups and trade outcomes on a composite centrality index (first principal component of network measures), with state and month fixed effects and rich dealer/trade/bond controls. Within-dealer variation comes from dealer fixed effects and state-specific centrality based on tax-segmented state markets. Additional tests address alternative channels: adverse selection, bond riskiness and inventory.", "key_assumptions": ["Conditional on controls, centrality is not a proxy for dealer size or unobserved dealer quality.", "Central dealers do not systematically intermediate a priori riskier or more illiquid bonds.", "State tax segmentation creates local submarkets in which state-specific centrality varies independently of aggregate dealer attributes."], "threats_addressed": ["dealer size / scale effects", "bond riskiness and illiquidity selection", "inventory risk channel", "asymmetric information / adverse selection", "time-invariant dealer heterogeneity (dealer fixed effects)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "other"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "subsample", "alternative_measure", "falsification", "other"]}, "tables": {"convention_notes": "Long descriptive table notes state the estimator (panel OLS with state and month fixed effects, Probit, Tobit), point to Appendix D for control definitions, define the centrality index as a first principal component, and report markups in percent. Tables progress from univariate tier sorts to multivariate regressions and then subsample tests.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' theory predictions", "'suggests that'", "framing of results as tests of hypotheses derived from models"]}, "claim_evidence": [{"claim": "The interdealer network has a stable core-periphery structure with a small set of hubs and thousands of peripheral dealers.", "evidence_type": "figure", "evidence_ref": "Figure 1, Figure 2, Table II"}, {"claim": "Central dealers charge investors higher round-trip markups than peripheral dealers.", "evidence_type": "table", "evidence_ref": "Table IV, Table V"}, {"claim": "Central dealers sell more directly to end customers and are associated with shorter intermediation chains.", "evidence_type": "table", "evidence_ref": "Table VI, Table VII"}, {"claim": "Central dealers prearrange fewer trades and unwind inventory faster, implying more immediacy.", "evidence_type": "table", "evidence_ref": "Table VIII, Table XIII"}, {"claim": "Investors turn more to central dealers in stress times.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "The adverse-selection component of markups is lower for central dealers.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "State-specific centrality affects markups more than aggregate centrality and adds explanatory power beyond dealer fixed effects.", "evidence_type": "table", "evidence_ref": "Table XIV"}], "structural_summary": "The introduction motivates the role of dealers in OTC markets and notes how little empirical work exists for lack of data. It previews the regulatory data, the main results, a state-segmentation identification refinement, tests ruling out alternative explanations, and the contribution relative to the literature. A separate section then derives hypotheses from network models. The paper next describes the data and the round-trip chain construction. Successive results sections cover network structure, the centrality premium in markups, search efficiency, and execution speed and stress-time trading. A section on alternative theories (adverse selection, riskiness, inventory, local information, subsamples) follows. The conclusion restates the cost–speed trade-off and the systemic role of central dealers.", "writing_profile_keywords": ["hypothesis-driven empirics", "theory-motivated tests", "network centrality", "regulatory audit trail data", "alternative explanations section", "OTC market microstructure", "descriptive-then-multivariate tables"], "confidence": 0.8, "notes": "Section VII is labelled 'theory' by the role rule but functions as robustness/alternative explanations, so it was assigned robustness. Standard error treatment and significance notation are not visible in the excerpts. No roadmap is visible in the excerpted introduction paragraphs."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.98, "usage": {"input": 9337, "output": 2419, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086522, "request_id": "req_011CggqEUyCaGmKuyx6hjE3n", "cached": false, "ts": "2026-11-03T23:05:53+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/740e13c7b987eff9e95784627fc42bc9325408b587b2032e295f09d8a4766656.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of a bank's proprietary sales of a stock on its retail customers' purchases of that stock, and the resulting performance consequences for those customers\",\"strategy\":\"Bank-stock-quarter panel regressions relating changes in customer holdings to changes in bank holdings. Causal interpretation is strengthened by forced sales at banks under 2008 SoFFin rescue conditions. Performance is compared through trading profit regressions and portfolio return tests across banks with and without proprietary trading desks.\",\"key_assumptions\":[\"Sales forced by SoFFin state-aid conditions are exogenous to stock-specific information\",\"Correlated bank and customer flows are not driven by market-making or by retail herding\",\"Customers of banks with and without trading desks are otherwise comparable (education compared)\"],\"threats_addressed\":[\"market-making inventory\",\"retail herding / aggregate retail demand\",\"endogeneity of bank sales\",\"financial stocks\",\"customer composition differences\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":2},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_measure\",\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes describe each column's specification. Descriptive tables report counts, means, SDs and percentiles. Performance tables use t-tests with unequal variances and Wilcoxon tests. A variable-definitions table is placed last.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggest' / 'evidence suggesting'\",\"'potential conflict of interest'\",\"'can be detrimental'\",\"admission that flows are not directly observed\",\"'reasonably exogenous'\"]},\"claim_evidence\":[{\"claim\":\"Banks' proprietary stock sales coincide with purchases of the same stocks by their retail customers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, V\"},{\"claim\":\"Forced sales under the 2008 state rescue conditions yield similar estimates, supporting a causal link\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The effect is stronger for illiquid stocks, in illiquid periods and for large positions, consistent with a price-impact motive\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Customers incur trading losses on stocks their bank sells, and these losses rise with the bank's profit on the same stock\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Stocks banks sell to customers underperform other stocks in bank and customer portfolios\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Customers of banks with proprietary trading desks underperform customers of banks without such desks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"}],\"structural_summary\":\"The introduction motivates the question with households' reliance on bank advice and banks' informational advantages. It then previews identification through forced sales under rescue conditions and previews the price-impact motive and performance results. It closes by linking the findings to bank-separation regulation and to the literatures on conflicts of interest and financial advice, followed by a roadmap. The body has a data section and two large results sections: stock flows (baseline, causality, robustness, channels, motives) and customer performance. A single-paragraph conclusion restates the findings.\",\"writing_profile_keywords\":[\"conflict of interest\",\"universal banks\",\"proprietary trading\",\"retail investors\",\"German regulatory data\",\"forced sales shock\",\"policy relevance\",\"question-driven abstract\"],\"confidence\":0.82,\"notes\":\"The abstract poses three numbered questions and answers all of them affirmatively. The conclusion excerpt is truncated, so whether it discusses policy implications there cannot be determined. Regulatory implications do appear in the introduction. The form of significance reporting in the regression tables is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "household_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of a bank's proprietary sales of a stock on its retail customers' purchases of that stock, and the resulting performance consequences for those customers", "strategy": "Bank-stock-quarter panel regressions relating changes in customer holdings to changes in bank holdings. Causal interpretation is strengthened by forced sales at banks under 2008 SoFFin rescue conditions. Performance is compared through trading profit regressions and portfolio return tests across banks with and without proprietary trading desks.", "key_assumptions": ["Sales forced by SoFFin state-aid conditions are exogenous to stock-specific information", "Correlated bank and customer flows are not driven by market-making or by retail herding", "Customers of banks with and without trading desks are otherwise comparable (education compared)"], "threats_addressed": ["market-making inventory", "retail herding / aggregate retail demand", "endogeneity of bank sales", "financial stocks", "customer composition differences"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 2}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_measure", "subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes describe each column's specification. Descriptive tables report counts, means, SDs and percentiles. Performance tables use t-tests with unequal variances and Wilcoxon tests. A variable-definitions table is placed last.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggest' / 'evidence suggesting'", "'potential conflict of interest'", "'can be detrimental'", "admission that flows are not directly observed", "'reasonably exogenous'"]}, "claim_evidence": [{"claim": "Banks' proprietary stock sales coincide with purchases of the same stocks by their retail customers", "evidence_type": "table", "evidence_ref": "Tables III, V"}, {"claim": "Forced sales under the 2008 state rescue conditions yield similar estimates, supporting a causal link", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The effect is stronger for illiquid stocks, in illiquid periods and for large positions, consistent with a price-impact motive", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Customers incur trading losses on stocks their bank sells, and these losses rise with the bank's profit on the same stock", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Stocks banks sell to customers underperform other stocks in bank and customer portfolios", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Customers of banks with proprietary trading desks underperform customers of banks without such desks", "evidence_type": "table", "evidence_ref": "Table XII"}], "structural_summary": "The introduction motivates the question with households' reliance on bank advice and banks' informational advantages. It then previews identification through forced sales under rescue conditions and previews the price-impact motive and performance results. It closes by linking the findings to bank-separation regulation and to the literatures on conflicts of interest and financial advice, followed by a roadmap. The body has a data section and two large results sections: stock flows (baseline, causality, robustness, channels, motives) and customer performance. A single-paragraph conclusion restates the findings.", "writing_profile_keywords": ["conflict of interest", "universal banks", "proprietary trading", "retail investors", "German regulatory data", "forced sales shock", "policy relevance", "question-driven abstract"], "confidence": 0.82, "notes": "The abstract poses three numbered questions and answers all of them affirmatively. The conclusion excerpt is truncated, so whether it discusses policy implications there cannot be determined. Regulatory implications do appear in the introduction. The form of significance reporting in the regression tables is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.74, "usage": {"input": 6420, "output": 2070, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067874, "request_id": "req_011CggpwKxzWwis9vkDqr1f2", "cached": false, "ts": "2026-11-03T23:01:58+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/754321de1a80c2a20a1d3c8305d744299a04b63fcbadb6df448f3cd8d1905201.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"international_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Country-level exposure (beta) of local long-run growth news to global long-run growth news, and how currency risk premia, interest rates and NFA positions vary with that exposure\",\"strategy\":\"Two-step approach: the predictable part of GDP growth from lagged p/d ratios measures long-run growth, which is regressed on the cross-sectional average to obtain exposures. These are fed into a calibrated recursive-preference multicountry model, whose moments are compared with data. A GMM test of Euler equation and other moment restrictions is added.\",\"key_assumptions\":[\"The p/d ratio captures the predictable long-run growth component\",\"Frictionless complete-markets risk sharing\",\"Epstein-Zin recursive preferences\",\"Exposure heterogeneity is in long-run shocks only, not short-run shocks\"],\"threats_addressed\":[\"Model rejection test against CRRA preferences\",\"Time-series test of NFA and exchange-rate responses to global long-run shocks\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables mix empirical estimates with calibration and simulated moments, setting data moments beside model moments under homogeneous and heterogeneous exposure and under CRRA. GMM standard errors are reported in parentheses, and Wald-test p-values in brackets.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\",\"figure_note\":\"none\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'good, but not perfect, alignment'\",\"explicit acknowledgment of unmatched cross-country spread\",\"'broadly consistent with the data'\",\"stated directions for future research\"]},\"claim_evidence\":[{\"claim\":\"G-10 countries differ substantially in exposure to global long-run growth news, but not to short-run shocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Homogeneous-exposure recursive risk sharing generates the forward premium anomaly but not carry-trade returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Heterogeneous exposure replicates HML-FX and HML-NFA carry-trade properties and cross-sectional patterns\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 3, 5, 6; Table III\"},{\"claim\":\"Low-exposure countries' NFA deteriorates and high-exposure currencies appreciate after negative global long-run shocks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 7, 8; Table IV\"},{\"claim\":\"GMM estimates are close to the calibration; the EZ model is not rejected while CRRA is strongly rejected\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"}],\"structural_summary\":\"The paper opens with an empirical stylized fact: G-10 countries are exposed heterogeneously to global long-run growth news. It then builds a multicountry Epstein-Zin risk-sharing model with long- and short-run shocks and calibrates it. A homogeneous-exposure benchmark is shown to fail on carry-trade returns, and heterogeneous exposure resolves that failure. Further testable predictions on NFA dynamics and exchange-rate responses are checked against the data, followed by a GMM test of the moment restrictions. The conclusion acknowledges imperfect fit and sets out a future research agenda.\",\"writing_profile_keywords\":[\"structural general equilibrium\",\"long-run risk\",\"recursive preferences\",\"carry trade\",\"model-data moment comparison\",\"news shocks\",\"net foreign assets\"],\"confidence\":0.8,\"notes\":\"Introduction excerpts are truncated, so the explicit list of contributions and any roadmap may be cut off. Section I is empirical data and estimation, but it is labeled 'data' because its main role is measuring exposures. Figures 7–8 show 90% confidence intervals; impulse-response figures do not appear to show uncertainty.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "international_finance", "subfield_secondary": "asset_pricing", "primary_method": "calibration_simulation", "identification": {"estimand": "Country-level exposure (beta) of local long-run growth news to global long-run growth news, and how currency risk premia, interest rates and NFA positions vary with that exposure", "strategy": "Two-step approach: the predictable part of GDP growth from lagged p/d ratios measures long-run growth, which is regressed on the cross-sectional average to obtain exposures. These are fed into a calibrated recursive-preference multicountry model, whose moments are compared with data. A GMM test of Euler equation and other moment restrictions is added.", "key_assumptions": ["The p/d ratio captures the predictable long-run growth component", "Frictionless complete-markets risk sharing", "Epstein-Zin recursive preferences", "Exposure heterogeneity is in long-run shocks only, not short-run shocks"], "threats_addressed": ["Model rejection test against CRRA preferences", "Time-series test of NFA and exchange-rate responses to global long-run shocks"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "other"]}, "tables": {"convention_notes": "Tables mix empirical estimates with calibration and simulated moments, setting data moments beside model moments under homogeneous and heterogeneous exposure and under CRRA. GMM standard errors are reported in parentheses, and Wald-test p-values in brackets.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed", "figure_note": "none"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'good, but not perfect, alignment'", "explicit acknowledgment of unmatched cross-country spread", "'broadly consistent with the data'", "stated directions for future research"]}, "claim_evidence": [{"claim": "G-10 countries differ substantially in exposure to global long-run growth news, but not to short-run shocks", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Homogeneous-exposure recursive risk sharing generates the forward premium anomaly but not carry-trade returns", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Heterogeneous exposure replicates HML-FX and HML-NFA carry-trade properties and cross-sectional patterns", "evidence_type": "figure", "evidence_ref": "Figures 3, 5, 6; Table III"}, {"claim": "Low-exposure countries' NFA deteriorates and high-exposure currencies appreciate after negative global long-run shocks", "evidence_type": "figure", "evidence_ref": "Figures 7, 8; Table IV"}, {"claim": "GMM estimates are close to the calibration; the EZ model is not rejected while CRRA is strongly rejected", "evidence_type": "table", "evidence_ref": "Table V"}], "structural_summary": "The paper opens with an empirical stylized fact: G-10 countries are exposed heterogeneously to global long-run growth news. It then builds a multicountry Epstein-Zin risk-sharing model with long- and short-run shocks and calibrates it. A homogeneous-exposure benchmark is shown to fail on carry-trade returns, and heterogeneous exposure resolves that failure. Further testable predictions on NFA dynamics and exchange-rate responses are checked against the data, followed by a GMM test of the moment restrictions. The conclusion acknowledges imperfect fit and sets out a future research agenda.", "writing_profile_keywords": ["structural general equilibrium", "long-run risk", "recursive preferences", "carry trade", "model-data moment comparison", "news shocks", "net foreign assets"], "confidence": 0.8, "notes": "Introduction excerpts are truncated, so the explicit list of contributions and any roadmap may be cut off. Section I is empirical data and estimation, but it is labeled 'data' because its main role is measuring exposures. Figures 7–8 show 90% confidence intervals; impulse-response figures do not appear to show uncertainty."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.25, "usage": {"input": 6026, "output": 2077, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066438, "request_id": "req_011Cggq8MiTsSuWqgnAftuMh", "cached": false, "ts": "2026-11-03T23:04:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/756e184688f33b4b0fa370af648383ea52c9875e0022d861c7c6fc218056d579.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"commodity futures / financial_intermediation\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Predictive relation between lagged basis-momentum and subsequent commodity nearby/spreading returns, plus the factor risk premium on basis-momentum exposure\",\"strategy\":\"Univariate and double portfolio sorts, pooled predictive regressions with fixed effects, time-series spanning regressions against benchmark factor models, and cross-sectional asset pricing tests on portfolios and individual commodities; competing explanations assessed through conditioning and interaction tests\",\"key_assumptions\":[\"Characteristic is observable ex ante, so predictive regressions avoid look-ahead bias\",\"Test assets and benchmark factor models are appropriate for the cross-sectional pricing tests\",\"Proxies for volatility, liquidity, intermediary capital and trader positions capture the intended economic channels\"],\"threats_addressed\":[\"Effect subsumed by basis, momentum or volatility characteristics\",\"Storage/inventory explanation (double sorts, non-storable financial assets)\",\"Hedging pressure explanation\",\"Transaction costs\",\"Sensitivity across maturities and portfolio composition\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_data_or_measure\",\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"internet_appendix\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define each characteristic by formula, specify panels for nearby versus spreading returns, describe the portfolio construction (e.g., High4-minus-Low4, median splits) and report t-statistics in parentheses. Fixed effects are listed model by model.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"we argue\",\"is consistent with\",\"unlikely to explain\",\"suggest\",\"might\"]},\"claim_evidence\":[{\"claim\":\"Basis-momentum sorts produce strong nearby and spreading return spreads\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Basis-momentum predicts commodity returns in pooled regressions controlling for benchmark characteristics and fixed effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Basis-momentum factors earn large alphas relative to existing three-factor commodity models\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Exposure to basis-momentum nearby factor is priced, with Sharpe ratio 0.55-0.85; two-factor model fits well\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Effect is similar across inventory/storability groups and appears in currencies, stock and bond indexes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI\"},{\"claim\":\"Predictability is stronger with high speculator spreading pressure; robust to hedging pressure\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Basis-momentum returns increase with aggregate volatility; risk premium is consistent with volatility, liquidity, intermediary and downside risk pricing\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-X\"}],\"structural_summary\":\"The paper introduces a new futures-curve characteristic and establishes its return predictability through sorts and pooled regressions. It then tests whether the characteristic is a priced factor using spanning and cross-sectional asset pricing tests. Next, it adjudicates among competing economic explanations: storage, hedging pressure, volatility, liquidity, intermediary and downside risk, rejecting the classical theories in favor of a speculator/intermediary risk channel. Extensions are relegated to an internet appendix. The introduction is organized as three numbered contributions.\",\"writing_profile_keywords\":[\"new characteristic\",\"factor pricing\",\"commodity futures\",\"competing explanations\",\"numbered contributions\",\"horse race against benchmarks\",\"risk-based interpretation\"],\"confidence\":0.85,\"notes\":\"No figures. Standard-error method is not visible in the table excerpts. The abstract states no explicit question or gap; the explanation sentence (index 3) is interpretive.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "commodity futures / financial_intermediation", "primary_method": "reduced_form_regression", "identification": {"estimand": "Predictive relation between lagged basis-momentum and subsequent commodity nearby/spreading returns, plus the factor risk premium on basis-momentum exposure", "strategy": "Univariate and double portfolio sorts, pooled predictive regressions with fixed effects, time-series spanning regressions against benchmark factor models, and cross-sectional asset pricing tests on portfolios and individual commodities; competing explanations assessed through conditioning and interaction tests", "key_assumptions": ["Characteristic is observable ex ante, so predictive regressions avoid look-ahead bias", "Test assets and benchmark factor models are appropriate for the cross-sectional pricing tests", "Proxies for volatility, liquidity, intermediary capital and trader positions capture the intended economic channels"], "threats_addressed": ["Effect subsumed by basis, momentum or volatility characteristics", "Storage/inventory explanation (double sorts, non-storable financial assets)", "Hedging pressure explanation", "Transaction costs", "Sensitivity across maturities and portfolio composition"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "contribution_statement"}], "contribution_paragraph_index": 1, "contribution_types": ["new_data_or_measure", "new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "internet_appendix", "types": ["alternative_specification", "subsample", "alternative_measure", "other"]}, "tables": {"convention_notes": "Table notes define each characteristic by formula, specify panels for nearby versus spreading returns, describe the portfolio construction (e.g., High4-minus-Low4, median splits) and report t-statistics in parentheses. Fixed effects are listed model by model.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["we argue", "is consistent with", "unlikely to explain", "suggest", "might"]}, "claim_evidence": [{"claim": "Basis-momentum sorts produce strong nearby and spreading return spreads", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Basis-momentum predicts commodity returns in pooled regressions controlling for benchmark characteristics and fixed effects", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Basis-momentum factors earn large alphas relative to existing three-factor commodity models", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Exposure to basis-momentum nearby factor is priced, with Sharpe ratio 0.55-0.85; two-factor model fits well", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Effect is similar across inventory/storability groups and appears in currencies, stock and bond indexes", "evidence_type": "table", "evidence_ref": "Tables V, VI"}, {"claim": "Predictability is stronger with high speculator spreading pressure; robust to hedging pressure", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Basis-momentum returns increase with aggregate volatility; risk premium is consistent with volatility, liquidity, intermediary and downside risk pricing", "evidence_type": "table", "evidence_ref": "Tables VIII-X"}], "structural_summary": "The paper introduces a new futures-curve characteristic and establishes its return predictability through sorts and pooled regressions. It then tests whether the characteristic is a priced factor using spanning and cross-sectional asset pricing tests. Next, it adjudicates among competing economic explanations: storage, hedging pressure, volatility, liquidity, intermediary and downside risk, rejecting the classical theories in favor of a speculator/intermediary risk channel. Extensions are relegated to an internet appendix. The introduction is organized as three numbered contributions.", "writing_profile_keywords": ["new characteristic", "factor pricing", "commodity futures", "competing explanations", "numbered contributions", "horse race against benchmarks", "risk-based interpretation"], "confidence": 0.85, "notes": "No figures. Standard-error method is not visible in the table excerpts. The abstract states no explicit question or gap; the explanation sentence (index 3) is interpretive."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.39, "usage": {"input": 6589, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07219, "request_id": "req_011CggqB4XSsXu4eySxF7xSi", "cached": false, "ts": "2026-11-03T23:05:05+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/77504260fc5c8bf98949cb11272bb50f1bb60d5e4b0dc9c1d0fa08e7160779e4.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"regression_discontinuity\",\"identification\":{\"estimand\":\"Effect of an information-free credit rating downgrade on consumers' subsequent borrowing, borrower actions and market exit, plus the share of the post-default credit decline attributable to the credit-reporting downgrade itself.\",\"strategy\":\"Regression discontinuity around Peru's 'rule of twenty' rating-alignment regulation. The running variable is an exchange-rate-adjusted debt share, so currency movements push a bank's share across the 20% threshold. Complete-default outcomes are then decomposed by interacting default with the expected exogenous classification change.\",\"key_assumptions\":[\"Currency movements are exogenous to individual consumers\",\"No manipulation of the running variable around the 20% threshold\",\"Borrowers just above and just below the threshold are comparable\",\"Threshold crossing affects outcomes only through the rating alignment\"],\"threats_addressed\":[\"Endogeneity of current-month balance changes (handled by using lagged, exchange-rate-adjusted balances)\",\"Covariate balance around the threshold\",\"Density manipulation (McCrary test)\",\"Placebo sample of borrowers with homogeneous loan ratings who should be unaffected\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"placebo\",\"alternative_specification\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Notes give the observation level (client-bank-month), the sample definition, and the equation estimated. Later tables point back to the baseline specification of Table IV. Stars mark significance levels. Outcomes are measured at 12-, 24- and 36-month horizons. Some variables are stated in soles.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggest\",\"may\",\"appears to\",\"potentially\",\"in our sample\"]},\"claim_evidence\":[{\"claim\":\"Complete default is associated with medium-term reductions in borrowing in a random sample of the full data set\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Exchange-rate-adjusted share predicts actual crossing of the 20% threshold\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Figure 1\"},{\"claim\":\"Borrowers just above and below the threshold look similar on observables, with no density manipulation\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Table V, Figures 2-3\"},{\"claim\":\"Crossing the threshold produces a moderate-duration rating downgrade only for borrowers with heterogeneous ratings\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 4\"},{\"claim\":\"Shocked borrowers see lower loan balances, less new financing and fewer new relationships for three years\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Figure 5\"},{\"claim\":\"Shocked borrowers pay down judicial-collection loans and zero out credit cards\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII, Figure 6\"},{\"claim\":\"Shocked borrowers are more likely to exit the consumer loan market at two- and three-year horizons\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX, Figure 7\"},{\"claim\":\"The reporting downgrade accounts for 25-65% of the post-default credit decline\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X-XI\"}],\"structural_summary\":\"Opens from the macro debate on weak post-2008 recovery and household borrowing constraints, then asks why negative credit events have lasting effects. Establishes a descriptive default–credit association first. It then uses a regulatory threshold combined with currency-driven variation as an RD to isolate information-free downgrades. The design is validated through first stage, covariate balance, density and placebo checks. Effects are traced on financing, borrower actions and broader outcomes, and the paper ends with a decomposition of the post-default decline. Policy implications about macro-contingent credit-history forgiveness close both the introduction and the conclusion.\",\"writing_profile_keywords\":[\"regression discontinuity\",\"regulatory threshold\",\"credit reporting\",\"consumer credit\",\"Peru\",\"exchange rate variation\",\"decomposition\",\"policy implication\"],\"confidence\":0.85,\"notes\":\"Standard-error type is not visible in the excerpts. Robustness checks are embedded in results subsections rather than a separate section. An Internet Appendix exists, but its content is not visible.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "regression_discontinuity", "identification": {"estimand": "Effect of an information-free credit rating downgrade on consumers' subsequent borrowing, borrower actions and market exit, plus the share of the post-default credit decline attributable to the credit-reporting downgrade itself.", "strategy": "Regression discontinuity around Peru's 'rule of twenty' rating-alignment regulation. The running variable is an exchange-rate-adjusted debt share, so currency movements push a bank's share across the 20% threshold. Complete-default outcomes are then decomposed by interacting default with the expected exogenous classification change.", "key_assumptions": ["Currency movements are exogenous to individual consumers", "No manipulation of the running variable around the 20% threshold", "Borrowers just above and just below the threshold are comparable", "Threshold crossing affects outcomes only through the rating alignment"], "threats_addressed": ["Endogeneity of current-month balance changes (handled by using lagged, exchange-rate-adjusted balances)", "Covariate balance around the threshold", "Density manipulation (McCrary test)", "Placebo sample of borrowers with homogeneous loan ratings who should be unaffected"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "policy_or_broader_implication"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["placebo", "alternative_specification", "falsification"]}, "tables": {"convention_notes": "Notes give the observation level (client-bank-month), the sample definition, and the equation estimated. Later tables point back to the baseline specification of Table IV. Stars mark significance levels. Outcomes are measured at 12-, 24- and 36-month horizons. Some variables are stated in soles.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggest", "may", "appears to", "potentially", "in our sample"]}, "claim_evidence": [{"claim": "Complete default is associated with medium-term reductions in borrowing in a random sample of the full data set", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Exchange-rate-adjusted share predicts actual crossing of the 20% threshold", "evidence_type": "table", "evidence_ref": "Table IV, Figure 1"}, {"claim": "Borrowers just above and below the threshold look similar on observables, with no density manipulation", "evidence_type": "figure", "evidence_ref": "Table V, Figures 2-3"}, {"claim": "Crossing the threshold produces a moderate-duration rating downgrade only for borrowers with heterogeneous ratings", "evidence_type": "table", "evidence_ref": "Table VI, Figure 4"}, {"claim": "Shocked borrowers see lower loan balances, less new financing and fewer new relationships for three years", "evidence_type": "table", "evidence_ref": "Table VII, Figure 5"}, {"claim": "Shocked borrowers pay down judicial-collection loans and zero out credit cards", "evidence_type": "table", "evidence_ref": "Table VIII, Figure 6"}, {"claim": "Shocked borrowers are more likely to exit the consumer loan market at two- and three-year horizons", "evidence_type": "table", "evidence_ref": "Table IX, Figure 7"}, {"claim": "The reporting downgrade accounts for 25-65% of the post-default credit decline", "evidence_type": "table", "evidence_ref": "Tables X-XI"}], "structural_summary": "Opens from the macro debate on weak post-2008 recovery and household borrowing constraints, then asks why negative credit events have lasting effects. Establishes a descriptive default–credit association first. It then uses a regulatory threshold combined with currency-driven variation as an RD to isolate information-free downgrades. The design is validated through first stage, covariate balance, density and placebo checks. Effects are traced on financing, borrower actions and broader outcomes, and the paper ends with a decomposition of the post-default decline. Policy implications about macro-contingent credit-history forgiveness close both the introduction and the conclusion.", "writing_profile_keywords": ["regression discontinuity", "regulatory threshold", "credit reporting", "consumer credit", "Peru", "exchange rate variation", "decomposition", "policy implication"], "confidence": 0.85, "notes": "Standard-error type is not visible in the excerpts. Robustness checks are embedded in results subsections rather than a separate section. An Internet Appendix exists, but its content is not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.73, "usage": {"input": 7041, "output": 2274, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074438, "request_id": "req_011CggpjPkMd3HEbAgrGHxnf", "cached": false, "ts": "2026-11-03T22:59:17+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7dbd22b2e37326f592e271c10deb1c4c7e38d9fc11816feaa681d9a362ccdcf4.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"analyst research / information intermediaries (capital markets)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in forecast accuracy, all-star likelihood, and market reactions between analysts with industry-related preanalyst experience and other analysts; DiD change in information asymmetry and prices after losing an expert versus a non-expert analyst\",\"strategy\":\"OLS and logit regressions with controls for analyst characteristics. Difference-in-differences around plausibly exogenous coverage terminations caused by brokerage house closures. Reg FD split used to probe the private-information channel.\",\"key_assumptions\":[\"Brokerage closures are exogenous to covered firms' information environment\",\"Controls capture other analyst skill attributes (experience, portfolio complexity, broker prestige)\",\"Self-selection of analysts into related industries does not drive results\"],\"threats_addressed\":[\"private information via social connections (Reg FD test)\",\"sample selection/self-selection\",\"industry classification choice\",\"alternative performance measures and fixed effects\",\"analyst effort\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes name the estimator (OLS, logistic, DiD means) and give dependent variable definitions in detail, pointing to Appendix C for other variables. Data sources and the sample period are stated. Notes are truncated, so the reporting of standard errors cannot be seen.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"plausibly exogenous\",\"consistent with\",\"cannot completely rule out\",\"unlikely to be a primary explanation\",\"suggesting\"]},\"claim_evidence\":[{\"claim\":\"Analysts with preanalyst experience in related industries forecast more accurately; the effect holds only for related-industry firms.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Odds of all-star status are 75% higher for related-experienced analysts than for unrelated-experienced analysts.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Forecast revisions by related-experienced analysts trigger about 0.43–0.45% stronger abnormal reactions than those of inexperienced analysts.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Losing an industry-expert analyst through broker closure raises information asymmetry more than losing other analysts.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The 3-day CAR difference-in-differences is 0.77% lower when an expert analyst is lost, and it does not reverse.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The industry-experience effect persists after Reg FD, so a private-information channel is unlikely to be the main driver.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Introduction paragraph 1\"},{\"claim\":\"Results are robust to GICS classification and to alternative accuracy measures.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"Recommendations of expert analysts have higher investment value.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"The paper opens by citing practitioner surveys that rank industry knowledge as the most important analyst trait, then notes the lack of systematic evidence on it. It introduces hand-collected biographical data and previews results in sequence: forecast accuracy, a Reg FD channel test, all-star career outcomes, market reactions to revisions, and DiD tests on broker-closure terminations. A contribution paragraph and a roadmap close the introduction. The body then runs through a literature section, data, main results, analyst allocation and real effects, and robustness (largely in the Internet Appendix). The conclusion restates the findings.\",\"writing_profile_keywords\":[\"sell-side analysts\",\"hand-collected data\",\"practitioner motivation\",\"industry expertise\",\"difference-in-differences\",\"broker closures\",\"Internet Appendix robustness\",\"sequential results preview\"],\"confidence\":0.8,\"notes\":\"Paragraphs of the introduction appear to be partially missing between paragraphs 0 and 1, where the data and accuracy results are presumably introduced. The form of significance reporting is not visible in the excerpts. The Reg FD finding is stated in the introduction without a table reference, so its evidence type is recorded as none_stated.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "analyst research / information intermediaries (capital markets)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in forecast accuracy, all-star likelihood, and market reactions between analysts with industry-related preanalyst experience and other analysts; DiD change in information asymmetry and prices after losing an expert versus a non-expert analyst", "strategy": "OLS and logit regressions with controls for analyst characteristics. Difference-in-differences around plausibly exogenous coverage terminations caused by brokerage house closures. Reg FD split used to probe the private-information channel.", "key_assumptions": ["Brokerage closures are exogenous to covered firms' information environment", "Controls capture other analyst skill attributes (experience, portfolio complexity, broker prestige)", "Self-selection of analysts into related industries does not drive results"], "threats_addressed": ["private information via social connections (Reg FD test)", "sample selection/self-selection", "industry classification choice", "alternative performance measures and fixed effects", "analyst effort"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Table notes name the estimator (OLS, logistic, DiD means) and give dependent variable definitions in detail, pointing to Appendix C for other variables. Data sources and the sample period are stated. Notes are truncated, so the reporting of standard errors cannot be seen.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["plausibly exogenous", "consistent with", "cannot completely rule out", "unlikely to be a primary explanation", "suggesting"]}, "claim_evidence": [{"claim": "Analysts with preanalyst experience in related industries forecast more accurately; the effect holds only for related-industry firms.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Odds of all-star status are 75% higher for related-experienced analysts than for unrelated-experienced analysts.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Forecast revisions by related-experienced analysts trigger about 0.43–0.45% stronger abnormal reactions than those of inexperienced analysts.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Losing an industry-expert analyst through broker closure raises information asymmetry more than losing other analysts.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The 3-day CAR difference-in-differences is 0.77% lower when an expert analyst is lost, and it does not reverse.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The industry-experience effect persists after Reg FD, so a private-information channel is unlikely to be the main driver.", "evidence_type": "none_stated", "evidence_ref": "Introduction paragraph 1"}, {"claim": "Results are robust to GICS classification and to alternative accuracy measures.", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "Recommendations of expert analysts have higher investment value.", "evidence_type": "table", "evidence_ref": "Table X"}], "structural_summary": "The paper opens by citing practitioner surveys that rank industry knowledge as the most important analyst trait, then notes the lack of systematic evidence on it. It introduces hand-collected biographical data and previews results in sequence: forecast accuracy, a Reg FD channel test, all-star career outcomes, market reactions to revisions, and DiD tests on broker-closure terminations. A contribution paragraph and a roadmap close the introduction. The body then runs through a literature section, data, main results, analyst allocation and real effects, and robustness (largely in the Internet Appendix). The conclusion restates the findings.", "writing_profile_keywords": ["sell-side analysts", "hand-collected data", "practitioner motivation", "industry expertise", "difference-in-differences", "broker closures", "Internet Appendix robustness", "sequential results preview"], "confidence": 0.8, "notes": "Paragraphs of the introduction appear to be partially missing between paragraphs 0 and 1, where the data and accuracy results are presumably introduced. The form of significance reporting is not visible in the excerpts. The Reg FD finding is stated in the introduction without a table reference, so its evidence type is recorded as none_stated."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.76, "usage": {"input": 7330, "output": 2286, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075834, "request_id": "req_011CggpXpfyceYJzHT1Gh13g", "cached": false, "ts": "2026-11-03T22:56:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7e76c79e10c38abac5883cccbc15304a911b3db307b394c68156f9a08f746458.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"experimental finance; corporate finance (capital structure irrelevance); behavioral_finance\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Effect of perfect versus zero return correlation (riskless versus risky arbitrage) on price deviations from MM value parity between leveraged and unleveraged shares; association of trader cognitive acuity with price discrepancies\",\"strategy\":\"Laboratory asset markets with between-session treatments (perfect-correlation versus no-correlation dividends), four repeated markets per session, plus single-asset and single-period follow-up treatments; CRT scores measure trader acuity\",\"key_assumptions\":[\"Experimental control holds all features constant except return correlation across treatments\",\"Costless short-selling and borrowing make riskless arbitrage feasible in the perfect-correlation treatment\",\"CRT score is a valid measure of trader acuity\"],\"threats_addressed\":[\"Bubble or mispricing dynamics specific to multi-asset trading (single-asset treatment)\",\"Multi-period dynamics (single-period treatment)\",\"Learning and experience (repeated markets)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"Tables report session-level averages in percent by market and treatment. Significance comes mainly from nonparametric tests (Wilcoxon signed-rank, Mann-Whitney), with p-values, z-statistics in brackets and stars. One table reports cluster regressions with robust SEs and Spearman correlations. Notes reference measure equations by number.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'results suggest'\",\"'in our setting' scoping\",\"'appear to be driven by'\",\"'hint at'\",\"explicit admission of only qualitative, not full quantitative, support\"]},\"claim_evidence\":[{\"claim\":\"Under perfectly correlated returns, average prices are close to MM parity, supporting value invariance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 1; Figure 2\"},{\"claim\":\"Price discrepancies are much larger when returns are uncorrelated; average parity deviation is about 22.5% vs 3.7%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Table III\"},{\"claim\":\"Discrepancies and deviations from fundamentals shrink across consecutive markets but persist with experienced traders\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figures 2-3\"},{\"claim\":\"The U-share is relatively overvalued versus the L-share in the no-correlation treatment\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Higher trader acuity (CRT) is associated with smaller price discrepancies\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Discrepancies are reduced by market equilibrating dynamics rather than specialized arbitrageurs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Single-asset markets show no significant relative deviations\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"}],\"structural_summary\":\"The paper opens by noting that the MM invariance theorem cannot be tested on field data. It then motivates a laboratory test and previews a treatment comparison of perfect versus zero return correlation. A literature review and the experimental design come next, followed by a restatement of the MM proof applied to the design. Formal measures and hypotheses are then laid out. Results proceed through parity tests, discrepancies, bubbles, smart-trader (CRT) analysis, equilibrating dynamics and diversification, with a summary. Supplementary single-asset and single-period treatments follow. The conclusion restates qualitative support, acknowledges incomplete quantitative convergence, and comments on real-world limits to arbitrage.\",\"writing_profile_keywords\":[\"experimental finance\",\"treatment comparison\",\"law of one price\",\"limits to arbitrage\",\"nonparametric tests\",\"hypothesis-driven\",\"qualitative support framing\"],\"confidence\":0.8,\"notes\":\"Section 3 is an artifact heading from the quoted MM theorem and belongs to the theory section. Section 6 is the continuation of the Results section. The no-correlation condition is framed as a control treatment. The acuity findings are partly correlational, although the introduction uses 'leads to' wording.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "experimental finance; corporate finance (capital structure irrelevance); behavioral_finance", "primary_method": "survey_or_experiment", "identification": {"estimand": "Effect of perfect versus zero return correlation (riskless versus risky arbitrage) on price deviations from MM value parity between leveraged and unleveraged shares; association of trader cognitive acuity with price discrepancies", "strategy": "Laboratory asset markets with between-session treatments (perfect-correlation versus no-correlation dividends), four repeated markets per session, plus single-asset and single-period follow-up treatments; CRT scores measure trader acuity", "key_assumptions": ["Experimental control holds all features constant except return correlation across treatments", "Costless short-selling and borrowing make riskless arbitrage feasible in the perfect-correlation treatment", "CRT score is a valid measure of trader acuity"], "threats_addressed": ["Bubble or mispricing dynamics specific to multi-asset trading (single-asset treatment)", "Multi-period dynamics (single-period treatment)", "Learning and experience (repeated markets)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "contribution_statement"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["other"]}, "tables": {"convention_notes": "Tables report session-level averages in percent by market and treatment. Significance comes mainly from nonparametric tests (Wilcoxon signed-rank, Mann-Whitney), with p-values, z-statistics in brackets and stars. One table reports cluster regressions with robust SEs and Spearman correlations. Notes reference measure equations by number.", "units_stated_in_notes": true, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'results suggest'", "'in our setting' scoping", "'appear to be driven by'", "'hint at'", "explicit admission of only qualitative, not full quantitative, support"]}, "claim_evidence": [{"claim": "Under perfectly correlated returns, average prices are close to MM parity, supporting value invariance", "evidence_type": "table", "evidence_ref": "Table II; Figure 1; Figure 2"}, {"claim": "Price discrepancies are much larger when returns are uncorrelated; average parity deviation is about 22.5% vs 3.7%", "evidence_type": "table", "evidence_ref": "Table II; Table III"}, {"claim": "Discrepancies and deviations from fundamentals shrink across consecutive markets but persist with experienced traders", "evidence_type": "table", "evidence_ref": "Table III; Figures 2-3"}, {"claim": "The U-share is relatively overvalued versus the L-share in the no-correlation treatment", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Higher trader acuity (CRT) is associated with smaller price discrepancies", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Discrepancies are reduced by market equilibrating dynamics rather than specialized arbitrageurs", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Single-asset markets show no significant relative deviations", "evidence_type": "table", "evidence_ref": "Table VII"}], "structural_summary": "The paper opens by noting that the MM invariance theorem cannot be tested on field data. It then motivates a laboratory test and previews a treatment comparison of perfect versus zero return correlation. A literature review and the experimental design come next, followed by a restatement of the MM proof applied to the design. Formal measures and hypotheses are then laid out. Results proceed through parity tests, discrepancies, bubbles, smart-trader (CRT) analysis, equilibrating dynamics and diversification, with a summary. Supplementary single-asset and single-period treatments follow. The conclusion restates qualitative support, acknowledges incomplete quantitative convergence, and comments on real-world limits to arbitrage.", "writing_profile_keywords": ["experimental finance", "treatment comparison", "law of one price", "limits to arbitrage", "nonparametric tests", "hypothesis-driven", "qualitative support framing"], "confidence": 0.8, "notes": "Section 3 is an artifact heading from the quoted MM theorem and belongs to the theory section. Section 6 is the continuation of the Results section. The no-correlation condition is framed as a control treatment. The acuity findings are partly correlational, although the introduction uses 'leads to' wording."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 27.29, "usage": {"input": 6769, "output": 3182, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.09151, "request_id": "req_011CggqDYm8FTsHzqcoMy92w", "cached": false, "ts": "2026-11-03T23:05:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/80b4001d204f9b31d629d29ab3efed964154a654f9940d1e9516a6f452f32ef8.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"information intermediaries / sell-side analysts\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Effect of industry-level drops in the number of analysts on the forecast accuracy, bias, effort and informativeness of the remaining analysts covering that industry, incremental to firm-level coverage effects\",\"strategy\":\"Uses analyst drops caused by brokerage house mergers and closures as plausibly exogenous shocks to industry-level analyst coverage, estimated with industry panel regressions; complemented by firm-level OLS and difference-in-differences regressions extending Hong and Kacperczyk (2010), and by analyses of all coverage changes\",\"key_assumptions\":[\"Analyst drops from brokerage mergers and closures are unrelated to the industry's underlying information environment\",\"Industry-level competition is the channel through which coverage changes affect other analysts\"],\"threats_addressed\":[\"endogeneity of analyst coverage changes\",\"external validity limits of drop-only tests (addressed by examining all changes, including increases)\",\"confounding of industry-level effects with firm-level coverage effects\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"robustness\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Panel-regression tables in which notes define the dependent variables (e.g., absolute and signed forecast error as proxies for accuracy and bias), the treatment variable (analyst drops from mergers and closures) and interaction indicators. Notes cross-reference variable definitions in Appendix A and earlier tables. Several tables are split into panels.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"our findings suggest\",\"plausibly exogenous\",\"consistent with\",\"we argue\",\"can potentially\",\"appears to\"]},\"claim_evidence\":[{\"claim\":\"Number of analysts in an industry rises with industry IPO activity and trading volume changes; large brokers are most sensitive to IPOs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Analyst headcount grew moderately over 1990-2010, with small brokers increasing their share\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Industry analyst drops from broker mergers/closures reduce remaining analysts' forecast accuracy and increase optimism bias\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Effects are stronger in industries with weaker information environments and when All-Star analysts are dropped\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-V\"},{\"claim\":\"After drops, remaining analysts exert less effort and forecast dispersion rises\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VII\"},{\"claim\":\"Analyst report informativeness falls about 0.28% per analyst lost, while earnings announcement informativeness rises about 0.42%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Industry-level coverage effects are incremental to firm-level effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX-X\"}],\"structural_summary\":\"The introduction motivates the role of analysts and identifies a gap: no aggregate, industry-level view of the analyst industry exists. It previews descriptive evidence on how analyst scope varies over time and its economic correlates, then sketches a competition and spillover mechanism. It previews results from merger- and closure-induced analyst drops on forecast quality, effort, convergence and price informativeness, and closes with firm- versus industry-level comparisons and contribution claims. In the body, Section I gives descriptive time-series evidence. Section II presents the quasi-experimental results with mechanism tests. Section III extends to all coverage changes and firm-level DID. The conclusion restates the findings and draws implications for the literature and regulation.\",\"writing_profile_keywords\":[\"sell-side analysts\",\"industry competition\",\"spillover effects\",\"brokerage mergers\",\"quasi-natural experiment\",\"forecast accuracy\",\"information dissemination\"],\"confidence\":0.8,\"notes\":\"Section I is labeled 'other' in the source but functions as data and descriptive analysis. The source says Section II's first paragraph refers to changes 'documented in Section II'; this appears to be a numbering mismatch. The excerpts do not state how standard errors are computed. The internet appendix is mentioned.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "information intermediaries / sell-side analysts", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Effect of industry-level drops in the number of analysts on the forecast accuracy, bias, effort and informativeness of the remaining analysts covering that industry, incremental to firm-level coverage effects", "strategy": "Uses analyst drops caused by brokerage house mergers and closures as plausibly exogenous shocks to industry-level analyst coverage, estimated with industry panel regressions; complemented by firm-level OLS and difference-in-differences regressions extending Hong and Kacperczyk (2010), and by analyses of all coverage changes", "key_assumptions": ["Analyst drops from brokerage mergers and closures are unrelated to the industry's underlying information environment", "Industry-level competition is the channel through which coverage changes affect other analysts"], "threats_addressed": ["endogeneity of analyst coverage changes", "external validity limits of drop-only tests (addressed by examining all changes, including increases)", "confounding of industry-level effects with firm-level coverage effects"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "robustness"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Panel-regression tables in which notes define the dependent variables (e.g., absolute and signed forecast error as proxies for accuracy and bias), the treatment variable (analyst drops from mergers and closures) and interaction indicators. Notes cross-reference variable definitions in Appendix A and earlier tables. Several tables are split into panels.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["our findings suggest", "plausibly exogenous", "consistent with", "we argue", "can potentially", "appears to"]}, "claim_evidence": [{"claim": "Number of analysts in an industry rises with industry IPO activity and trading volume changes; large brokers are most sensitive to IPOs", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Analyst headcount grew moderately over 1990-2010, with small brokers increasing their share", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Industry analyst drops from broker mergers/closures reduce remaining analysts' forecast accuracy and increase optimism bias", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Effects are stronger in industries with weaker information environments and when All-Star analysts are dropped", "evidence_type": "table", "evidence_ref": "Tables III-V"}, {"claim": "After drops, remaining analysts exert less effort and forecast dispersion rises", "evidence_type": "table", "evidence_ref": "Tables VI-VII"}, {"claim": "Analyst report informativeness falls about 0.28% per analyst lost, while earnings announcement informativeness rises about 0.42%", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Industry-level coverage effects are incremental to firm-level effects", "evidence_type": "table", "evidence_ref": "Tables IX-X"}], "structural_summary": "The introduction motivates the role of analysts and identifies a gap: no aggregate, industry-level view of the analyst industry exists. It previews descriptive evidence on how analyst scope varies over time and its economic correlates, then sketches a competition and spillover mechanism. It previews results from merger- and closure-induced analyst drops on forecast quality, effort, convergence and price informativeness, and closes with firm- versus industry-level comparisons and contribution claims. In the body, Section I gives descriptive time-series evidence. Section II presents the quasi-experimental results with mechanism tests. Section III extends to all coverage changes and firm-level DID. The conclusion restates the findings and draws implications for the literature and regulation.", "writing_profile_keywords": ["sell-side analysts", "industry competition", "spillover effects", "brokerage mergers", "quasi-natural experiment", "forecast accuracy", "information dissemination"], "confidence": 0.8, "notes": "Section I is labeled 'other' in the source but functions as data and descriptive analysis. The source says Section II's first paragraph refers to changes 'documented in Section II'; this appears to be a numbering mismatch. The excerpts do not state how standard errors are computed. The internet appendix is mentioned."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.23, "usage": {"input": 7584, "output": 2310, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07733, "request_id": "req_011CggpbUMxCwXjHoWuYjC65", "cached": false, "ts": "2026-11-03T22:57:29+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/814265f77bcdec2ff2f329c07d96eefbdebfcf63c2ad007e45397355f6197975.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"political economy of finance; household finance (mortgages and foreclosures)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in the hazard of, and time to, foreclosure start for 90-day delinquent mortgages located in districts of House Financial Services Committee members versus other districts during 2009–2010.\",\"strategy\":\"Loan-level exponential hazard and accelerated-failure-time models that use committee membership as a political factor argued to be orthogonal to local economic conditions. Models include loan and zip-code controls and state-by-time fixed effects. Placebos use other committees and past periods, and delinquency onset is checked for self-selection.\",\"key_assumptions\":[\"Committee membership is not tied to districts and does not proxy for economic factors affecting servicers' foreclosure decisions\",\"No self-selection of members into the committee based on local delinquency conditions\",\"Controls and state-time fixed effects absorb local economic shocks\"],\"threats_addressed\":[\"Servicer capacity constraints from differing delinquency volumes\",\"Self-selection into committee membership\",\"Spurious committee effects (placebo committees)\",\"Time-invariant district effects (past-period placebos)\",\"Stimulus-fund allocation as an alternative explanation\",\"Econometric model specification\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"placebo\",\"alternative_sample_period\",\"alternative_specification\",\"subsample\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Table notes name the estimator (exponential hazard, maximum-likelihood AFT with right-censoring), the sample period, the sample definition and the binary committee variable. Separate tables cover placebos, delinquency, senior members and alternative models.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'unlikely to be driven by'\",\"'conservatively estimate'\",\"'may be more effective'\",\"explicit statement of what the paper does not show\",\"'quite plausible' assumptions\"]},\"claim_evidence\":[{\"claim\":\"Servicers delayed foreclosure starts on delinquent loans in districts of HFSC members.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Foreclosure starts take about half a month longer in committee districts than the roughly 12-month average elsewhere.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Delinquency onset does not differ between committee and noncommittee districts.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"No comparable effect appears for the Transportation & Infrastructure or Armed Services committees, or for HFSC in past periods.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The cost of delay (~$30M) is comparable to lobbying and an order of magnitude above PAC contributions to members.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Introduction paragraph 2; foreclosure cost estimates from prior literature\"},{\"claim\":\"Results hold under alternative periods and econometric models and for senior members.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"}],\"structural_summary\":\"Opens with a political anecdote about foreclosure moratorium calls to motivate whether banks' foreclosure timing responds to political concerns. The introduction previews the main hazard result, uses null delinquency differences to rule out capacity constraints, and quantifies the cost of delay against lobbying and PAC spending. It then frames foreclosure delay as a novel in-kind political channel. The body moves through institutional background, loan-level data and hazard regressions, then a robustness section of placebos, self-selection checks and alternative methods. A welfare section tests borrowers' nonmortgage credit outcomes. The conclusion restates the findings, draws political-economy implications, notes assumptions, and leaves banks' returns as an open question.\",\"writing_profile_keywords\":[\"political economy\",\"hazard models\",\"placebo tests\",\"loan-level data\",\"anecdotal motivation\",\"cost comparison\",\"first-to-document claim\"],\"confidence\":0.8,\"notes\":\"Abstract is results-focused, with no explicit gap or method sentence. Standard-error treatment is not visible in the excerpts. Section V is labeled robustness by the rule, but it functions as welfare and mechanism analysis.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "political economy of finance; household finance (mortgages and foreclosures)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in the hazard of, and time to, foreclosure start for 90-day delinquent mortgages located in districts of House Financial Services Committee members versus other districts during 2009–2010.", "strategy": "Loan-level exponential hazard and accelerated-failure-time models that use committee membership as a political factor argued to be orthogonal to local economic conditions. Models include loan and zip-code controls and state-by-time fixed effects. Placebos use other committees and past periods, and delinquency onset is checked for self-selection.", "key_assumptions": ["Committee membership is not tied to districts and does not proxy for economic factors affecting servicers' foreclosure decisions", "No self-selection of members into the committee based on local delinquency conditions", "Controls and state-time fixed effects absorb local economic shocks"], "threats_addressed": ["Servicer capacity constraints from differing delinquency volumes", "Self-selection into committee membership", "Spurious committee effects (placebo committees)", "Time-invariant district effects (past-period placebos)", "Stimulus-fund allocation as an alternative explanation", "Econometric model specification"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["placebo", "alternative_sample_period", "alternative_specification", "subsample", "falsification"]}, "tables": {"convention_notes": "Table notes name the estimator (exponential hazard, maximum-likelihood AFT with right-censoring), the sample period, the sample definition and the binary committee variable. Separate tables cover placebos, delinquency, senior members and alternative models.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'unlikely to be driven by'", "'conservatively estimate'", "'may be more effective'", "explicit statement of what the paper does not show", "'quite plausible' assumptions"]}, "claim_evidence": [{"claim": "Servicers delayed foreclosure starts on delinquent loans in districts of HFSC members.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Foreclosure starts take about half a month longer in committee districts than the roughly 12-month average elsewhere.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Delinquency onset does not differ between committee and noncommittee districts.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "No comparable effect appears for the Transportation & Infrastructure or Armed Services committees, or for HFSC in past periods.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The cost of delay (~$30M) is comparable to lobbying and an order of magnitude above PAC contributions to members.", "evidence_type": "descriptive_statistic", "evidence_ref": "Introduction paragraph 2; foreclosure cost estimates from prior literature"}, {"claim": "Results hold under alternative periods and econometric models and for senior members.", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}], "structural_summary": "Opens with a political anecdote about foreclosure moratorium calls to motivate whether banks' foreclosure timing responds to political concerns. The introduction previews the main hazard result, uses null delinquency differences to rule out capacity constraints, and quantifies the cost of delay against lobbying and PAC spending. It then frames foreclosure delay as a novel in-kind political channel. The body moves through institutional background, loan-level data and hazard regressions, then a robustness section of placebos, self-selection checks and alternative methods. A welfare section tests borrowers' nonmortgage credit outcomes. The conclusion restates the findings, draws political-economy implications, notes assumptions, and leaves banks' returns as an open question.", "writing_profile_keywords": ["political economy", "hazard models", "placebo tests", "loan-level data", "anecdotal motivation", "cost comparison", "first-to-document claim"], "confidence": 0.8, "notes": "Abstract is results-focused, with no explicit gap or method sentence. Standard-error treatment is not visible in the excerpts. Section V is labeled robustness by the rule, but it functions as welfare and mechanism analysis."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.59, "usage": {"input": 6963, "output": 2203, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072706, "request_id": "req_011Cggq8BZX8HcirHFuhjYh9", "cached": false, "ts": "2026-11-03T23:04:26+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/8461f53eeaacbf7c7cec9c17fb0cb2c73bde1ce5c15213403740f88c6ab09eda.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Predictive association between ex-ante bank LMI and crisis outcomes (stock crash probability, borrowing from Fed facilities/TARP); aggregate liquidity mismatch over time\",\"strategy\":\"Construct a theory-based measure from balance sheets weighted by market-based liquidity factors; validate it through time-series aggregates, a stress test, and cross-sectional probability regressions on pre-crisis LMI with controls; compare against Basel LCR/NSFR and the Berger-Bouwman measure\",\"key_assumptions\":[\"Repo haircuts and the OIS-Tbill spread proxy asset and funding liquidity\",\"Calibrated liquidity weights (e.g., kappa) are appropriate\",\"Measuring LMI ex ante makes it predetermined relative to crisis outcomes\"],\"threats_addressed\":[\"Fixed versus time-varying weights\",\"Comparison with alternative liquidity measures\",\"Bank characteristic controls\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_data_or_measure\",\"new_theory_or_model\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_measure\",\"sensitivity_to_parameters\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Table notes state the sample, the regression equation in probability form, dependent-variable definitions, and the quarters covered. Stress-test table reports scenarios at 1σ, 2σ and 3σ shocks. Standard errors are not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"acknowledging the measure is not a finished product\",\"noting calibration choices drive results\",\"suggesting future data to pin down weights\"]},\"claim_evidence\":[{\"claim\":\"Aggregate LMI falls from about +$4 trillion before the crisis to about -$6 trillion in 2008\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"A stress test based on LMI signals banking fragility in early 2007\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Figure 9\"},{\"claim\":\"Banks with lower pre-crisis LMI were more likely to borrow from Fed facilities and TARP\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Lower ex-ante LMI predicts higher stock crash probability during the crisis\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"With fixed liquidity weights, LMI varies little between normal and crisis periods\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 8\"},{\"claim\":\"Federal Reserve liquidity injections correlate with the increase in LMI in 2008\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"}],\"structural_summary\":\"The paper opens with liquidity crises and post-crisis regulation, then identifies the lack of an agreed liquidity measurement framework. It derives the LMI from a recursive theoretical framework and builds the empirical measure using haircut-based and spread-based liquidity weights. It then validates the measure on macro criteria (aggregate time series, decomposition, stress test) and on micro criteria (cross-sectional prediction of crash risk and government borrowing), benchmarking against the Basel ratios and Berger-Bouwman. The conclusion restates the measure's advantages and notes that the calibration is still open to refinement.\",\"writing_profile_keywords\":[\"measurement paper\",\"macroprudential\",\"stress test\",\"benchmark comparison\",\"theory-grounded index\",\"validation criteria\"],\"confidence\":0.8,\"notes\":\"Validation is predictive rather than causal. How statistical significance is reported is not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Predictive association between ex-ante bank LMI and crisis outcomes (stock crash probability, borrowing from Fed facilities/TARP); aggregate liquidity mismatch over time", "strategy": "Construct a theory-based measure from balance sheets weighted by market-based liquidity factors; validate it through time-series aggregates, a stress test, and cross-sectional probability regressions on pre-crisis LMI with controls; compare against Basel LCR/NSFR and the Berger-Bouwman measure", "key_assumptions": ["Repo haircuts and the OIS-Tbill spread proxy asset and funding liquidity", "Calibrated liquidity weights (e.g., kappa) are appropriate", "Measuring LMI ex ante makes it predetermined relative to crisis outcomes"], "threats_addressed": ["Fixed versus time-varying weights", "Comparison with alternative liquidity measures", "Bank characteristic controls"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_data_or_measure", "new_theory_or_model", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_measure", "sensitivity_to_parameters", "alternative_sample_period"]}, "tables": {"convention_notes": "Table notes state the sample, the regression equation in probability form, dependent-variable definitions, and the quarters covered. Stress-test table reports scenarios at 1σ, 2σ and 3σ shocks. Standard errors are not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["acknowledging the measure is not a finished product", "noting calibration choices drive results", "suggesting future data to pin down weights"]}, "claim_evidence": [{"claim": "Aggregate LMI falls from about +$4 trillion before the crisis to about -$6 trillion in 2008", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "A stress test based on LMI signals banking fragility in early 2007", "evidence_type": "table", "evidence_ref": "Table III, Figure 9"}, {"claim": "Banks with lower pre-crisis LMI were more likely to borrow from Fed facilities and TARP", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Lower ex-ante LMI predicts higher stock crash probability during the crisis", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "With fixed liquidity weights, LMI varies little between normal and crisis periods", "evidence_type": "figure", "evidence_ref": "Figure 8"}, {"claim": "Federal Reserve liquidity injections correlate with the increase in LMI in 2008", "evidence_type": "figure", "evidence_ref": "Figure 6"}], "structural_summary": "The paper opens with liquidity crises and post-crisis regulation, then identifies the lack of an agreed liquidity measurement framework. It derives the LMI from a recursive theoretical framework and builds the empirical measure using haircut-based and spread-based liquidity weights. It then validates the measure on macro criteria (aggregate time series, decomposition, stress test) and on micro criteria (cross-sectional prediction of crash risk and government borrowing), benchmarking against the Basel ratios and Berger-Bouwman. The conclusion restates the measure's advantages and notes that the calibration is still open to refinement.", "writing_profile_keywords": ["measurement paper", "macroprudential", "stress test", "benchmark comparison", "theory-grounded index", "validation criteria"], "confidence": 0.8, "notes": "Validation is predictive rather than causal. How statistical significance is reported is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.14, "usage": {"input": 7420, "output": 2013, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070734, "request_id": "req_011CggppjiYreP4Y1q26AeVm", "cached": false, "ts": "2026-11-03T23:00:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/84971cd34ee70f51d1d784652edb2486e4348059191347f18e56b095f941a3bd.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"corporate_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of an increase in housing collateral value on the probability of starting a firm and on postentry size, debt, growth and survival\",\"strategy\":\"Difference-in-differences-like comparison: local house-price appreciation interacted with homeownership. Treated homeowners are compared with renters, and with homeowners carrying an outstanding mortgage (who cannot take a second mortgage in France), within region-by-year fixed effects. A regional-level analysis interacts price growth with the homeownership share.\",\"key_assumptions\":[\"Local demand shocks affect owners and controls (renters, mortgaged owners) in the same region similarly\",\"French institutions prevent home equity withdrawal for mortgaged owners\",\"House-price variation is not correlated with owner-specific entrepreneurial opportunities\"],\"threats_addressed\":[\"local economic or demand shocks\",\"wealth versus liquidity effects (full versus partial owners)\",\"individual characteristics\",\"crowding out at the regional level\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes describe the regression model (linear probability or OLS), the interaction term between house-price growth and owner status, the fixed effects and controls, and the sample period. Summary statistics tables precede regression tables for each analysis.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"yes\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"may, in turn\",\"suggests\",\"improving identification\"]},\"claim_evidence\":[{\"claim\":\"Collateral value increases raise homeowners' probability of entering entrepreneurship relative to renters and mortgaged owners\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III\"},{\"claim\":\"Treated entrepreneurs start larger firms and use more debt at creation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Size differences persist through 2006\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Table VI, Figure 2\"},{\"claim\":\"Effects are stronger in industries with more prevalent credit constraints\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Regional firm creation is more sensitive to house prices where the homeownership share is higher\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Moving from the 25th to the 75th percentile of house-price growth raises firm assets by about 13%\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Intro paragraph 1\"}],\"structural_summary\":\"The introduction opens with the claim of the paper, notes that the existing wealth–entrepreneurship correlation is ambiguous, and stresses its policy relevance. It then sets out a three-part design (extensive margin, intensive margin, aggregate) with results previewed for each part, and ends with literature positioning. An institutional section on French mortgages justifies the control groups. Three empirical sections follow a parallel structure: data, specification, identification, results and sample splits. A short conclusion restates the contribution, contrasts it with Mian and Sufi, and leaves quantification of the relative channels for future research.\",\"writing_profile_keywords\":[\"institutional setting as identification\",\"parallel section structure\",\"extensive and intensive margin\",\"administrative data\",\"policy motivation\",\"interaction design\"],\"confidence\":0.85,\"notes\":\"The excerpts give no detail on standard errors or significance notation. Robustness checks appear within results via sample splits and controls.\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "corporate_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of an increase in housing collateral value on the probability of starting a firm and on postentry size, debt, growth and survival", "strategy": "Difference-in-differences-like comparison: local house-price appreciation interacted with homeownership. Treated homeowners are compared with renters, and with homeowners carrying an outstanding mortgage (who cannot take a second mortgage in France), within region-by-year fixed effects. A regional-level analysis interacts price growth with the homeownership share.", "key_assumptions": ["Local demand shocks affect owners and controls (renters, mortgaged owners) in the same region similarly", "French institutions prevent home equity withdrawal for mortgaged owners", "House-price variation is not correlated with owner-specific entrepreneurial opportunities"], "threats_addressed": ["local economic or demand shocks", "wealth versus liquidity effects (full versus partial owners)", "individual characteristics", "crowding out at the regional level"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["subsample", "alternative_specification"]}, "tables": {"convention_notes": "Table notes describe the regression model (linear probability or OLS), the interaction term between house-price growth and owner status, the fixed effects and controls, and the sample period. Summary statistics tables precede regression tables for each analysis.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "yes"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "may, in turn", "suggests", "improving identification"]}, "claim_evidence": [{"claim": "Collateral value increases raise homeowners' probability of entering entrepreneurship relative to renters and mortgaged owners", "evidence_type": "table", "evidence_ref": "Tables II, III"}, {"claim": "Treated entrepreneurs start larger firms and use more debt at creation", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Size differences persist through 2006", "evidence_type": "figure", "evidence_ref": "Table VI, Figure 2"}, {"claim": "Effects are stronger in industries with more prevalent credit constraints", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Regional firm creation is more sensitive to house prices where the homeownership share is higher", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Moving from the 25th to the 75th percentile of house-price growth raises firm assets by about 13%", "evidence_type": "none_stated", "evidence_ref": "Intro paragraph 1"}], "structural_summary": "The introduction opens with the claim of the paper, notes that the existing wealth–entrepreneurship correlation is ambiguous, and stresses its policy relevance. It then sets out a three-part design (extensive margin, intensive margin, aggregate) with results previewed for each part, and ends with literature positioning. An institutional section on French mortgages justifies the control groups. Three empirical sections follow a parallel structure: data, specification, identification, results and sample splits. A short conclusion restates the contribution, contrasts it with Mian and Sufi, and leaves quantification of the relative channels for future research.", "writing_profile_keywords": ["institutional setting as identification", "parallel section structure", "extensive and intensive margin", "administrative data", "policy motivation", "interaction design"], "confidence": 0.85, "notes": "The excerpts give no detail on standard errors or significance notation. Robustness checks appear within results via sample splits and controls."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.88, "usage": {"input": 5832, "output": 1913, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062382, "request_id": "req_011CggpVEKSPA7VTNM8ZPije", "cached": false, "ts": "2026-11-03T22:56:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/85b14a267230438b12476adf4ab3309d3ac8a4c789d9eefb045ab8e1b481e2f9.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of the previous trading day's value-weighted earnings surprise of large firms on the announcement return reaction of firms announcing today, holding their own surprise constant\",\"strategy\":\"Panel regressions of announcement-window returns on the prior-day salient surprise. Identification rests on announcement dates being scheduled weeks in advance, which makes announcement sequencing plausibly unrelated to fundamentals. Supplemented by timing tests, long-run reversal tests and tests ruling out information transmission.\",\"key_assumptions\":[\"Announcement dates are set in advance and are not strategically timed relative to other firms' surprises\",\"The prior-day surprise does not carry information relevant to today's announcers\",\"Return predictability reflects mispricing rather than risk compensation\"],\"threats_addressed\":[\"information transmission from earlier announcements\",\"expectational errors versus perception errors\",\"risk changes\",\"trading frictions and capital constraints\",\"strategic timing of announcements\",\"analyst forecast bias or staleness\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":7,\"move\":\"motivation\"},{\"paragraph_index\":8,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"identification_defense\"},{\"paragraph_index\":12,\"move\":\"preview_of_results\"},{\"paragraph_index\":13,\"move\":\"contribution_statement\"},{\"paragraph_index\":13,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":13,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"robustness\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"subsample\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define variables and return windows and refer back to Table II for definitions. Standard errors are clustered by date and reported in parentheses, with stars marking the 10/5/1% significance levels. Specification differences are described column by column.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' phrasing\",\"'suggestive evidence' for heterogeneity results\",\"'potentially' and 'may' for extrapolation beyond markets\",\"'unlikely to be correlated' in the identification argument\"],\"claim_evidence\":[{\"claim\":\"Announcement returns fall as the prior day's large-firm surprise rises; moving from the worst to the best decile lowers today's return response by about 53 bp\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Table II\"},{\"claim\":\"The effect is a level shift that does not depend on an interaction with today's own surprise\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 3\"},{\"claim\":\"The return distortion reverses within roughly 50 trading days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"A long-short strategy in large firms based on the prior-day surprise earns about 15% per year abnormal return\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figure 4\"},{\"claim\":\"The prior-day surprise does not predict today's surprises and does not move today's announcers' prices on day t-1\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The results are not explained by risk, trading frictions or strategic timing of announcements\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"},{\"claim\":\"Only the t-1 surprise matters; t-2, t-3 and lead surprises do not\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"The paper opens with a literary epigraph and lab evidence on contrast effects, then states the theoretical prediction. It previews the main result with a figure and quantifies it. It argues against information-transmission explanations by previewing the trading-strategy and reversal results, and positions contribution against the limited-attention and expectational-error literatures. The body moves through data, baseline results (reversals, trading strategy), a long section on alternative explanations, robustness to alternative surprise measures, and further exploration of timing, heterogeneity and industry effects. The conclusion restates the finding, argues generality and implications outside finance, and suggests extensions such as habit formation.\",\"writing_profile_keywords\":[\"behavioral bias in the field\",\"epigraph opening\",\"figure-led preview\",\"alternative explanations section\",\"trading strategy\",\"event study returns\",\"lab-to-field contribution\"],\"confidence\":0.85,\"notes\":\"Introduction paragraphs 0-6 are fragments of a Plato epigraph. The 'Further Exploration' section is mapped to the results role. Figure captions in the excerpts do not mention confidence bands, so uncertainty in figures is marked unclear.\"}", "json": null, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.77, "usage": {"input": 8269, "output": 2130, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07647, "request_id": "req_011CggpyixtzBwxJLpmxpdtR", "cached": false, "ts": "2026-11-03T23:02:31+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/87edcde20968ed5b8e8eb341e154d9b6d9c1c7812d0387498764153fc0bc1ee8.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-city dispersion in financial misconduct rates beyond sampling variation, and the association of city FM rates with measures of local social norms (non-FM misbehavior).\",\"strategy\":\"Negative binomial and Poisson count regressions at the city and city-year level testing excess variance and city fixed effects; cross-sectional regressions of FM on a misbehavior index; controls for regional characteristics, enforcement proxies, and firm characteristics; peer logit regressions.\",\"key_assumptions\":[\"Non-FM misbehavior measures proxy for city-level ethical norms\",\"Detected FM reflects committed FM after controlling for enforcement proxies\",\"Observable controls capture relevant firm and regional heterogeneity\"],\"threats_addressed\":[\"sampling variation / chance clustering\",\"regional enforcement differences (detection and deterrence)\",\"firm heterogeneity in benefits of FM\",\"regional demographic and economic characteristics\",\"reflection problem (acknowledged caveat)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":0},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"roadmap\"},{\"paragraph_index\":10,\"move\":\"preview_of_approach\"},{\"paragraph_index\":11,\"move\":\"preview_of_approach\"},{\"paragraph_index\":12,\"move\":\"preview_of_results\"},{\"paragraph_index\":13,\"move\":\"preview_of_results\"},{\"paragraph_index\":14,\"move\":\"identification_defense\"},{\"paragraph_index\":14,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":15,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":15,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":16,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":16,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":15,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Table notes open by naming the estimator (negative binomial, Poisson, panel logit) and defining the dependent variable (area-level count of FM firm-years). Appendix tables repeat the regressions with alternative constructions of the misbehavior index, area-year units, and a large-firm subsample.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"appear to\",\"suggest\",\"cannot definitively rule out\",\"remain agnostic about ultimate causes\",\"explicit caveat on reflection problem\"]},\"claim_evidence\":[{\"claim\":\"FM rates cluster in some cities more than chance predicts, with roughly threefold differences across cities.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Table II\"},{\"claim\":\"City fixed effects remain significant in panel regressions at the city-year level.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Each of five non-FM misbehavior measures is positively related to city FM rates; the combined index has a t-statistic above four.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"City FM rates and the misbehavior index ranking are strongly correlated (rho of about 0.60).\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Including the misbehavior index eliminates significant excess clustering of FM across cities.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Enforcement proxies and firm characteristics do not explain the city differences in FM.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Table VIII\"}],\"structural_summary\":\"Opens with the gap between Becker-style crime models and observed spatial variation in crime, then previews numbered findings and walks through the sections. A stylized cost-benefit model motivates three candidate explanations: norms, enforcement, and firm heterogeneity. The analysis first measures excess clustering of FM with count models, then relates FM to observable regional traits and to an index of non-FM misbehavior. Separate sections rule out enforcement and firm-characteristic explanations. The paper closes by acknowledging that the sources of norms remain unidentified and flags them for future research.\",\"writing_profile_keywords\":[\"social norms\",\"geography\",\"count models\",\"measurement-first\",\"alternative explanations\",\"rare events\",\"culture and finance\"],\"confidence\":0.82,\"notes\":\"The introduction's numbered findings list is split across paragraphs 2-9. Causal language is cautious: norms 'appear' to matter, and the authors explicitly acknowledge the reflection problem and that the sources of norms are unidentified.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-city dispersion in financial misconduct rates beyond sampling variation, and the association of city FM rates with measures of local social norms (non-FM misbehavior).", "strategy": "Negative binomial and Poisson count regressions at the city and city-year level testing excess variance and city fixed effects; cross-sectional regressions of FM on a misbehavior index; controls for regional characteristics, enforcement proxies, and firm characteristics; peer logit regressions.", "key_assumptions": ["Non-FM misbehavior measures proxy for city-level ethical norms", "Detected FM reflects committed FM after controlling for enforcement proxies", "Observable controls capture relevant firm and regional heterogeneity"], "threats_addressed": ["sampling variation / chance clustering", "regional enforcement differences (detection and deterrence)", "firm heterogeneity in benefits of FM", "regional demographic and economic characteristics", "reflection problem (acknowledged caveat)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 0}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "roadmap"}, {"paragraph_index": 10, "move": "preview_of_approach"}, {"paragraph_index": 11, "move": "preview_of_approach"}, {"paragraph_index": 12, "move": "preview_of_results"}, {"paragraph_index": 13, "move": "preview_of_results"}, {"paragraph_index": 14, "move": "identification_defense"}, {"paragraph_index": 14, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 15, "move": "related_literature_positioning"}, {"paragraph_index": 15, "move": "policy_or_broader_implication"}, {"paragraph_index": 16, "move": "related_literature_positioning"}, {"paragraph_index": 16, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 15, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "alternative_measure", "alternative_sample_period"]}, "tables": {"convention_notes": "Table notes open by naming the estimator (negative binomial, Poisson, panel logit) and defining the dependent variable (area-level count of FM firm-years). Appendix tables repeat the regressions with alternative constructions of the misbehavior index, area-year units, and a large-firm subsample.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["appear to", "suggest", "cannot definitively rule out", "remain agnostic about ultimate causes", "explicit caveat on reflection problem"]}, "claim_evidence": [{"claim": "FM rates cluster in some cities more than chance predicts, with roughly threefold differences across cities.", "evidence_type": "table", "evidence_ref": "Table I, Table II"}, {"claim": "City fixed effects remain significant in panel regressions at the city-year level.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Each of five non-FM misbehavior measures is positively related to city FM rates; the combined index has a t-statistic above four.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "City FM rates and the misbehavior index ranking are strongly correlated (rho of about 0.60).", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Including the misbehavior index eliminates significant excess clustering of FM across cities.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Enforcement proxies and firm characteristics do not explain the city differences in FM.", "evidence_type": "table", "evidence_ref": "Table VII, Table VIII"}], "structural_summary": "Opens with the gap between Becker-style crime models and observed spatial variation in crime, then previews numbered findings and walks through the sections. A stylized cost-benefit model motivates three candidate explanations: norms, enforcement, and firm heterogeneity. The analysis first measures excess clustering of FM with count models, then relates FM to observable regional traits and to an index of non-FM misbehavior. Separate sections rule out enforcement and firm-characteristic explanations. The paper closes by acknowledging that the sources of norms remain unidentified and flags them for future research.", "writing_profile_keywords": ["social norms", "geography", "count models", "measurement-first", "alternative explanations", "rare events", "culture and finance"], "confidence": 0.82, "notes": "The introduction's numbered findings list is split across paragraphs 2-9. Causal language is cautious: norms 'appear' to matter, and the authors explicitly acknowledge the reflection problem and that the sources of norms are unidentified."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.04, "usage": {"input": 8535, "output": 2236, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079654, "request_id": "req_011Cggq3fD3bwhUtKUdbbWbB", "cached": false, "ts": "2026-11-03T23:03:25+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/8930267bbf8a8193a4525d0abe3b5938c92ae998360d0a3a65f6b6fa919d694f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between an individual's self-efficacy (Pearlin Mastery score percentile) and indicators of financial delinquency, its consequences, and financial preparedness, plus heterogeneity after job-loss and health shocks\",\"strategy\":\"Linear probability models on NLSY panel data. These condition on cognitive ability, risk tolerance, time preference, education, income, net worth and Big-Five traits. Sibling-group (mother) fixed effects absorb shared family background, and interactions with shocks and parental net worth capture heterogeneity.\",\"key_assumptions\":[\"Conditional on observed abilities, preferences and socioeconomic controls, remaining variation in self-efficacy is not driven by omitted factors that also affect distress\",\"Within-sibling differences in self-efficacy are not correlated with unobserved determinants of delinquency\",\"Self-efficacy is relatively stable over time, so it is not mechanically caused by distress\"],\"threats_addressed\":[\"confounding by cognitive ability and preferences\",\"confounding by education, earnings and net worth\",\"shared parental support (sibling fixed effects)\",\"other personality traits (Big-Five)\",\"higher indebtedness as alternative explanation\",\"optimism and time preferences as alternative hypotheses\",\"sample construction and measurement of self-efficacy\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"other\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes open by stating what the table reports, then define the dependent-variable indicators and sample restrictions. Most coefficient tables are linear probability models with standard errors clustered by individual. Self-efficacy enters as a percentile rank. The correlation table marks 5% significance with an asterisk. Robustness tables are organized in panels covering survey weights, person-level collapse and alternative measures.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"predicts/correlated wording alongside effect wording\",\"explicit caveat that relationships may not be causal\",\"'consistent with' framing of mechanism\",\"acknowledgment that omitted variables bias is possible\",\"'appear to' qualifiers\"]},\"claim_evidence\":[{\"claim\":\"Higher self-efficacy is associated with lower delinquency on debt and bill payments\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 2\"},{\"claim\":\"Higher self-efficacy is associated with fewer collections, foreclosures, repossessions and credit-related consequences\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV and V\"},{\"claim\":\"The negative relation survives sibling-group fixed effects, so shared parental support does not drive it\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The self-efficacy–default relation is strongest for individuals from the poorest parental backgrounds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"A one-SD increase in self-efficacy raises the likelihood of emergency savings by about 2 percentage points\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Older adults with higher self-efficacy save, plan for retirement and buy long-term care insurance more\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Low self-efficacy individuals default much more after job loss or illness; high self-efficacy individuals show little increase\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Results are robust to Big-Five traits, optimism, time preferences, survey weights and alternative self-efficacy measures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XI and XII\"},{\"claim\":\"Self-efficacy is relatively stable over the life course\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Gecas (1989); Section II.C\"}],\"structural_summary\":\"The paper motivates the question with intertemporal trade-offs, where perceived effectiveness of costly actions shapes decisions. It then sketches an effort-choice mechanism and previews the approach: survey panel regressions with rich controls and sibling fixed effects. Results are previewed with magnitudes and an explicit non-causality caveat. Positioning is against the cognitive-ability and subjective-expectations literatures, followed by a roadmap. A conceptual section defines and measures self-efficacy, and the data section documents the stability of the measure. Results then proceed from delinquency to its consequences, the parental-support mechanism, preparedness, and shock interactions. A separate section addresses alternative hypotheses and robustness. The conclusion restates the findings and notes that noncognitive skills may be malleable.\",\"writing_profile_keywords\":[\"household finance\",\"noncognitive skills\",\"survey panel data\",\"linear probability models\",\"sibling fixed effects\",\"heterogeneity by shocks\",\"explicit causality caveat\",\"psychology-informed\"],\"confidence\":0.82,\"notes\":\"Causal wording ('effects', 'determinant', 'raises') coexists with an explicit caveat that the estimated relationships may not be causal. I therefore coded strength 3. Section IV's rule-based role was 'theory', but its content is alternative hypotheses and robustness. Section I is a conceptual/mechanism section without a formal model and is coded as theory. Magnitudes appear in the introduction but not in the abstract.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between an individual's self-efficacy (Pearlin Mastery score percentile) and indicators of financial delinquency, its consequences, and financial preparedness, plus heterogeneity after job-loss and health shocks", "strategy": "Linear probability models on NLSY panel data. These condition on cognitive ability, risk tolerance, time preference, education, income, net worth and Big-Five traits. Sibling-group (mother) fixed effects absorb shared family background, and interactions with shocks and parental net worth capture heterogeneity.", "key_assumptions": ["Conditional on observed abilities, preferences and socioeconomic controls, remaining variation in self-efficacy is not driven by omitted factors that also affect distress", "Within-sibling differences in self-efficacy are not correlated with unobserved determinants of delinquency", "Self-efficacy is relatively stable over time, so it is not mechanically caused by distress"], "threats_addressed": ["confounding by cognitive ability and preferences", "confounding by education, earnings and net worth", "shared parental support (sibling fixed effects)", "other personality traits (Big-Five)", "higher indebtedness as alternative explanation", "optimism and time preferences as alternative hypotheses", "sample construction and measurement of self-efficacy"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "other"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Notes open by stating what the table reports, then define the dependent-variable indicators and sample restrictions. Most coefficient tables are linear probability models with standard errors clustered by individual. Self-efficacy enters as a percentile rank. The correlation table marks 5% significance with an asterisk. Robustness tables are organized in panels covering survey weights, person-level collapse and alternative measures.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["predicts/correlated wording alongside effect wording", "explicit caveat that relationships may not be causal", "'consistent with' framing of mechanism", "acknowledgment that omitted variables bias is possible", "'appear to' qualifiers"]}, "claim_evidence": [{"claim": "Higher self-efficacy is associated with lower delinquency on debt and bill payments", "evidence_type": "table", "evidence_ref": "Table III; Figure 2"}, {"claim": "Higher self-efficacy is associated with fewer collections, foreclosures, repossessions and credit-related consequences", "evidence_type": "table", "evidence_ref": "Tables IV and V"}, {"claim": "The negative relation survives sibling-group fixed effects, so shared parental support does not drive it", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The self-efficacy–default relation is strongest for individuals from the poorest parental backgrounds", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "A one-SD increase in self-efficacy raises the likelihood of emergency savings by about 2 percentage points", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Older adults with higher self-efficacy save, plan for retirement and buy long-term care insurance more", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Low self-efficacy individuals default much more after job loss or illness; high self-efficacy individuals show little increase", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Results are robust to Big-Five traits, optimism, time preferences, survey weights and alternative self-efficacy measures", "evidence_type": "table", "evidence_ref": "Tables XI and XII"}, {"claim": "Self-efficacy is relatively stable over the life course", "evidence_type": "prior_literature", "evidence_ref": "Gecas (1989); Section II.C"}], "structural_summary": "The paper motivates the question with intertemporal trade-offs, where perceived effectiveness of costly actions shapes decisions. It then sketches an effort-choice mechanism and previews the approach: survey panel regressions with rich controls and sibling fixed effects. Results are previewed with magnitudes and an explicit non-causality caveat. Positioning is against the cognitive-ability and subjective-expectations literatures, followed by a roadmap. A conceptual section defines and measures self-efficacy, and the data section documents the stability of the measure. Results then proceed from delinquency to its consequences, the parental-support mechanism, preparedness, and shock interactions. A separate section addresses alternative hypotheses and robustness. The conclusion restates the findings and notes that noncognitive skills may be malleable.", "writing_profile_keywords": ["household finance", "noncognitive skills", "survey panel data", "linear probability models", "sibling fixed effects", "heterogeneity by shocks", "explicit causality caveat", "psychology-informed"], "confidence": 0.82, "notes": "Causal wording ('effects', 'determinant', 'raises') coexists with an explicit caveat that the estimated relationships may not be causal. I therefore coded strength 3. Section IV's rule-based role was 'theory', but its content is alternative hypotheses and robustness. Section I is a conceptual/mechanism section without a formal model and is coded as theory. Magnitudes appear in the introduction but not in the abstract."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 28.36, "usage": {"input": 7318, "output": 3355, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.097166, "request_id": "req_011Cggq9ZRCDpHtQmLfbSoSD", "cached": false, "ts": "2026-11-03T23:04:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/8b377d5e496a28154da743125af2cdb1fddf102d8d553b9e6ba5d716293dead6.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"political economy / public procurement\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Change in procurement contract volume and contract performance for private firms whose CEO shares a network with the state firm CEO, before vs. after the presidential election\",\"strategy\":\"Difference-in-differences around Lee Myung Bak's 2007 nomination and election, which shifted firms' political connectedness. Connections are defined through CEO networks. Contracts from connected vs. nonconnected state firms to the same private firm are compared, supplemented by stock-return event responses.\",\"key_assumptions\":[\"The election and nomination provide exogenous variation in connectedness\",\"Without the new connections, connected and nonconnected firms would have followed parallel trends\",\"State firm CEO appointments are not driven by the expected allocation of contracts\",\"Adverse outcomes are reported equally for connected and nonconnected contracts\"],\"threats_addressed\":[\"the new president's political agenda\",\"endogenous CEO appointments in state firms\",\"differences in observable contract characteristics\",\"differences in reporting of adverse outcomes\",\"placebo test using state firms that already had a KU-network CEO\",\"effects of other (non-presidential) alumni networks\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"placebo\",\"alternative_measure\",\"alternative_specification\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define dummy variables, event windows (pre- and post-election periods by quarter) and dependent variables, such as annualized contract volume changes scaled by assets and binary performance indicators. Results are organized by panels. Standard error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggest\",\"appear\",\"consistent with\",\"rough approximations only\",\"it is likely that\"]},\"claim_evidence\":[{\"claim\":\"Connected firms' stock returns fell after the scandal news and rose after the nomination\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figures 3-4\"},{\"claim\":\"Connected private firms received more procurement contracts after the election, driven by state firms with a same-network CEO\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 5\"},{\"claim\":\"No increase from state firms that already had a KU-network CEO before the election\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Same-alumni-network connections raise contract volume even for networks unrelated to the president\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Connected contracts are about 10 pp more likely to have adverse outcomes and about 5 pp more likely to have cost increases\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 6\"},{\"claim\":\"Performance differences are not explained by observable contract characteristics\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-IX\"}],\"structural_summary\":\"The introduction frames a debate over whether political connections create rents or efficiency gains. It then previews the Korean setting, the network-based allocation channel, efficiency results with magnitudes, robustness checks and contributions. The body moves through institutional background, data and empirical design, followed by market-value validation of the connection measure and a long results section with mechanism and robustness subsections. Back-of-the-envelope calculations of rents and social costs come next. The conclusion restates findings and draws procurement policy implications.\",\"writing_profile_keywords\":[\"political connections\",\"public procurement\",\"natural experiment\",\"social networks\",\"allocative efficiency\",\"single-author\",\"Korea\"],\"confidence\":0.8,\"notes\":\"Section IV validates the connection measure using stock-price reactions and real firm effects. The Institutional Background section is labeled 'literature' by the rule but functions as setting description, so it is annotated as 'other'. Standard error type is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "political economy / public procurement", "primary_method": "difference_in_differences", "identification": {"estimand": "Change in procurement contract volume and contract performance for private firms whose CEO shares a network with the state firm CEO, before vs. after the presidential election", "strategy": "Difference-in-differences around Lee Myung Bak's 2007 nomination and election, which shifted firms' political connectedness. Connections are defined through CEO networks. Contracts from connected vs. nonconnected state firms to the same private firm are compared, supplemented by stock-return event responses.", "key_assumptions": ["The election and nomination provide exogenous variation in connectedness", "Without the new connections, connected and nonconnected firms would have followed parallel trends", "State firm CEO appointments are not driven by the expected allocation of contracts", "Adverse outcomes are reported equally for connected and nonconnected contracts"], "threats_addressed": ["the new president's political agenda", "endogenous CEO appointments in state firms", "differences in observable contract characteristics", "differences in reporting of adverse outcomes", "placebo test using state firms that already had a KU-network CEO", "effects of other (non-presidential) alumni networks"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["placebo", "alternative_measure", "alternative_specification", "falsification", "other"]}, "tables": {"convention_notes": "Table notes define dummy variables, event windows (pre- and post-election periods by quarter) and dependent variables, such as annualized contract volume changes scaled by assets and binary performance indicators. Results are organized by panels. Standard error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggest", "appear", "consistent with", "rough approximations only", "it is likely that"]}, "claim_evidence": [{"claim": "Connected firms' stock returns fell after the scandal news and rose after the nomination", "evidence_type": "table", "evidence_ref": "Table II; Figures 3-4"}, {"claim": "Connected private firms received more procurement contracts after the election, driven by state firms with a same-network CEO", "evidence_type": "table", "evidence_ref": "Table IV; Figure 5"}, {"claim": "No increase from state firms that already had a KU-network CEO before the election", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Same-alumni-network connections raise contract volume even for networks unrelated to the president", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Connected contracts are about 10 pp more likely to have adverse outcomes and about 5 pp more likely to have cost increases", "evidence_type": "table", "evidence_ref": "Table VII; Figure 6"}, {"claim": "Performance differences are not explained by observable contract characteristics", "evidence_type": "table", "evidence_ref": "Tables VIII-IX"}], "structural_summary": "The introduction frames a debate over whether political connections create rents or efficiency gains. It then previews the Korean setting, the network-based allocation channel, efficiency results with magnitudes, robustness checks and contributions. The body moves through institutional background, data and empirical design, followed by market-value validation of the connection measure and a long results section with mechanism and robustness subsections. Back-of-the-envelope calculations of rents and social costs come next. The conclusion restates findings and draws procurement policy implications.", "writing_profile_keywords": ["political connections", "public procurement", "natural experiment", "social networks", "allocative efficiency", "single-author", "Korea"], "confidence": 0.8, "notes": "Section IV validates the connection measure using stock-price reactions and real firm effects. The Institutional Background section is labeled 'literature' by the rule but functions as setting description, so it is annotated as 'other'. Standard error type is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.69, "usage": {"input": 8134, "output": 2145, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07623, "request_id": "req_011CggqFpKpAmYDmVkifPVwV", "cached": false, "ts": "2026-11-03T23:06:09+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/8b459b7ea7ad1ee09aa4bb129b2368013603ce147a83b1c985ecdb3ded528cd6.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"international_finance; macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of foreign (U.S., U.K., Eurozone) monetary policy rate shocks and QE on credit supply of the corresponding foreign banks in Mexico, on borrower risk-taking, and on firm-level real outcomes\",\"strategy\":\"Loan-level supervisory credit register panel with borrower or borrower×month fixed effects (Khwaja–Mian style), interacting foreign policy shocks with bank nationality. Foreign policy is treated as exogenous to Mexico. Firm×bank and state×industry×month fixed effects are used for single-bank borrowers. Firm-level regressions capture real effects and credit substitution; bank-level regressions capture liabilities and risk.\",\"key_assumptions\":[\"Foreign monetary policy is exogenous to Mexican economic conditions\",\"Borrower×time fixed effects absorb credit demand and firm fundamentals\",\"Taylor-rule residuals proxy for policy rate shocks\",\"Central bank balance sheet to GDP proxies for QE\"],\"threats_addressed\":[\"Credit demand confounding\",\"Foreign banks lending to different types of firms\",\"Global shocks (period fixed effects)\",\"Firms substituting toward other credit sources\",\"Endogeneity of local monetary policy\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":7,\"contribution_types\":[\"new_identification\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Table notes give the estimator (OLS), sample period, and level of aggregation (firm-bank-month or firm-year). They define dependent variables and sample classification rules. Appendix tables define variables with units (thousands of pesos, months). Lagged-policy panels are reported separately.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"results suggest\",\"consistent with\",\"proxy for\",\"may\",\"mainly/mostly\"]},\"claim_evidence\":[{\"claim\":\"Foreign policy rate cuts raise credit supply mainly via banks of the same nationality in Mexico, e.g., about 2.1% higher volume per one-SD cut.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II panels\"},{\"claim\":\"The Mexican policy rate affects lending by all banks regardless of nationality, with smaller magnitude.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II Panel A\"},{\"claim\":\"Foreign monetary softening produces firm-level real effects on assets, investment, employment and survival, net of credit substitution.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Credit expansion and ex post defaults rise for borrowers with high ex ante loan rates, consistent with reach-for-yield.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Under foreign softening, foreign subsidiaries increase foreign and short-term liabilities, expand balance sheets and experience more defaults.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"bank-level results; table not visible in excerpts\"},{\"claim\":\"Strong credit growth financed by foreign liabilities predicts financial crises.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Jorda, Schularick, and Taylor (2011); Gourinchas and Obstfeld (2012)\"}],\"structural_summary\":\"Opens with macro-financial and policy motivation, citing policymakers' concerns about spillovers. Describes identification via loan-level Mexican register data with fixed effects and exogenous foreign policy shocks. Previews loan-level, risk-taking, bank-level and real-effect results with magnitudes. Sketches a reach-for-yield mechanism, states the identification contribution against the international and risk-taking literatures, and gives a roadmap. A single empirical strategy section is followed by a results section organized by loan, firm, and risk-taking outcomes. The conclusion restates results and offers policy coordination implications and future research.\",\"writing_profile_keywords\":[\"credit register\",\"Khwaja-Mian fixed effects\",\"international spillovers\",\"risk-taking channel\",\"policy-oriented\",\"magnitude-rich introduction\",\"QE\"],\"confidence\":0.8,\"notes\":\"Section 3 appears to be a parsing artifact (a table panel title) that continues the results section. Standard error treatment is not visible in the excerpts. A robustness mention ('robust results') appears, and lagged-policy specifications are in Table IIC. An internet appendix exists.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "international_finance; macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of foreign (U.S., U.K., Eurozone) monetary policy rate shocks and QE on credit supply of the corresponding foreign banks in Mexico, on borrower risk-taking, and on firm-level real outcomes", "strategy": "Loan-level supervisory credit register panel with borrower or borrower×month fixed effects (Khwaja–Mian style), interacting foreign policy shocks with bank nationality. Foreign policy is treated as exogenous to Mexico. Firm×bank and state×industry×month fixed effects are used for single-bank borrowers. Firm-level regressions capture real effects and credit substitution; bank-level regressions capture liabilities and risk.", "key_assumptions": ["Foreign monetary policy is exogenous to Mexican economic conditions", "Borrower×time fixed effects absorb credit demand and firm fundamentals", "Taylor-rule residuals proxy for policy rate shocks", "Central bank balance sheet to GDP proxies for QE"], "threats_addressed": ["Credit demand confounding", "Foreign banks lending to different types of firms", "Global shocks (period fixed effects)", "Firms substituting toward other credit sources", "Endogeneity of local monetary policy"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 7, "contribution_types": ["new_identification", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_sample_period"]}, "tables": {"convention_notes": "Table notes give the estimator (OLS), sample period, and level of aggregation (firm-bank-month or firm-year). They define dependent variables and sample classification rules. Appendix tables define variables with units (thousands of pesos, months). Lagged-policy panels are reported separately.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["results suggest", "consistent with", "proxy for", "may", "mainly/mostly"]}, "claim_evidence": [{"claim": "Foreign policy rate cuts raise credit supply mainly via banks of the same nationality in Mexico, e.g., about 2.1% higher volume per one-SD cut.", "evidence_type": "table", "evidence_ref": "Table II panels"}, {"claim": "The Mexican policy rate affects lending by all banks regardless of nationality, with smaller magnitude.", "evidence_type": "table", "evidence_ref": "Table II Panel A"}, {"claim": "Foreign monetary softening produces firm-level real effects on assets, investment, employment and survival, net of credit substitution.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Credit expansion and ex post defaults rise for borrowers with high ex ante loan rates, consistent with reach-for-yield.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Under foreign softening, foreign subsidiaries increase foreign and short-term liabilities, expand balance sheets and experience more defaults.", "evidence_type": "none_stated", "evidence_ref": "bank-level results; table not visible in excerpts"}, {"claim": "Strong credit growth financed by foreign liabilities predicts financial crises.", "evidence_type": "prior_literature", "evidence_ref": "Jorda, Schularick, and Taylor (2011); Gourinchas and Obstfeld (2012)"}], "structural_summary": "Opens with macro-financial and policy motivation, citing policymakers' concerns about spillovers. Describes identification via loan-level Mexican register data with fixed effects and exogenous foreign policy shocks. Previews loan-level, risk-taking, bank-level and real-effect results with magnitudes. Sketches a reach-for-yield mechanism, states the identification contribution against the international and risk-taking literatures, and gives a roadmap. A single empirical strategy section is followed by a results section organized by loan, firm, and risk-taking outcomes. The conclusion restates results and offers policy coordination implications and future research.", "writing_profile_keywords": ["credit register", "Khwaja-Mian fixed effects", "international spillovers", "risk-taking channel", "policy-oriented", "magnitude-rich introduction", "QE"], "confidence": 0.8, "notes": "Section 3 appears to be a parsing artifact (a table panel title) that continues the results section. Standard error treatment is not visible in the excerpts. A robustness mention ('robust results') appears, and lagged-policy specifications are in Table IIC. An internet appendix exists."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.24, "usage": {"input": 7469, "output": 2316, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07699, "request_id": "req_011CggqEP9CqRshJLfZq9a2i", "cached": false, "ts": "2026-11-03T23:05:51+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/902f017174cc08dac668fe95224dd85483e898a2974b1f799c961db6145a4690.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"organizational economics / internal capital markets; contract and mechanism design\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; the paper does an analytical mechanism-design analysis of a principal-agent model with private information\",\"key_assumptions\":[\"risk-neutral economy\",\"worker (agent) is penniless, has limited liability and a zero reservation wage\",\"agent's effort is unobservable\",\"executive privately observes a binary productivity shock\",\"effort is more productive when the firm's prospects are better\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":2},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"There are no tables. Results are stated as propositions, with proofs in Appendix A.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"modal verbs ('may distort', 'can arise')\",\"'suggests' wording for model implications\",\"'consistent with' when linking predictions to prior evidence\",\"conditional framing ('in some situations')\"]},\"claim_evidence\":[{\"claim\":\"When executives are privately informed, an unconstrained executive tends to overinvest relative to first best, because higher investment signals better prospects and induces more worker effort.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B.3; Appendix A proofs\"},{\"claim\":\"Investment rigidities, implemented as pooling mechanisms, can be an optimal response to the tendency to overinvest.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B.4-B.5; Appendix A\"},{\"claim\":\"Multidivisional firms show investment distortions that differ from those of single-division firms, with partial pooling mechanisms examined.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III\"},{\"claim\":\"In a continuum-of-types extension, overinvestment persists and the optimal mechanism pools types at the bottom of the distribution.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"Minimum investment thresholds, lumpy investment and expenditure caps observed in firms are consistent with the model's predicted rigidities.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section V; Cooper and Haltiwanger (2006), Ross (1986)\"}],\"structural_summary\":\"A pure theory paper on top-down capital budgeting.\\n- The introduction motivates the question, outlines the principal-agent setting (an executive whose investment choices signal information to a worker), and previews the main results.\\n- The model section sets up the game. The analysis section first solves a known-type benchmark, then the private-information case: optimal separating vs. pooling mechanisms and when each is preferred.\\n- A separate section extends the analysis to a two-division firm and characterizes partial-pooling mechanisms.\\n- Short sections then discuss extensions (e.g., a continuum of types) and empirical implications tied to prior evidence.\\n- The conclusion restates the contribution. Proofs are in a long appendix.\",\"writing_profile_keywords\":[\"mechanism design\",\"signaling\",\"principal-agent\",\"capital budgeting\",\"overinvestment\",\"pooling vs separating\",\"proposition-proof structure\",\"empirical implications section\"],\"confidence\":0.8,\"notes\":\"The introduction is given as a single truncated paragraph, so moves beyond it (e.g., literature positioning, roadmap) are not visible. The literature-positioning value is assigned on the assumption that it sits inside the introduction, since no separate literature section exists. The Extensions section is coded 'other' because it extends the theory rather than performing empirical robustness checks. Figures are model illustrations without uncertainty bands.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "organizational economics / internal capital markets; contract and mechanism design", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; the paper does an analytical mechanism-design analysis of a principal-agent model with private information", "key_assumptions": ["risk-neutral economy", "worker (agent) is penniless, has limited liability and a zero reservation wage", "agent's effort is unobservable", "executive privately observes a binary productivity shock", "effort is more productive when the firm's prospects are better"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 2}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 0, "move": "preview_of_results"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["other"]}, "tables": {"convention_notes": "There are no tables. Results are stated as propositions, with proofs in Appendix A.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["modal verbs ('may distort', 'can arise')", "'suggests' wording for model implications", "'consistent with' when linking predictions to prior evidence", "conditional framing ('in some situations')"]}, "claim_evidence": [{"claim": "When executives are privately informed, an unconstrained executive tends to overinvest relative to first best, because higher investment signals better prospects and induces more worker effort.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B.3; Appendix A proofs"}, {"claim": "Investment rigidities, implemented as pooling mechanisms, can be an optimal response to the tendency to overinvest.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B.4-B.5; Appendix A"}, {"claim": "Multidivisional firms show investment distortions that differ from those of single-division firms, with partial pooling mechanisms examined.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III"}, {"claim": "In a continuum-of-types extension, overinvestment persists and the optimal mechanism pools types at the bottom of the distribution.", "evidence_type": "none_stated", "evidence_ref": "Section IV"}, {"claim": "Minimum investment thresholds, lumpy investment and expenditure caps observed in firms are consistent with the model's predicted rigidities.", "evidence_type": "prior_literature", "evidence_ref": "Section V; Cooper and Haltiwanger (2006), Ross (1986)"}], "structural_summary": "A pure theory paper on top-down capital budgeting.\n- The introduction motivates the question, outlines the principal-agent setting (an executive whose investment choices signal information to a worker), and previews the main results.\n- The model section sets up the game. The analysis section first solves a known-type benchmark, then the private-information case: optimal separating vs. pooling mechanisms and when each is preferred.\n- A separate section extends the analysis to a two-division firm and characterizes partial-pooling mechanisms.\n- Short sections then discuss extensions (e.g., a continuum of types) and empirical implications tied to prior evidence.\n- The conclusion restates the contribution. Proofs are in a long appendix.", "writing_profile_keywords": ["mechanism design", "signaling", "principal-agent", "capital budgeting", "overinvestment", "pooling vs separating", "proposition-proof structure", "empirical implications section"], "confidence": 0.8, "notes": "The introduction is given as a single truncated paragraph, so moves beyond it (e.g., literature positioning, roadmap) are not visible. The literature-positioning value is assigned on the assumption that it sits inside the introduction, since no separate literature section exists. The Extensions section is coded 'other' because it extends the theory rather than performing empirical robustness checks. Figures are model illustrations without uncertainty bands."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.72, "usage": {"input": 3773, "output": 1852, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.052926, "request_id": "req_011Cggpi94nVZTmVD7qfL9Fi", "cached": false, "ts": "2026-11-03T22:58:58+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9116af05cca6c59a9864a9965c8298b9413155362f97a718c4c537eb5da94e81.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of the intensity of a CEO's early-life exposure to fatal natural disasters (none vs. medium vs. extreme fatality) on corporate risk-taking policies and outcomes.\",\"strategy\":\"Panel OLS and linear probability models using early-life disaster exposure inferred from birth county and year. Specifications include firm, year, CEO birth-year and CEO birth-state fixed effects and control for the county's historical disaster incidence. Supplemented by a difference-in-differences comparison with neighboring unaffected counties, a placebo with random reassignment of birth county, and an analysis of exogenous CEO turnovers.\",\"key_assumptions\":[\"Disaster exposure during the formative years is exogenous to later corporate outcomes, conditional on fixed effects and historical county risk.\",\"Birth county proxies for where the CEO lived during childhood.\",\"Disaster experience does not affect selection into becoming a CEO, nor CEO-firm matching in ways that drive results.\"],\"threats_addressed\":[\"cohort effects\",\"growing up in a high-risk county vs. actually experiencing a disaster\",\"measurement error in childhood residence\",\"omitted variables\",\"CEO-firm matching\",\"spurious correlation (placebo)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"literature_gap\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":10,\"move\":\"identification_defense\"},{\"paragraph_index\":10,\"move\":\"robustness_mention\"},{\"paragraph_index\":11,\"move\":\"contribution_statement\"},{\"paragraph_index\":12,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":13,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":11,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"placebo\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state the estimator (OLS or LPM), the dependent variable by column, and the fixed-effect set (firm, year, CEO birth year, CEO birth state). Descriptive tables come first, followed by policy-by-policy regression tables, then robustness tables (placebo, DiD-style treated vs. control, exogenous turnovers). The excerpts do not show how standard errors are computed.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'results support the view/hypothesis'\",\"'appear to be more cautious'\",\"'suggests that our results may have broader implications'\",\"'all else equal' framing\"]},\"claim_evidence\":[{\"claim\":\"Inverse U-shaped relation between the intensity of CEO fatal disaster exposure and firm leverage and cash policy.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Leverage differences reflect active financing choices: moderate-disaster CEOs fill financing deficits with debt.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Moderate-disaster CEOs face higher interest costs and loan spreads, lower ratings and more bankruptcy; extreme-disaster CEOs show the reverse.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"Acquisition frequency, payment method, diversification and announcement returns vary nonmonotonically with disaster intensity.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"Moderate-disaster CEOs run firms with higher total and idiosyncratic volatility.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"CEOs' childhood disaster exposure resembles that of the general U.S. population, suggesting no selection into CEO role.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Results survive random reassignment placebo, treated vs. neighboring control counties, and exogenous turnovers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XI, XII, XIII\"}],\"structural_summary\":\"The paper opens with an epigraph and a literature-based observation that prior work assumes a monotonic link between experience and managerial risk-taking, then poses a dosage (intensity) hypothesis. The introduction previews the measure, policy-by-policy results with magnitudes, and the interpretation (desensitization vs. caution). It then gives an identification defense (fixed effects, historical-risk control, DiD), contribution claims, broader implications and a roadmap. A short motivation section grounds the hypothesis in the literature, followed by data construction, main results across financing, credit, acquisitions and volatility, and a dedicated robustness section. The conclusion restates the findings and points to investor-behavior research.\",\"writing_profile_keywords\":[\"managerial style\",\"early-life experience\",\"nonmonotonic effect\",\"natural disasters\",\"fixed effects\",\"dosage metaphor\",\"epigraph opening\",\"policy-by-policy results\"],\"confidence\":0.8,\"notes\":\"The Motivation section (I) functions as literature-based hypothesis development; it is coded as literature. The excerpts do not show standard errors or how significance is reported. The introduction's first five paragraph entries are fragments of the epigraph, so the substantive intro text begins at index 5.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of the intensity of a CEO's early-life exposure to fatal natural disasters (none vs. medium vs. extreme fatality) on corporate risk-taking policies and outcomes.", "strategy": "Panel OLS and linear probability models using early-life disaster exposure inferred from birth county and year. Specifications include firm, year, CEO birth-year and CEO birth-state fixed effects and control for the county's historical disaster incidence. Supplemented by a difference-in-differences comparison with neighboring unaffected counties, a placebo with random reassignment of birth county, and an analysis of exogenous CEO turnovers.", "key_assumptions": ["Disaster exposure during the formative years is exogenous to later corporate outcomes, conditional on fixed effects and historical county risk.", "Birth county proxies for where the CEO lived during childhood.", "Disaster experience does not affect selection into becoming a CEO, nor CEO-firm matching in ways that drive results."], "threats_addressed": ["cohort effects", "growing up in a high-risk county vs. actually experiencing a disaster", "measurement error in childhood residence", "omitted variables", "CEO-firm matching", "spurious correlation (placebo)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": true, "sentence_index": 0}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "literature_gap"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 10, "move": "identification_defense"}, {"paragraph_index": 10, "move": "robustness_mention"}, {"paragraph_index": 11, "move": "contribution_statement"}, {"paragraph_index": 12, "move": "policy_or_broader_implication"}, {"paragraph_index": 13, "move": "roadmap"}], "contribution_paragraph_index": 11, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_measure", "placebo", "alternative_specification", "other"]}, "tables": {"convention_notes": "Notes state the estimator (OLS or LPM), the dependent variable by column, and the fixed-effect set (firm, year, CEO birth year, CEO birth state). Descriptive tables come first, followed by policy-by-policy regression tables, then robustness tables (placebo, DiD-style treated vs. control, exogenous turnovers). The excerpts do not show how standard errors are computed.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'results support the view/hypothesis'", "'appear to be more cautious'", "'suggests that our results may have broader implications'", "'all else equal' framing"]}, "claim_evidence": [{"claim": "Inverse U-shaped relation between the intensity of CEO fatal disaster exposure and firm leverage and cash policy.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Leverage differences reflect active financing choices: moderate-disaster CEOs fill financing deficits with debt.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Moderate-disaster CEOs face higher interest costs and loan spreads, lower ratings and more bankruptcy; extreme-disaster CEOs show the reverse.", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "Acquisition frequency, payment method, diversification and announcement returns vary nonmonotonically with disaster intensity.", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "Moderate-disaster CEOs run firms with higher total and idiosyncratic volatility.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "CEOs' childhood disaster exposure resembles that of the general U.S. population, suggesting no selection into CEO role.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Results survive random reassignment placebo, treated vs. neighboring control counties, and exogenous turnovers.", "evidence_type": "table", "evidence_ref": "Tables XI, XII, XIII"}], "structural_summary": "The paper opens with an epigraph and a literature-based observation that prior work assumes a monotonic link between experience and managerial risk-taking, then poses a dosage (intensity) hypothesis. The introduction previews the measure, policy-by-policy results with magnitudes, and the interpretation (desensitization vs. caution). It then gives an identification defense (fixed effects, historical-risk control, DiD), contribution claims, broader implications and a roadmap. A short motivation section grounds the hypothesis in the literature, followed by data construction, main results across financing, credit, acquisitions and volatility, and a dedicated robustness section. The conclusion restates the findings and points to investor-behavior research.", "writing_profile_keywords": ["managerial style", "early-life experience", "nonmonotonic effect", "natural disasters", "fixed effects", "dosage metaphor", "epigraph opening", "policy-by-policy results"], "confidence": 0.8, "notes": "The Motivation section (I) functions as literature-based hypothesis development; it is coded as literature. The excerpts do not show standard errors or how significance is reported. The introduction's first five paragraph entries are fragments of the epigraph, so the substantive intro text begins at index 5."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.99, "usage": {"input": 7888, "output": 2408, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080506, "request_id": "req_011CggpSEWGEgMr5GvRdYip3", "cached": false, "ts": "2026-11-03T22:55:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/91406d7ea71f4af36f093c845e488c680ac089006700f5254bce5b1f76854275.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"term structure modeling / fixed income derivatives\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Structural parameters of a linear-rational term structure model, plus model fit to swap rates and swaption implied volatilities\",\"strategy\":\"Quasi-maximum likelihood with the Kalman filter on a weekly panel of swaps and swaptions, followed by simulation-based checks of the model-implied dynamics against the data\",\"key_assumptions\":[\"Factor process has a linear drift and the state price density is linear in the factors\",\"Three term structure factors with one to three USV factors\",\"Quasi-ML approximation with the Kalman filter is adequate\"],\"threats_addressed\":[\"Model misspecification (compared across numbers of USV factors)\",\"Fit near the zero lower bound\",\"Benchmarking against an exponential-affine model\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Notes define variables, units (basis points, percent, annualized) and data periods. Parameter estimates are reported with asymptotic standard errors in parentheses and regression slopes with t-statistics in parentheses. Data and model-simulated panels are shown side by side. One table compares frameworks qualitatively.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'arguably' when describing the dominant literature\",\"qualified statements such as 'a very good fit' and 'many features'\",\"explicit statement of the model's focus or scope\"]},\"claim_evidence\":[{\"claim\":\"The linear-rational framework guarantees nonnegative short rates, accommodates USV, and gives semi-analytical swaption prices.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I theorems; Appendix A proofs\"},{\"claim\":\"No exponential-affine model can jointly ensure nonnegative rates and USV, or price swaptions semi-analytically.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Duffie and Kan (1996), Dai and Singleton (2000), Joslin (2014); Table I\"},{\"claim\":\"The preferred specification with three USV factors fits swaps and swaptions well, including near the ZLB.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III; Figure 2\"},{\"claim\":\"Volatility becomes compressed and level-dependent near the ZLB, with R2 around 0.5, but shows little level-dependence at moderate rates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figures 1, 7\"},{\"claim\":\"Near the ZLB the model generates asymmetric short-rate distributions and shifts the first PC from a level toward a slope factor.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5, 6\"},{\"claim\":\"The model captures unconditional and conditional swap risk premium features.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI; Figure 8\"},{\"claim\":\"On swaps alone, LRSQ pricing performance is virtually identical to an exponential-affine square-root model.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section IV.J\"}],\"structural_summary\":\"Opens by motivating the problem that low rates create for existing term structure models and states a twofold contribution: a new model class and an empirical analysis. It then sketches the state-price-density construction and previews the estimation design and the simulated results on term structure, volatility and risk premium dynamics near the ZLB. Positioning against the LG and exponential-affine frameworks follows, then a roadmap. The body develops the theory (framework, LRSQ model, permanent-component extension) before a long empirical section covering estimation, specification comparison, ZLB dynamics, risk premia and an affine benchmark. Proofs are in the appendix, and the conclusion restates the contribution and lists extensions.\",\"writing_profile_keywords\":[\"model class introduction\",\"tractability advantages enumerated\",\"Kalman filter quasi-ML\",\"simulation-based model evaluation\",\"zero lower bound\",\"unspanned stochastic volatility\",\"benchmark comparison\"],\"confidence\":0.85,\"notes\":\"Model-introduction paper with substantial estimation, so the genre is mixed. Causal identification is not applicable. Section III's role was mapped to theory.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "term structure modeling / fixed income derivatives", "primary_method": "structural_estimation", "identification": {"estimand": "Structural parameters of a linear-rational term structure model, plus model fit to swap rates and swaption implied volatilities", "strategy": "Quasi-maximum likelihood with the Kalman filter on a weekly panel of swaps and swaptions, followed by simulation-based checks of the model-implied dynamics against the data", "key_assumptions": ["Factor process has a linear drift and the state price density is linear in the factors", "Three term structure factors with one to three USV factors", "Quasi-ML approximation with the Kalman filter is adequate"], "threats_addressed": ["Model misspecification (compared across numbers of USV factors)", "Fit near the zero lower bound", "Benchmarking against an exponential-affine model"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "subsample"]}, "tables": {"convention_notes": "Notes define variables, units (basis points, percent, annualized) and data periods. Parameter estimates are reported with asymptotic standard errors in parentheses and regression slopes with t-statistics in parentheses. Data and model-simulated panels are shown side by side. One table compares frameworks qualitatively.", "units_stated_in_notes": true, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'arguably' when describing the dominant literature", "qualified statements such as 'a very good fit' and 'many features'", "explicit statement of the model's focus or scope"]}, "claim_evidence": [{"claim": "The linear-rational framework guarantees nonnegative short rates, accommodates USV, and gives semi-analytical swaption prices.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I theorems; Appendix A proofs"}, {"claim": "No exponential-affine model can jointly ensure nonnegative rates and USV, or price swaptions semi-analytically.", "evidence_type": "prior_literature", "evidence_ref": "Duffie and Kan (1996), Dai and Singleton (2000), Joslin (2014); Table I"}, {"claim": "The preferred specification with three USV factors fits swaps and swaptions well, including near the ZLB.", "evidence_type": "table", "evidence_ref": "Tables II, III; Figure 2"}, {"claim": "Volatility becomes compressed and level-dependent near the ZLB, with R2 around 0.5, but shows little level-dependence at moderate rates.", "evidence_type": "table", "evidence_ref": "Table IV; Figures 1, 7"}, {"claim": "Near the ZLB the model generates asymmetric short-rate distributions and shifts the first PC from a level toward a slope factor.", "evidence_type": "figure", "evidence_ref": "Figures 5, 6"}, {"claim": "The model captures unconditional and conditional swap risk premium features.", "evidence_type": "table", "evidence_ref": "Tables V, VI; Figure 8"}, {"claim": "On swaps alone, LRSQ pricing performance is virtually identical to an exponential-affine square-root model.", "evidence_type": "none_stated", "evidence_ref": "Section IV.J"}], "structural_summary": "Opens by motivating the problem that low rates create for existing term structure models and states a twofold contribution: a new model class and an empirical analysis. It then sketches the state-price-density construction and previews the estimation design and the simulated results on term structure, volatility and risk premium dynamics near the ZLB. Positioning against the LG and exponential-affine frameworks follows, then a roadmap. The body develops the theory (framework, LRSQ model, permanent-component extension) before a long empirical section covering estimation, specification comparison, ZLB dynamics, risk premia and an affine benchmark. Proofs are in the appendix, and the conclusion restates the contribution and lists extensions.", "writing_profile_keywords": ["model class introduction", "tractability advantages enumerated", "Kalman filter quasi-ML", "simulation-based model evaluation", "zero lower bound", "unspanned stochastic volatility", "benchmark comparison"], "confidence": 0.85, "notes": "Model-introduction paper with substantial estimation, so the genre is mixed. Causal identification is not applicable. Section III's role was mapped to theory."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.64, "usage": {"input": 7657, "output": 2259, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076602, "request_id": "req_011CggpXR9ceQyG8aAaTdHLQ", "cached": false, "ts": "2026-11-03T22:56:36+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/91a7b3a82e114ac7f442bc63c10ac9f63e8457cc63cbaf27394c19b6c3b0bf05.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"asset_pricing; corporate_finance (debt overhang, OTC derivatives valuation)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical: marginal shareholder and creditor value effects of debt-financed dealer positions)\",\"strategy\":\"Two-period structural balance-sheet model of a dealer with marginal valuation propositions, extended to a continuous-time reduced-form model calibrated for swap XVA numerics\",\"key_assumptions\":[\"risk-neutral pricing measure for claims on dealer assets\",\"frictional financial distress costs present (Modigliani-Miller holds without them)\",\"dealer quotes to maximize shareholder value\",\"finite expectations / regularity conditions on asset and liability distributions\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"},{\"idx\":13,\"role\":\"appendix\"},{\"idx\":14,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Two tables. One lists dealer FVA disclosures in currency units by bank and quarter. The other reports numerical XVAs for a 10-year swap in thousands of dollars, with paired payer|receiver cells and running-spread equivalents in basis points shown in parentheses. No regression output.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'essentially' / 'roughly' qualifiers\",\"conditional statements tied to distress-cost assumptions\",\"'may have arisen' / 'presumably' in interpretive passages\",\"explicit deferral of topics to future research\"]},\"claim_evidence\":[{\"claim\":\"FVAs equal the debt overhang cost to dealer shareholders of financing positions; they are transfers to creditors, not components of market value\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II, Propositions 1-2, Appendix B\"},{\"claim\":\"Shareholder-value-maximizing dealer quotes must widen bid-ask spreads to cover FVA\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.B\"},{\"claim\":\"Arbitraging a CIP violation benefits shareholders only if the basis roughly exceeds the dealer's credit spread\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"FVAs are economically large for plain-vanilla interest rate swaps\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Section V\"},{\"claim\":\"Major dealers report FVA charges of hundreds of millions of dollars\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I\"},{\"claim\":\"CDS big-bang up-front payments widened bid-ask spreads, consistent with the model\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Wang et al. (2016)\"},{\"claim\":\"A pecking order for financing methods favors certain sources from the shareholder viewpoint\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Appendix A, Proposition A1\"},{\"claim\":\"XVAs are nearly linear in the dealer's credit spread\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Appendix C\"}],\"structural_summary\":\"The paper opens with its central claim: FVAs are debt overhang costs to shareholders rather than market-value components. A toy T-bill example illustrates the inconsistency in current practice. The introduction then identifies the missing model as the gap and positions the paper relative to empirical evidence (CDS up-fronts, CIP). Section I reviews the practitioner controversy over FVA. A two-period structural marginal-valuation model follows, applied to unsecured swaps and dealer quotation and then to CIP arbitrage. A reduced-form continuous-time framework then delivers calibrated XVA magnitudes. A long concluding discussion summarizes results and covers incentives, competitive advantage, regulatory capital (KVA), and future research. Proofs and computational details are in the appendices.\",\"writing_profile_keywords\":[\"theory-driven\",\"practitioner controversy\",\"debt overhang\",\"structural plus reduced-form\",\"numerical illustration\",\"OTC dealer pricing\",\"proofs in appendix\",\"extended concluding discussion\"],\"confidence\":0.82,\"notes\":\"Theory paper with a calibrated numerical illustration; no estimation. Sections II-IV are theory (role 'other' in the input). Section V is classed as results because it reports numerical magnitudes. Appendix section headings are parsing artifacts. Robustness takes the form of parameter-sensitivity analysis (term structure, coupon, credit spread).\"}", "json": {"genre": "theoretical", "subfield": "financial_intermediation_banking", "subfield_secondary": "asset_pricing; corporate_finance (debt overhang, OTC derivatives valuation)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical: marginal shareholder and creditor value effects of debt-financed dealer positions)", "strategy": "Two-period structural balance-sheet model of a dealer with marginal valuation propositions, extended to a continuous-time reduced-form model calibrated for swap XVA numerics", "key_assumptions": ["risk-neutral pricing measure for claims on dealer assets", "frictional financial distress costs present (Modigliani-Miller holds without them)", "dealer quotes to maximize shareholder value", "finite expectations / regularity conditions on asset and liability distributions"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 5}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}, {"idx": 13, "role": "appendix"}, {"idx": 14, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters"]}, "tables": {"convention_notes": "Two tables. One lists dealer FVA disclosures in currency units by bank and quarter. The other reports numerical XVAs for a 10-year swap in thousands of dollars, with paired payer|receiver cells and running-spread equivalents in basis points shown in parentheses. No regression output.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'essentially' / 'roughly' qualifiers", "conditional statements tied to distress-cost assumptions", "'may have arisen' / 'presumably' in interpretive passages", "explicit deferral of topics to future research"]}, "claim_evidence": [{"claim": "FVAs equal the debt overhang cost to dealer shareholders of financing positions; they are transfers to creditors, not components of market value", "evidence_type": "equation_or_proof", "evidence_ref": "Section II, Propositions 1-2, Appendix B"}, {"claim": "Shareholder-value-maximizing dealer quotes must widen bid-ask spreads to cover FVA", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.B"}, {"claim": "Arbitraging a CIP violation benefits shareholders only if the basis roughly exceeds the dealer's credit spread", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV"}, {"claim": "FVAs are economically large for plain-vanilla interest rate swaps", "evidence_type": "table", "evidence_ref": "Table II, Section V"}, {"claim": "Major dealers report FVA charges of hundreds of millions of dollars", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I"}, {"claim": "CDS big-bang up-front payments widened bid-ask spreads, consistent with the model", "evidence_type": "prior_literature", "evidence_ref": "Wang et al. (2016)"}, {"claim": "A pecking order for financing methods favors certain sources from the shareholder viewpoint", "evidence_type": "equation_or_proof", "evidence_ref": "Appendix A, Proposition A1"}, {"claim": "XVAs are nearly linear in the dealer's credit spread", "evidence_type": "figure", "evidence_ref": "Appendix C"}], "structural_summary": "The paper opens with its central claim: FVAs are debt overhang costs to shareholders rather than market-value components. A toy T-bill example illustrates the inconsistency in current practice. The introduction then identifies the missing model as the gap and positions the paper relative to empirical evidence (CDS up-fronts, CIP). Section I reviews the practitioner controversy over FVA. A two-period structural marginal-valuation model follows, applied to unsecured swaps and dealer quotation and then to CIP arbitrage. A reduced-form continuous-time framework then delivers calibrated XVA magnitudes. A long concluding discussion summarizes results and covers incentives, competitive advantage, regulatory capital (KVA), and future research. Proofs and computational details are in the appendices.", "writing_profile_keywords": ["theory-driven", "practitioner controversy", "debt overhang", "structural plus reduced-form", "numerical illustration", "OTC dealer pricing", "proofs in appendix", "extended concluding discussion"], "confidence": 0.82, "notes": "Theory paper with a calibrated numerical illustration; no estimation. Sections II-IV are theory (role 'other' in the input). Section V is classed as results because it reports numerical magnitudes. Appendix section headings are parsing artifacts. Robustness takes the form of parameter-sensitivity analysis (term structure, coupon, credit spread)."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.75, "usage": {"input": 8775, "output": 2443, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084754, "request_id": "req_011CggqAifM61etHqPSKysKs", "cached": false, "ts": "2026-11-03T23:05:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/93ff9eb2c8766125b7caa3faa8bd577d003be9ced11f5f9843c630d421eb6304.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"public/municipal finance; fixed income valuation\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"NPV of advance refundings measured with option-adjusted callable bond valuation; associations between aggregate refunding activity or NPV losses and fiscal and macroeconomic conditions; cross-sectional probability of refunding\",\"strategy\":\"Each transaction is valued using municipal and SLGS term structures and Black-model volatilities. Aggregate time-series regressions use lagged explanatory variables, and a cross-sectional probit relates refunding probability to potential NPV and cash flow savings.\",\"key_assumptions\":[\"Valuation model and term structures correctly measure the continuation value of the callable bond\",\"Lagged fiscal variables proxy for financial constraints\"],\"threats_addressed\":[\"Covenant-elimination motive\",\"Informational frictions about the implicit option value (option ignored entirely)\",\"Transaction costs omitted, so losses are likely understated\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":3},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"data\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"results\"},{\"idx\":9,\"role\":\"conclusion\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"sensitivity_to_parameters\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Notes define variables and units (dollars, percentages, per $100 par, per CUSIP). Regression tables report lagged explanatory variables and include a probit model. A numerical example table illustrates the valuation logic. Standard error treatment is not visible.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggest' and 'consistent with the view'\",\"'may' for mechanisms\",\"explicit note that losses are likely underestimated\",\"'it is tempting to conclude ... however'\"]},\"claim_evidence\":[{\"claim\":\"About 85% of advance refundings occur at an NPV loss, with aggregate losses exceeding $15 billion\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figures 2-3\"},{\"claim\":\"More than 96% of refundings give short-term cash flow savings, averaging 141 bp of par\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 4\"},{\"claim\":\"Even if the implicit option is ignored, over half of refundings occur at a loss\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Refunding activity responds to potential cash flow savings but not to potential NPV\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Refunding activity rises with tax revenue shortfalls and budget deficits\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"NPV losses grow as tax collections fall and unfunded pension liabilities rise\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Potential cash flow savings drive the cross-sectional probability of refunding\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Average refunding NPV is negative in all 50 states, with heterogeneity across states\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The paper opens with the rapid growth of advance refundings and frames each one as an early synthetic call akin to American option exercise. It sets out an optimal-refunding valuation framework and gives a case study of political context. It then values over 206,000 transactions and documents widespread NPV losses alongside short-term cash flow savings. Alternative frictions (covenants, informational problems) are ruled out. Aggregate time-series and cross-sectional probit regressions link refunding activity to fiscal constraints. The conclusion restates the findings and discusses which other parties benefit from suboptimal refundings.\",\"writing_profile_keywords\":[\"municipal bonds\",\"option valuation\",\"NPV losses\",\"financial constraints\",\"time-series regressions\",\"case study\",\"policy relevance\"],\"confidence\":0.8,\"notes\":\"No explicit contribution paragraph is visible. Standard error and significance conventions are not visible in the excerpts. Section VI is treated as the robustness section because it rules out alternative explanations.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "public/municipal finance; fixed income valuation", "primary_method": "reduced_form_regression", "identification": {"estimand": "NPV of advance refundings measured with option-adjusted callable bond valuation; associations between aggregate refunding activity or NPV losses and fiscal and macroeconomic conditions; cross-sectional probability of refunding", "strategy": "Each transaction is valued using municipal and SLGS term structures and Black-model volatilities. Aggregate time-series regressions use lagged explanatory variables, and a cross-sectional probit relates refunding probability to potential NPV and cash flow savings.", "key_assumptions": ["Valuation model and term structures correctly measure the continuation value of the callable bond", "Lagged fiscal variables proxy for financial constraints"], "threats_addressed": ["Covenant-elimination motive", "Informational frictions about the implicit option value (option ignored entirely)", "Transaction costs omitted, so losses are likely understated"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 3}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "policy_or_broader_implication"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "data"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "results"}, {"idx": 9, "role": "conclusion"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["sensitivity_to_parameters", "alternative_measure"]}, "tables": {"convention_notes": "Notes define variables and units (dollars, percentages, per $100 par, per CUSIP). Regression tables report lagged explanatory variables and include a probit model. A numerical example table illustrates the valuation logic. Standard error treatment is not visible.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggest' and 'consistent with the view'", "'may' for mechanisms", "explicit note that losses are likely underestimated", "'it is tempting to conclude ... however'"]}, "claim_evidence": [{"claim": "About 85% of advance refundings occur at an NPV loss, with aggregate losses exceeding $15 billion", "evidence_type": "table", "evidence_ref": "Table III; Figures 2-3"}, {"claim": "More than 96% of refundings give short-term cash flow savings, averaging 141 bp of par", "evidence_type": "table", "evidence_ref": "Table III; Figure 4"}, {"claim": "Even if the implicit option is ignored, over half of refundings occur at a loss", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Refunding activity responds to potential cash flow savings but not to potential NPV", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Refunding activity rises with tax revenue shortfalls and budget deficits", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "NPV losses grow as tax collections fall and unfunded pension liabilities rise", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Potential cash flow savings drive the cross-sectional probability of refunding", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Average refunding NPV is negative in all 50 states, with heterogeneity across states", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The paper opens with the rapid growth of advance refundings and frames each one as an early synthetic call akin to American option exercise. It sets out an optimal-refunding valuation framework and gives a case study of political context. It then values over 206,000 transactions and documents widespread NPV losses alongside short-term cash flow savings. Alternative frictions (covenants, informational problems) are ruled out. Aggregate time-series and cross-sectional probit regressions link refunding activity to fiscal constraints. The conclusion restates the findings and discusses which other parties benefit from suboptimal refundings.", "writing_profile_keywords": ["municipal bonds", "option valuation", "NPV losses", "financial constraints", "time-series regressions", "case study", "policy relevance"], "confidence": 0.8, "notes": "No explicit contribution paragraph is visible. Standard error and significance conventions are not visible in the excerpts. Section VI is treated as the robustness section because it rules out alternative explanations."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.32, "usage": {"input": 8026, "output": 2170, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076298, "request_id": "req_011CggpeZe3Vsp9duoyQgftp", "cached": false, "ts": "2026-11-03T22:58:10+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/96812bbe77ead8b9ba355452945bee90075875e35e4f0968b86a2a0e2aa91748.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"survey_or_address\",\"subfield\":\"financial_econometrics\",\"subfield_secondary\":\"research methodology / meta-science of publication bias in asset pricing\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"not applicable (methodological address); the target quantity is the posterior probability that the null is true given a reported p-value and prior odds\",\"strategy\":\"not applicable; conceptual argument, a hypothetical ticker-symbol data-mining example, and Bayesian conversion of p-values using minimum Bayes factors\",\"key_assumptions\":[\"prior odds on the null vs. alternative can be specified from economic plausibility\",\"minimum Bayes factor bounds the evidence against the null\"],\"threats_addressed\":[\"multiple testing\",\"p-hacking\",\"publication bias / file drawer effect\",\"low prior probability of true effects (rare incidence)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":4},\"gap\":{\"present\":true,\"sentence_index\":3},\"method\":{\"present\":true,\"sentence_index\":5},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":6}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":2,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_method\",\"synthesis\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"other\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"literature\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"theory\"},{\"idx\":8,\"role\":\"theory\"},{\"idx\":9,\"role\":\"theory\"},{\"idx\":10,\"role\":\"theory\"},{\"idx\":11,\"role\":\"robustness\"},{\"idx\":12,\"role\":\"other\"},{\"idx\":13,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Tables are illustrative computation grids. Rows are significance levels or z-scores/p-values, and columns are prior odds labeled from 'long shot' to 'even odds'. Notes define the MBF and SD-MBF formulas. One table re-expresses p-values from published papers as Bayesianized p-values.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":false},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"modal verbs (may, likely)\",\"framing as a conversation starter\",\"disclaimer that past research quality is high\",\"extended quotations from authorities (Hubble, Russell)\"]},\"claim_evidence\":[{\"claim\":\"Searching many specifications can produce a seemingly significant factor from economically meaningless sorting such as ticker letters.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Introduction ticker example (t=3.23)\"},{\"claim\":\"Published factor t-statistics cluster just above conventional thresholds, consistent with publication bias.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Figure 6 (HLZ 2016)\"},{\"claim\":\"Economics and business publish supportive results at rates close to psychology and above the physical sciences.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2 (Fanelli 2010)\"},{\"claim\":\"The share of positive results has risen over time across sciences.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3 (Fanelli 2012)\"},{\"claim\":\"With low prior odds, even strict significance thresholds leave a high probability that the null is true.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Table II\"},{\"claim\":\"The required t-statistic threshold depends on the prior plausibility of the hypothesis.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Reported p-values in published factor papers translate into varied Bayesianized p-values depending on prior odds.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"}],\"structural_summary\":\"A presidential address that opens with a provocative hypothetical data-mining example and then frames the problem as an agency conflict between editors and authors. It brings in meta-scientific evidence from other fields and reviews what p-values mean and their limitations. It then discusses p-hacking and the low base rates of true effects, and makes the case for Bayesian reasoning. As a compromise, it proposes minimum Bayes factors and illustrates them with tables and applications to published papers. It notes limitations, closes with research-culture recommendations, and ends with a reflective, quotation-rich conclusion.\",\"writing_profile_keywords\":[\"presidential address\",\"first-person voice\",\"meta-science\",\"p-hacking\",\"Bayesian inference\",\"minimum Bayes factor\",\"publication bias\",\"normative recommendations\",\"illustrative examples\"],\"confidence\":0.8,\"notes\":\"This is not an empirical paper with identification. The figures are mostly reproduced from other studies. Section headings idx 2–4 are artifacts of figure panel labels; idx 4 continues the discussion of Fanelli's explanations. Robustness checks are not applicable, and Section VIII discusses limitations of the method rather than reporting robustness tests.\"}", "json": {"genre": "survey_or_address", "subfield": "financial_econometrics", "subfield_secondary": "research methodology / meta-science of publication bias in asset pricing", "primary_method": "other", "identification": {"estimand": "not applicable (methodological address); the target quantity is the posterior probability that the null is true given a reported p-value and prior odds", "strategy": "not applicable; conceptual argument, a hypothetical ticker-symbol data-mining example, and Bayesian conversion of p-values using minimum Bayes factors", "key_assumptions": ["prior odds on the null vs. alternative can be specified from economic plausibility", "minimum Bayes factor bounds the evidence against the null"], "threats_addressed": ["multiple testing", "p-hacking", "publication bias / file drawer effect", "low prior probability of true effects (rare incidence)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 4}, "gap": {"present": true, "sentence_index": 3}, "method": {"present": true, "sentence_index": 5}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 6}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 2, "move": "roadmap"}], "contribution_paragraph_index": 2, "contribution_types": ["new_method", "synthesis"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "other"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "literature"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "theory"}, {"idx": 8, "role": "theory"}, {"idx": 9, "role": "theory"}, {"idx": 10, "role": "theory"}, {"idx": 11, "role": "robustness"}, {"idx": 12, "role": "other"}, {"idx": 13, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Tables are illustrative computation grids. Rows are significance levels or z-scores/p-values, and columns are prior odds labeled from 'long shot' to 'even odds'. Notes define the MBF and SD-MBF formulas. One table re-expresses p-values from published papers as Bayesianized p-values.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": false}, "hedging": {"level": "moderate", "typical_devices": ["modal verbs (may, likely)", "framing as a conversation starter", "disclaimer that past research quality is high", "extended quotations from authorities (Hubble, Russell)"]}, "claim_evidence": [{"claim": "Searching many specifications can produce a seemingly significant factor from economically meaningless sorting such as ticker letters.", "evidence_type": "descriptive_statistic", "evidence_ref": "Introduction ticker example (t=3.23)"}, {"claim": "Published factor t-statistics cluster just above conventional thresholds, consistent with publication bias.", "evidence_type": "figure", "evidence_ref": "Figure 1, Figure 6 (HLZ 2016)"}, {"claim": "Economics and business publish supportive results at rates close to psychology and above the physical sciences.", "evidence_type": "figure", "evidence_ref": "Figure 2 (Fanelli 2010)"}, {"claim": "The share of positive results has risen over time across sciences.", "evidence_type": "figure", "evidence_ref": "Figure 3 (Fanelli 2012)"}, {"claim": "With low prior odds, even strict significance thresholds leave a high probability that the null is true.", "evidence_type": "table", "evidence_ref": "Table I, Table II"}, {"claim": "The required t-statistic threshold depends on the prior plausibility of the hypothesis.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Reported p-values in published factor papers translate into varied Bayesianized p-values depending on prior odds.", "evidence_type": "table", "evidence_ref": "Table IV"}], "structural_summary": "A presidential address that opens with a provocative hypothetical data-mining example and then frames the problem as an agency conflict between editors and authors. It brings in meta-scientific evidence from other fields and reviews what p-values mean and their limitations. It then discusses p-hacking and the low base rates of true effects, and makes the case for Bayesian reasoning. As a compromise, it proposes minimum Bayes factors and illustrates them with tables and applications to published papers. It notes limitations, closes with research-culture recommendations, and ends with a reflective, quotation-rich conclusion.", "writing_profile_keywords": ["presidential address", "first-person voice", "meta-science", "p-hacking", "Bayesian inference", "minimum Bayes factor", "publication bias", "normative recommendations", "illustrative examples"], "confidence": 0.8, "notes": "This is not an empirical paper with identification. The figures are mostly reproduced from other studies. Section headings idx 2–4 are artifacts of figure panel labels; idx 4 continues the discussion of Fanelli's explanations. Robustness checks are not applicable, and Section VIII discusses limitations of the method rather than reporting robustness tests."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.36, "usage": {"input": 7476, "output": 2100, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072698, "request_id": "req_011CggpcpmpCGxdqD1gXthDB", "cached": false, "ts": "2026-11-03T22:57:47+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/99d41201d3e5b0bbd488490b868926d7ebc47018290597924f1059567ffdd399.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"organizational economics / trade credit and supply chains\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of restricting trade credit maturity to 30 days on the likelihood of trade with affected suppliers, on vertical integration, and on total procurement\",\"strategy\":\"Within-product difference-in-differences comparing suppliers just below and just above a revenue cutoff (UF 100,000), before and after the Agreement, with product-time and firm-product fixed effects. Also exploits product-level exposure and time-varying firm fixed effects.\",\"key_assumptions\":[\"Parallel trends between Treated and Control suppliers in the absence of the Agreement\",\"No manipulation of the revenue cutoff\",\"Cutoff determination not confounded by contemporaneous legislation\",\"Enforcement of the Agreement\"],\"threats_addressed\":[\"Differential trends between small and large firms\",\"Differential exit of smaller firms\",\"Manipulation around the threshold\",\"Contemporaneous legislation\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Table notes state the estimating equation with its fixed effects and define the Treated/Control groups and exposure measures. Summary statistics compare the two groups in the preperiod. The excerpts do not show how standard errors are computed.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"suggests\",\"likely\",\"we interpret this as\",\"most consistent with\",\"in this setting\"]},\"claim_evidence\":[{\"claim\":\"The trade credit maturity restriction lowers the probability that a Treated supplier trades a product with the retailer by 11%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 2\"},{\"claim\":\"The retailer becomes more likely to source from its own subsidiaries for highly exposed products\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 3\"},{\"claim\":\"Total procurement falls for products mostly bought from affected suppliers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Results are not driven by differential trends or differential exit of small firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 1\"},{\"claim\":\"Exclusive or high-share relationships mitigate the negative effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Effects concentrate in durable/nonperishable goods and non-factoring firms, consistent with trade credit as a quality guarantee\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The introduction frames the question in organizational economics, notes that prior evidence is limited to case studies and industry-level work, and claims first causal evidence. It previews a regulatory difference-in-differences design and three margins of adjustment with magnitudes, then robustness checks, relational-contract heterogeneity and the mechanism. It closes by contrasting the results with French evidence and giving a roadmap. The body describes the setting and data, then presents the trade-margin results with graphical evidence and robustness. Later sections cover vertical integration and procurement, exclusivity as a mitigating factor, and tests of competing trade credit theories. The conclusion restates the three margins and notes welfare implications.\",\"writing_profile_keywords\":[\"regulatory natural experiment\",\"within-product DiD\",\"trade credit\",\"vertical integration\",\"mechanism discrimination\",\"proprietary data\",\"first causal evidence\"],\"confidence\":0.85,\"notes\":\"The section role rules were overridden: Section I is a setting/data section and Section V is a mechanism-testing results section. The excerpts do not show how standard errors are reported.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "organizational economics / trade credit and supply chains", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of restricting trade credit maturity to 30 days on the likelihood of trade with affected suppliers, on vertical integration, and on total procurement", "strategy": "Within-product difference-in-differences comparing suppliers just below and just above a revenue cutoff (UF 100,000), before and after the Agreement, with product-time and firm-product fixed effects. Also exploits product-level exposure and time-varying firm fixed effects.", "key_assumptions": ["Parallel trends between Treated and Control suppliers in the absence of the Agreement", "No manipulation of the revenue cutoff", "Cutoff determination not confounded by contemporaneous legislation", "Enforcement of the Agreement"], "threats_addressed": ["Differential trends between small and large firms", "Differential exit of smaller firms", "Manipulation around the threshold", "Contemporaneous legislation"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_specification", "falsification"]}, "tables": {"convention_notes": "Table notes state the estimating equation with its fixed effects and define the Treated/Control groups and exposure measures. Summary statistics compare the two groups in the preperiod. The excerpts do not show how standard errors are computed.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "suggests", "likely", "we interpret this as", "most consistent with", "in this setting"]}, "claim_evidence": [{"claim": "The trade credit maturity restriction lowers the probability that a Treated supplier trades a product with the retailer by 11%", "evidence_type": "table", "evidence_ref": "Table II; Figure 2"}, {"claim": "The retailer becomes more likely to source from its own subsidiaries for highly exposed products", "evidence_type": "table", "evidence_ref": "Table IV; Figure 3"}, {"claim": "Total procurement falls for products mostly bought from affected suppliers", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Results are not driven by differential trends or differential exit of small firms", "evidence_type": "table", "evidence_ref": "Table III; Figure 1"}, {"claim": "Exclusive or high-share relationships mitigate the negative effects", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Effects concentrate in durable/nonperishable goods and non-factoring firms, consistent with trade credit as a quality guarantee", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The introduction frames the question in organizational economics, notes that prior evidence is limited to case studies and industry-level work, and claims first causal evidence. It previews a regulatory difference-in-differences design and three margins of adjustment with magnitudes, then robustness checks, relational-contract heterogeneity and the mechanism. It closes by contrasting the results with French evidence and giving a roadmap. The body describes the setting and data, then presents the trade-margin results with graphical evidence and robustness. Later sections cover vertical integration and procurement, exclusivity as a mitigating factor, and tests of competing trade credit theories. The conclusion restates the three margins and notes welfare implications.", "writing_profile_keywords": ["regulatory natural experiment", "within-product DiD", "trade credit", "vertical integration", "mechanism discrimination", "proprietary data", "first causal evidence"], "confidence": 0.85, "notes": "The section role rules were overridden: Section I is a setting/data section and Section V is a mechanism-testing results section. The excerpts do not show how standard errors are reported."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.64, "usage": {"input": 6306, "output": 2029, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066598, "request_id": "req_011CggpTW7QiEoBdCikcmxkN", "cached": false, "ts": "2026-11-03T22:55:39+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9aef7f4d5fab154669e45ab4728b30a9de73c2ae1f9acdcd14cd940f611fa2b3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Coefficient of relative risk aversion implied by consumption covariances with stock returns (GMM Euler equations), and cross-sectional consumption risk premia\",\"strategy\":\"A Kalman-filter model of NIPA consumption estimation is used to reverse (unfilter) reported NIPA consumption. Stock returns are time-aggregated to address time-aggregation bias. The resulting series is used in GMM Euler-equation estimation and Fama-MacBeth cross-sectional tests. A simulation experiment validates the procedure.\",\"key_assumptions\":[\"BEA estimation process can be approximated by an optimal (Kalman) filter\",\"Measurement error in observable consumption is uncorrelated with stock returns\",\"Calibrated filter parameters (signal-to-noise, consumption volatility) are appropriate\"],\"threats_addressed\":[\"time-aggregation bias\",\"heterogeneity between nondurables and services\",\"choice of consumption volatility measure in the unfilter\",\"conditioning information and long-horizon returns\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"literature_gap\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_data_or_measure\",\"resolution_of_puzzle\",\"new_evidence\",\"new_method\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Notes define moment restrictions and estimators and fix the discount factor. They state sample periods (full sample and postwar). Some tables scale moments (×100). Simulation tables report medians of 10,000 simulations.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'might be attributed, at least in part'\",\"'may underestimate/overstate'\",\"'provides a first attempt'\",\"explicit acknowledgment of failures (profitability premium, return predictability)\"]},\"claim_evidence\":[{\"claim\":\"Unfiltered NIPA consumption explains the equity premium with a reasonable relative risk aversion\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Unfiltering recovers properties of observable consumption in a simulated economy\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Table I; Figure 1\"},{\"claim\":\"Unfiltered consumption prices size, value, and investment portfolios with low pricing errors but fails for profitability\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"The filter model explains differences between NIPA nondurables and services\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Calibrating to unfiltered consumption changes the implications of habit and long-run risk models\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X; Figure 5\"}],\"structural_summary\":\"The introduction frames the paper around the puzzle of why garbage beats NIPA consumption in the consumption CAPM. It proposes statistical filtering by NIPA as the explanation and enumerates findings, including failures. Literature positioning (real estate unsmoothing, time aggregation) closes the introduction with a roadmap. Section I builds a Kalman-filter model, inverts it, and validates it by simulation. Section II compares consumption measures and estimates risk aversion by GMM, with several robustness variants. Section III runs cross-sectional tests. Section IV recalibrates habit and long-run risk models. A short conclusion restates the contribution and acknowledges the unexplained return predictability.\",\"writing_profile_keywords\":[\"measurement-error explanation\",\"puzzle resolution\",\"Kalman unfiltering\",\"GMM Euler equations\",\"simulation validation\",\"model calibration implications\",\"enumerated findings\"],\"confidence\":0.8,\"notes\":\"Primary method is a hybrid of filter modeling, calibration/simulation, and GMM estimation. Structural_estimation was chosen as the closest label. The excerpts do not show how statistical significance is reported.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "structural_estimation", "identification": {"estimand": "Coefficient of relative risk aversion implied by consumption covariances with stock returns (GMM Euler equations), and cross-sectional consumption risk premia", "strategy": "A Kalman-filter model of NIPA consumption estimation is used to reverse (unfilter) reported NIPA consumption. Stock returns are time-aggregated to address time-aggregation bias. The resulting series is used in GMM Euler-equation estimation and Fama-MacBeth cross-sectional tests. A simulation experiment validates the procedure.", "key_assumptions": ["BEA estimation process can be approximated by an optimal (Kalman) filter", "Measurement error in observable consumption is uncorrelated with stock returns", "Calibrated filter parameters (signal-to-noise, consumption volatility) are appropriate"], "threats_addressed": ["time-aggregation bias", "heterogeneity between nondurables and services", "choice of consumption volatility measure in the unfilter", "conditioning information and long-horizon returns"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "literature_gap"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "policy_or_broader_implication"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_data_or_measure", "resolution_of_puzzle", "new_evidence", "new_method"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_measure", "alternative_specification", "subsample", "alternative_sample_period"]}, "tables": {"convention_notes": "Notes define moment restrictions and estimators and fix the discount factor. They state sample periods (full sample and postwar). Some tables scale moments (×100). Simulation tables report medians of 10,000 simulations.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'might be attributed, at least in part'", "'may underestimate/overstate'", "'provides a first attempt'", "explicit acknowledgment of failures (profitability premium, return predictability)"]}, "claim_evidence": [{"claim": "Unfiltered NIPA consumption explains the equity premium with a reasonable relative risk aversion", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Unfiltering recovers properties of observable consumption in a simulated economy", "evidence_type": "figure", "evidence_ref": "Table I; Figure 1"}, {"claim": "Unfiltered consumption prices size, value, and investment portfolios with low pricing errors but fails for profitability", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "The filter model explains differences between NIPA nondurables and services", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Calibrating to unfiltered consumption changes the implications of habit and long-run risk models", "evidence_type": "table", "evidence_ref": "Table X; Figure 5"}], "structural_summary": "The introduction frames the paper around the puzzle of why garbage beats NIPA consumption in the consumption CAPM. It proposes statistical filtering by NIPA as the explanation and enumerates findings, including failures. Literature positioning (real estate unsmoothing, time aggregation) closes the introduction with a roadmap. Section I builds a Kalman-filter model, inverts it, and validates it by simulation. Section II compares consumption measures and estimates risk aversion by GMM, with several robustness variants. Section III runs cross-sectional tests. Section IV recalibrates habit and long-run risk models. A short conclusion restates the contribution and acknowledges the unexplained return predictability.", "writing_profile_keywords": ["measurement-error explanation", "puzzle resolution", "Kalman unfiltering", "GMM Euler equations", "simulation validation", "model calibration implications", "enumerated findings"], "confidence": 0.8, "notes": "Primary method is a hybrid of filter modeling, calibration/simulation, and GMM estimation. Structural_estimation was chosen as the closest label. The excerpts do not show how statistical significance is reported."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.92, "usage": {"input": 9163, "output": 2237, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082186, "request_id": "req_011CggpUwX99UMZ6CSdXUqQv", "cached": false, "ts": "2026-11-03T22:56:00+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9f2a363b55a34ce4f3517f2702249ee2cd5b2e30b0361849ea4034ae5ee84858.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"law_and_finance / bankruptcy; asset reallocation\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of Chapter 7 conversion (liquidation) versus remaining in Chapter 11 (reorganization) on five-year utilization of plant real estate (occupancy, continuation, holding time, employment) for marginal Chapter 11 filers\",\"strategy\":\"Within court-division, filers are randomly assigned to judges by blind rotation. The judge's leave-out share of Chapter 11 cases converted to Chapter 7 instruments for conversion. Second stage compares asset outcomes, with heterogeneity splits by market thickness and access to finance.\",\"key_assumptions\":[\"Random (blind rotation) assignment of judges within division-year\",\"Judge leniency affects outcomes only through conversion (exclusion restriction)\",\"Relevance: judges differ in their conversion propensities\",\"Results describe marginal (complier) firms\"],\"threats_addressed\":[\"Selection into bankruptcy chapter (handled by restricting to Chapter 11 filers)\",\"Endogeneity of conversion\",\"Non-random judge assignment (randomization tests)\",\"Violation of exclusion restriction\",\"Pre-trends (reduced-form dynamics)\",\"Local utilization benchmarks\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"contribution_statement\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":10,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_specification\",\"placebo\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable and the unit of observation (filing vs. plant), describe the sample splits (above/below median), and state that observation counts are rounded because of Census disclosure rules. Tables run in order: summary statistics, first stage, randomization test, main IV results, then heterogeneity tables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' phrasing\",\"'suggest' for efficiency interpretations\",\"explicit admission that asset efficiency cannot be directly measured\",\"'most consistent with the interpretation'\",\"weighing competing explanations\"]},\"claim_evidence\":[{\"claim\":\"Liquidated plants are 17.4% less likely to be occupied five years after filing than reorganized plants\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Judge conversion propensity strongly predicts Chapter 11-to-7 conversion\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 2\"},{\"claim\":\"Judge assignment is random within division with respect to observable characteristics\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Utilization effects are concentrated in thin asset markets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Utilization effects are concentrated in areas with low access to small business finance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Liquidation reduces utilization even after controlling for benchmark local utilization rates\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Redeployed assets usually go to local firms in the same industry, consistent with asset specificity\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"Motivates the paper with bankruptcy's role in reallocating assets and poses unanswered questions. Sketches the frictions under each bankruptcy regime, then introduces Census-linked tracking of real estate and stylized facts. Defends a random-judge IV design and previews the main effects. Sets out competing interpretations (inefficient liquidation vs. continuation bias) and adjudicates between them with heterogeneity tests. Positions the paper in the literature and closes the introduction with a roadmap. The body covers institutions, data, measurement, the identification section (including exclusion restriction and compliers), results, an efficiency discussion, and a brief conclusion.\",\"writing_profile_keywords\":[\"random judge assignment\",\"instrumental variables\",\"Census LBD\",\"asset reallocation\",\"bankruptcy\",\"heterogeneity by market frictions\",\"competing interpretations\"],\"confidence\":0.85,\"notes\":\"Standard-error details are not visible in the table note excerpts. The Efficiency Discussion section was rule-labeled as conclusion but is treated here as an interpretive 'other' section. An internet appendix is mentioned, so some robustness material is likely located there.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "law_and_finance / bankruptcy; asset reallocation", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of Chapter 7 conversion (liquidation) versus remaining in Chapter 11 (reorganization) on five-year utilization of plant real estate (occupancy, continuation, holding time, employment) for marginal Chapter 11 filers", "strategy": "Within court-division, filers are randomly assigned to judges by blind rotation. The judge's leave-out share of Chapter 11 cases converted to Chapter 7 instruments for conversion. Second stage compares asset outcomes, with heterogeneity splits by market thickness and access to finance.", "key_assumptions": ["Random (blind rotation) assignment of judges within division-year", "Judge leniency affects outcomes only through conversion (exclusion restriction)", "Relevance: judges differ in their conversion propensities", "Results describe marginal (complier) firms"], "threats_addressed": ["Selection into bankruptcy chapter (handled by restricting to Chapter 11 filers)", "Endogeneity of conversion", "Non-random judge assignment (randomization tests)", "Violation of exclusion restriction", "Pre-trends (reduced-form dynamics)", "Local utilization benchmarks"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "tension_or_puzzle"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "contribution_statement"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "roadmap"}], "contribution_paragraph_index": 10, "contribution_types": ["new_evidence", "new_data_or_measure", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_specification", "placebo"]}, "tables": {"convention_notes": "Table notes define the dependent variable and the unit of observation (filing vs. plant), describe the sample splits (above/below median), and state that observation counts are rounded because of Census disclosure rules. Tables run in order: summary statistics, first stage, randomization test, main IV results, then heterogeneity tables.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' phrasing", "'suggest' for efficiency interpretations", "explicit admission that asset efficiency cannot be directly measured", "'most consistent with the interpretation'", "weighing competing explanations"]}, "claim_evidence": [{"claim": "Liquidated plants are 17.4% less likely to be occupied five years after filing than reorganized plants", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Judge conversion propensity strongly predicts Chapter 11-to-7 conversion", "evidence_type": "table", "evidence_ref": "Table III; Figure 2"}, {"claim": "Judge assignment is random within division with respect to observable characteristics", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Utilization effects are concentrated in thin asset markets", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Utilization effects are concentrated in areas with low access to small business finance", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Liquidation reduces utilization even after controlling for benchmark local utilization rates", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Redeployed assets usually go to local firms in the same industry, consistent with asset specificity", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "Motivates the paper with bankruptcy's role in reallocating assets and poses unanswered questions. Sketches the frictions under each bankruptcy regime, then introduces Census-linked tracking of real estate and stylized facts. Defends a random-judge IV design and previews the main effects. Sets out competing interpretations (inefficient liquidation vs. continuation bias) and adjudicates between them with heterogeneity tests. Positions the paper in the literature and closes the introduction with a roadmap. The body covers institutions, data, measurement, the identification section (including exclusion restriction and compliers), results, an efficiency discussion, and a brief conclusion.", "writing_profile_keywords": ["random judge assignment", "instrumental variables", "Census LBD", "asset reallocation", "bankruptcy", "heterogeneity by market frictions", "competing interpretations"], "confidence": 0.85, "notes": "Standard-error details are not visible in the table note excerpts. The Efficiency Discussion section was rule-labeled as conclusion but is treated here as an interpretive 'other' section. An internet appendix is mentioned, so some robustness material is likely located there."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 33.41, "usage": {"input": 8261, "output": 2357, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080978, "request_id": "req_011CggqDgmaR5zMGopWMED9N", "cached": false, "ts": "2026-11-03T23:05:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9fb6d560ab8a1055eb85907ce2ed3d1e4f70014e62a69e4182bd2805e6b940a7.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"asset_pricing (housing, disagreement/speculation)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Not a causal estimand; the evidence describes cross-city patterns in 2000-2006 house and land price growth and home builder behavior that are consistent with model predictions.\",\"strategy\":\"A two-period disagreement model with developers, land constraints and owner-occupancy frictions. Its predictions are compared with descriptive stylized facts: land vs. construction cost growth across cities, home builder land holdings and short interest, and anomalous cities vs. predicted demand shocks.\",\"key_assumptions\":[\"short-sale constraints in land and housing\",\"rental frictions make owning preferred to renting\",\"fixed amount of developable land\",\"competitive developers with constant conversion cost\"],\"threats_addressed\":[\"alternative explanation: differential local demand shocks (Figure 6)\",\"alternative theory of nonmonotonic booms that is agnostic on land vs. construction costs\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":3},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"other\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"No tables. Empirical evidence is presented through figures whose captions give data sources (10-K filings, R.S. Means), variable construction and city groupings.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' framing of evidence\",\"'we present some evidence... further work is needed'\",\"acknowledging what is not systematically examined\",\"'helps explain' / 'offers an alternative explanation'\"],\"typical_devices_note\":null},\"claim_evidence\":[{\"claim\":\"Disagreement amplifies house price booms most in cities at intermediate development levels, i.e., with elastic supply that faces a future barrier.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III, Figure 2, Appendix A\"},{\"claim\":\"Land price growth 2000-2006 was high and tracked house price growth across cities, while construction cost growth did not.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Public home builders amassed land beyond construction needs during the boom while their stocks were heavily short-sold.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"Anomalous sand-state cities had booms exceeding those predicted by demographic demand shocks.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6\"},{\"claim\":\"Las Vegas illustrates a long-run development barrier and concentrated land speculation.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Figure 1, intro paragraph 5\"},{\"claim\":\"Disagreement reduces welfare by reallocating space from high-utility pessimists to optimists and developers.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.E\"}],\"structural_summary\":\"The paper opens by asking whether disagreement-based asset pricing theory extends to housing. It highlights two distinctive features of housing (elastic supply, non-transferable flow utility) and sketches a two-period model. It then previews extensions and a puzzle about cross-city booms, previews the empirical stylized facts, and closes the introduction with a Las Vegas anecdote and a roadmap. The body builds the model, solves the agreement benchmark, analyzes equilibrium with disagreement (including the cross-city variation and welfare), and adds extensions. A descriptive empirical section uses figures to show patterns consistent with the predictions. The conclusion restates the contributions and flags open empirical questions.\",\"writing_profile_keywords\":[\"theory-first\",\"disagreement models\",\"housing supply elasticity\",\"stylized facts\",\"figure-based evidence\",\"anecdotal illustration\",\"proofs in appendix\"],\"confidence\":0.8,\"notes\":\"Section IV is labeled robustness by rule, but it consists of theoretical extensions, so it is coded as theory. The empirical section is descriptive and contains no tables, so statistical significance language cannot be determined from the excerpts. The abstract states no method explicitly.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "macro_finance", "subfield_secondary": "asset_pricing (housing, disagreement/speculation)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Not a causal estimand; the evidence describes cross-city patterns in 2000-2006 house and land price growth and home builder behavior that are consistent with model predictions.", "strategy": "A two-period disagreement model with developers, land constraints and owner-occupancy frictions. Its predictions are compared with descriptive stylized facts: land vs. construction cost growth across cities, home builder land holdings and short interest, and anomalous cities vs. predicted demand shocks.", "key_assumptions": ["short-sale constraints in land and housing", "rental frictions make owning preferred to renting", "fixed amount of developable land", "competitive developers with constant conversion cost"], "threats_addressed": ["alternative explanation: differential local demand shocks (Figure 6)", "alternative theory of nonmonotonic booms that is agnostic on land vs. construction costs"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 3}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "other"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_theory_or_model", "new_evidence", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["other"]}, "tables": {"convention_notes": "No tables. Empirical evidence is presented through figures whose captions give data sources (10-K filings, R.S. Means), variable construction and city groupings.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' framing of evidence", "'we present some evidence... further work is needed'", "acknowledging what is not systematically examined", "'helps explain' / 'offers an alternative explanation'"], "typical_devices_note": null}, "claim_evidence": [{"claim": "Disagreement amplifies house price booms most in cities at intermediate development levels, i.e., with elastic supply that faces a future barrier.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III, Figure 2, Appendix A"}, {"claim": "Land price growth 2000-2006 was high and tracked house price growth across cities, while construction cost growth did not.", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Public home builders amassed land beyond construction needs during the boom while their stocks were heavily short-sold.", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "Anomalous sand-state cities had booms exceeding those predicted by demographic demand shocks.", "evidence_type": "figure", "evidence_ref": "Figures 5-6"}, {"claim": "Las Vegas illustrates a long-run development barrier and concentrated land speculation.", "evidence_type": "descriptive_statistic", "evidence_ref": "Figure 1, intro paragraph 5"}, {"claim": "Disagreement reduces welfare by reallocating space from high-utility pessimists to optimists and developers.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.E"}], "structural_summary": "The paper opens by asking whether disagreement-based asset pricing theory extends to housing. It highlights two distinctive features of housing (elastic supply, non-transferable flow utility) and sketches a two-period model. It then previews extensions and a puzzle about cross-city booms, previews the empirical stylized facts, and closes the introduction with a Las Vegas anecdote and a roadmap. The body builds the model, solves the agreement benchmark, analyzes equilibrium with disagreement (including the cross-city variation and welfare), and adds extensions. A descriptive empirical section uses figures to show patterns consistent with the predictions. The conclusion restates the contributions and flags open empirical questions.", "writing_profile_keywords": ["theory-first", "disagreement models", "housing supply elasticity", "stylized facts", "figure-based evidence", "anecdotal illustration", "proofs in appendix"], "confidence": 0.8, "notes": "Section IV is labeled robustness by rule, but it consists of theoretical extensions, so it is coded as theory. The empirical section is descriptive and contains no tables, so statistical significance language cannot be determined from the excerpts. The abstract states no method explicitly."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.95, "usage": {"input": 5966, "output": 2162, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067898, "request_id": "req_011Cggq7vhJrD6RGCMKiVQKQ", "cached": false, "ts": "2026-11-03T23:04:23+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a2a48a79b5b1b0b53cfad15b34fe527af4950abac3bb4ab7db032802af034d16.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Model-implied, then empirical, relations between analyst-forecast dispersion and future excess returns, and conditional time-series momentum in high-dispersion periods and recessions.\",\"strategy\":\"Equilibrium model with heterogeneous forecasting models, calibrated to US data. Simulated predictions are compared with the same regressions run on observed data using empirical proxies such as analyst-forecast dispersion.\",\"key_assumptions\":[\"Investors use different forecasting models (continuous-state vs. discrete-state)\",\"Calibrated parameters satisfy restrictions that generate countercyclical disagreement spikes\",\"Analyst-forecast dispersion proxies for squared disagreement\"],\"threats_addressed\":[\"Linear trend in dispersion\",\"12-month seasonality in the fundamental\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":6}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Tables report calibrated parameters with delta-method standard errors in brackets, a regime definition table, and regression outputs. Regression standard errors appear in parentheses, with stars marking the 10/5/1% significance levels.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"mixed\",\"convention_notes_placeholder\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":false},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'we provide empirical support' framing\",\"'suggests' wording\",\"model-conditional statements ('in our model')\"]},\"claim_evidence\":[{\"claim\":\"Under the calibrated parameters, disagreement between the two agents spikes countercyclically, with beliefs polarizing in bad times.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section II, Figure 4, Table I\"},{\"claim\":\"Contemporaneous excess returns decrease with squared disagreement because disagreement variation is priced.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A\"},{\"claim\":\"The model generates time-series momentum that is strongest in bad times.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 8\"},{\"claim\":\"Empirically, future excess returns are positively related to analyst-forecast dispersion.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 10\"},{\"claim\":\"Empirical time-series momentum is stronger in high-dispersion periods and in recessions.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 11-12\"},{\"claim\":\"Empirical dispersion is higher in NBER recessions and co-moves with model-implied dispersion.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"}],\"structural_summary\":\"The paper opens with the puzzle that return predictability concentrates in bad times and proposes a heterogeneous-model learning mechanism to explain it. It then builds a two-agent equilibrium model and calibrates it to the US business cycle, showing that disagreement spikes countercyclically. From the model it derives predictions on return predictability and momentum. Finally, it tests these predictions by setting simulated statistics beside the same statistics computed on empirical proxies, and closes with avenues for future research.\",\"writing_profile_keywords\":[\"equilibrium model\",\"heterogeneous beliefs\",\"disagreement\",\"calibration\",\"model vs. data simulations\",\"time-series momentum\",\"business cycle\"],\"confidence\":0.8,\"notes\":\"Theory-led paper with an empirical validation section. Robustness checks are not visible in the excerpts. Figures report t-statistics, and Figure 4 shows confidence bands.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Model-implied, then empirical, relations between analyst-forecast dispersion and future excess returns, and conditional time-series momentum in high-dispersion periods and recessions.", "strategy": "Equilibrium model with heterogeneous forecasting models, calibrated to US data. Simulated predictions are compared with the same regressions run on observed data using empirical proxies such as analyst-forecast dispersion.", "key_assumptions": ["Investors use different forecasting models (continuous-state vs. discrete-state)", "Calibrated parameters satisfy restrictions that generate countercyclical disagreement spikes", "Analyst-forecast dispersion proxies for squared disagreement"], "threats_addressed": ["Linear trend in dispersion", "12-month seasonality in the fundamental"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 6}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Tables report calibrated parameters with delta-method standard errors in brackets, a regime definition table, and regression outputs. Regression standard errors appear in parentheses, with stars marking the 10/5/1% significance levels.", "units_stated_in_notes": false, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "mixed", "convention_notes_placeholder": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": false}, "hedging": {"level": "moderate", "typical_devices": ["'we provide empirical support' framing", "'suggests' wording", "model-conditional statements ('in our model')"]}, "claim_evidence": [{"claim": "Under the calibrated parameters, disagreement between the two agents spikes countercyclically, with beliefs polarizing in bad times.", "evidence_type": "figure", "evidence_ref": "Section II, Figure 4, Table I"}, {"claim": "Contemporaneous excess returns decrease with squared disagreement because disagreement variation is priced.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A"}, {"claim": "The model generates time-series momentum that is strongest in bad times.", "evidence_type": "figure", "evidence_ref": "Figure 8"}, {"claim": "Empirically, future excess returns are positively related to analyst-forecast dispersion.", "evidence_type": "figure", "evidence_ref": "Figure 10"}, {"claim": "Empirical time-series momentum is stronger in high-dispersion periods and in recessions.", "evidence_type": "figure", "evidence_ref": "Figures 11-12"}, {"claim": "Empirical dispersion is higher in NBER recessions and co-moves with model-implied dispersion.", "evidence_type": "table", "evidence_ref": "Table IV"}], "structural_summary": "The paper opens with the puzzle that return predictability concentrates in bad times and proposes a heterogeneous-model learning mechanism to explain it. It then builds a two-agent equilibrium model and calibrates it to the US business cycle, showing that disagreement spikes countercyclically. From the model it derives predictions on return predictability and momentum. Finally, it tests these predictions by setting simulated statistics beside the same statistics computed on empirical proxies, and closes with avenues for future research.", "writing_profile_keywords": ["equilibrium model", "heterogeneous beliefs", "disagreement", "calibration", "model vs. data simulations", "time-series momentum", "business cycle"], "confidence": 0.8, "notes": "Theory-led paper with an empirical validation section. Robustness checks are not visible in the excerpts. Figures report t-statistics, and Figure 4 shows confidence bands."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.79, "usage": {"input": 6638, "output": 1929, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065926, "request_id": "req_011Cggpn76LRKtVDJEp4cDgQ", "cached": false, "ts": "2026-11-03T22:59:51+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a390f5144faf928d8992516699729c0deeaa1426c380fe8de73fed958abc745f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model of trading volume and prices under cursed expectations)\",\"strategy\":\"not applicable; analytical comparison of a CARA-normal noisy rational expectations model under cursedness, overconfidence and dismissiveness\",\"key_assumptions\":[\"CARA preferences\",\"normally distributed payoffs and signals\",\"signals conditionally independent given payoff\",\"cursed traders do not fully infer information from price\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"literature\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"No tables; results are delivered as propositions and equations.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"modal verbs (may, can)\",\"conditional qualifiers (given moderate noninformational motives)\",\"'in many people's estimation' framing\"]},\"claim_evidence\":[{\"claim\":\"Cursedness generates substantial trade, with per-trader volume rising in market size and aggregate volume unbounded as traders grow numerous\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.A\"},{\"claim\":\"Cursedness and dismissiveness imply price underreaction while overconfidence implies overreaction\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B\"},{\"claim\":\"Under dismissiveness or overconfidence, per-trader volume may vanish as market grows\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B\"},{\"claim\":\"Cursedness enables overconfidence to have large volume effects\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.C\"},{\"claim\":\"Mixed markets of rational and cursed traders can produce more volume than fully cursed markets\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C\"},{\"claim\":\"Public revelation of private signals raises rational/dismissive volume but cuts cursed volume when hedging motives are moderate\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections III.A-III.B\"},{\"claim\":\"Neglect of information in others' actions is documented in auctions and laboratory settings\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section IV\"}],\"structural_summary\":\"The paper opens with the no-trade puzzle and excess-volume motivation, then situates itself against heterogeneous-prior and overconfidence models. It introduces a competitive analog of cursed equilibrium in a CARA-normal noisy REE setup. The baseline symmetric equilibrium derives volume and price predictions and contrasts cursedness with overconfidence and dismissiveness. Extensions add a public signal, random endowments and heterogeneous traders. A short section reviews experimental and empirical evidence on cursedness, and the conclusion restates the main comparative results.\",\"writing_profile_keywords\":[\"behavioral equilibrium concept\",\"noisy rational expectations\",\"trading volume\",\"comparative statics\",\"model comparison\",\"propositions\",\"extensions\"],\"confidence\":0.85,\"notes\":\"Section IV discusses existing evidence (prior literature) rather than new empirical tests, so it is assigned the literature role. Section III is labeled robustness by the rule but consists of theoretical extensions, so it is assigned the theory role.\"}", "json": {"genre": "theoretical", "subfield": "behavioral_finance", "subfield_secondary": "market_microstructure", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model of trading volume and prices under cursed expectations)", "strategy": "not applicable; analytical comparison of a CARA-normal noisy rational expectations model under cursedness, overconfidence and dismissiveness", "key_assumptions": ["CARA preferences", "normally distributed payoffs and signals", "signals conditionally independent given payoff", "cursed traders do not fully infer information from price"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "related_literature_positioning"}], "contribution_paragraph_index": 1, "contribution_types": ["new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "literature"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "No tables; results are delivered as propositions and equations.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["modal verbs (may, can)", "conditional qualifiers (given moderate noninformational motives)", "'in many people's estimation' framing"]}, "claim_evidence": [{"claim": "Cursedness generates substantial trade, with per-trader volume rising in market size and aggregate volume unbounded as traders grow numerous", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.A"}, {"claim": "Cursedness and dismissiveness imply price underreaction while overconfidence implies overreaction", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B"}, {"claim": "Under dismissiveness or overconfidence, per-trader volume may vanish as market grows", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B"}, {"claim": "Cursedness enables overconfidence to have large volume effects", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.C"}, {"claim": "Mixed markets of rational and cursed traders can produce more volume than fully cursed markets", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C"}, {"claim": "Public revelation of private signals raises rational/dismissive volume but cuts cursed volume when hedging motives are moderate", "evidence_type": "equation_or_proof", "evidence_ref": "Sections III.A-III.B"}, {"claim": "Neglect of information in others' actions is documented in auctions and laboratory settings", "evidence_type": "prior_literature", "evidence_ref": "Section IV"}], "structural_summary": "The paper opens with the no-trade puzzle and excess-volume motivation, then situates itself against heterogeneous-prior and overconfidence models. It introduces a competitive analog of cursed equilibrium in a CARA-normal noisy REE setup. The baseline symmetric equilibrium derives volume and price predictions and contrasts cursedness with overconfidence and dismissiveness. Extensions add a public signal, random endowments and heterogeneous traders. A short section reviews experimental and empirical evidence on cursedness, and the conclusion restates the main comparative results.", "writing_profile_keywords": ["behavioral equilibrium concept", "noisy rational expectations", "trading volume", "comparative statics", "model comparison", "propositions", "extensions"], "confidence": 0.85, "notes": "Section IV discusses existing evidence (prior literature) rather than new empirical tests, so it is assigned the literature role. Section III is labeled robustness by the rule but consists of theoretical extensions, so it is assigned the theory role."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 30.16, "usage": {"input": 4725, "output": 1922, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.058134, "request_id": "req_011CggqCFm9Tn1hBEYYb3Bbo", "cached": false, "ts": "2026-11-03T23:05:33+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a6304ba055c6f043a96eef5d545d710d2cf2c770eeb03b8b61a03f43b131cb2a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"The share of cross-sectional variation in client portfolio risky share and home bias that is attributable to advisor-specific effects, compared with the share explained by investor attributes.\",\"strategy\":\"Panel regressions with advisor fixed effects and year fixed effects. A two-way fixed effects design adds investor fixed effects and uses clients forced to switch advisors after the advisor's death, retirement or resignation, then tracks convergence toward the new advisor's allocation. Advisor fixed effects are then regressed on advisor attributes, including the advisor's own portfolio.\",\"key_assumptions\":[\"Forced switches are unrelated to changes in client preferences.\",\"Investor fixed effects absorb persistent unobserved client preferences.\",\"Investors are not matched to advisors on unobserved characteristics, or such matching is controlled for by the switcher design.\"],\"threats_addressed\":[\"Matching between investors and advisors on omitted characteristics\",\"Unobserved client preferences\",\"Self-directed investments initiated by the client\",\"Client-initiated switches that coincide with preference changes\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":4},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"other\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"Notes define the dependent variables (risky share, home bias), the sample period and the fixed effects included. Tables are organized in panels by outcome (risky share vs. home bias). Descriptive statistics are reported as percentiles.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"Offering alternative interpretations of the advisor fixed effects\",\"An explicit caveats-and-robustness subsection\",\"Framing the matching concern as a hypothesis that is then tested\"]},\"claim_evidence\":[{\"claim\":\"Investor characteristics explain only about 12% of the variation in risky share.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Advisor fixed effects raise adjusted R2 from 12% to 30% for risky share, and add 24% for home bias.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Moving from the 25th to the 75th percentile of advisor effects shifts risky share by 20 pp and home bias by 32 pp.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Advisor effects remain important after controlling for investor fixed effects among forced switchers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"An advisor's own risky share is the strongest predictor of the advisor fixed effect.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 5\"},{\"claim\":\"Advised portfolios cost about 2.5% per year, roughly 1.5% more than life cycle funds.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6; Section IV\"},{\"claim\":\"Advisors' gross and net alphas are estimated against passive factor benchmarks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"}],\"structural_summary\":\"The introduction motivates the topic with how widely households use advisors, then sets the competing views of underperformance and \\\"interior decoration\\\" against each other. It previews variance-decomposition results with R2 magnitudes and raises the matching interpretation. It addresses that interpretation with a forced-switcher two-way fixed effects design, then states three contributions and gives a roadmap. The body proceeds through the data, customization regressions (with caveats, investor fixed effects, and advisor attributes), performance alphas, and a fee decomposition. The conclusion restates three findings with magnitudes.\",\"writing_profile_keywords\":[\"administrative data\",\"variance decomposition\",\"advisor fixed effects\",\"two-way fixed effects\",\"forced switchers\",\"adjusted R2 magnitudes\",\"fee decomposition\",\"one-size-fits-all\"],\"confidence\":0.8,\"notes\":\"The conclusion excerpt is truncated, so policy implications may appear later in it but are not visible. The standard error type is not visible in the table note excerpts. Some figures show 95% confidence intervals (Figures 1 and 4), while others show distributions or averages.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "The share of cross-sectional variation in client portfolio risky share and home bias that is attributable to advisor-specific effects, compared with the share explained by investor attributes.", "strategy": "Panel regressions with advisor fixed effects and year fixed effects. A two-way fixed effects design adds investor fixed effects and uses clients forced to switch advisors after the advisor's death, retirement or resignation, then tracks convergence toward the new advisor's allocation. Advisor fixed effects are then regressed on advisor attributes, including the advisor's own portfolio.", "key_assumptions": ["Forced switches are unrelated to changes in client preferences.", "Investor fixed effects absorb persistent unobserved client preferences.", "Investors are not matched to advisors on unobserved characteristics, or such matching is controlled for by the switcher design."], "threats_addressed": ["Matching between investors and advisors on omitted characteristics", "Unobserved client preferences", "Self-directed investments initiated by the client", "Client-initiated switches that coincide with preference changes"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 4}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "other"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["other"]}, "tables": {"convention_notes": "Notes define the dependent variables (risky share, home bias), the sample period and the fixed effects included. Tables are organized in panels by outcome (risky share vs. home bias). Descriptive statistics are reported as percentiles.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["Offering alternative interpretations of the advisor fixed effects", "An explicit caveats-and-robustness subsection", "Framing the matching concern as a hypothesis that is then tested"]}, "claim_evidence": [{"claim": "Investor characteristics explain only about 12% of the variation in risky share.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Advisor fixed effects raise adjusted R2 from 12% to 30% for risky share, and add 24% for home bias.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Moving from the 25th to the 75th percentile of advisor effects shifts risky share by 20 pp and home bias by 32 pp.", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Advisor effects remain important after controlling for investor fixed effects among forced switchers.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "An advisor's own risky share is the strongest predictor of the advisor fixed effect.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 5"}, {"claim": "Advised portfolios cost about 2.5% per year, roughly 1.5% more than life cycle funds.", "evidence_type": "figure", "evidence_ref": "Figure 6; Section IV"}, {"claim": "Advisors' gross and net alphas are estimated against passive factor benchmarks.", "evidence_type": "table", "evidence_ref": "Table V"}], "structural_summary": "The introduction motivates the topic with how widely households use advisors, then sets the competing views of underperformance and \"interior decoration\" against each other. It previews variance-decomposition results with R2 magnitudes and raises the matching interpretation. It addresses that interpretation with a forced-switcher two-way fixed effects design, then states three contributions and gives a roadmap. The body proceeds through the data, customization regressions (with caveats, investor fixed effects, and advisor attributes), performance alphas, and a fee decomposition. The conclusion restates three findings with magnitudes.", "writing_profile_keywords": ["administrative data", "variance decomposition", "advisor fixed effects", "two-way fixed effects", "forced switchers", "adjusted R2 magnitudes", "fee decomposition", "one-size-fits-all"], "confidence": 0.8, "notes": "The conclusion excerpt is truncated, so policy implications may appear later in it but are not visible. The standard error type is not visible in the table note excerpts. Some figures show 95% confidence intervals (Figures 1 and 4), while others show distributions or averages."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.07, "usage": {"input": 6266, "output": 2147, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068798, "request_id": "req_011Cggpd87721PchFpgzdAHV", "cached": false, "ts": "2026-11-03T22:57:51+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a68b40a852f093ad27958461df974eefe8b948153e6d853e108e6ee44543ee14.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"corporate_finance; asset_pricing (time-varying discount rates)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between the predicted aggregate equity risk premium and buyout activity at the aggregate and firm level, plus cross-sectional differences in that sensitivity\",\"strategy\":\"Time-series regressions of buyout volume and value on predicted risk premium (OLS predictive and VAR/Campbell-Shiller estimates) versus credit factors. Firm panel deal-likelihood regressions with firm or time fixed effects. Interactions of the risk premium with firm risk, agency proxies and exit liquidity to test model-specific predictions.\",\"key_assumptions\":[\"Predicted excess return from D/P, cay and the T-bill rate (or a VAR) captures the aggregate risk premium\",\"Credit-spread measures adequately proxy for credit-market explanations\",\"Cross-sectional predictions (beta versus idiosyncratic risk) separate the risk-premium mechanism from debt-capacity stories\"],\"threats_addressed\":[\"Credit market conditions as an alternative explanation\",\"Debt-equity mispricing alternative\",\"Persistence of regressors (Newey-West SEs)\",\"Seasonality (quarter dummies)\",\"Firm heterogeneity (firm fixed effects)\",\"Time-varying confounds in cross-section (time fixed effects)\",\"Expected growth as a separate valuation component\"]},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"results\"},{\"idx\":9,\"role\":\"conclusion\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_measure\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes give the sample period (1982Q4–2011Q4) and define variables, including the risk premium estimators. Dollar values are deflated to 2010 dollars and activity is scaled in basis points of public firms. Aggregate tables report summary statistics first, then coefficient tables with panels or columns for alternative dependent variables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'consistent with' phrasing\",\"'broadly match' for extension results\",\"acknowledging the simplified view of fund participants\"]},\"claim_evidence\":[{\"claim\":\"Buyout activity falls when the aggregate risk premium is high and rises when it is low\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Risk premium explains about 30% of activity variation versus up to 10% for credit factors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Expected market growth is positively related to buyout activity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Firms with high beta or idiosyncratic volatility are less likely buyout targets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Target beta distribution shifts lower in high risk-premium periods while idiosyncratic risk does not\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table VIII\"},{\"claim\":\"Poor-governance and free-cash-flow firms are more sensitive, and more liquid industries less sensitive, to the risk premium\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Low risk premium raises the NPV of performance gains and lowers the cost of GP incentive risk\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II model; Appendix A–B\"},{\"claim\":\"M&A and IPO activity also co-move with the risk premium\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"Opens with large swings in buyout activity and offers a discount-rate explanation, set against credit-centric literature. Aggregate time-series evidence comes first (risk premium versus credit spreads, then a valuation decomposition). A two-period agency model follows, with performance and illiquidity channels. Firm-panel tests check model-specific cross-sectional predictions (beta versus idiosyncratic risk, agency proxies, exit liquidity) to rule out debt-capacity stories. The paper extends the analysis to M&A and IPOs and closes with heterogeneity caveats and a research agenda.\",\"writing_profile_keywords\":[\"discount-rate channel\",\"contrast with credit-market view\",\"model-derived cross-sectional predictions\",\"evidence-first then theory\",\"confident tone\",\"unified explanation\"],\"confidence\":0.8,\"notes\":[\"Causal language strength rated 4: the abstract says booms 'form in response to' and the conclusion says the risk premium 'explains' waves, but the evidence is regression-based with no exogenous variation.\",\"Standard errors are Newey-West in the aggregate regressions and two-way clustered by firm and quarter in the panel; the schema value 'mixed' covers both.\",\"Several 'sections' are actually equation fragments produced by parsing.\"]}", "json": {"genre": "mixed_theory_empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "corporate_finance; asset_pricing (time-varying discount rates)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between the predicted aggregate equity risk premium and buyout activity at the aggregate and firm level, plus cross-sectional differences in that sensitivity", "strategy": "Time-series regressions of buyout volume and value on predicted risk premium (OLS predictive and VAR/Campbell-Shiller estimates) versus credit factors. Firm panel deal-likelihood regressions with firm or time fixed effects. Interactions of the risk premium with firm risk, agency proxies and exit liquidity to test model-specific predictions.", "key_assumptions": ["Predicted excess return from D/P, cay and the T-bill rate (or a VAR) captures the aggregate risk premium", "Credit-spread measures adequately proxy for credit-market explanations", "Cross-sectional predictions (beta versus idiosyncratic risk) separate the risk-premium mechanism from debt-capacity stories"], "threats_addressed": ["Credit market conditions as an alternative explanation", "Debt-equity mispricing alternative", "Persistence of regressors (Newey-West SEs)", "Seasonality (quarter dummies)", "Firm heterogeneity (firm fixed effects)", "Time-varying confounds in cross-section (time fixed effects)", "Expected growth as a separate valuation component"]}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 5}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "results"}, {"idx": 9, "role": "conclusion"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_measure", "alternative_specification"]}, "tables": {"convention_notes": "Notes give the sample period (1982Q4–2011Q4) and define variables, including the risk premium estimators. Dollar values are deflated to 2010 dollars and activity is scaled in basis points of public firms. Aggregate tables report summary statistics first, then coefficient tables with panels or columns for alternative dependent variables.", "units_stated_in_notes": true, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'consistent with' phrasing", "'broadly match' for extension results", "acknowledging the simplified view of fund participants"]}, "claim_evidence": [{"claim": "Buyout activity falls when the aggregate risk premium is high and rises when it is low", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Risk premium explains about 30% of activity variation versus up to 10% for credit factors", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Expected market growth is positively related to buyout activity", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Firms with high beta or idiosyncratic volatility are less likely buyout targets", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Target beta distribution shifts lower in high risk-premium periods while idiosyncratic risk does not", "evidence_type": "figure", "evidence_ref": "Figure 3; Table VIII"}, {"claim": "Poor-governance and free-cash-flow firms are more sensitive, and more liquid industries less sensitive, to the risk premium", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Low risk premium raises the NPV of performance gains and lowers the cost of GP incentive risk", "evidence_type": "equation_or_proof", "evidence_ref": "Section II model; Appendix A–B"}, {"claim": "M&A and IPO activity also co-move with the risk premium", "evidence_type": "table", "evidence_ref": "Table X"}], "structural_summary": "Opens with large swings in buyout activity and offers a discount-rate explanation, set against credit-centric literature. Aggregate time-series evidence comes first (risk premium versus credit spreads, then a valuation decomposition). A two-period agency model follows, with performance and illiquidity channels. Firm-panel tests check model-specific cross-sectional predictions (beta versus idiosyncratic risk, agency proxies, exit liquidity) to rule out debt-capacity stories. The paper extends the analysis to M&A and IPOs and closes with heterogeneity caveats and a research agenda.", "writing_profile_keywords": ["discount-rate channel", "contrast with credit-market view", "model-derived cross-sectional predictions", "evidence-first then theory", "confident tone", "unified explanation"], "confidence": 0.8, "notes": ["Causal language strength rated 4: the abstract says booms 'form in response to' and the conclusion says the risk premium 'explains' waves, but the evidence is regression-based with no exogenous variation.", "Standard errors are Newey-West in the aggregate regressions and two-way clustered by firm and quarter in the panel; the schema value 'mixed' covers both.", "Several 'sections' are actually equation fragments produced by parsing."]}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.49, "usage": {"input": 8616, "output": 2568, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086618, "request_id": "req_011CggpThPZKLuuR58E9Q9sk", "cached": false, "ts": "2026-11-03T22:55:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a9967e4cd6bc9065fc8d350988734904461d72b9a4bed0778a3dc780cc04b88e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in crisis-period stock returns and operating performance between firms with high and low pre-crisis CSR (social capital), interpreted as the payoff to trust during a trust shock\",\"strategy\":\"Cross-sectional regressions of crisis-period raw and abnormal returns on pre-crisis (2006) CSR with industry dummies and controls. Complemented by panel firm fixed effects models interacting CSR with crisis and post-crisis dummies, and a continuous-treatment difference-in-differences design for operating outcomes.\",\"key_assumptions\":[\"The 2008-2009 crisis was an unexpected shock to the importance of trust\",\"Pre-crisis CSR is not proxying for omitted firm characteristics that independently drive crisis returns\",\"CSR ratings capture firm-level social capital\"],\"threats_addressed\":[\"confounding by the credit supply shock (tested July 2007-July 2008)\",\"CSR firms outperforming generally rather than only in crises (tested via pre/post comparison with firm fixed effects)\",\"time-invariant unobservables (firm fixed effects)\",\"alternative trust shock as external validity check (Enron/Worldcom)\",\"measurement timing of CSR (2005, 2007, 2008 measures)\",\"industry differences (industry dummies)\",\"governance and transparency controls\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":true,\"sentence_index\":0},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":13,\"move\":\"motivation\"},{\"paragraph_index\":13,\"move\":\"literature_gap\"},{\"paragraph_index\":13,\"move\":\"identification_defense\"},{\"paragraph_index\":14,\"move\":\"preview_of_approach\"},{\"paragraph_index\":14,\"move\":\"preview_of_results\"},{\"paragraph_index\":15,\"move\":\"identification_defense\"},{\"paragraph_index\":15,\"move\":\"robustness_mention\"},{\"paragraph_index\":16,\"move\":\"identification_defense\"},{\"paragraph_index\":16,\"move\":\"preview_of_results\"},{\"paragraph_index\":16,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":16,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":16,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_sample_period\",\"falsification\",\"alternative_measure\",\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes define the sample, the CSR construction (strengths minus concerns), return windows, and the abnormal return model, and they write out the panel regression equations. Dependent variables are given as both raw and market-model-adjusted returns.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggests\",\"consistent with\",\"indicate\",\"to alleviate concerns\"]},\"claim_evidence\":[{\"claim\":\"High-CSR firms earned 4-7 percentage points higher returns during the 2008-09 crisis\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"High-CSR firms also outperformed during the Enron/Worldcom trust shock\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"The CSR return effect appears only in the crisis period, not in pre- or post-crisis periods, under firm fixed effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Results are robust to measuring CSR in other years\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"High-CSR firms had higher profitability, sales growth, sales per employee, and debt raising during the crisis\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The relation between CSR and returns is stronger in high-trust regions\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section III.E\"}],\"structural_summary\":\"The paper opens with epigraphs on the crisis as a collapse of trust, then gives macro-level literature on social capital and identifies a gap at the firm level. The introduction previews the main return result with magnitudes and three tests that rule out alternatives, followed by panel and mechanism results. A separate conceptual section links trust, social capital, and CSR. After the data section, a long results section presents baseline regressions, alternative shock episodes, the credit-shock placebo, regional trust heterogeneity, CSR components, and robustness. A difference-in-differences section on operating outcomes and capital raising follows, and a brief conclusion restates the findings.\",\"writing_profile_keywords\":[\"epigraph opening\",\"trust shock framing\",\"alternative-explanation tests\",\"mechanism via real outcomes\",\"CSR measurement\",\"crisis-period returns\"],\"confidence\":0.8,\"notes\":\"The introduction paragraphs at indices 0-12 are epigraph fragments, so the substantive introduction text begins at index 13. Significance reporting conventions are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "asset_pricing", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in crisis-period stock returns and operating performance between firms with high and low pre-crisis CSR (social capital), interpreted as the payoff to trust during a trust shock", "strategy": "Cross-sectional regressions of crisis-period raw and abnormal returns on pre-crisis (2006) CSR with industry dummies and controls. Complemented by panel firm fixed effects models interacting CSR with crisis and post-crisis dummies, and a continuous-treatment difference-in-differences design for operating outcomes.", "key_assumptions": ["The 2008-2009 crisis was an unexpected shock to the importance of trust", "Pre-crisis CSR is not proxying for omitted firm characteristics that independently drive crisis returns", "CSR ratings capture firm-level social capital"], "threats_addressed": ["confounding by the credit supply shock (tested July 2007-July 2008)", "CSR firms outperforming generally rather than only in crises (tested via pre/post comparison with firm fixed effects)", "time-invariant unobservables (firm fixed effects)", "alternative trust shock as external validity check (Enron/Worldcom)", "measurement timing of CSR (2005, 2007, 2008 measures)", "industry differences (industry dummies)", "governance and transparency controls"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": true, "sentence_index": 0}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 13, "move": "motivation"}, {"paragraph_index": 13, "move": "literature_gap"}, {"paragraph_index": 13, "move": "identification_defense"}, {"paragraph_index": 14, "move": "preview_of_approach"}, {"paragraph_index": 14, "move": "preview_of_results"}, {"paragraph_index": 15, "move": "identification_defense"}, {"paragraph_index": 15, "move": "robustness_mention"}, {"paragraph_index": 16, "move": "identification_defense"}, {"paragraph_index": 16, "move": "preview_of_results"}, {"paragraph_index": 16, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 16, "move": "contribution_statement"}], "contribution_paragraph_index": 16, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_sample_period", "falsification", "alternative_measure", "subsample", "alternative_specification"]}, "tables": {"convention_notes": "Table notes define the sample, the CSR construction (strengths minus concerns), return windows, and the abnormal return model, and they write out the panel regression equations. Dependent variables are given as both raw and market-model-adjusted returns.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggests", "consistent with", "indicate", "to alleviate concerns"]}, "claim_evidence": [{"claim": "High-CSR firms earned 4-7 percentage points higher returns during the 2008-09 crisis", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "High-CSR firms also outperformed during the Enron/Worldcom trust shock", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "The CSR return effect appears only in the crisis period, not in pre- or post-crisis periods, under firm fixed effects", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Results are robust to measuring CSR in other years", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "High-CSR firms had higher profitability, sales growth, sales per employee, and debt raising during the crisis", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The relation between CSR and returns is stronger in high-trust regions", "evidence_type": "none_stated", "evidence_ref": "Section III.E"}], "structural_summary": "The paper opens with epigraphs on the crisis as a collapse of trust, then gives macro-level literature on social capital and identifies a gap at the firm level. The introduction previews the main return result with magnitudes and three tests that rule out alternatives, followed by panel and mechanism results. A separate conceptual section links trust, social capital, and CSR. After the data section, a long results section presents baseline regressions, alternative shock episodes, the credit-shock placebo, regional trust heterogeneity, CSR components, and robustness. A difference-in-differences section on operating outcomes and capital raising follows, and a brief conclusion restates the findings.", "writing_profile_keywords": ["epigraph opening", "trust shock framing", "alternative-explanation tests", "mechanism via real outcomes", "CSR measurement", "crisis-period returns"], "confidence": 0.8, "notes": "The introduction paragraphs at indices 0-12 are epigraph fragments, so the substantive introduction text begins at index 13. Significance reporting conventions are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.81, "usage": {"input": 5540, "output": 2121, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065374, "request_id": "req_011CggpfhfGLy8SvPtJsyyAY", "cached": false, "ts": "2026-11-03T22:58:26+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/af98ad51ec92a67a895b0b99ac3b5a5420693f2dc0bd7b43346b2e3180fe645a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"asset_pricing (transaction costs, investor clienteles); policy/taxation\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Average treatment effect of the 2012 French FTT on market quality measures of large French stocks, and on trading by trader type and on fund holdings and turnover\",\"strategy\":\"Difference-in-differences comparing French stocks above the 1bn EUR threshold with non-French control stocks, using stock and time fixed effects. Separate splits for SLP and non-SLP stocks. Exploits the exemption of intraday (HFT) trading to isolate indirect liquidity effects. Fund-level treatment effects are regressed on fund characteristics.\",\"key_assumptions\":[\"parallel trends between French and non-French control stocks\",\"control stocks not affected by the French tax\",\"HFTs affected only indirectly because intraday trading is exempt\"],\"threats_addressed\":[\"mechanical decrease in HFT activity (Table VIII)\",\"heterogeneity across liquidity-provision regimes (SLP vs non-SLP)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"falsification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Table notes restate the estimating regression equation, define the treated and control groups, and state the sample period. Summary tables report standard deviations in parentheses. Some units are given, such as bps and billions of USD. Several table titles use the word 'Causal Impact'.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'in line with' / 'consistent with' theory\",\"'we find no support/evidence'\",\"'suggest'\",\"'it is unclear whether'\",\"'to our knowledge'\"]},\"claim_evidence\":[{\"claim\":\"No evidence that the FTT improves market quality through a change in volume composition; market quality declines via a liquidity effect.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV; Figure 1\"},{\"claim\":\"Negative effects are muted for the most liquid stocks and more severe for less liquid (non-SLP) stocks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Exempt HFTs are indirectly affected through the liquidity channel.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII, IX\"},{\"claim\":\"High-turnover funds reduced holdings of French stocks while low-turnover funds increased them, and funds reduced turnover.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"FTT revenue fell short of expectations.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"official revenue figures, footnote 3\"}],\"structural_summary\":\"The introduction frames competing composition vs liquidity theories of FTTs, notes mixed aggregate evidence, and motivates granular data. Section I describes the policy experiment, develops competing hypotheses, and sets out the DiD design. Section II estimates aggregate market quality effects, with heterogeneity across SLP and non-SLP stocks. Section III examines trader types, especially exempt HFTs, as direct evidence on the liquidity channel. Section IV tests the extensive vs intensive margins of adjustment using fund holdings. The conclusion draws out policy implications and future research.\",\"writing_profile_keywords\":[\"policy experiment\",\"competing hypotheses\",\"difference-in-differences\",\"theory-guided testing\",\"granular data\",\"heterogeneity\",\"policy debate\"],\"confidence\":0.8,\"notes\":\"Standard error type and significance notation are not visible in the excerpts. Robustness is inferred from Table VIII and the internet appendix mention. The paper uses explicit causal phrasing in its table titles.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "asset_pricing (transaction costs, investor clienteles); policy/taxation", "primary_method": "difference_in_differences", "identification": {"estimand": "Average treatment effect of the 2012 French FTT on market quality measures of large French stocks, and on trading by trader type and on fund holdings and turnover", "strategy": "Difference-in-differences comparing French stocks above the 1bn EUR threshold with non-French control stocks, using stock and time fixed effects. Separate splits for SLP and non-SLP stocks. Exploits the exemption of intraday (HFT) trading to isolate indirect liquidity effects. Fund-level treatment effects are regressed on fund characteristics.", "key_assumptions": ["parallel trends between French and non-French control stocks", "control stocks not affected by the French tax", "HFTs affected only indirectly because intraday trading is exempt"], "threats_addressed": ["mechanical decrease in HFT activity (Table VIII)", "heterogeneity across liquidity-provision regimes (SLP vs non-SLP)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 2, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["falsification", "subsample"]}, "tables": {"convention_notes": "Table notes restate the estimating regression equation, define the treated and control groups, and state the sample period. Summary tables report standard deviations in parentheses. Some units are given, such as bps and billions of USD. Several table titles use the word 'Causal Impact'.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'in line with' / 'consistent with' theory", "'we find no support/evidence'", "'suggest'", "'it is unclear whether'", "'to our knowledge'"]}, "claim_evidence": [{"claim": "No evidence that the FTT improves market quality through a change in volume composition; market quality declines via a liquidity effect.", "evidence_type": "table", "evidence_ref": "Tables III, IV; Figure 1"}, {"claim": "Negative effects are muted for the most liquid stocks and more severe for less liquid (non-SLP) stocks.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Exempt HFTs are indirectly affected through the liquidity channel.", "evidence_type": "table", "evidence_ref": "Tables VII, VIII, IX"}, {"claim": "High-turnover funds reduced holdings of French stocks while low-turnover funds increased them, and funds reduced turnover.", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "FTT revenue fell short of expectations.", "evidence_type": "descriptive_statistic", "evidence_ref": "official revenue figures, footnote 3"}], "structural_summary": "The introduction frames competing composition vs liquidity theories of FTTs, notes mixed aggregate evidence, and motivates granular data. Section I describes the policy experiment, develops competing hypotheses, and sets out the DiD design. Section II estimates aggregate market quality effects, with heterogeneity across SLP and non-SLP stocks. Section III examines trader types, especially exempt HFTs, as direct evidence on the liquidity channel. Section IV tests the extensive vs intensive margins of adjustment using fund holdings. The conclusion draws out policy implications and future research.", "writing_profile_keywords": ["policy experiment", "competing hypotheses", "difference-in-differences", "theory-guided testing", "granular data", "heterogeneity", "policy debate"], "confidence": 0.8, "notes": "Standard error type and significance notation are not visible in the excerpts. Robustness is inferred from Table VIII and the internet appendix mention. The paper uses explicit causal phrasing in its table titles."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.99, "usage": {"input": 6779, "output": 1998, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06787, "request_id": "req_011Cggpn756kc6isVapqRaMf", "cached": false, "ts": "2026-11-03T22:59:51+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b3d4ec80ca625e673836f522671ade7a9b9a1bf8b2b34accfed4d18a16f7bfa6.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Model-implied reverse mortgage take-up rates and welfare gains, and how they respond to counterfactual changes in risks, bequest motives, contract terms and Great Recession shocks\",\"strategy\":\"Life-cycle retirement model with uninsurable health, mortality, medical expense and house price risk. It is calibrated by two-stage estimation (Gourinchas–Parker style) on HRS data; RMLs are added afterward and counterfactual experiments are run.\",\"key_assumptions\":[\"Model structure, including preferences, bequest motive and risk processes, describes retiree behavior\",\"No aggregate uncertainty in the baseline model\",\"Parameters estimated without RMLs remain valid once RMLs are introduced\",\"Retirees aged 65+ with no labor supply decision\"],\"threats_addressed\":[\"Model fit checked against untargeted moments: distribution of debt, portfolio shares, and age distribution of RML borrowers\",\"Sensitivity analysis reported in the Internet Appendix\",\"Additional experiments with aggregate house price dynamics\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":4},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"internet_appendix\",\"types\":[\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Tables report the calibrated parameters and model-generated take-up rates and welfare gains under baseline and counterfactual scenarios. Notes define each column and the welfare measure (a one-time payment at age 65 equating expected lifetime utility), cross-reference Table II's notes, and state parameter frequency (annual vs. biennial). No standard errors are shown.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"unclear\",\"figures_note_placeholder\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'according to the model' / 'our model indicates' attribution\",\"modal verbs such as 'may' and 'is likely to'\",\"conditioning on the existing RML contract\",\"acknowledging that aggregate house price dynamics are absent from the model\"]},\"claim_evidence\":[{\"claim\":\"Average welfare gain from RMLs is small per homeowner but larger for borrowers, low-income households and the oldest households\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Bequest motives and uncertainty about health and expenses dampen RML demand; without bequest motives, take-up rises sharply\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Eliminating up-front costs, or making RMLs recourse loans, roughly triples demand\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The Great Recession lowers aggregate take-up in the short run but triples demand among the oldest, lowest-income households\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The RML take-up share among older homeowners is low but has grown over time\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"The calibrated model matches asset age profiles, the debt distribution and the age distribution of RML borrowers\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-5, 7\"}],\"structural_summary\":\"The introduction opens with the product, the gap in research, and the low-take-up puzzle, then poses four research questions. It previews heterogeneous welfare gains, counterfactual findings and Great Recession effects, and closes with contributions and positioning against the literatures on retirement saving and mortgage choice. The body proceeds through: institutional overview; a structural life-cycle model covering renters, owners, and owners with an RML; two-stage calibration with checks of fit on untargeted moments; and results built on counterfactual experiments. The conclusion restates the findings, notes policy trade-offs and a model limitation, and points to future work.\",\"writing_profile_keywords\":[\"quantitative structural model\",\"life-cycle retirement\",\"counterfactual experiments\",\"welfare gains\",\"policy reform evaluation\",\"model fit validation\",\"heterogeneity across income and age\"],\"confidence\":0.85,\"notes\":\"Calibration uses a two-stage estimation in the style of Gourinchas–Parker. It was labeled structural_estimation, though the paper calls itself calibrated. Section I is an institutional overview. The 0.89% model take-up rate differs from the 1.9% data figure.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "macro_finance", "primary_method": "structural_estimation", "identification": {"estimand": "Model-implied reverse mortgage take-up rates and welfare gains, and how they respond to counterfactual changes in risks, bequest motives, contract terms and Great Recession shocks", "strategy": "Life-cycle retirement model with uninsurable health, mortality, medical expense and house price risk. It is calibrated by two-stage estimation (Gourinchas–Parker style) on HRS data; RMLs are added afterward and counterfactual experiments are run.", "key_assumptions": ["Model structure, including preferences, bequest motive and risk processes, describes retiree behavior", "No aggregate uncertainty in the baseline model", "Parameters estimated without RMLs remain valid once RMLs are introduced", "Retirees aged 65+ with no labor supply decision"], "threats_addressed": ["Model fit checked against untargeted moments: distribution of debt, portfolio shares, and age distribution of RML borrowers", "Sensitivity analysis reported in the Internet Appendix", "Additional experiments with aggregate house price dynamics"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 4}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_theory_or_model", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "internet_appendix", "types": ["sensitivity_to_parameters"]}, "tables": {"convention_notes": "Tables report the calibrated parameters and model-generated take-up rates and welfare gains under baseline and counterfactual scenarios. Notes define each column and the welfare measure (a one-time payment at age 65 equating expected lifetime utility), cross-reference Table II's notes, and state parameter frequency (annual vs. biennial). No standard errors are shown.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "unclear", "figures_note_placeholder": null}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'according to the model' / 'our model indicates' attribution", "modal verbs such as 'may' and 'is likely to'", "conditioning on the existing RML contract", "acknowledging that aggregate house price dynamics are absent from the model"]}, "claim_evidence": [{"claim": "Average welfare gain from RMLs is small per homeowner but larger for borrowers, low-income households and the oldest households", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Bequest motives and uncertainty about health and expenses dampen RML demand; without bequest motives, take-up rises sharply", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Eliminating up-front costs, or making RMLs recourse loans, roughly triples demand", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The Great Recession lowers aggregate take-up in the short run but triples demand among the oldest, lowest-income households", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "The RML take-up share among older homeowners is low but has grown over time", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "The calibrated model matches asset age profiles, the debt distribution and the age distribution of RML borrowers", "evidence_type": "figure", "evidence_ref": "Figures 2-5, 7"}], "structural_summary": "The introduction opens with the product, the gap in research, and the low-take-up puzzle, then poses four research questions. It previews heterogeneous welfare gains, counterfactual findings and Great Recession effects, and closes with contributions and positioning against the literatures on retirement saving and mortgage choice. The body proceeds through: institutional overview; a structural life-cycle model covering renters, owners, and owners with an RML; two-stage calibration with checks of fit on untargeted moments; and results built on counterfactual experiments. The conclusion restates the findings, notes policy trade-offs and a model limitation, and points to future work.", "writing_profile_keywords": ["quantitative structural model", "life-cycle retirement", "counterfactual experiments", "welfare gains", "policy reform evaluation", "model fit validation", "heterogeneity across income and age"], "confidence": 0.85, "notes": "Calibration uses a two-stage estimation in the style of Gourinchas–Parker. It was labeled structural_estimation, though the paper calls itself calibrated. Section I is an institutional overview. The 0.89% model take-up rate differs from the 1.9% data figure."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.05, "usage": {"input": 7036, "output": 2229, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073518, "request_id": "req_011CggpYvguM4oRi2dVp599B", "cached": false, "ts": "2026-11-03T22:56:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b4476c02a9753adea77f974a6ac5f7fd6d6d556be8eb7b0b54cd6337c3a5d02b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Model-implied cross-sectional dispersion of returns (size and value premia) when capital adjustment costs are disciplined by firm-level investment moments\",\"strategy\":\"Calibrate a one-factor neoclassical production-based model to Compustat investment moments, the labor share and risk-free rate moments, then compare simulated portfolio sorts to the data; extend with operating and labor leverage\",\"key_assumptions\":[\"one-factor stochastic discount factor\",\"mean-reverting idiosyncratic productivity\",\"adjustment cost parameterization matched to investment volatility, autocorrelation and the share of negative investment\",\"operating cost calibrated to the firm exit rate\"],\"threats_addressed\":[\"operating leverage as an amplification channel\",\"labor leverage via CES production and wage dynamics\",\"operating costs that scale with capital\",\"reconciliation with Zhang (2005)\",\"asymmetric adjustment costs\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":7,\"move\":\"robustness_mention\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":11,\"move\":\"other\"},{\"paragraph_index\":12,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"robustness\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Tables present data and model-simulated moments side by side, with targeted moments flagged. Calibrated parameters are listed by symbol, with Cobb-Douglas and CES specifications shown in separate columns. Portfolio-sort tables report model excess returns in percent, with the data top-minus-bottom spread on a separate row. Comparative-statics tables vary the adjustment-cost parameters.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"scope restriction to one-factor models\",\"phrases such as 'do not necessarily extend'\",\"'is likely to' and 'does not appear promising'\",\"explicit acknowledgment of what remains unclear in the two-factor literature\"]},\"claim_evidence\":[{\"claim\":\"Public firms divest capital frequently after adverse shocks, with negative investment in about 18% of quarterly observations.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I, Figure 1\"},{\"claim\":\"With adjustment costs disciplined by investment data, the model's value premium is about one-third of the empirical premium.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Table V\"},{\"claim\":\"Operating costs that scale with capital do not raise return dispersion once investment targets are re-matched.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Labor leverage via CES production fails to close the gap in the value premium.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII-VIII\"},{\"claim\":\"Zhang (2005) generates a value premium only with investment volatility far below the data.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table C1, Appendix C\"},{\"claim\":\"Analytical intuition for how productivity, size, and book-to-market relate to returns.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Appendix B\"}],\"structural_summary\":\"The paper first documents firm-level investment facts from Compustat. It then builds and calibrates a one-factor neoclassical production model to those facts and shows that the model's size and value premia fall well short of the data. Next, it tests whether operating leverage and labor leverage, the leading amplification fixes in the literature, can close the gap, and finds they cannot once investment moments are matched. Appendices add a three-period analytical model and a reconciliation with Zhang (2005). The argument is framed as a tension between the parameter restrictions implied by quantities and by prices.\",\"writing_profile_keywords\":[\"production-based asset pricing\",\"calibration\",\"negative result\",\"investment moments discipline\",\"value premium\",\"critique of paradigm\"],\"confidence\":0.8,\"notes\":\"This is a quantitative-theory critique that rests on calibrated simulations; there is no econometric identification. The assignment of Sections III and IV to robustness is interpretive: they serve as model extensions that test the robustness of the main conclusion.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "calibration_simulation", "identification": {"estimand": "Model-implied cross-sectional dispersion of returns (size and value premia) when capital adjustment costs are disciplined by firm-level investment moments", "strategy": "Calibrate a one-factor neoclassical production-based model to Compustat investment moments, the labor share and risk-free rate moments, then compare simulated portfolio sorts to the data; extend with operating and labor leverage", "key_assumptions": ["one-factor stochastic discount factor", "mean-reverting idiosyncratic productivity", "adjustment cost parameterization matched to investment volatility, autocorrelation and the share of negative investment", "operating cost calibrated to the firm exit rate"], "threats_addressed": ["operating leverage as an amplification channel", "labor leverage via CES production and wage dynamics", "operating costs that scale with capital", "reconciliation with Zhang (2005)", "asymmetric adjustment costs"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "tension_or_puzzle"}, {"paragraph_index": 7, "move": "robustness_mention"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "policy_or_broader_implication"}, {"paragraph_index": 11, "move": "other"}, {"paragraph_index": 12, "move": "roadmap"}], "contribution_paragraph_index": 8, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "robustness"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Tables present data and model-simulated moments side by side, with targeted moments flagged. Calibrated parameters are listed by symbol, with Cobb-Douglas and CES specifications shown in separate columns. Portfolio-sort tables report model excess returns in percent, with the data top-minus-bottom spread on a separate row. Comparative-statics tables vary the adjustment-cost parameters.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["scope restriction to one-factor models", "phrases such as 'do not necessarily extend'", "'is likely to' and 'does not appear promising'", "explicit acknowledgment of what remains unclear in the two-factor literature"]}, "claim_evidence": [{"claim": "Public firms divest capital frequently after adverse shocks, with negative investment in about 18% of quarterly observations.", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I, Figure 1"}, {"claim": "With adjustment costs disciplined by investment data, the model's value premium is about one-third of the empirical premium.", "evidence_type": "table", "evidence_ref": "Table IV, Table V"}, {"claim": "Operating costs that scale with capital do not raise return dispersion once investment targets are re-matched.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Labor leverage via CES production fails to close the gap in the value premium.", "evidence_type": "table", "evidence_ref": "Tables VII-VIII"}, {"claim": "Zhang (2005) generates a value premium only with investment volatility far below the data.", "evidence_type": "table", "evidence_ref": "Table C1, Appendix C"}, {"claim": "Analytical intuition for how productivity, size, and book-to-market relate to returns.", "evidence_type": "equation_or_proof", "evidence_ref": "Appendix B"}], "structural_summary": "The paper first documents firm-level investment facts from Compustat. It then builds and calibrates a one-factor neoclassical production model to those facts and shows that the model's size and value premia fall well short of the data. Next, it tests whether operating leverage and labor leverage, the leading amplification fixes in the literature, can close the gap, and finds they cannot once investment moments are matched. Appendices add a three-period analytical model and a reconciliation with Zhang (2005). The argument is framed as a tension between the parameter restrictions implied by quantities and by prices.", "writing_profile_keywords": ["production-based asset pricing", "calibration", "negative result", "investment moments discipline", "value premium", "critique of paradigm"], "confidence": 0.8, "notes": "This is a quantitative-theory critique that rests on calibrated simulations; there is no econometric identification. The assignment of Sections III and IV to robustness is interpretive: they serve as model extensions that test the robustness of the main conclusion."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.33, "usage": {"input": 8040, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077994, "request_id": "req_011CggqBesg1TV1RPRbtWdBg", "cached": false, "ts": "2026-11-03T23:05:13+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b98c63f6848a3834d63cb0a3918eaf1b485d7875ac981e68b8ce1b12b0c351da.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of executives' personal lending relationships on lender choice and loan terms (spreads, amounts), and downstream effects on investment and employment during downturns\",\"strategy\":\"Executive deaths and quasi-exogenous retirements at other firms instrument for own-firm executive turnover, which shocks the firm's set of personal bank connections. Within-firm comparison across lenders, before and after turnover, using LPM/OLS/IV regressions.\",\"key_assumptions\":[\"Exclusion restriction: deaths and retirements at other firms do not directly affect a firm's loan terms across connected versus unconnected lenders\",\"Relevance: a geographically constrained executive market with industry-specific expertise links outside deaths to own-firm turnover\",\"Relationship-intensity measures capture the ex ante expected number of transactions between borrower and lender\"],\"threats_addressed\":[\"Executive-firm matching on lending relationships (effects absent before turnover)\",\"Endogeneity of turnover\",\"Firm-level confounds (within-firm, across-lender comparison)\",\"Compensating differentials in other loan terms\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_identification\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes describe the estimator (fixed-effect LPM, OLS vs. IV), the dependent variable definition, sample size and period (9,458 Dealscan loans, 1994-2012), and the panel structure. Variable construction is referred to the text and the Internet Appendix. The aggregation table explicitly states its partial-equilibrium ceteris paribus assumption.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"plausibly exogenous / quasi-exogenous\",\"suggests\",\"most consistent with\",\"may alleviate\",\"under certain conditions\",\"back-of-the-envelope / partial equilibrium caveat\"]},\"claim_evidence\":[{\"claim\":\"Outside executive deaths and retirements predict own-firm executive turnover\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"After turnover, firms favor lenders tied to new executives and get lower spreads and larger loans, with no pre-turnover effect\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V; Figure 2\"},{\"claim\":\"Personal relationship intensity has a larger, more precise effect on lender choice than institutional intensity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Personal relationship effects on loan terms are stronger in NBER recessions and the 2008-09 crisis\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Firms with intense personal relationships invest more and cut fewer jobs in downturns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII; Figure 3\"},{\"claim\":\"Aggregate crisis investment would be at least $32.6bn lower without intense personal relationships\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The introduction opens with Coasean relationship theory and the mixed evidence on firm-bank relationships, then proposes that relationships sit with executives rather than firms. It previews an instrument based on executive deaths and retirements at other firms and defends the exclusion restriction. Results follow with magnitudes, a within-firm design illustrated by a figure, and evidence on downturns. Contributions and positioning against the social-ties papers come next, then a roadmap. The body proceeds from measurement and sample, to turnover-based lender choice and loan terms, to downturn heterogeneity and real effects, then a back-of-the-envelope aggregation. It closes with polynomial, placebo and robustness tests and a short conclusion restating the findings with a dollar figure.\",\"writing_profile_keywords\":[\"first-person singular\",\"instrumental variables\",\"executive turnover shock\",\"economic magnitudes in dollars\",\"within-firm across-lender comparison\",\"financial crisis real effects\",\"theory-motivated puzzle\"],\"confidence\":0.8,\"notes\":\"How significance is reported (stars, t-stats, clustering) is not visible in the excerpts. Section V's opening paragraph refers to 'Sections IV and V', apparently a typo for Sections III and IV.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of executives' personal lending relationships on lender choice and loan terms (spreads, amounts), and downstream effects on investment and employment during downturns", "strategy": "Executive deaths and quasi-exogenous retirements at other firms instrument for own-firm executive turnover, which shocks the firm's set of personal bank connections. Within-firm comparison across lenders, before and after turnover, using LPM/OLS/IV regressions.", "key_assumptions": ["Exclusion restriction: deaths and retirements at other firms do not directly affect a firm's loan terms across connected versus unconnected lenders", "Relevance: a geographically constrained executive market with industry-specific expertise links outside deaths to own-firm turnover", "Relationship-intensity measures capture the ex ante expected number of transactions between borrower and lender"], "threats_addressed": ["Executive-firm matching on lending relationships (effects absent before turnover)", "Endogeneity of turnover", "Firm-level confounds (within-firm, across-lender comparison)", "Compensating differentials in other loan terms"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_data_or_measure", "new_identification", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes describe the estimator (fixed-effect LPM, OLS vs. IV), the dependent variable definition, sample size and period (9,458 Dealscan loans, 1994-2012), and the panel structure. Variable construction is referred to the text and the Internet Appendix. The aggregation table explicitly states its partial-equilibrium ceteris paribus assumption.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["plausibly exogenous / quasi-exogenous", "suggests", "most consistent with", "may alleviate", "under certain conditions", "back-of-the-envelope / partial equilibrium caveat"]}, "claim_evidence": [{"claim": "Outside executive deaths and retirements predict own-firm executive turnover", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "After turnover, firms favor lenders tied to new executives and get lower spreads and larger loans, with no pre-turnover effect", "evidence_type": "table", "evidence_ref": "Tables IV, V; Figure 2"}, {"claim": "Personal relationship intensity has a larger, more precise effect on lender choice than institutional intensity", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Personal relationship effects on loan terms are stronger in NBER recessions and the 2008-09 crisis", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Firms with intense personal relationships invest more and cut fewer jobs in downturns", "evidence_type": "table", "evidence_ref": "Table VIII; Figure 3"}, {"claim": "Aggregate crisis investment would be at least $32.6bn lower without intense personal relationships", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The introduction opens with Coasean relationship theory and the mixed evidence on firm-bank relationships, then proposes that relationships sit with executives rather than firms. It previews an instrument based on executive deaths and retirements at other firms and defends the exclusion restriction. Results follow with magnitudes, a within-firm design illustrated by a figure, and evidence on downturns. Contributions and positioning against the social-ties papers come next, then a roadmap. The body proceeds from measurement and sample, to turnover-based lender choice and loan terms, to downturn heterogeneity and real effects, then a back-of-the-envelope aggregation. It closes with polynomial, placebo and robustness tests and a short conclusion restating the findings with a dollar figure.", "writing_profile_keywords": ["first-person singular", "instrumental variables", "executive turnover shock", "economic magnitudes in dollars", "within-firm across-lender comparison", "financial crisis real effects", "theory-motivated puzzle"], "confidence": 0.8, "notes": "How significance is reported (stars, t-stats, clustering) is not visible in the excerpts. Section V's opening paragraph refers to 'Sections IV and V', apparently a typo for Sections III and IV."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.23, "usage": {"input": 7870, "output": 2347, "cache_read": 0, "cache_write": 3968}, "cost_usd": 0.09826, "request_id": "req_011CggpMRmJ56eBjjbFYjXy9", "cached": false, "ts": "2026-11-03T22:54:20+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b9f37a3f6aa5677023a96b0ecd19ea3352ae0b5fff1d7d25fe7bfe2599468a09.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"credit derivatives / auction design\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model of equilibrium prices and allocations)\",\"strategy\":\"not applicable; equilibrium analysis of a stylized two-stage CDS auction compared against a competitive/double-auction benchmark\",\"key_assumptions\":[\"infinitesimal price-taking traders\",\"symmetric distribution of CDS benefits\",\"price caps/floors not formally modeled in core analysis\"],\"threats_addressed\":[\"imperfect competition handled in a generalized double auction appendix (finite n traders)\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"appendix\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"Tables are qualitative summaries: one maps states and open interest to the direction of price bias and to allocation errors by trader type; the other compares the paper's economic channels with prior models using yes/no entries.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"explicit model-scope caveats (infinitesimal traders)\",\"acknowledging double auction inefficiency under imperfect competition\",\"'in principle'/'may' phrasing on practical motivations\"]},\"claim_evidence\":[{\"claim\":\"Current two-stage CDS auction design yields biased prices and inefficient allocations of defaulted bonds\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV, Propositions; Appendix A proofs; Table I\"},{\"claim\":\"A double auction with unconstrained demand schedules attains the competitive first-best outcome\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III\"},{\"claim\":\"Price caps/floors formed by dealer quotes give dealers incentive and ability to manipulate the final price\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section V (informal discussion)\"},{\"claim\":\"With finitely many traders the double auction keeps the competitive price and its allocation inefficiency is O(1/n)\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Appendix B\"},{\"claim\":\"Auction-generated information matters for postauction price formation\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Gupta and Sundaram (2012)\"},{\"claim\":\"Numerical example illustrates CDS positions, physical requests, auction prices and allocations\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 3-6\"}],\"structural_summary\":\"Opens with the size of the CDS market and its policy relevance, then states that recovery value set by the auction matters for pricing and settlement. Institutional background on the two-stage auction follows, illustrated with Lehman figures. A stylized model with infinitesimal traders is set up and benchmarked against competitive equilibrium. Equilibrium analysis shows price bias and misallocation, supported by a numerical example and propositions. A short discussion covers price caps/floors and dealer manipulation, followed by empirical predictions. The paper closes by proposing a double auction design and assessing why the current restrictions exist. Proofs and a finite-trader generalization are in the appendices.\",\"writing_profile_keywords\":[\"mechanism design\",\"auction theory\",\"policy-oriented\",\"propositions and proofs\",\"numerical illustration\",\"institutional detail\",\"design proposal\"],\"confidence\":0.75,\"notes\":\"The introduction is given as a single truncated paragraph, so contribution and roadmap moves are not visible. Section V is labeled 'conclusion' by the rule, but it is analytical, so it is annotated as theory. Section I is institutional background and is annotated as other. Empirical predictions appear in Section VI, but the paper contains no data analysis.\"}", "json": {"genre": "theoretical", "subfield": "market_microstructure", "subfield_secondary": "credit derivatives / auction design", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model of equilibrium prices and allocations)", "strategy": "not applicable; equilibrium analysis of a stylized two-stage CDS auction compared against a competitive/double-auction benchmark", "key_assumptions": ["infinitesimal price-taking traders", "symmetric distribution of CDS benefits", "price caps/floors not formally modeled in core analysis"], "threats_addressed": ["imperfect competition handled in a generalized double auction appendix (finite n traders)"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}], "contribution_paragraph_index": -1, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "appendix", "types": ["other"]}, "tables": {"convention_notes": "Tables are qualitative summaries: one maps states and open interest to the direction of price bias and to allocation errors by trader type; the other compares the paper's economic channels with prior models using yes/no entries.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["explicit model-scope caveats (infinitesimal traders)", "acknowledging double auction inefficiency under imperfect competition", "'in principle'/'may' phrasing on practical motivations"]}, "claim_evidence": [{"claim": "Current two-stage CDS auction design yields biased prices and inefficient allocations of defaulted bonds", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV, Propositions; Appendix A proofs; Table I"}, {"claim": "A double auction with unconstrained demand schedules attains the competitive first-best outcome", "evidence_type": "equation_or_proof", "evidence_ref": "Section III"}, {"claim": "Price caps/floors formed by dealer quotes give dealers incentive and ability to manipulate the final price", "evidence_type": "none_stated", "evidence_ref": "Section V (informal discussion)"}, {"claim": "With finitely many traders the double auction keeps the competitive price and its allocation inefficiency is O(1/n)", "evidence_type": "equation_or_proof", "evidence_ref": "Appendix B"}, {"claim": "Auction-generated information matters for postauction price formation", "evidence_type": "prior_literature", "evidence_ref": "Gupta and Sundaram (2012)"}, {"claim": "Numerical example illustrates CDS positions, physical requests, auction prices and allocations", "evidence_type": "figure", "evidence_ref": "Figures 3-6"}], "structural_summary": "Opens with the size of the CDS market and its policy relevance, then states that recovery value set by the auction matters for pricing and settlement. Institutional background on the two-stage auction follows, illustrated with Lehman figures. A stylized model with infinitesimal traders is set up and benchmarked against competitive equilibrium. Equilibrium analysis shows price bias and misallocation, supported by a numerical example and propositions. A short discussion covers price caps/floors and dealer manipulation, followed by empirical predictions. The paper closes by proposing a double auction design and assessing why the current restrictions exist. Proofs and a finite-trader generalization are in the appendices.", "writing_profile_keywords": ["mechanism design", "auction theory", "policy-oriented", "propositions and proofs", "numerical illustration", "institutional detail", "design proposal"], "confidence": 0.75, "notes": "The introduction is given as a single truncated paragraph, so contribution and roadmap moves are not visible. Section V is labeled 'conclusion' by the rule, but it is analytical, so it is annotated as theory. Section I is institutional background and is annotated as other. Empirical predictions appear in Section VI, but the paper contains no data analysis."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.81, "usage": {"input": 6074, "output": 1826, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06161, "request_id": "req_011CggpkwVcEBK8aSNfuickp", "cached": false, "ts": "2026-11-03T22:59:35+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ba6e81c8ca490d18b286e9587c704cb0f1674e4a734bb0e3564dcce389193f04.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"international_finance (law and finance, CSR/ESG)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between a country's legal origin (common vs. civil, by family) and firm-level CSR ratings, plus the differential responsiveness of CSR to demand shocks by legal origin.\",\"strategy\":\"Cross-country panel regressions (pooled OLS, random-effects GLS, random-effects ordered probit) of CSR ratings on legal origin dummies with year and industry fixed effects and country- and firm-level controls. Quasi-natural experiments (Chinese milk scandal, Indian Ocean earthquake, Deepwater Horizon) estimated as difference-in-differences with country fixed effects, plus placebo tests and mechanism tests.\",\"key_assumptions\":[\"Legal origin is predetermined and does not merely proxy for omitted country characteristics once controls are included\",\"Scandals and disasters shift CSR demand exogenously and comparably across legal origins\",\"Placebo industries are unaffected by the shocks\"],\"threats_addressed\":[\"alternative explanations: doing good by doing well, ownership, political institutions, globalization, culture\",\"country-level omitted channels, handled with country fixed effects in the shock experiments\",\"changes in market shares or consumer demand driving the responsiveness results\",\"choice of CSR measure or data provider\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"placebo\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables open with descriptive and nonparametric comparisons by legal origin, followed by the main regression table across multiple estimators. Later robustness tables are framed as repeating one benchmark model with alternative controls, dependent variables or samples. Notes state the estimation method and fixed effects and use a 10/5/1% star convention. Descriptive tables report standard deviations in brackets.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'appears to be' and 'suggests' attached to findings\",\"framing results as correlations or associations\",\"explicit caveat that higher CSR does not imply higher total social responsibility\",\"'may reflect' alternative interpretations such as constraints rather than managerial objectives\"]},\"claim_evidence\":[{\"claim\":\"Firms from common law countries have lower CSR ratings than civil law firms, and Scandinavian-origin firms score highest.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-IV\"},{\"claim\":\"Legal origin outperforms ownership, political institutions, culture and firm financials as an explanation of CSR.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV-V\"},{\"claim\":\"The legal origin result holds across CSR subindices and alternative ESG data providers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VII\"},{\"claim\":\"Civil law firms respond more strongly in CSR to scandals and disasters.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-IX\"},{\"claim\":\"The greater responsiveness of civil law firms is not explained by changes in market shares.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Civil law firms face lower shareholder litigation risk and more regulation, which serve as channels.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Countries with higher sustainability ratings tend to be civil law countries.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"The paper opens with the tension between shareholder primacy and widespread CSR activity, then argues that the doing good/doing well literature cannot explain cross-country variation. It proposes legal origin as the explanation and grounds this in law-and-finance theory in a dedicated section. Global CSR data feed descriptive and nonparametric comparisons, followed by panel regressions using several estimators. Robustness is organized as tests of alternative theories, alternative dependent variables and alternative samples. The paper then adds difference-in-differences shock experiments with placebo tests and a market-share check, and closes with tests of mechanisms. The conclusion restates macro- and micro-level contributions and caveats against welfare interpretations.\",\"writing_profile_keywords\":[\"law and finance\",\"cross-country panel\",\"horse race against alternative explanations\",\"quasi-natural experiments\",\"mechanism tests\",\"multiple data sources\",\"associational framing\"],\"confidence\":0.8,\"notes\":\"The section titled Economic Mechanisms was labeled robustness by the rule; I assigned it the results role because it directly tests channels. The excerpts do not state how standard errors are computed or clustered, so that field is unclear. Causal wording is mostly associational ('strongly correlated', 'predictor'), although the shock experiments are framed as quasi-natural experiments; I therefore rated causal language strength at 2.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "international_finance (law and finance, CSR/ESG)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between a country's legal origin (common vs. civil, by family) and firm-level CSR ratings, plus the differential responsiveness of CSR to demand shocks by legal origin.", "strategy": "Cross-country panel regressions (pooled OLS, random-effects GLS, random-effects ordered probit) of CSR ratings on legal origin dummies with year and industry fixed effects and country- and firm-level controls. Quasi-natural experiments (Chinese milk scandal, Indian Ocean earthquake, Deepwater Horizon) estimated as difference-in-differences with country fixed effects, plus placebo tests and mechanism tests.", "key_assumptions": ["Legal origin is predetermined and does not merely proxy for omitted country characteristics once controls are included", "Scandals and disasters shift CSR demand exogenously and comparably across legal origins", "Placebo industries are unaffected by the shocks"], "threats_addressed": ["alternative explanations: doing good by doing well, ownership, political institutions, globalization, culture", "country-level omitted channels, handled with country fixed effects in the shock experiments", "changes in market shares or consumer demand driving the responsiveness results", "choice of CSR measure or data provider"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure", "placebo", "other"]}, "tables": {"convention_notes": "Tables open with descriptive and nonparametric comparisons by legal origin, followed by the main regression table across multiple estimators. Later robustness tables are framed as repeating one benchmark model with alternative controls, dependent variables or samples. Notes state the estimation method and fixed effects and use a 10/5/1% star convention. Descriptive tables report standard deviations in brackets.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'appears to be' and 'suggests' attached to findings", "framing results as correlations or associations", "explicit caveat that higher CSR does not imply higher total social responsibility", "'may reflect' alternative interpretations such as constraints rather than managerial objectives"]}, "claim_evidence": [{"claim": "Firms from common law countries have lower CSR ratings than civil law firms, and Scandinavian-origin firms score highest.", "evidence_type": "table", "evidence_ref": "Tables II-IV"}, {"claim": "Legal origin outperforms ownership, political institutions, culture and firm financials as an explanation of CSR.", "evidence_type": "table", "evidence_ref": "Tables IV-V"}, {"claim": "The legal origin result holds across CSR subindices and alternative ESG data providers.", "evidence_type": "table", "evidence_ref": "Tables VI-VII"}, {"claim": "Civil law firms respond more strongly in CSR to scandals and disasters.", "evidence_type": "table", "evidence_ref": "Tables VIII-IX"}, {"claim": "The greater responsiveness of civil law firms is not explained by changes in market shares.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Civil law firms face lower shareholder litigation risk and more regulation, which serve as channels.", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Countries with higher sustainability ratings tend to be civil law countries.", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "The paper opens with the tension between shareholder primacy and widespread CSR activity, then argues that the doing good/doing well literature cannot explain cross-country variation. It proposes legal origin as the explanation and grounds this in law-and-finance theory in a dedicated section. Global CSR data feed descriptive and nonparametric comparisons, followed by panel regressions using several estimators. Robustness is organized as tests of alternative theories, alternative dependent variables and alternative samples. The paper then adds difference-in-differences shock experiments with placebo tests and a market-share check, and closes with tests of mechanisms. The conclusion restates macro- and micro-level contributions and caveats against welfare interpretations.", "writing_profile_keywords": ["law and finance", "cross-country panel", "horse race against alternative explanations", "quasi-natural experiments", "mechanism tests", "multiple data sources", "associational framing"], "confidence": 0.8, "notes": "The section titled Economic Mechanisms was labeled robustness by the rule; I assigned it the results role because it directly tests channels. The excerpts do not state how standard errors are computed or clustered, so that field is unclear. Causal wording is mostly associational ('strongly correlated', 'predictor'), although the shock experiments are framed as quasi-natural experiments; I therefore rated causal language strength at 2."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.05, "usage": {"input": 8623, "output": 2443, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084146, "request_id": "req_011CggpYpd1yJoXawyuvzD99", "cached": false, "ts": "2026-11-03T22:56:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/bf8953e88697fe7709378c0602f3c5411b7328d76ac77b5b377f7565a9fdec7a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Predictive association between the labor-share and term-structure moments: variance ratios of wages and dividends, and the risk, premium and slope of dividend strips.\",\"strategy\":\"Time-series predictive regressions of variance ratios and dividend strip returns on the labor-share, with controls and Newey-West t-statistics, combined with a calibrated general equilibrium model whose predictions are compared with the data.\",\"key_assumptions\":[\"Income insurance concerns transitory shocks only.\",\"Workers and shareholders share permanent shocks, with cointegrated levels.\",\"Limited market participation: workers consume their wages.\"],\"threats_addressed\":[\"Financial leverage as an alternative channel for downward-sloping dividend risk (EBIT term structure shown to be similar).\",\"Other sources of transitory dividend risk not attributable to income insurance.\",\"Belief-formation channel as an alternative explanation.\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"sentence_index\":4},\"method\":{\"present\":true,\"sentence_index\":4},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"other\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state the regression equations, report coefficients with Newey-West (and Hansen-Hodrick) t-statistics in parentheses and adjusted R2; calibration tables set data moments against model moments.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'supports the idea'\",\"'consistent with'\",\"'can have only a limited effect'\",\"explicit acknowledgment of model limitations\"]},\"claim_evidence\":[{\"claim\":\"Variance-ratio term structures of output, wages and dividends are flat, increasing and decreasing respectively.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Labor-share predicts the wage-dividend variance-ratio gap and the dividend variance ratios.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Figure 2\"},{\"claim\":\"Labor-share forecasts dividend strip volatility, premium and slope, with high correlations.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-III; Figure 3\"},{\"claim\":\"Labor-share positively forecasts dividend and consumption growth.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The transitory dividend component not tied to income insurance does not explain the slope of dividend risk.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The calibrated model matches cash-flow moments and standard asset pricing moments.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII, IX\"},{\"claim\":\"The model generates downward-sloping dividend strip volatility and premia.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"}],\"structural_summary\":\"The introduction frames a puzzle: leading models imply long-run priced risk, while the data show downward-sloping equity term structures. It poses questions, previews four empirical findings, sketches the income-insurance model, contrasts the paper with competing mechanisms, admits a limitation on interest rates, and gives a roadmap. Section I presents the regression evidence on variance ratios and dividend strips. Section II builds the general equilibrium model. Section III calibrates it and compares asset pricing results with the data. Section IV extends the model to time-varying uncertainty. The conclusion restates the contribution and points to possible extensions.\",\"writing_profile_keywords\":[\"term structure of equity\",\"variance ratios\",\"income insurance\",\"labor-share\",\"dividend strips\",\"general equilibrium calibration\",\"recursive preferences\"],\"confidence\":0.8,\"notes\":\"The empirical part precedes the model. Figure 1 shows confidence bands, but the model figures appear not to show uncertainty. Robustness is located within Section I.D. The magnitudes stated in the introduction are correlations of 86%, 57% and 52%.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "calibration_simulation", "identification": {"estimand": "Predictive association between the labor-share and term-structure moments: variance ratios of wages and dividends, and the risk, premium and slope of dividend strips.", "strategy": "Time-series predictive regressions of variance ratios and dividend strip returns on the labor-share, with controls and Newey-West t-statistics, combined with a calibrated general equilibrium model whose predictions are compared with the data.", "key_assumptions": ["Income insurance concerns transitory shocks only.", "Workers and shareholders share permanent shocks, with cointegrated levels.", "Limited market participation: workers consume their wages."], "threats_addressed": ["Financial leverage as an alternative channel for downward-sloping dividend risk (EBIT term structure shown to be similar).", "Other sources of transitory dividend risk not attributable to income insurance.", "Belief-formation channel as an alternative explanation."], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": true, "sentence_index": 4}, "method": {"present": true, "sentence_index": 4}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "other"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Notes state the regression equations, report coefficients with Newey-West (and Hansen-Hodrick) t-statistics in parentheses and adjusted R2; calibration tables set data moments against model moments.", "units_stated_in_notes": false, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'supports the idea'", "'consistent with'", "'can have only a limited effect'", "explicit acknowledgment of model limitations"]}, "claim_evidence": [{"claim": "Variance-ratio term structures of output, wages and dividends are flat, increasing and decreasing respectively.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Labor-share predicts the wage-dividend variance-ratio gap and the dividend variance ratios.", "evidence_type": "table", "evidence_ref": "Table I; Figure 2"}, {"claim": "Labor-share forecasts dividend strip volatility, premium and slope, with high correlations.", "evidence_type": "table", "evidence_ref": "Tables II-III; Figure 3"}, {"claim": "Labor-share positively forecasts dividend and consumption growth.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The transitory dividend component not tied to income insurance does not explain the slope of dividend risk.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "The calibrated model matches cash-flow moments and standard asset pricing moments.", "evidence_type": "table", "evidence_ref": "Tables VI, VII, IX"}, {"claim": "The model generates downward-sloping dividend strip volatility and premia.", "evidence_type": "figure", "evidence_ref": "Figure 6"}], "structural_summary": "The introduction frames a puzzle: leading models imply long-run priced risk, while the data show downward-sloping equity term structures. It poses questions, previews four empirical findings, sketches the income-insurance model, contrasts the paper with competing mechanisms, admits a limitation on interest rates, and gives a roadmap. Section I presents the regression evidence on variance ratios and dividend strips. Section II builds the general equilibrium model. Section III calibrates it and compares asset pricing results with the data. Section IV extends the model to time-varying uncertainty. The conclusion restates the contribution and points to possible extensions.", "writing_profile_keywords": ["term structure of equity", "variance ratios", "income insurance", "labor-share", "dividend strips", "general equilibrium calibration", "recursive preferences"], "confidence": 0.8, "notes": "The empirical part precedes the model. Figure 1 shows confidence bands, but the model figures appear not to show uncertainty. Robustness is located within Section I.D. The magnitudes stated in the introduction are correlations of 86%, 57% and 52%."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.93, "usage": {"input": 9197, "output": 2236, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082302, "request_id": "req_011Cggphu79gT5xDcCNv15Fo", "cached": false, "ts": "2026-11-03T22:58:57+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c06ea2a7f8eb82e1a09e257e73230929468ddc19a71a51f067676055dc86772b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"corporate_finance\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Effect on investor click-through (interest) of disclosing each information category (team, traction, current investors) about a fund-raising start-up\",\"strategy\":\"Randomized field experiment: among the categories that pass the platform's disclosure threshold, which ones appear in featured emails is randomized; linear regressions of clicks on category indicators with start-up fixed effects, using opened emails only\",\"key_assumptions\":[\"Randomization of displayed categories is independent of investor and start-up characteristics\",\"Investors read a missing category as signaling low quality\",\"Click-through captures investor interest at the initial screening stage\"],\"threats_addressed\":[\"Omitted unobservable start-up characteristics (start-up fixed effects plus randomization)\",\"Investors not seeing the email (sample restricted to opened emails)\",\"External validity of the 21 experiment firms (compared with the broad AngelList sample)\",\"Ordering of information categories\",\"Differences in signal-to-noise ratios across categories\",\"Interpretation of click rates as interest\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"other\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"contribution_statement\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"subsample\",\"alternative_measure\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the binary click outcome and the information-category indicators and state that only opened emails are used. Descriptive tables cover emails, start-ups and investors, and a final table compares experiment firms with the broad platform sample. No coefficient values or standard-error details are visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggestive evidence\",\"indirect evidence\",\"explicit statement of limitations (long-run outcomes not observed)\",\"results consistent with an existing model\",\"disclaimer that nonhuman assets may still matter\"]},\"claim_evidence\":[{\"claim\":\"Investors' click rates respond strongly to founding-team information but not significantly to traction or current-investor information\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Investors specialized in the start-up's sector react as strongly to team information, indicating that operational ability matters beyond signaling\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"More experienced and successful investors react only to team information, indirectly supporting team-based selection as a viable strategy\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, IX\"},{\"claim\":\"Experiment start-ups are compared with the broad AngelList population to assess external validity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Successful early-stage investing reflects skill, as shown by persistence of VC returns\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Kaplan and Schoar (2005) et al.\"}],\"structural_summary\":\"The introduction motivates the question of how early-stage investors select start-ups and notes scarce evidence. It frames three competing hypotheses drawn from theories of the firm and herding, explains the omitted-variable and selected-sample problems, and introduces a randomized email experiment on AngelList. The body describes the platform, the experimental design and summary statistics, then reports main click-rate regressions. Two follow-up sections test mechanisms (signal versus operational capability) and the rationality of team-based selection. A discussion section addresses robustness and external validity, and a short conclusion restates the findings and links them to Rajan (2012).\",\"writing_profile_keywords\":[\"randomized field experiment\",\"competing hypotheses framing\",\"online platform data\",\"mechanism exploration\",\"explicit limitations\",\"first experimental evidence claim\",\"entrepreneurial finance\"],\"confidence\":0.85,\"notes\":\"The section role labeled 'conclusion' for Summary Statistics has been corrected to 'data'. The Discussion section serves as the robustness section. Standard-error treatment and significance notation are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "corporate_finance", "primary_method": "survey_or_experiment", "identification": {"estimand": "Effect on investor click-through (interest) of disclosing each information category (team, traction, current investors) about a fund-raising start-up", "strategy": "Randomized field experiment: among the categories that pass the platform's disclosure threshold, which ones appear in featured emails is randomized; linear regressions of clicks on category indicators with start-up fixed effects, using opened emails only", "key_assumptions": ["Randomization of displayed categories is independent of investor and start-up characteristics", "Investors read a missing category as signaling low quality", "Click-through captures investor interest at the initial screening stage"], "threats_addressed": ["Omitted unobservable start-up characteristics (start-up fixed effects plus randomization)", "Investors not seeing the email (sample restricted to opened emails)", "External validity of the 21 experiment firms (compared with the broad AngelList sample)", "Ordering of information categories", "Differences in signal-to-noise ratios across categories", "Interpretation of click rates as interest"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "other"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "contribution_statement"}, {"paragraph_index": 11, "move": "related_literature_positioning"}, {"paragraph_index": 12, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["subsample", "alternative_measure", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes define the binary click outcome and the information-category indicators and state that only opened emails are used. Descriptive tables cover emails, start-ups and investors, and a final table compares experiment firms with the broad platform sample. No coefficient values or standard-error details are visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggestive evidence", "indirect evidence", "explicit statement of limitations (long-run outcomes not observed)", "results consistent with an existing model", "disclaimer that nonhuman assets may still matter"]}, "claim_evidence": [{"claim": "Investors' click rates respond strongly to founding-team information but not significantly to traction or current-investor information", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Investors specialized in the start-up's sector react as strongly to team information, indicating that operational ability matters beyond signaling", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "More experienced and successful investors react only to team information, indirectly supporting team-based selection as a viable strategy", "evidence_type": "table", "evidence_ref": "Tables VII, IX"}, {"claim": "Experiment start-ups are compared with the broad AngelList population to assess external validity", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Successful early-stage investing reflects skill, as shown by persistence of VC returns", "evidence_type": "prior_literature", "evidence_ref": "Kaplan and Schoar (2005) et al."}], "structural_summary": "The introduction motivates the question of how early-stage investors select start-ups and notes scarce evidence. It frames three competing hypotheses drawn from theories of the firm and herding, explains the omitted-variable and selected-sample problems, and introduces a randomized email experiment on AngelList. The body describes the platform, the experimental design and summary statistics, then reports main click-rate regressions. Two follow-up sections test mechanisms (signal versus operational capability) and the rationality of team-based selection. A discussion section addresses robustness and external validity, and a short conclusion restates the findings and links them to Rajan (2012).", "writing_profile_keywords": ["randomized field experiment", "competing hypotheses framing", "online platform data", "mechanism exploration", "explicit limitations", "first experimental evidence claim", "entrepreneurial finance"], "confidence": 0.85, "notes": "The section role labeled 'conclusion' for Summary Statistics has been corrected to 'data'. The Discussion section serves as the robustness section. Standard-error treatment and significance notation are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.38, "usage": {"input": 7796, "output": 2245, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076878, "request_id": "req_011CggpW3Hwckb3yVmdXQ6tj", "cached": false, "ts": "2026-11-03T22:56:16+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c3c47e8ed7f92d8e4c908e7c6374f831a6098b8a382a9ff882eafd6050b4b094.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Effect of nonincentivized rank feedback and of payout-relevant tournament incentives on the share invested in a risky asset, conditional on prior rank, compared with a linear-incentive baseline.\",\"strategy\":\"Randomized between-subject treatments (baseline, ranking, tournament) run as lab-in-the-field experiments with professionals, lab experiments with students, and further treatments varying frame, payoff recipient, and identity prime. Analysis uses random-effects panel regressions with AR(1) disturbances and probit models.\",\"key_assumptions\":[\"Random assignment of participants to treatments\",\"Experimental investment choices reflect real-world preferences\",\"Rank feedback is anonymous and does not affect payoffs in the ranking treatment\"],\"threats_addressed\":[\"Professional identity importing (identity priming treatments)\",\"Investment framing (abstract lottery frame)\",\"Payoff consequences (family return treatment)\",\"Gender composition\",\"Subject-pool differences (student comparison)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"robustness\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Each table reports random-effects panel regressions with AR(1) disturbances, or probit estimates, of risky-asset share on treatment and rank variables. Notes define the dependent variable and the lagged return controls.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"results suggest\",\"could\",\"may indicate\",\"consistent with\",\"less clear-cut\"]},\"claim_evidence\":[{\"claim\":\"Underperforming professionals take more risk when a nonincentivized ranking is shown, relative to baseline\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Figure 1\"},{\"claim\":\"Students respond only to payout-relevant tournament rankings, not to pure rank feedback\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 2\"},{\"claim\":\"Rank-driven risk-taking differs between professionals and students\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"The rank effect is robust to an abstract lottery frame and to family-return payoffs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 4\"},{\"claim\":\"The rank effect persists under private-identity priming, with no gender differences\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figure 5\"},{\"claim\":\"Professionals report stronger preferences for relative performance than students\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"}],\"structural_summary\":\"The introduction motivates the paper with tournament incentives and the financial crisis, then splits incentives into monetary and nonmonetary rank components. It previews the professional experiment, a student comparison, and identity-transfer tests, followed by three literature contributions and a list of policy implications. The body has one section per experiment (professionals, students), then a section of additional experiments on identity and framing. The conclusion restates findings, discusses possible sources of the effect (self-selection, industry culture), gives implications for regulators and employers, and calls for future research.\",\"writing_profile_keywords\":[\"lab-in-the-field experiment\",\"financial professionals\",\"rank incentives\",\"subject-pool comparison\",\"social identity\",\"policy implications\"],\"confidence\":0.8,\"notes\":\"The abstract text is truncated and includes the first line of the introduction. Introduction paragraph 4 is a fragment split into single-word items. The significance reporting convention is not visible in the table notes shown.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "survey_or_experiment", "identification": {"estimand": "Effect of nonincentivized rank feedback and of payout-relevant tournament incentives on the share invested in a risky asset, conditional on prior rank, compared with a linear-incentive baseline.", "strategy": "Randomized between-subject treatments (baseline, ranking, tournament) run as lab-in-the-field experiments with professionals, lab experiments with students, and further treatments varying frame, payoff recipient, and identity prime. Analysis uses random-effects panel regressions with AR(1) disturbances and probit models.", "key_assumptions": ["Random assignment of participants to treatments", "Experimental investment choices reflect real-world preferences", "Rank feedback is anonymous and does not affect payoffs in the ranking treatment"], "threats_addressed": ["Professional identity importing (identity priming treatments)", "Investment framing (abstract lottery frame)", "Payoff consequences (family return treatment)", "Gender composition", "Subject-pool differences (student comparison)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": true, "sentence_index": 0}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "robustness"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Each table reports random-effects panel regressions with AR(1) disturbances, or probit estimates, of risky-asset share on treatment and rank variables. Notes define the dependent variable and the lagged return controls.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["results suggest", "could", "may indicate", "consistent with", "less clear-cut"]}, "claim_evidence": [{"claim": "Underperforming professionals take more risk when a nonincentivized ranking is shown, relative to baseline", "evidence_type": "table", "evidence_ref": "Table I; Figure 1"}, {"claim": "Students respond only to payout-relevant tournament rankings, not to pure rank feedback", "evidence_type": "table", "evidence_ref": "Table II; Figure 2"}, {"claim": "Rank-driven risk-taking differs between professionals and students", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "The rank effect is robust to an abstract lottery frame and to family-return payoffs", "evidence_type": "table", "evidence_ref": "Table IV; Figure 4"}, {"claim": "The rank effect persists under private-identity priming, with no gender differences", "evidence_type": "table", "evidence_ref": "Table V; Figure 5"}, {"claim": "Professionals report stronger preferences for relative performance than students", "evidence_type": "figure", "evidence_ref": "Figure 3"}], "structural_summary": "The introduction motivates the paper with tournament incentives and the financial crisis, then splits incentives into monetary and nonmonetary rank components. It previews the professional experiment, a student comparison, and identity-transfer tests, followed by three literature contributions and a list of policy implications. The body has one section per experiment (professionals, students), then a section of additional experiments on identity and framing. The conclusion restates findings, discusses possible sources of the effect (self-selection, industry culture), gives implications for regulators and employers, and calls for future research.", "writing_profile_keywords": ["lab-in-the-field experiment", "financial professionals", "rank incentives", "subject-pool comparison", "social identity", "policy implications"], "confidence": 0.8, "notes": "The abstract text is truncated and includes the first line of the introduction. Introduction paragraph 4 is a fragment split into single-word items. The significance reporting convention is not visible in the table notes shown."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.38, "usage": {"input": 7582, "output": 1993, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070982, "request_id": "req_011Cggq5TJEKhimziquDM3Kf", "cached": false, "ts": "2026-11-03T23:03:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c583cea899d2dcd81ac3456c675850c66b02633461c2d7b95aa287898b8588c6.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"experimental finance / collateral and leverage\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Price spread between two assets with identical payoffs that differ only in whether they can be used as collateral, compared against model-implied equilibrium values\",\"strategy\":\"Controlled laboratory double-auction experiment implementing a collateral general equilibrium model. The two assets have identical payoffs, so the price spread isolates collateral value. Observed outcomes are compared to equilibrium benchmarks calibrated to elicited risk aversion.\",\"key_assumptions\":[\"Laboratory implementation faithfully represents the model economy\",\"Subjects have CRRA preferences, which allows the Holt-Laury elicitation to map choices to risk-aversion parameters\",\"Induced payoffs control subjects' valuations\"],\"threats_addressed\":[\"Heterogeneity in risk aversion, handled through Holt-Laury elicitation and an equilibrium computed for different beta values\",\"Learning and convergence, addressed by comparing practice rounds, early paid rounds and late paid rounds\",\"Session-level differences, addressed with by-session tables and Wilcoxon tests\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":4},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"p_values\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Notes report descriptive statistics (means, medians, standard deviations in parentheses), both pooled and by session. Nonparametric Wilcoxon signed-rank and rank-sum tests are reported with p-values. Theory tables give equilibrium values across a range of risk-aversion parameters. Prices are in experimental currency units, and the notes describe the variables but do not consistently state units.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'seems to arise' when stating the mechanism\",\"'quantitatively close to' theoretical predictions\",\"'consistent with' the theory\"]},\"claim_evidence\":[{\"claim\":\"The collateralizeable asset trades at a higher price than the identical noncollateralizeable asset, a significant deviation from the Law of One Price\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII; Figure 2\"},{\"claim\":\"Observed prices and spread are not statistically different from the model's predictions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Most buyers borrow the maximum and are collateral constrained, consistent with the theoretical mechanism\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X, XI, XIII; Figure 3\"},{\"claim\":\"Final holdings match the theory: high-valuation agents hold the collateralized asset but share the other, so the allocation is not Pareto efficient\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XV, XVI\"},{\"claim\":\"Prices, spread and constrained shares converge toward the collateral equilibrium across rounds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XVII, XVIII; Figures 5-8\"},{\"claim\":\"Attempts to arbitrage the price difference decline as the experiment progresses\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XIX\"},{\"claim\":\"In the model, collateral value raises the price of the leverageable asset, and the spread depends on risk aversion\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Tables II, III\"}],\"structural_summary\":\"The paper motivates the question with the crisis and the theoretical literature on collateral value. It then builds a lab-implementable general equilibrium model with two identical assets, only one of which can serve as collateral. Next it describes the experimental design and procedures. Results begin by calibrating the theoretical benchmark to elicited risk aversion, then test prices, borrowing, constraint status, allocations and convergence across rounds, using nonparametric tests against the model's predictions. The conclusion restates the first-of-its-kind contribution and summarizes the findings.\",\"writing_profile_keywords\":[\"laboratory experiment\",\"theory-testing\",\"Law of One Price\",\"collateral value\",\"nonparametric tests\",\"convergence to equilibrium\",\"double auction\"],\"confidence\":0.85,\"notes\":\"Section II (experiment design) is mapped to empirical_design. No explicit contribution paragraph is visible in the intro excerpt; the conclusion claims to be the first lab study. Robustness work appears as by-session breakdowns and risk-aversion sensitivity within the results section.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "experimental finance / collateral and leverage", "primary_method": "survey_or_experiment", "identification": {"estimand": "Price spread between two assets with identical payoffs that differ only in whether they can be used as collateral, compared against model-implied equilibrium values", "strategy": "Controlled laboratory double-auction experiment implementing a collateral general equilibrium model. The two assets have identical payoffs, so the price spread isolates collateral value. Observed outcomes are compared to equilibrium benchmarks calibrated to elicited risk aversion.", "key_assumptions": ["Laboratory implementation faithfully represents the model economy", "Subjects have CRRA preferences, which allows the Holt-Laury elicitation to map choices to risk-aversion parameters", "Induced payoffs control subjects' valuations"], "threats_addressed": ["Heterogeneity in risk aversion, handled through Holt-Laury elicitation and an equilibrium computed for different beta values", "Learning and convergence, addressed by comparing practice rounds, early paid rounds and late paid rounds", "Session-level differences, addressed with by-session tables and Wilcoxon tests"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 4}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "p_values", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["subsample", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Notes report descriptive statistics (means, medians, standard deviations in parentheses), both pooled and by session. Nonparametric Wilcoxon signed-rank and rank-sum tests are reported with p-values. Theory tables give equilibrium values across a range of risk-aversion parameters. Prices are in experimental currency units, and the notes describe the variables but do not consistently state units.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'seems to arise' when stating the mechanism", "'quantitatively close to' theoretical predictions", "'consistent with' the theory"]}, "claim_evidence": [{"claim": "The collateralizeable asset trades at a higher price than the identical noncollateralizeable asset, a significant deviation from the Law of One Price", "evidence_type": "table", "evidence_ref": "Tables VII, VIII; Figure 2"}, {"claim": "Observed prices and spread are not statistically different from the model's predictions", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Most buyers borrow the maximum and are collateral constrained, consistent with the theoretical mechanism", "evidence_type": "table", "evidence_ref": "Tables X, XI, XIII; Figure 3"}, {"claim": "Final holdings match the theory: high-valuation agents hold the collateralized asset but share the other, so the allocation is not Pareto efficient", "evidence_type": "table", "evidence_ref": "Tables XV, XVI"}, {"claim": "Prices, spread and constrained shares converge toward the collateral equilibrium across rounds", "evidence_type": "table", "evidence_ref": "Tables XVII, XVIII; Figures 5-8"}, {"claim": "Attempts to arbitrage the price difference decline as the experiment progresses", "evidence_type": "table", "evidence_ref": "Table XIX"}, {"claim": "In the model, collateral value raises the price of the leverageable asset, and the spread depends on risk aversion", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Tables II, III"}], "structural_summary": "The paper motivates the question with the crisis and the theoretical literature on collateral value. It then builds a lab-implementable general equilibrium model with two identical assets, only one of which can serve as collateral. Next it describes the experimental design and procedures. Results begin by calibrating the theoretical benchmark to elicited risk aversion, then test prices, borrowing, constraint status, allocations and convergence across rounds, using nonparametric tests against the model's predictions. The conclusion restates the first-of-its-kind contribution and summarizes the findings.", "writing_profile_keywords": ["laboratory experiment", "theory-testing", "Law of One Price", "collateral value", "nonparametric tests", "convergence to equilibrium", "double auction"], "confidence": 0.85, "notes": "Section II (experiment design) is mapped to empirical_design. No explicit contribution paragraph is visible in the intro excerpt; the conclusion claims to be the first lab study. Robustness work appears as by-session breakdowns and risk-aversion sensitivity within the results section."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 24.17, "usage": {"input": 8169, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07691, "request_id": "req_011Cggq9H7BDxjvv2Ve7dStn", "cached": false, "ts": "2026-11-03T23:04:47+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c712d6a87b2a4e82767caf965f1f50746e4ceec0b45e066e4fbb40cbe9ee805f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"asset_pricing (ex-dividend pricing, institutional investor skill)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Abnormal profits of institutional dividend capture trades after transaction costs, and the cross-sectional relation between those profits and prior trade execution skill\",\"strategy\":\"Transaction-level institutional data (Abel Noser) used to compute realized ex-day returns net of commissions; portfolio sorts by abnormal buy intensity and by skill deciles; regressions of buy intensity on event and trader characteristics; comparison against placebo non-ex-days and non-dividend trades\",\"key_assumptions\":[\"Prior execution-skill measures (AIPV) capture persistent trader ability rather than stock-picking skill\",\"Placebo non-ex-days are a valid comparison for ex-days\",\"Benchmark-adjusted abnormal returns (DGTW) correctly remove risk-related returns\"],\"threats_addressed\":[\"Skill-profit relation could be generic rather than ex-day specific (placebo non-ex-day test)\",\"Profits could come from stock-picking rather than execution (stock-picking skill and undervaluation checks)\",\"Transaction costs eliminating average ex-day returns (actual execution prices and commissions used)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":true,\"sentence_index\":7}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"other\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"subsample\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes define variables and benchmarks in detail: abnormal volume formulas, VWAP execution prices before and after commissions, DGTW-adjusted holding-period returns, and quintile/decile sort construction. Results are shown as panels of means by sort and as regressions. Standard-error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"results suggest\",\"one potential interpretation is\",\"may be as important\",\"evidence suggests\"]},\"claim_evidence\":[{\"claim\":\"Average ex-day returns are not significantly positive once actual execution prices and commissions are incorporated\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Dividend capture buys earn significantly positive post-commission abnormal returns, higher than other buys\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Dividend capture trades are under 6% of buys but contribute about 15% of abnormal returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Institutional dividend capture buying intensity persists for at least two subsequent quarters\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 1\"},{\"claim\":\"Profitability is higher for ex-day events with high abnormal cum-day buy intensity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"High- vs low-skill institutions differ by about 40 bps in dividend capture profitability, with no relation on placebo days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Buy intensity is related to yield, idiosyncratic risk, transaction costs and lagged ex-day returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"}],\"structural_summary\":\"The paper opens by citing evidence that execution skill earns profits and asks whether skilled institutions benefit uniformly or target specific opportunities. It proposes dividend capture as the setting and states three research questions. Results are previewed with magnitudes, followed by a placebo-based defense and contributions to two literatures. A short theory/literature section precedes the data and descriptive statistics, which show that net ex-day returns vanish after costs. Two results sections follow: profitability and persistence at the client-manager level, then the link to trader skill with cross-sectional characteristics. The conclusion is structured around the three research questions and closes with a future-research direction.\",\"writing_profile_keywords\":[\"proprietary transaction data\",\"question-driven structure\",\"placebo test\",\"portfolio sorts\",\"transaction costs\",\"institutional skill\",\"economic magnitudes\"],\"confidence\":0.82,\"notes\":\"Robustness material appears within the results and in an Internet Appendix. Significance reporting conventions are not visible in the excerpts. A single equation appears in Section IV.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "asset_pricing (ex-dividend pricing, institutional investor skill)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Abnormal profits of institutional dividend capture trades after transaction costs, and the cross-sectional relation between those profits and prior trade execution skill", "strategy": "Transaction-level institutional data (Abel Noser) used to compute realized ex-day returns net of commissions; portfolio sorts by abnormal buy intensity and by skill deciles; regressions of buy intensity on event and trader characteristics; comparison against placebo non-ex-days and non-dividend trades", "key_assumptions": ["Prior execution-skill measures (AIPV) capture persistent trader ability rather than stock-picking skill", "Placebo non-ex-days are a valid comparison for ex-days", "Benchmark-adjusted abnormal returns (DGTW) correctly remove risk-related returns"], "threats_addressed": ["Skill-profit relation could be generic rather than ex-day specific (placebo non-ex-day test)", "Profits could come from stock-picking rather than execution (stock-picking skill and undervaluation checks)", "Transaction costs eliminating average ex-day returns (actual execution prices and commissions used)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": true, "sentence_index": 7}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "other"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "subsample", "alternative_measure"]}, "tables": {"convention_notes": "Table notes define variables and benchmarks in detail: abnormal volume formulas, VWAP execution prices before and after commissions, DGTW-adjusted holding-period returns, and quintile/decile sort construction. Results are shown as panels of means by sort and as regressions. Standard-error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["results suggest", "one potential interpretation is", "may be as important", "evidence suggests"]}, "claim_evidence": [{"claim": "Average ex-day returns are not significantly positive once actual execution prices and commissions are incorporated", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Dividend capture buys earn significantly positive post-commission abnormal returns, higher than other buys", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Dividend capture trades are under 6% of buys but contribute about 15% of abnormal returns", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Institutional dividend capture buying intensity persists for at least two subsequent quarters", "evidence_type": "table", "evidence_ref": "Table VI; Figure 1"}, {"claim": "Profitability is higher for ex-day events with high abnormal cum-day buy intensity", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "High- vs low-skill institutions differ by about 40 bps in dividend capture profitability, with no relation on placebo days", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Buy intensity is related to yield, idiosyncratic risk, transaction costs and lagged ex-day returns", "evidence_type": "table", "evidence_ref": "Table VII"}], "structural_summary": "The paper opens by citing evidence that execution skill earns profits and asks whether skilled institutions benefit uniformly or target specific opportunities. It proposes dividend capture as the setting and states three research questions. Results are previewed with magnitudes, followed by a placebo-based defense and contributions to two literatures. A short theory/literature section precedes the data and descriptive statistics, which show that net ex-day returns vanish after costs. Two results sections follow: profitability and persistence at the client-manager level, then the link to trader skill with cross-sectional characteristics. The conclusion is structured around the three research questions and closes with a future-research direction.", "writing_profile_keywords": ["proprietary transaction data", "question-driven structure", "placebo test", "portfolio sorts", "transaction costs", "institutional skill", "economic magnitudes"], "confidence": 0.82, "notes": "Robustness material appears within the results and in an Internet Appendix. Significance reporting conventions are not visible in the excerpts. A single equation appears in Section IV."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.76, "usage": {"input": 6460, "output": 2139, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069414, "request_id": "req_011CggpUj8WQTmW1Q6F8gkaQ", "cached": false, "ts": "2026-11-03T22:55:57+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c8b9faf575e0b5f39d5591adc1f31b80b881e65ce1670ed5b18a9b8e662ef37d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance (heterogeneous beliefs)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model; comparative statics of equilibrium price, mean return, volatility and volume with respect to belief dispersion)\",\"strategy\":\"not applicable; closed-form general equilibrium analysis with numerical illustrations calibrated to empirical dispersion moments\",\"key_assumptions\":[\"continuum of CRRA investors with heterogeneous dogmatic or Bayesian beliefs\",\"single consumption date\",\"pure-exchange, dynamically complete market, finite horizon\"],\"threats_addressed\":[\"results not driven by learning (dogmatic case vs. Bayesian extension)\",\"generality across multiple stocks\",\"comparison with two-investor setups\"],\"causal_language_strength\":3,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"theory\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"No tables. Results are shown as figures plotting equilibrium quantities against cash-flow news or belief dispersion, each compared with a no-dispersion benchmark line. Parameters are calibrated in Appendix B.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'qualitatively matches' empirical regularities\",\"'do not necessarily generate'\",\"'does not appear to be possible'\"]},\"claim_evidence\":[{\"claim\":\"Stock price is convex in cash-flow news and rises with belief dispersion under optimistic views\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2; Section III.A\"},{\"claim\":\"Mean return falls with dispersion when the view is optimistic and rises when pessimistic\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4; Section III.B\"},{\"claim\":\"Belief dispersion raises stock volatility and trading volume, which are positively related\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6; Section III.C\"},{\"claim\":\"Otherwise identical two-investor economies need not deliver the main results\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7; Section IV\"},{\"claim\":\"Equilibrium quantities admit closed-form expressions in average bias and dispersion\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Appendix A\"},{\"claim\":\"Empirical evidence on dispersion and returns is mixed\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction paragraph 0\"}],\"structural_summary\":\"Opens by framing mixed empirical evidence on belief dispersion as a puzzle that existing theory does not resolve. Then builds a continuum-investor equilibrium model and reduces beliefs to two sufficient statistics, average bias and dispersion. Derives closed-form price, return, volatility and volume implications, illustrated with calibrated figures. Contrasts the model with a two-investor benchmark and extends it to multiple stocks and Bayesian learning. Concludes by restating the findings, with proofs and calibration in appendices.\",\"writing_profile_keywords\":[\"closed-form equilibrium\",\"continuum of investors\",\"heterogeneous beliefs\",\"comparative statics figures\",\"benchmark comparison\",\"extensions as robustness\",\"calibration appendix\"],\"confidence\":0.85,\"notes\":\"The extension sections (IV-VI) work as theoretical robustness and generality checks rather than empirical robustness. The introduction is a single long block in the excerpt, so moves are assigned only to paragraph 0; any roadmap is not visible.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance (heterogeneous beliefs)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model; comparative statics of equilibrium price, mean return, volatility and volume with respect to belief dispersion)", "strategy": "not applicable; closed-form general equilibrium analysis with numerical illustrations calibrated to empirical dispersion moments", "key_assumptions": ["continuum of CRRA investors with heterogeneous dogmatic or Bayesian beliefs", "single consumption date", "pure-exchange, dynamically complete market, finite horizon"], "threats_addressed": ["results not driven by learning (dogmatic case vs. Bayesian extension)", "generality across multiple stocks", "comparison with two-investor setups"], "causal_language_strength": 3, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "contribution_statement"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "theory"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["other"]}, "tables": {"convention_notes": "No tables. Results are shown as figures plotting equilibrium quantities against cash-flow news or belief dispersion, each compared with a no-dispersion benchmark line. Parameters are calibrated in Appendix B.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'qualitatively matches' empirical regularities", "'do not necessarily generate'", "'does not appear to be possible'"]}, "claim_evidence": [{"claim": "Stock price is convex in cash-flow news and rises with belief dispersion under optimistic views", "evidence_type": "figure", "evidence_ref": "Figure 2; Section III.A"}, {"claim": "Mean return falls with dispersion when the view is optimistic and rises when pessimistic", "evidence_type": "figure", "evidence_ref": "Figure 4; Section III.B"}, {"claim": "Belief dispersion raises stock volatility and trading volume, which are positively related", "evidence_type": "figure", "evidence_ref": "Figures 5-6; Section III.C"}, {"claim": "Otherwise identical two-investor economies need not deliver the main results", "evidence_type": "figure", "evidence_ref": "Figure 7; Section IV"}, {"claim": "Equilibrium quantities admit closed-form expressions in average bias and dispersion", "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Appendix A"}, {"claim": "Empirical evidence on dispersion and returns is mixed", "evidence_type": "prior_literature", "evidence_ref": "Introduction paragraph 0"}], "structural_summary": "Opens by framing mixed empirical evidence on belief dispersion as a puzzle that existing theory does not resolve. Then builds a continuum-investor equilibrium model and reduces beliefs to two sufficient statistics, average bias and dispersion. Derives closed-form price, return, volatility and volume implications, illustrated with calibrated figures. Contrasts the model with a two-investor benchmark and extends it to multiple stocks and Bayesian learning. Concludes by restating the findings, with proofs and calibration in appendices.", "writing_profile_keywords": ["closed-form equilibrium", "continuum of investors", "heterogeneous beliefs", "comparative statics figures", "benchmark comparison", "extensions as robustness", "calibration appendix"], "confidence": 0.85, "notes": "The extension sections (IV-VI) work as theoretical robustness and generality checks rather than empirical robustness. The introduction is a single long block in the excerpt, so moves are assigned only to paragraph 0; any roadmap is not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.69, "usage": {"input": 5855, "output": 1877, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061754, "request_id": "req_011CggpvurA5PMW2cneVnGTZ", "cached": false, "ts": "2026-11-03T23:01:49+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c8da58e67a3b5da6893b6169200fa3158c7dd7d578d4b05c5146947dc10f9a4e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Degree of expectation stickiness (lambda) in analyst forecasts, and how strongly profitability-anomaly returns vary with stickiness and with profit persistence\",\"strategy\":\"Model-derived tests. Forecast errors are regressed on forecast revisions (Coibion-Gorodnichenko style) to estimate stickiness. Forecast errors are regressed on profitability signals. Portfolios are double-sorted on firm-level stickiness or persistence and on the anomaly signal.\",\"key_assumptions\":[\"Analyst forecasts represent market participants' beliefs\",\"Profits are predictable from persistent public signals\",\"Sticky-expectation updating nests rational expectations\"],\"threats_addressed\":[\"look-ahead bias in firm-level stickiness estimation\",\"mechanical correlation under rational-expectations null (simulations)\",\"past trend in profits as alternative explanation\",\"risk adjustment via factor models\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_theory_or_model\",\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"placebo\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define variables using model notation and describe the regression or sort procedure. Portfolio tables report excess returns and CAPM, three-factor and four-factor alphas, plus double-sorts. Units are partially stated in the notes (returns/alphas are given in some unit, but the excerpt is truncated).\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"conditional assumptions (assuming analyst forecasts are representative)\",\"'in line with' interpretations\",\"'suggests'\"]},\"claim_evidence\":[{\"claim\":\"The profitability anomaly carries large economic significance: high Sharpe ratio, no crash risk, high capacity\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Novy-Marx (2013, 2015); Lemperiere et al. (2017); Landier, Simon, Thesmar (2015)\"},{\"claim\":\"Analyst expectations are sticky: forecast errors are positively related to past forecast revisions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Figure 1\"},{\"claim\":\"Analysts are too pessimistic about future profits of high-profit firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 2\"},{\"claim\":\"The profitability anomaly is stronger for stocks followed by stickier analysts\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The profitability anomaly is stronger for firms with more persistent profits\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Double-sort results are not mechanical under the rational-expectations null\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Less experienced and busier analysts have stickier beliefs, consistent with limited attention\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"In the sticky-belief model, past profits forecast future returns\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendix proofs\"}],\"structural_summary\":\"The paper opens with the risk-versus-behavioral debate over return anomalies and narrows to the profitability anomaly. It builds a simple sticky-expectations pricing model borrowed from the macro information-rigidity literature, which yields testable propositions. The empirical part then follows the propositions in order. First, it measures stickiness from I/B/E/S analyst forecasts in pooled, analyst-level and firm-level estimates. Second, it shows that past profits predict forecast errors. Third, it double-sorts anomaly returns on stickiness and on profit persistence. A separate robustness section addresses look-ahead bias and uses simulations under the rational-expectations null. A brief conclusion restates the model, its predictions, and the limited-attention determinants of stickiness.\",\"writing_profile_keywords\":[\"proposition-driven testing\",\"behavioral asset pricing\",\"analyst forecasts\",\"portfolio sorts\",\"model-to-data mapping\",\"sticky expectations\"],\"confidence\":0.8,\"notes\":\"Section III is labeled 'theory' by the role rule but contains the empirical tests, so it is annotated as results. Significance-reporting conventions are not visible in the excerpts. The introduction excerpt is truncated and no roadmap is visible.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Degree of expectation stickiness (lambda) in analyst forecasts, and how strongly profitability-anomaly returns vary with stickiness and with profit persistence", "strategy": "Model-derived tests. Forecast errors are regressed on forecast revisions (Coibion-Gorodnichenko style) to estimate stickiness. Forecast errors are regressed on profitability signals. Portfolios are double-sorted on firm-level stickiness or persistence and on the anomaly signal.", "key_assumptions": ["Analyst forecasts represent market participants' beliefs", "Profits are predictable from persistent public signals", "Sticky-expectation updating nests rational expectations"], "threats_addressed": ["look-ahead bias in firm-level stickiness estimation", "mechanical correlation under rational-expectations null (simulations)", "past trend in profits as alternative explanation", "risk adjustment via factor models"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "contribution_statement"}], "contribution_paragraph_index": 2, "contribution_types": ["new_theory_or_model", "new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["placebo", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes define variables using model notation and describe the regression or sort procedure. Portfolio tables report excess returns and CAPM, three-factor and four-factor alphas, plus double-sorts. Units are partially stated in the notes (returns/alphas are given in some unit, but the excerpt is truncated).", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["conditional assumptions (assuming analyst forecasts are representative)", "'in line with' interpretations", "'suggests'"]}, "claim_evidence": [{"claim": "The profitability anomaly carries large economic significance: high Sharpe ratio, no crash risk, high capacity", "evidence_type": "prior_literature", "evidence_ref": "Novy-Marx (2013, 2015); Lemperiere et al. (2017); Landier, Simon, Thesmar (2015)"}, {"claim": "Analyst expectations are sticky: forecast errors are positively related to past forecast revisions", "evidence_type": "table", "evidence_ref": "Table III, Figure 1"}, {"claim": "Analysts are too pessimistic about future profits of high-profit firms", "evidence_type": "table", "evidence_ref": "Table VI, Figure 2"}, {"claim": "The profitability anomaly is stronger for stocks followed by stickier analysts", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The profitability anomaly is stronger for firms with more persistent profits", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Double-sort results are not mechanical under the rational-expectations null", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Less experienced and busier analysts have stickier beliefs, consistent with limited attention", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "In the sticky-belief model, past profits forecast future returns", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendix proofs"}], "structural_summary": "The paper opens with the risk-versus-behavioral debate over return anomalies and narrows to the profitability anomaly. It builds a simple sticky-expectations pricing model borrowed from the macro information-rigidity literature, which yields testable propositions. The empirical part then follows the propositions in order. First, it measures stickiness from I/B/E/S analyst forecasts in pooled, analyst-level and firm-level estimates. Second, it shows that past profits predict forecast errors. Third, it double-sorts anomaly returns on stickiness and on profit persistence. A separate robustness section addresses look-ahead bias and uses simulations under the rational-expectations null. A brief conclusion restates the model, its predictions, and the limited-attention determinants of stickiness.", "writing_profile_keywords": ["proposition-driven testing", "behavioral asset pricing", "analyst forecasts", "portfolio sorts", "model-to-data mapping", "sticky expectations"], "confidence": 0.8, "notes": "Section III is labeled 'theory' by the role rule but contains the empirical tests, so it is annotated as results. Significance-reporting conventions are not visible in the excerpts. The introduction excerpt is truncated and no roadmap is visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.07, "usage": {"input": 5901, "output": 2185, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068098, "request_id": "req_011CggqGABeYE9wTYTzhNLup", "cached": false, "ts": "2026-11-03T23:06:13+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/cadc21897c580856f56100a89ea1be9199dbc415f8c7cba43aba876407109318.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"shareholder activism / M&A risk arbitrage / corporate governance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between activist arbitrageur involvement and deal outcomes (completion probability, premium revision, duration) and arbitrage returns.\",\"strategy\":\"Cross-sectional regressions comparing activist-involved deals with passive-arbitrage and no-arbitrage control deals. Supplemented by determinants regressions for selection, manual classification of the contributors to deal failure, and CAR/calendar-time portfolio return analysis.\",\"key_assumptions\":[\"Observable deal characteristics capture activists' selection of targets\",\"Passive-arbitrage and no-arbitrage deals form valid comparison groups\"],\"threats_addressed\":[\"Selection of deals by activists (selection effect section)\",\"Attributing deal failure to activists versus other causes (direct manual classification)\",\"Risk adjustment of returns (four-factor model, calendar-time portfolios, deal-characteristic adjustment)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes state the sample size and period (2000–2015 deals) and refer readers to the Appendix for variable definitions. They also describe how control samples are constructed and which methodological sources are followed, such as Hsieh–Walkling for identifying passive arbitrageurs and Cao et al. for calendar-time portfolios. Standard-error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'are associated with' phrasing\",\"'appear to play a role'\",\"'conditional on the deals they select'\",\"'suggest'\"]},\"claim_evidence\":[{\"claim\":\"Activists target deals with low premiums and deals exposed to managerial conflicts of interest.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III, IV\"},{\"claim\":\"Activist presence is associated with lower completion probability and higher premium revisions (3.4 pp across all deals, 4.3 pp among completed deals).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"A negative ISS recommendation is associated with a 21 pp lower probability of deal completion.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Introduction paragraph 1\"},{\"claim\":\"Activist arbitrage earns four-factor CARs of 4.9–5.4% per deal, versus 2.2% for passive arbitrage.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The calendar-time monthly alpha is 1.60% for activist arbitrage versus 0.51% for passive arbitrage.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Deal failures are classified manually by their underlying contributors to assess activists' direct impact.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The introduction defines the phenomenon and documents its rise, then previews results on ISS's intermediary role, deal outcomes and returns, before positioning the paper within the takeover-governance literature. The data section builds a hand-collected sample of activist events alongside SDC M&A deals and identifies passive arbitrageurs. Three results sections follow: target selection, deal resolution (with a selection-effect discussion and a direct classification of deal failures), and returns (long-only, long-short, calendar-time, and characteristic-adjusted). A brief conclusion restates the findings in hedged terms.\",\"writing_profile_keywords\":[\"phenomenon-defining\",\"first-study claim\",\"hand-collected sample\",\"associational language\",\"returns analysis\",\"governance framing\",\"control-sample comparisons\"],\"confidence\":0.8,\"notes\":\"The section labeled 'Appendix' contains only a pointer to the Internet Appendix. Robustness checks appear within the results sections, and further material is likely in the Internet Appendix. One result is significant at 5%, but the significance notation style is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "shareholder activism / M&A risk arbitrage / corporate governance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between activist arbitrageur involvement and deal outcomes (completion probability, premium revision, duration) and arbitrage returns.", "strategy": "Cross-sectional regressions comparing activist-involved deals with passive-arbitrage and no-arbitrage control deals. Supplemented by determinants regressions for selection, manual classification of the contributors to deal failure, and CAR/calendar-time portfolio return analysis.", "key_assumptions": ["Observable deal characteristics capture activists' selection of targets", "Passive-arbitrage and no-arbitrage deals form valid comparison groups"], "threats_addressed": ["Selection of deals by activists (selection effect section)", "Attributing deal failure to activists versus other causes (direct manual classification)", "Risk adjustment of returns (four-factor model, calendar-time portfolios, deal-characteristic adjustment)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification"]}, "tables": {"convention_notes": "Table notes state the sample size and period (2000–2015 deals) and refer readers to the Appendix for variable definitions. They also describe how control samples are constructed and which methodological sources are followed, such as Hsieh–Walkling for identifying passive arbitrageurs and Cao et al. for calendar-time portfolios. Standard-error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'are associated with' phrasing", "'appear to play a role'", "'conditional on the deals they select'", "'suggest'"]}, "claim_evidence": [{"claim": "Activists target deals with low premiums and deals exposed to managerial conflicts of interest.", "evidence_type": "table", "evidence_ref": "Tables II, III, IV"}, {"claim": "Activist presence is associated with lower completion probability and higher premium revisions (3.4 pp across all deals, 4.3 pp among completed deals).", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "A negative ISS recommendation is associated with a 21 pp lower probability of deal completion.", "evidence_type": "none_stated", "evidence_ref": "Introduction paragraph 1"}, {"claim": "Activist arbitrage earns four-factor CARs of 4.9–5.4% per deal, versus 2.2% for passive arbitrage.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The calendar-time monthly alpha is 1.60% for activist arbitrage versus 0.51% for passive arbitrage.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Deal failures are classified manually by their underlying contributors to assess activists' direct impact.", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The introduction defines the phenomenon and documents its rise, then previews results on ISS's intermediary role, deal outcomes and returns, before positioning the paper within the takeover-governance literature. The data section builds a hand-collected sample of activist events alongside SDC M&A deals and identifies passive arbitrageurs. Three results sections follow: target selection, deal resolution (with a selection-effect discussion and a direct classification of deal failures), and returns (long-only, long-short, calendar-time, and characteristic-adjusted). A brief conclusion restates the findings in hedged terms.", "writing_profile_keywords": ["phenomenon-defining", "first-study claim", "hand-collected sample", "associational language", "returns analysis", "governance framing", "control-sample comparisons"], "confidence": 0.8, "notes": "The section labeled 'Appendix' contains only a pointer to the Internet Appendix. Robustness checks appear within the results sections, and further material is likely in the Internet Appendix. One result is significant at 5%, but the significance notation style is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.67, "usage": {"input": 6688, "output": 2040, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068346, "request_id": "req_011Cggq83S8BAPiaB23GCfXq", "cached": false, "ts": "2026-11-03T23:04:22+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/cbc338de4c33690ff77323b03f16135d822e0098cfc993e635dc010d794e9cf3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance (governance), entrepreneurial_finance_private_equity, political economy\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between the share of each trustee category on public pension fund boards and the net IRR / multiple of PE investments made by the fund\",\"strategy\":\"Cross-sectional variation in board composition, largely set by statute decades earlier, regressed on LP-investment performance with controls and fixed effects (vintage, state). Robustness uses no-change and old-board subsamples and pension fund fixed effects that exploit changes in composition. Channel tests rely on fund types, in-state overweighting, investment selection, political contributions and trustee experience.\",\"key_assumptions\":[\"Board composition, fixed by statute long ago, is not driven by the fund's past investment styles or outcomes\",\"Within-state differences in board charters are unrelated to unobserved determinants of PE performance\",\"Observed controls capture differences in consultants, risk-taking and PE experience\"],\"threats_addressed\":[\"reverse causality / endogenous board structure\",\"state-level confounders (state fixed effects)\",\"changes in board structure during the sample (no-change and old-board subsamples)\",\"investment consultant choice, risk-taking, PE experience\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_specification\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes give sample period, observation level (LP-investment or LP-vintage), the performance measure used (net IRR vs. multiple), and sample restrictions. Alternative-measure versions are often moved to the Internet Appendix. Standard-error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggestive evidence/support'\",\"'appears to be partly driven'\",\"'consistent with'\",\"explicit caveat that channels are hard to disentangle\",\"'may' in hypothesis statements\"]},\"claim_evidence\":[{\"claim\":\"Each 10pp increase in state-appointed (ex officio) trustees lowers PE net IRR by about 0.9 (0.5) percentage points relative to participant-appointed trustees.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Underperformance is robust to excluding funds with board changes, to state fixed effects, and to pension fund fixed effects.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V\"},{\"claim\":\"Fund-type controls explain 20–30% of the gap, mainly through state-heavy boards' tilt toward real estate; the rest arises within VC and RE.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Political boards overweight in-state investments and select small funds with few co-investors and inexperienced GPs.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII, IX\"},{\"claim\":\"Underperformance is positively related to finance-industry political contributions to board members.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Lower financial experience explains the underperformance of participant-elected trustees but not that of state officials.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"}],\"structural_summary\":\"The paper opens with a question about how politicians shape governance and frames public pension boards as an unexplored channel of political influence. The introduction previews the baseline magnitudes, maps Shleifer's Control, Corruption and Confusion channels onto the setting, and defends identification by pointing to statutory board composition and fixed-effects tests. It then positions the work against four literatures. After an institutional overview and a data section, four results sections follow: baseline performance with robustness, the Control channel via investment selection, the Corruption channel via political contributions, and the Confusion channel via trustee experience. The conclusion restates the findings and frames them as suggestive evidence.\",\"writing_profile_keywords\":[\"political economy of finance\",\"pension fund governance\",\"private equity performance\",\"channel decomposition\",\"Shleifer three-channel framework\",\"suggestive causal framing\",\"results-first abstract\"],\"confidence\":0.8,\"notes\":\"The abstract consists only of results statements, with no explicit method or question sentence. The robustness table sits inside the results section, and some alternative-measure results are placed in the Internet Appendix. Standard-error clustering is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance (governance), entrepreneurial_finance_private_equity, political economy", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between the share of each trustee category on public pension fund boards and the net IRR / multiple of PE investments made by the fund", "strategy": "Cross-sectional variation in board composition, largely set by statute decades earlier, regressed on LP-investment performance with controls and fixed effects (vintage, state). Robustness uses no-change and old-board subsamples and pension fund fixed effects that exploit changes in composition. Channel tests rely on fund types, in-state overweighting, investment selection, political contributions and trustee experience.", "key_assumptions": ["Board composition, fixed by statute long ago, is not driven by the fund's past investment styles or outcomes", "Within-state differences in board charters are unrelated to unobserved determinants of PE performance", "Observed controls capture differences in consultants, risk-taking and PE experience"], "threats_addressed": ["reverse causality / endogenous board structure", "state-level confounders (state fixed effects)", "changes in board structure during the sample (no-change and old-board subsamples)", "investment consultant choice, risk-taking, PE experience"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 0}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "robustness_mention"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_specification", "alternative_measure"]}, "tables": {"convention_notes": "Table notes give sample period, observation level (LP-investment or LP-vintage), the performance measure used (net IRR vs. multiple), and sample restrictions. Alternative-measure versions are often moved to the Internet Appendix. Standard-error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggestive evidence/support'", "'appears to be partly driven'", "'consistent with'", "explicit caveat that channels are hard to disentangle", "'may' in hypothesis statements"]}, "claim_evidence": [{"claim": "Each 10pp increase in state-appointed (ex officio) trustees lowers PE net IRR by about 0.9 (0.5) percentage points relative to participant-appointed trustees.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Underperformance is robust to excluding funds with board changes, to state fixed effects, and to pension fund fixed effects.", "evidence_type": "table", "evidence_ref": "Tables IV, V"}, {"claim": "Fund-type controls explain 20–30% of the gap, mainly through state-heavy boards' tilt toward real estate; the rest arises within VC and RE.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Political boards overweight in-state investments and select small funds with few co-investors and inexperienced GPs.", "evidence_type": "table", "evidence_ref": "Tables VII, VIII, IX"}, {"claim": "Underperformance is positively related to finance-industry political contributions to board members.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Lower financial experience explains the underperformance of participant-elected trustees but not that of state officials.", "evidence_type": "table", "evidence_ref": "Table XI"}], "structural_summary": "The paper opens with a question about how politicians shape governance and frames public pension boards as an unexplored channel of political influence. The introduction previews the baseline magnitudes, maps Shleifer's Control, Corruption and Confusion channels onto the setting, and defends identification by pointing to statutory board composition and fixed-effects tests. It then positions the work against four literatures. After an institutional overview and a data section, four results sections follow: baseline performance with robustness, the Control channel via investment selection, the Corruption channel via political contributions, and the Confusion channel via trustee experience. The conclusion restates the findings and frames them as suggestive evidence.", "writing_profile_keywords": ["political economy of finance", "pension fund governance", "private equity performance", "channel decomposition", "Shleifer three-channel framework", "suggestive causal framing", "results-first abstract"], "confidence": 0.8, "notes": "The abstract consists only of results statements, with no explicit method or question sentence. The robustness table sits inside the results section, and some alternative-measure results are placed in the Internet Appendix. Standard-error clustering is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 27.36, "usage": {"input": 7788, "output": 2293, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077806, "request_id": "req_011Cggq3Rx51rEycaWHgmR1J", "cached": false, "ts": "2026-11-03T23:03:30+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/cbddda01fd301f3269b4d754fd6444f236420becbe9ca7313a26033256f265fa.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Intent-to-treat effects of randomized SMS message content (overdraft discount, availability, bundled discounts, frequency, duration) on overdraft, debit, and autopay usage during and after the campaign\",\"strategy\":\"Randomized field experiment with a Turkish bank over 108,000 checking accounts; OLS of usage on treatment-arm indicators, comparing messages that mention price with those that do not (no pure control group)\",\"key_assumptions\":[\"Random assignment of message arms is balanced across baseline characteristics\",\"No other bank communications to the sample during the campaign\",\"Comparisons across promotion arms identify the effect of mentioning price, since a pure control group was not used\"],\"threats_addressed\":[\"Baseline imbalance (orthogonality checks)\",\"Mediation by an earlier common message\",\"Classical/mechanical explanations such as bundled transactions raising balance variance\",\"Persistence vs. transitory effects (post-experiment window)\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":5,\"move\":\"motivation\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":8,\"move\":\"identification_defense\"},{\"paragraph_index\":9,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":10,\"move\":\"identification_defense\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"preview_of_results\"},{\"paragraph_index\":12,\"move\":\"preview_of_results\"},{\"paragraph_index\":13,\"move\":\"preview_of_results\"},{\"paragraph_index\":14,\"move\":\"contribution_statement\"},{\"paragraph_index\":15,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":16,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":17,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":14,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_specification\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Each column is one OLS regression of a usage measure on treatment-arm indicators, with Huber–White SEs in parentheses. Notes explain that some rows give additive effects of discount arms relative to reference arms. Table I is a balance table reporting proportions with SEs and joint-test p-values; stars are used. Appendix tables cover balances, products, mediation, frequency, and duration.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"robust\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' framing\",\"'some evidence'\",\"'suggests'\",\"explicit statement that no welfare analysis is done\",\"acknowledging confidence intervals do not rule out heterogeneity\"]},\"claim_evidence\":[{\"claim\":\"Overdraft discount messages lower overdraft likelihood by about 1.2 pp versus availability-only messages, on a 31% baseline\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Bundling the overdraft discount with debit or autopay discounts reduces overdrafting more than the discount alone\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Debit card and autopay discounts weakly increase use of those services\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Overdraft availability messages raise overdraft likelihood by about 0.9 pp\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Treatment effects do not persist after messaging ends\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 3\"},{\"claim\":\"No heterogeneity by prior overdraft experience\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Randomization produced balanced arms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"}],\"structural_summary\":\"Opens with a rhetorical question, then motivates add-on/shrouding business models using retail overdrafts. It poses the theoretical puzzle of why firms shroud prices and states the empirical gap. A bank field experiment follows, with one-sided hypothesis tests framed as stacking the deck against the behavioral prediction. Point estimates with SEs are previewed in the introduction, followed by literature positioning and a roadmap. The body covers the institutional setting (including ethics), experimental design with balance checks, and results with heterogeneity and frequency/duration variations. A discussion section weighs classical alternatives, and the conclusion gives cautious policy implications, a welfare caveat, and future design refinements.\",\"writing_profile_keywords\":[\"field experiment\",\"shrouded add-on pricing\",\"limited attention\",\"behavioral vs classical tests\",\"one-sided hypothesis\",\"SE reported in text\",\"policy caveats\"],\"confidence\":0.85,\"notes\":\"The intro says Section V discusses alternatives, but the heading list puts that discussion in Section IV. The rule-based tag 'theory' for Section I is wrong; it describes the institutional setting and ethics, so it is labeled 'other'. The Discussion section is labeled 'other' rather than conclusion.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "survey_or_experiment", "identification": {"estimand": "Intent-to-treat effects of randomized SMS message content (overdraft discount, availability, bundled discounts, frequency, duration) on overdraft, debit, and autopay usage during and after the campaign", "strategy": "Randomized field experiment with a Turkish bank over 108,000 checking accounts; OLS of usage on treatment-arm indicators, comparing messages that mention price with those that do not (no pure control group)", "key_assumptions": ["Random assignment of message arms is balanced across baseline characteristics", "No other bank communications to the sample during the campaign", "Comparisons across promotion arms identify the effect of mentioning price, since a pure control group was not used"], "threats_addressed": ["Baseline imbalance (orthogonality checks)", "Mediation by an earlier common message", "Classical/mechanical explanations such as bundled transactions raising balance variance", "Persistence vs. transitory effects (post-experiment window)"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 5}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 5, "move": "motivation"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 8, "move": "identification_defense"}, {"paragraph_index": 9, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 10, "move": "identification_defense"}, {"paragraph_index": 10, "move": "preview_of_results"}, {"paragraph_index": 11, "move": "preview_of_results"}, {"paragraph_index": 12, "move": "preview_of_results"}, {"paragraph_index": 13, "move": "preview_of_results"}, {"paragraph_index": 14, "move": "contribution_statement"}, {"paragraph_index": 15, "move": "related_literature_positioning"}, {"paragraph_index": 16, "move": "related_literature_positioning"}, {"paragraph_index": 17, "move": "roadmap"}], "contribution_paragraph_index": 14, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_specification", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Each column is one OLS regression of a usage measure on treatment-arm indicators, with Huber–White SEs in parentheses. Notes explain that some rows give additive effects of discount arms relative to reference arms. Table I is a balance table reporting proportions with SEs and joint-test p-values; stars are used. Appendix tables cover balances, products, mediation, frequency, and duration.", "units_stated_in_notes": false, "standard_errors_description": "robust"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' framing", "'some evidence'", "'suggests'", "explicit statement that no welfare analysis is done", "acknowledging confidence intervals do not rule out heterogeneity"]}, "claim_evidence": [{"claim": "Overdraft discount messages lower overdraft likelihood by about 1.2 pp versus availability-only messages, on a 31% baseline", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Bundling the overdraft discount with debit or autopay discounts reduces overdrafting more than the discount alone", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Debit card and autopay discounts weakly increase use of those services", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Overdraft availability messages raise overdraft likelihood by about 0.9 pp", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Treatment effects do not persist after messaging ends", "evidence_type": "table", "evidence_ref": "Table IV; Figure 3"}, {"claim": "No heterogeneity by prior overdraft experience", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Randomization produced balanced arms", "evidence_type": "table", "evidence_ref": "Table I"}], "structural_summary": "Opens with a rhetorical question, then motivates add-on/shrouding business models using retail overdrafts. It poses the theoretical puzzle of why firms shroud prices and states the empirical gap. A bank field experiment follows, with one-sided hypothesis tests framed as stacking the deck against the behavioral prediction. Point estimates with SEs are previewed in the introduction, followed by literature positioning and a roadmap. The body covers the institutional setting (including ethics), experimental design with balance checks, and results with heterogeneity and frequency/duration variations. A discussion section weighs classical alternatives, and the conclusion gives cautious policy implications, a welfare caveat, and future design refinements.", "writing_profile_keywords": ["field experiment", "shrouded add-on pricing", "limited attention", "behavioral vs classical tests", "one-sided hypothesis", "SE reported in text", "policy caveats"], "confidence": 0.85, "notes": "The intro says Section V discusses alternatives, but the heading list puts that discussion in Section IV. The rule-based tag 'theory' for Section I is wrong; it describes the institutional setting and ethics, so it is labeled 'other'. The Discussion section is labeled 'other' rather than conclusion."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.97, "usage": {"input": 9828, "output": 2396, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088026, "request_id": "req_011CggpqXTsTjn69nRgraWhz", "cached": false, "ts": "2026-11-03T23:00:43+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/cfec39edd59c95fd48cf93d0b7d85b9690cd34c2716e0dfee0e096f149a176b0.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Conditional association between survey-measured financial literacy and household portfolio returns and dynamics: risk-taking, inertia and rebalancing direction.\",\"strategy\":\"OLS, probit and IV regressions on matched administrative and survey panel data with demographic and wealth controls. IV is used only to decompose active versus passive changes in the risky share, instrumented with the zero-rebalancing passive change; it is not used to instrument literacy.\",\"key_assumptions\":[\"Controls for risk, wealth and demographics capture confounders correlated with literacy\",\"No reverse causality from returns to literacy acquisition\",\"Validity of the zero-rebalancing passive change as instrument for the observed passive change\"],\"threats_addressed\":[\"reverse causality (returns inducing literacy acquisition)\",\"omitted wealth and demographic confounders\",\"differences in portfolio risk\",\"crisis period and delegated management (robustness table)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"OLS, probit (marginal effects) and IV tables with robust t-statistics in parentheses and conventional 10/5/1% stars. Variable definitions are deferred to the appendix, and returns are stated in percentage points.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"robust\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"associational verbs ('is associated with', 'relates to')\",\"'suggestive' qualifiers on descriptive figures\",\"explicit acknowledgment of lacking exogenous variation in literacy\",\"reliance on prior IV literature to support causal reading\"]},\"claim_evidence\":[{\"claim\":\"Controlling for risk, the most literate households earn about 0.4% higher annual returns than the least literate.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 2\"},{\"claim\":\"Literate households raise their risky share more when expected excess returns of risky funds are high.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figures 3-4\"},{\"claim\":\"Passive changes account for a larger share of risky-share changes for low-literacy households (64% vs 30%).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 5\"},{\"claim\":\"More literate households rebalance in a contrarian direction, keeping risk exposure stable.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The literacy-returns relationship survives crisis-period and delegation controls.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Literate households do not experience more extreme pre-survey returns, easing reverse causality concerns.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Internet Appendix (mentioned in conclusion)\"}],\"structural_summary\":\"Opens with established heterogeneity in household returns and literacy, then frames mechanisms as poorly understood because suitable data are scarce. Introduces a novel matched administrative-survey dataset and previews three dimensions of portfolio dynamics (risk timing, inertia, rebalancing direction), each with a quantitative finding and links to prior claims. It closes the introduction with a link to returns and literature positioning. Body sections cover data, construction of the literacy measure, returns, and portfolio choices organized by the three dimensions. The conclusion concedes the lack of exogenous variation and offers mitigating evidence and policy relevance.\",\"writing_profile_keywords\":[\"matched administrative-survey data\",\"mechanism decomposition\",\"household portfolio dynamics\",\"associational with causal caveats\",\"three-result preview structure\",\"financial literacy\"],\"confidence\":0.85,\"notes\":\"Single-author slip ('my') appears in a multi-author intro. IV regressions instrument the passive change, not literacy. No roadmap paragraph is visible in the intro excerpt. Robustness appears in Table IV within the results and in the Internet Appendix.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "asset_pricing", "primary_method": "reduced_form_regression", "identification": {"estimand": "Conditional association between survey-measured financial literacy and household portfolio returns and dynamics: risk-taking, inertia and rebalancing direction.", "strategy": "OLS, probit and IV regressions on matched administrative and survey panel data with demographic and wealth controls. IV is used only to decompose active versus passive changes in the risky share, instrumented with the zero-rebalancing passive change; it is not used to instrument literacy.", "key_assumptions": ["Controls for risk, wealth and demographics capture confounders correlated with literacy", "No reverse causality from returns to literacy acquisition", "Validity of the zero-rebalancing passive change as instrument for the observed passive change"], "threats_addressed": ["reverse causality (returns inducing literacy acquisition)", "omitted wealth and demographic confounders", "differences in portfolio risk", "crisis period and delegated management (robustness table)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "tension_or_puzzle"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}], "contribution_paragraph_index": 2, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "OLS, probit (marginal effects) and IV tables with robust t-statistics in parentheses and conventional 10/5/1% stars. Variable definitions are deferred to the appendix, and returns are stated in percentage points.", "units_stated_in_notes": true, "standard_errors_description": "robust"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["associational verbs ('is associated with', 'relates to')", "'suggestive' qualifiers on descriptive figures", "explicit acknowledgment of lacking exogenous variation in literacy", "reliance on prior IV literature to support causal reading"]}, "claim_evidence": [{"claim": "Controlling for risk, the most literate households earn about 0.4% higher annual returns than the least literate.", "evidence_type": "table", "evidence_ref": "Table III; Figure 2"}, {"claim": "Literate households raise their risky share more when expected excess returns of risky funds are high.", "evidence_type": "table", "evidence_ref": "Table V; Figures 3-4"}, {"claim": "Passive changes account for a larger share of risky-share changes for low-literacy households (64% vs 30%).", "evidence_type": "table", "evidence_ref": "Table VI; Figure 5"}, {"claim": "More literate households rebalance in a contrarian direction, keeping risk exposure stable.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The literacy-returns relationship survives crisis-period and delegation controls.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Literate households do not experience more extreme pre-survey returns, easing reverse causality concerns.", "evidence_type": "none_stated", "evidence_ref": "Internet Appendix (mentioned in conclusion)"}], "structural_summary": "Opens with established heterogeneity in household returns and literacy, then frames mechanisms as poorly understood because suitable data are scarce. Introduces a novel matched administrative-survey dataset and previews three dimensions of portfolio dynamics (risk timing, inertia, rebalancing direction), each with a quantitative finding and links to prior claims. It closes the introduction with a link to returns and literature positioning. Body sections cover data, construction of the literacy measure, returns, and portfolio choices organized by the three dimensions. The conclusion concedes the lack of exogenous variation and offers mitigating evidence and policy relevance.", "writing_profile_keywords": ["matched administrative-survey data", "mechanism decomposition", "household portfolio dynamics", "associational with causal caveats", "three-result preview structure", "financial literacy"], "confidence": 0.85, "notes": "Single-author slip ('my') appears in a multi-author intro. IV regressions instrument the passive change, not literacy. No roadmap paragraph is visible in the intro excerpt. Robustness appears in Table IV within the results and in the Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.66, "usage": {"input": 6023, "output": 2133, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067546, "request_id": "req_011CggptREu5Xmii87CayiFh", "cached": false, "ts": "2026-11-03T23:01:19+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d76871d030d2d80c7ef4d99f74d2b9ee0f49ee8cfc6d02bb8218e2b399a98bb4.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How an investor's IPO allocation, measured as normalized rationing, relates to the revenues that investor generates for the bookrunner and to proxies for its information production during bookbuilding.\",\"strategy\":\"Allocation regressions on revenue quartile rankings and bid/behavior proxies. The regressions are run pooled and bank by bank, then with investor fixed effects and investor-bank fixed effects to absorb unobserved characteristics and relationships. Expected profitability of bids is also tested.\",\"key_assumptions\":[\"After investor and investor-bank fixed effects, no time-varying unobserved information production remains that is correlated with revenue rankings.\",\"The observable proxies (bid type, meeting attendance, bidding frequency) adequately capture information revelation.\"],\"threats_addressed\":[\"Unobserved investor expertise or efficiency in producing information\",\"Closer investor-bank relationships that lead to more information revelation\",\"Allocation differences reflecting expected profitability rather than allocation levels alone\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_data_or_measure\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"robustness\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable (normalized rationing, i.e., percentage allotted divided by percentage bid) and the regressors (bid-size quartiles, price-sensitive vs money bids, revenue quartiles). Descriptive tables are split into panels. The expected-profitability model is referenced to an Internet Appendix table.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"evidence suggests\",\"may mean\",\"alternative explanations exist\",\"welfare implications ambiguous\",\"generalizability caveat\"]},\"claim_evidence\":[{\"claim\":\"Price-sensitive bids receive slightly higher allocations, with considerable variation across banks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"Top revenue-quartile investors receive allocations about 60% higher relative to bid than non-clients.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The revenue-allocation link persists with investor and investor-bank fixed effects.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"The revenue-allocation relationship is concentrated in hot IPOs.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"High-revenue investors obtain higher expected profits from their bids.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Prior studies lacked allocation books and could only show indirect quid pro quo links.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Reuter (2006); Goldstein, Irvine, and Puckett (2011)\"}],\"structural_summary\":\"The paper motivates the question with past IPO scandals and a regulator's market study that supplied unique data. The introduction frames two non-exclusive hypotheses, information revelation and quid pro quo, and previews how unobservables are handled. It then lists findings in an ordered sequence and positions itself against indirect prior evidence. Section I describes the institutional setting and data. Section II reports pooled and bank-level allocation regressions. Section III challenges the quid pro quo interpretation with fixed effects and expected-profitability tests and discusses welfare. The conclusion restates the findings with caveats about generalizability.\",\"writing_profile_keywords\":[\"regulatory data\",\"proprietary allocation books\",\"competing hypotheses\",\"fixed effects\",\"welfare ambiguity\",\"European IPOs\"],\"confidence\":0.8,\"notes\":\"The data come from an FCA market study. Significance conventions are not visible in the excerpts. The Internet Appendix is referenced for the expected-price model.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "How an investor's IPO allocation, measured as normalized rationing, relates to the revenues that investor generates for the bookrunner and to proxies for its information production during bookbuilding.", "strategy": "Allocation regressions on revenue quartile rankings and bid/behavior proxies. The regressions are run pooled and bank by bank, then with investor fixed effects and investor-bank fixed effects to absorb unobserved characteristics and relationships. Expected profitability of bids is also tested.", "key_assumptions": ["After investor and investor-bank fixed effects, no time-varying unobserved information production remains that is correlated with revenue rankings.", "The observable proxies (bid type, meeting attendance, bidding frequency) adequately capture information revelation."], "threats_addressed": ["Unobserved investor expertise or efficiency in producing information", "Closer investor-bank relationships that lead to more information revelation", "Allocation differences reflecting expected profitability rather than allocation levels alone"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_data_or_measure", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "robustness"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "subsample", "alternative_measure"]}, "tables": {"convention_notes": "Table notes define the dependent variable (normalized rationing, i.e., percentage allotted divided by percentage bid) and the regressors (bid-size quartiles, price-sensitive vs money bids, revenue quartiles). Descriptive tables are split into panels. The expected-profitability model is referenced to an Internet Appendix table.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "evidence suggests", "may mean", "alternative explanations exist", "welfare implications ambiguous", "generalizability caveat"]}, "claim_evidence": [{"claim": "Price-sensitive bids receive slightly higher allocations, with considerable variation across banks.", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "Top revenue-quartile investors receive allocations about 60% higher relative to bid than non-clients.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The revenue-allocation link persists with investor and investor-bank fixed effects.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "The revenue-allocation relationship is concentrated in hot IPOs.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "High-revenue investors obtain higher expected profits from their bids.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Prior studies lacked allocation books and could only show indirect quid pro quo links.", "evidence_type": "prior_literature", "evidence_ref": "Reuter (2006); Goldstein, Irvine, and Puckett (2011)"}], "structural_summary": "The paper motivates the question with past IPO scandals and a regulator's market study that supplied unique data. The introduction frames two non-exclusive hypotheses, information revelation and quid pro quo, and previews how unobservables are handled. It then lists findings in an ordered sequence and positions itself against indirect prior evidence. Section I describes the institutional setting and data. Section II reports pooled and bank-level allocation regressions. Section III challenges the quid pro quo interpretation with fixed effects and expected-profitability tests and discusses welfare. The conclusion restates the findings with caveats about generalizability.", "writing_profile_keywords": ["regulatory data", "proprietary allocation books", "competing hypotheses", "fixed effects", "welfare ambiguity", "European IPOs"], "confidence": 0.8, "notes": "The data come from an FCA market study. Significance conventions are not visible in the excerpts. The Internet Appendix is referenced for the expected-price model."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.12, "usage": {"input": 7310, "output": 2014, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070314, "request_id": "req_011Cggq5aKuhSCAdwZ8XFe4J", "cached": false, "ts": "2026-11-03T23:03:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/daf786ca42c98c39ee19b47aade036d419cab2e757a13cea265dca52e24e06db.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model of shareholder incentives over leverage adjustments)\",\"strategy\":\"not applicable; analytical propositions derived in reduced-form static and dynamic trade-off models\",\"key_assumptions\":[\"shareholders cannot commit to future funding choices\",\"standard trade-off frictions: taxes, bankruptcy costs, agency costs\",\"creditors anticipate shareholder-controlled funding decisions\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"No tables. Results are stated as propositions with proofs, and numerical illustrations are given in figures.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"conditional statements (if shareholders cannot commit...)\",\"'generally'/'typically'\",\"'may' in policy suggestions\",\"'consistent with' when linking to evidence\"]},\"claim_evidence\":[{\"claim\":\"Shareholders lose from any leverage reduction once debt is in place, because the reduction transfers wealth to existing creditors\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Intro paragraph 1 algebra; Section II.B\"},{\"claim\":\"Resistance to leverage reduction persists even when the reduction alleviates bankruptcy or agency costs and raises firm value\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B-C\"},{\"claim\":\"Shareholders gain from issuing junior debt even when it destroys firm value\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section II.D; Figures 1-2\"},{\"claim\":\"Without commitment, equilibrium leverage takes stable levels that ratchet up with each shock\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section III; Figures 3-4\"},{\"claim\":\"When forced to delever, shareholders prefer asset sales over pure recapitalization\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section IV; Figures 6-7\"},{\"claim\":\"Firms respond asymmetrically to tax changes, consistent with the ratchet effect\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Conclusion citing Heider and Ljungqvist (2015)\"}],\"structural_summary\":\"A theory paper. The abstract states the core mechanism and names it. The introduction states the contribution and builds intuition with simple algebra. It positions the paper against the Black-Scholes result, previews static, dynamic and asset-transaction results, and ends with banking and policy implications. A separate literature section follows. The body has three model sections: a one-time adjustment with frictions, dynamic equilibrium with shocks, maturity and covenants, and alternative deleveraging modes. The conclusion lists bulleted testable predictions and discusses welfare and policy. Proofs are in the appendix.\",\"writing_profile_keywords\":[\"analytical theory\",\"named effect\",\"propositions and proofs\",\"intuition after results\",\"testable predictions\",\"policy implications\",\"banking application\"],\"confidence\":0.85,\"notes\":\"Section IV was rule-labeled as robustness but is an extension of the model. The intro appears truncated, so no roadmap is visible.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model of shareholder incentives over leverage adjustments)", "strategy": "not applicable; analytical propositions derived in reduced-form static and dynamic trade-off models", "key_assumptions": ["shareholders cannot commit to future funding choices", "standard trade-off frictions: taxes, bankruptcy costs, agency costs", "creditors anticipate shareholder-controlled funding decisions"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "No tables. Results are stated as propositions with proofs, and numerical illustrations are given in figures.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["conditional statements (if shareholders cannot commit...)", "'generally'/'typically'", "'may' in policy suggestions", "'consistent with' when linking to evidence"]}, "claim_evidence": [{"claim": "Shareholders lose from any leverage reduction once debt is in place, because the reduction transfers wealth to existing creditors", "evidence_type": "equation_or_proof", "evidence_ref": "Intro paragraph 1 algebra; Section II.B"}, {"claim": "Resistance to leverage reduction persists even when the reduction alleviates bankruptcy or agency costs and raises firm value", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B-C"}, {"claim": "Shareholders gain from issuing junior debt even when it destroys firm value", "evidence_type": "figure", "evidence_ref": "Section II.D; Figures 1-2"}, {"claim": "Without commitment, equilibrium leverage takes stable levels that ratchet up with each shock", "evidence_type": "figure", "evidence_ref": "Section III; Figures 3-4"}, {"claim": "When forced to delever, shareholders prefer asset sales over pure recapitalization", "evidence_type": "figure", "evidence_ref": "Section IV; Figures 6-7"}, {"claim": "Firms respond asymmetrically to tax changes, consistent with the ratchet effect", "evidence_type": "prior_literature", "evidence_ref": "Conclusion citing Heider and Ljungqvist (2015)"}], "structural_summary": "A theory paper. The abstract states the core mechanism and names it. The introduction states the contribution and builds intuition with simple algebra. It positions the paper against the Black-Scholes result, previews static, dynamic and asset-transaction results, and ends with banking and policy implications. A separate literature section follows. The body has three model sections: a one-time adjustment with frictions, dynamic equilibrium with shocks, maturity and covenants, and alternative deleveraging modes. The conclusion lists bulleted testable predictions and discusses welfare and policy. Proofs are in the appendix.", "writing_profile_keywords": ["analytical theory", "named effect", "propositions and proofs", "intuition after results", "testable predictions", "policy implications", "banking application"], "confidence": 0.85, "notes": "Section IV was rule-labeled as robustness but is an extension of the model. The intro appears truncated, so no roadmap is visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.85, "usage": {"input": 6832, "output": 1800, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064122, "request_id": "req_011Cggpo3gyrkHX2tNkBYVUv", "cached": false, "ts": "2026-11-03T23:00:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/db59df98a24e3e097eed12234df17027d1e3ffed50ad9453c97d169888559b15.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"corporate_finance (debt overhang theory applied to households); real estate\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in quarterly housing investment (home improvement/maintenance spending, unscheduled principal payments) associated with negative equity or estimated mortgage default risk\",\"strategy\":\"Cross-sectional regressions of spending on a negative-equity indicator or estimated default risk, with controls and state-year fixed effects. Supplemented by a propensity-score matched sample, subsamples of financially unconstrained households, a within-household comparison across durable categories, household fixed effects (changes over time and across properties of multiproperty owners), and recourse vs. nonrecourse state contrasts.\",\"key_assumptions\":[\"Conditional on observables, negative equity is not correlated with unobserved preferences for home investment\",\"Negative equity and estimated default risk proxy for perceived default risk\",\"High-income or high-asset households are not financially constrained\",\"Durables retained in default share the same household-level confounders as home improvements\"],\"threats_addressed\":[\"financial/liquidity constraints\",\"omitted household characteristics (household fixed effects)\",\"general spending decline or low expected income\",\"intertemporal substitution of past improvements\",\"selection on observables (matching)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"robustness\"},{\"idx\":9,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"alternative_specification\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes describe the sample and period, the dependent and key independent variables, and the fixed effects used. Mean spending is reported for the comparison group so coefficients can be read as percentages. Matching procedures are described in the notes. Standard-error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' language\",\"acknowledging that results do not rule out alternatives\",\"noting low power of fixed-effects tests\",\"'mixed' evidence admission\",\"'may create' deadweight loss\"]},\"claim_evidence\":[{\"claim\":\"Negative-equity homeowners cut improvement/maintenance spending and unscheduled principal payments by roughly 30%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV\"},{\"claim\":\"Housing investment declines with a continuous estimate of default risk\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figure 2\"},{\"claim\":\"Spending falls with loan-to-value in raw and residualized data\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Financially unconstrained households also cut housing investment when underwater\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Spending on durables retained in default is not reduced by negative-equity households\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Wealthy underwater homeowners cut investment more in nonrecourse states\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Household fixed-effects estimates broadly confirm improvement results; principal payment evidence mixed\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Results robust in a propensity-matched sample\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The paper opens by framing household behavior through the classic corporate and public debt overhang theory. It then motivates the question with post-2006 negative-equity statistics and previews a baseline finding from consumer expenditure survey data. It acknowledges a competing explanation, financial constraints, and presents a sequence of discriminating tests: unconstrained subsamples, recourse laws, cross-durable comparisons, and household fixed effects. Brief theory and data sections are followed by a research-design section, several results sections each addressing one alternative, and an additional-tests section referring to the internet appendix. The conclusion restates the findings, gives an aggregate magnitude, and suggests future work on social costs.\",\"writing_profile_keywords\":[\"debt overhang\",\"household finance\",\"alternative-hypothesis elimination\",\"sequential falsification tests\",\"survey data\",\"negative equity\",\"economic magnitudes\"],\"confidence\":0.8,\"notes\":\"How statistical significance is reported is not visible in the excerpts. The intro excerpt is truncated, so the main magnitude preview may be in the missing text; the 30% figure appears in the conclusion. Robustness appears both in the main text and in the internet appendix.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "corporate_finance (debt overhang theory applied to households); real estate", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in quarterly housing investment (home improvement/maintenance spending, unscheduled principal payments) associated with negative equity or estimated mortgage default risk", "strategy": "Cross-sectional regressions of spending on a negative-equity indicator or estimated default risk, with controls and state-year fixed effects. Supplemented by a propensity-score matched sample, subsamples of financially unconstrained households, a within-household comparison across durable categories, household fixed effects (changes over time and across properties of multiproperty owners), and recourse vs. nonrecourse state contrasts.", "key_assumptions": ["Conditional on observables, negative equity is not correlated with unobserved preferences for home investment", "Negative equity and estimated default risk proxy for perceived default risk", "High-income or high-asset households are not financially constrained", "Durables retained in default share the same household-level confounders as home improvements"], "threats_addressed": ["financial/liquidity constraints", "omitted household characteristics (household fixed effects)", "general spending decline or low expected income", "intertemporal substitution of past improvements", "selection on observables (matching)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "robustness"}, {"idx": 9, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "alternative_specification", "falsification", "other"]}, "tables": {"convention_notes": "Table notes describe the sample and period, the dependent and key independent variables, and the fixed effects used. Mean spending is reported for the comparison group so coefficients can be read as percentages. Matching procedures are described in the notes. Standard-error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' language", "acknowledging that results do not rule out alternatives", "noting low power of fixed-effects tests", "'mixed' evidence admission", "'may create' deadweight loss"]}, "claim_evidence": [{"claim": "Negative-equity homeowners cut improvement/maintenance spending and unscheduled principal payments by roughly 30%", "evidence_type": "table", "evidence_ref": "Tables III, IV"}, {"claim": "Housing investment declines with a continuous estimate of default risk", "evidence_type": "table", "evidence_ref": "Table V; Figure 2"}, {"claim": "Spending falls with loan-to-value in raw and residualized data", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Financially unconstrained households also cut housing investment when underwater", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Spending on durables retained in default is not reduced by negative-equity households", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Wealthy underwater homeowners cut investment more in nonrecourse states", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Household fixed-effects estimates broadly confirm improvement results; principal payment evidence mixed", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Results robust in a propensity-matched sample", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The paper opens by framing household behavior through the classic corporate and public debt overhang theory. It then motivates the question with post-2006 negative-equity statistics and previews a baseline finding from consumer expenditure survey data. It acknowledges a competing explanation, financial constraints, and presents a sequence of discriminating tests: unconstrained subsamples, recourse laws, cross-durable comparisons, and household fixed effects. Brief theory and data sections are followed by a research-design section, several results sections each addressing one alternative, and an additional-tests section referring to the internet appendix. The conclusion restates the findings, gives an aggregate magnitude, and suggests future work on social costs.", "writing_profile_keywords": ["debt overhang", "household finance", "alternative-hypothesis elimination", "sequential falsification tests", "survey data", "negative equity", "economic magnitudes"], "confidence": 0.8, "notes": "How statistical significance is reported is not visible in the excerpts. The intro excerpt is truncated, so the main magnitude preview may be in the missing text; the 30% figure appears in the conclusion. Robustness appears both in the main text and in the internet appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.89, "usage": {"input": 8442, "output": 2319, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080942, "request_id": "req_011CggpWTUj8goewU5fubnP1", "cached": false, "ts": "2026-11-03T22:56:22+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/dbcf558280519460269319f8f1da733a7a4d0d33d3adf3c7fd00ac4d762e0dc5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_intermediation_banking (sell-side analysts as information intermediaries)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How much analyst revisions move stock prices (two-day CAR and probability of an influential revision), plus forecast precision and analyst activity, in bad times versus good times\",\"strategy\":\"Panel and probit regressions of announcement CARs and influential-revision indicators on bad-times dummies (crises, NBER recessions, high policy uncertainty) with firm, analyst and recommendation controls; cross-sectional interactions with firm opacity; competing hypotheses tested against each other\",\"key_assumptions\":[\"Bad-times dummies capture exogenous increases in macro uncertainty\",\"Controls absorb differences in firm and analyst characteristics across periods\",\"The DGTW characteristics-matched benchmark correctly measures abnormal returns\"],\"threats_addressed\":[\"Firm-specific vs marketwide uncertainty driving results\",\"Market reacting more to all firm news in bad times\",\"Investor overreaction or price pressure (checked via post-revision drift and reversals)\",\"Conflicts of interest\",\"Alternative specifications and samples\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":7,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"robustness\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state variable definitions, sample periods and units (CARs and marginal effects in percent). Probit tables report marginal effects. Influential revisions are defined by a 1.96 threshold relative to expected volatility. Bad-times measures are defined once and cross-referenced across tables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' phrasing\",\"'likely explain'\",\"'might' for mechanisms\",\"'some evidence' for spillovers\",\"explicitly nonmutually exclusive hypotheses\"]},\"claim_evidence\":[{\"claim\":\"Recommendation changes have a larger two-day price impact and are more often influential in bad times\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-IV, Figure 1\"},{\"claim\":\"Earnings forecast revisions are more likely to be influential in bad times\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Uncertainty measures such as VIX and implied volatility rise sharply in bad times\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Traditional absolute forecast errors rise in bad times, but errors scaled by volatility fall\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The increase in analyst impact in bad times is larger for opaque firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Analysts produce more forecasts and longer reports in bad times, consistent with greater effort\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX-XI\"},{\"claim\":\"Increased crisis impact of downgrades corresponds to roughly $80 million in market value per typical change\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Conclusion, Table I Panel C\"},{\"claim\":\"Overreaction cannot explain results since post-revision drift shows no reversals\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section V.E\"}],\"structural_summary\":\"The introduction states a tension: higher uncertainty in bad times could make analyst output more valuable or make analysts' jobs harder. It then previews the main finding and a Bayesian argument motivating uncertainty-scaled precision, and walks through results for five competing explanations before positioning against prior work. The body proceeds in order: a hypothesis section with three 'less impact' and five 'more impact' hypotheses; main results on price impact using several bad-times proxies; a robustness section; forecast-precision analysis; and hypothesis-by-hypothesis tests of mechanisms. The conclusion restates findings and adds an economic magnitude.\",\"writing_profile_keywords\":[\"competing hypotheses\",\"horse-race of explanations\",\"bad-times proxies\",\"event-study CARs\",\"probit marginal effects\",\"uncertainty-scaled precision\",\"mechanism sections\"],\"confidence\":0.8,\"notes\":\"No roadmap paragraph is visible in the introduction excerpts. The second hypothesis (analyst effort) appears to be truncated from the excerpted introduction paragraphs. How significance is presented is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_intermediation_banking (sell-side analysts as information intermediaries)", "primary_method": "reduced_form_regression", "identification": {"estimand": "How much analyst revisions move stock prices (two-day CAR and probability of an influential revision), plus forecast precision and analyst activity, in bad times versus good times", "strategy": "Panel and probit regressions of announcement CARs and influential-revision indicators on bad-times dummies (crises, NBER recessions, high policy uncertainty) with firm, analyst and recommendation controls; cross-sectional interactions with firm opacity; competing hypotheses tested against each other", "key_assumptions": ["Bad-times dummies capture exogenous increases in macro uncertainty", "Controls absorb differences in firm and analyst characteristics across periods", "The DGTW characteristics-matched benchmark correctly measures abnormal returns"], "threats_addressed": ["Firm-specific vs marketwide uncertainty driving results", "Market reacting more to all firm news in bad times", "Investor overreaction or price pressure (checked via post-revision drift and reversals)", "Conflicts of interest", "Alternative specifications and samples"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "contribution_statement"}], "contribution_paragraph_index": 7, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "robustness"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "alternative_measure", "falsification", "other"]}, "tables": {"convention_notes": "Table notes state variable definitions, sample periods and units (CARs and marginal effects in percent). Probit tables report marginal effects. Influential revisions are defined by a 1.96 threshold relative to expected volatility. Bad-times measures are defined once and cross-referenced across tables.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' phrasing", "'likely explain'", "'might' for mechanisms", "'some evidence' for spillovers", "explicitly nonmutually exclusive hypotheses"]}, "claim_evidence": [{"claim": "Recommendation changes have a larger two-day price impact and are more often influential in bad times", "evidence_type": "table", "evidence_ref": "Tables II-IV, Figure 1"}, {"claim": "Earnings forecast revisions are more likely to be influential in bad times", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Uncertainty measures such as VIX and implied volatility rise sharply in bad times", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Traditional absolute forecast errors rise in bad times, but errors scaled by volatility fall", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The increase in analyst impact in bad times is larger for opaque firms", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Analysts produce more forecasts and longer reports in bad times, consistent with greater effort", "evidence_type": "table", "evidence_ref": "Tables IX-XI"}, {"claim": "Increased crisis impact of downgrades corresponds to roughly $80 million in market value per typical change", "evidence_type": "descriptive_statistic", "evidence_ref": "Conclusion, Table I Panel C"}, {"claim": "Overreaction cannot explain results since post-revision drift shows no reversals", "evidence_type": "none_stated", "evidence_ref": "Section V.E"}], "structural_summary": "The introduction states a tension: higher uncertainty in bad times could make analyst output more valuable or make analysts' jobs harder. It then previews the main finding and a Bayesian argument motivating uncertainty-scaled precision, and walks through results for five competing explanations before positioning against prior work. The body proceeds in order: a hypothesis section with three 'less impact' and five 'more impact' hypotheses; main results on price impact using several bad-times proxies; a robustness section; forecast-precision analysis; and hypothesis-by-hypothesis tests of mechanisms. The conclusion restates findings and adds an economic magnitude.", "writing_profile_keywords": ["competing hypotheses", "horse-race of explanations", "bad-times proxies", "event-study CARs", "probit marginal effects", "uncertainty-scaled precision", "mechanism sections"], "confidence": 0.8, "notes": "No roadmap paragraph is visible in the introduction excerpts. The second hypothesis (analyst effort) appears to be truncated from the excerpted introduction paragraphs. How significance is presented is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.33, "usage": {"input": 8143, "output": 2327, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079906, "request_id": "req_011Cggpxe6S8PS4aD3CqPySw", "cached": false, "ts": "2026-11-03T23:02:18+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/dc5af954604d2857fcc8d236f78d1544e524c5ecd48ccdd576d52bf3629f2562.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Equilibrium sorting between borrower bank-dependence and lender capital, plus counterfactual crisis loan supply under alternative matching assignments\",\"strategy\":\"Cross-sectional and within-bank regressions of lender capital on borrower characteristics; lead-versus-participant and revolver-versus-term-loan contrasts to narrow mechanisms; Fox (2016) semiparametric matching estimator for counterfactual matchings; Khwaja-Mian style crisis lending regressions\",\"key_assumptions\":[\"Lack of an S&P rating proxies for bank dependence\",\"Lead arrangers monitor while participants bear default risk without monitoring\",\"Each quarter is an independent matching market with one-to-one matching to a single lead arranger\",\"Observed matches reflect pairwise stability (maximum score)\"],\"threats_addressed\":[\"Mechanism ambiguity across competing theories, via lender roles and loan types\",\"Time-invariant bank heterogeneity, via within-bank variation\",\"Inconsistency of bootstrap for maximum score, via an alternative CI construction\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"other\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes open by stating the dependent variable and sample, then define key variables (market equity ratio as bank capital, unrated borrower as bank-dependent). Spreads are given in basis points. The counterfactual table states its estimation window.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"evidence suggests\",\"consistent with / inconsistent with\",\"explicit caveat that directional statements are descriptive only\",\"external validity discussion\",\"not isolating a specific channel\"]},\"claim_evidence\":[{\"claim\":\"Unrated (bank-dependent) firms borrow from lead arrangers with about 1.8 points higher equity capital than rated firms do.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Sorting by capital holds for lead arrangers but not for participants, favoring monitoring and relationship explanations over risk management.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Borrowers of poorly capitalized banks were more likely to issue bonds after the 2008 crisis.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"High-capital banks earn a spread premium when lending to small or bank-dependent firms.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Observed matching mitigates crisis loan supply cuts for bank-dependent firms relative to counterfactual assignments.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Competing banking theories predict opposite sortings, making the question an empirical one.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction paragraph 2; Section I\"}],\"structural_summary\":\"The introduction moves from the known effects of banking relationships to the unexplored question of what determines them. It previews the headline sorting result with magnitudes and draws out its aggregate implication for credit cyclicality. It then sets competing theoretical predictions against each other, adds an interpretive caveat, and closes with literature positioning. The body follows: a conceptual framework of rival hypotheses, data construction, regression evidence (cross-section, within-bank, mechanism tests via lender roles and loan types, pricing), and a semiparametric matching model used for crisis counterfactuals. The conclusion restates the findings, discusses external validity, and spells out implications for the credit channel.\",\"writing_profile_keywords\":[\"stylized fact\",\"two-sided matching\",\"competing hypotheses\",\"mechanism narrowing\",\"counterfactual\",\"credit channel\",\"first-person singular\"],\"confidence\":0.8,\"notes\":\"Single-author paper. The section labeled '1' continues the matching model section (a parsing artifact). Standard-error conventions are not visible in the excerpts; the matching model uses a non-bootstrap confidence interval construction.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Equilibrium sorting between borrower bank-dependence and lender capital, plus counterfactual crisis loan supply under alternative matching assignments", "strategy": "Cross-sectional and within-bank regressions of lender capital on borrower characteristics; lead-versus-participant and revolver-versus-term-loan contrasts to narrow mechanisms; Fox (2016) semiparametric matching estimator for counterfactual matchings; Khwaja-Mian style crisis lending regressions", "key_assumptions": ["Lack of an S&P rating proxies for bank dependence", "Lead arrangers monitor while participants bear default risk without monitoring", "Each quarter is an independent matching market with one-to-one matching to a single lead arranger", "Observed matches reflect pairwise stability (maximum score)"], "threats_addressed": ["Mechanism ambiguity across competing theories, via lender roles and loan types", "Time-invariant bank heterogeneity, via within-bank variation", "Inconsistency of bootstrap for maximum score, via an alternative CI construction"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": false, "sentence_index": -1}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 2}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "policy_or_broader_implication"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "other"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification"]}, "tables": {"convention_notes": "Notes open by stating the dependent variable and sample, then define key variables (market equity ratio as bank capital, unrated borrower as bank-dependent). Spreads are given in basis points. The counterfactual table states its estimation window.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["evidence suggests", "consistent with / inconsistent with", "explicit caveat that directional statements are descriptive only", "external validity discussion", "not isolating a specific channel"]}, "claim_evidence": [{"claim": "Unrated (bank-dependent) firms borrow from lead arrangers with about 1.8 points higher equity capital than rated firms do.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Sorting by capital holds for lead arrangers but not for participants, favoring monitoring and relationship explanations over risk management.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Borrowers of poorly capitalized banks were more likely to issue bonds after the 2008 crisis.", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "High-capital banks earn a spread premium when lending to small or bank-dependent firms.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Observed matching mitigates crisis loan supply cuts for bank-dependent firms relative to counterfactual assignments.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Competing banking theories predict opposite sortings, making the question an empirical one.", "evidence_type": "prior_literature", "evidence_ref": "Introduction paragraph 2; Section I"}], "structural_summary": "The introduction moves from the known effects of banking relationships to the unexplored question of what determines them. It previews the headline sorting result with magnitudes and draws out its aggregate implication for credit cyclicality. It then sets competing theoretical predictions against each other, adds an interpretive caveat, and closes with literature positioning. The body follows: a conceptual framework of rival hypotheses, data construction, regression evidence (cross-section, within-bank, mechanism tests via lender roles and loan types, pricing), and a semiparametric matching model used for crisis counterfactuals. The conclusion restates the findings, discusses external validity, and spells out implications for the credit channel.", "writing_profile_keywords": ["stylized fact", "two-sided matching", "competing hypotheses", "mechanism narrowing", "counterfactual", "credit channel", "first-person singular"], "confidence": 0.8, "notes": "Single-author paper. The section labeled '1' continues the matching model section (a parsing artifact). Standard-error conventions are not visible in the excerpts; the matching model uses a non-bootstrap confidence interval construction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.19, "usage": {"input": 6821, "output": 2166, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071398, "request_id": "req_011CggpsTJQAJL9BoHfCssfc", "cached": false, "ts": "2026-11-03T23:01:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/dce5b7fab6d1c8a0aec5d7f020178a76036a1e3c4577a40932b4497f39edb4e7.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"not applicable (theoretical model; equilibrium effects of nominal rate policy rules on leverage, risk premia, asset prices and growth)\",\"strategy\":\"Two-agent continuous-time general equilibrium model in which a liquidity premium links the nominal rate to the cost of leverage. Solved globally by Chebyshev collocation. Policy rules are compared across equilibria; impulse responses are computed in an extended production model.\",\"key_assumptions\":[\"Banks face funding shocks that force costly asset liquidation unless they hold liquid buffers\",\"Central bank controls the liquidity premium via the nominal rate and the supply of liquid assets\",\"Heterogeneous risk aversion between banks and depositors\"],\"threats_addressed\":[],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":4},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"other\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"separate_section\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Only two tables, both listing calibrated parameters with their symbols and values. Notes link each table to the sections and figures it generates. The extended-model table lists only the parameters that differ from the baseline.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'arguably' when interpreting historical episodes\",\"'consistent with' when linking model output to empirical findings\",\"'potentially greater instability'\"]},\"claim_evidence\":[{\"claim\":\"Nominal policy rate and the government-bond liquidity premium co-move strongly over time\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1 (correlation 78%, 1955-2010)\"},{\"claim\":\"Lower nominal rates raise bank leverage and lower risk premia and Sharpe ratios\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"High-rate policy lowers valuations despite a lower real rate, and affects volatility\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Forward guidance raises asset prices ahead of liftoff; small timing changes cause large price responses\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Greenspan put stabilizes prices locally but leads to larger crashes and volatility later\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"},{\"claim\":\"Positive nominal rate shock lowers investment and growth and permanently reduces output\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 7-8\"},{\"claim\":\"Model generates an upward-sloping term structure and a term premium that rises after rate shocks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"},{\"claim\":\"Equilibrium characterization and analytical expressions for key mechanisms\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Appendix A\"}],\"structural_summary\":\"The paper opens with stylized facts and policy debates suggesting a risk premium channel of monetary policy, then states that it builds a model of that channel. A single motivating figure supports the key mechanism linking the nominal rate to the liquidity premium. The introduction previews numerically solved results in order: static policy rules, liquidity holdings, forward guidance, Greenspan put, and an extended production model with impulse responses and term structure. A roadmap follows. The literature has its own section. The model and equilibrium sections derive analytical characterizations, with proofs in appendices. The results section calibrates the model and presents results mainly through figures. A further section adds production and nominal rate shocks. The conclusion restates the mechanism and suggests unconventional policy analysis for future work.\",\"writing_profile_keywords\":[\"continuous-time general equilibrium\",\"heterogeneous agents\",\"liquidity premium mechanism\",\"global numerical solution\",\"figure-driven results\",\"policy experiments\",\"impulse responses\",\"motivating stylized fact\"],\"confidence\":0.85,\"notes\":\"The genre is theoretical with a calibration/simulation method. The only empirical content is the motivating Figure 1. No robustness or sensitivity analysis is visible in the excerpts. Abstract sentence 0 combines the research question with the contribution.\"}", "json": {"genre": "theoretical", "subfield": "macro_finance", "subfield_secondary": "asset_pricing", "primary_method": "calibration_simulation", "identification": {"estimand": "not applicable (theoretical model; equilibrium effects of nominal rate policy rules on leverage, risk premia, asset prices and growth)", "strategy": "Two-agent continuous-time general equilibrium model in which a liquidity premium links the nominal rate to the cost of leverage. Solved globally by Chebyshev collocation. Policy rules are compared across equilibria; impulse responses are computed in an extended production model.", "key_assumptions": ["Banks face funding shocks that force costly asset liquidation unless they hold liquid buffers", "Central bank controls the liquidity premium via the nominal rate and the supply of liquid assets", "Heterogeneous risk aversion between banks and depositors"], "threats_addressed": [], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 4}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "other"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "policy_or_broader_implication"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "separate_section"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Only two tables, both listing calibrated parameters with their symbols and values. Notes link each table to the sections and figures it generates. The extended-model table lists only the parameters that differ from the baseline.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'arguably' when interpreting historical episodes", "'consistent with' when linking model output to empirical findings", "'potentially greater instability'"]}, "claim_evidence": [{"claim": "Nominal policy rate and the government-bond liquidity premium co-move strongly over time", "evidence_type": "figure", "evidence_ref": "Figure 1 (correlation 78%, 1955-2010)"}, {"claim": "Lower nominal rates raise bank leverage and lower risk premia and Sharpe ratios", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "High-rate policy lowers valuations despite a lower real rate, and affects volatility", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Forward guidance raises asset prices ahead of liftoff; small timing changes cause large price responses", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Greenspan put stabilizes prices locally but leads to larger crashes and volatility later", "evidence_type": "figure", "evidence_ref": "Figure 6"}, {"claim": "Positive nominal rate shock lowers investment and growth and permanently reduces output", "evidence_type": "figure", "evidence_ref": "Figures 7-8"}, {"claim": "Model generates an upward-sloping term structure and a term premium that rises after rate shocks", "evidence_type": "figure", "evidence_ref": "Figure 9"}, {"claim": "Equilibrium characterization and analytical expressions for key mechanisms", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Appendix A"}], "structural_summary": "The paper opens with stylized facts and policy debates suggesting a risk premium channel of monetary policy, then states that it builds a model of that channel. A single motivating figure supports the key mechanism linking the nominal rate to the liquidity premium. The introduction previews numerically solved results in order: static policy rules, liquidity holdings, forward guidance, Greenspan put, and an extended production model with impulse responses and term structure. A roadmap follows. The literature has its own section. The model and equilibrium sections derive analytical characterizations, with proofs in appendices. The results section calibrates the model and presents results mainly through figures. A further section adds production and nominal rate shocks. The conclusion restates the mechanism and suggests unconventional policy analysis for future work.", "writing_profile_keywords": ["continuous-time general equilibrium", "heterogeneous agents", "liquidity premium mechanism", "global numerical solution", "figure-driven results", "policy experiments", "impulse responses", "motivating stylized fact"], "confidence": 0.85, "notes": "The genre is theoretical with a calibration/simulation method. The only empirical content is the motivating Figure 1. No robustness or sensitivity analysis is visible in the excerpts. Abstract sentence 0 combines the research question with the contribution."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.58, "usage": {"input": 7341, "output": 2304, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076238, "request_id": "req_011Cggpp9jdXgcMqcfjA3nWQ", "cached": false, "ts": "2026-11-03T23:00:22+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/deedd6eb76f35d9eb54d790f4fbb57831883419abb05f25854aabc4d167f5aad.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"asset_pricing (life-cycle portfolio choice)\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Optimal default-fund DC equity share and the welfare gain (certainty-equivalent retirement consumption) relative to a 100-minus-age default, within a calibrated life-cycle model.\",\"strategy\":\"Registry data documents how passive and active investors differ. A life-cycle model with endogenous opt-out and costly stock market participation is calibrated by matching data moments. Counterfactual default designs are then simulated.\",\"key_assumptions\":[\"Model structure (preferences, labor income process, return process) is correctly specified.\",\"Participation and opt-out cost heterogeneity is captured by the assumed joint distribution.\",\"Partial-equilibrium setting.\"],\"threats_addressed\":[\"model misspecification of equity returns (low premium, skewness)\",\"suboptimal portfolio choice or frictions outside the pension system\",\"endogenous composition of default investors under alternative designs\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":23,\"role\":\"theory\"},{\"idx\":24,\"role\":\"results\"},{\"idx\":25,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Notes state the reporting currency (SEK) and the SEK/USD exchange rate, and define investor categories. Model tables report parameters, matched moments, and regressions run on simulated data, with simulation sizes (economies, investors) given.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"model-conditional phrasing (according to the model, our model suggests)\",\"'may' and 'seem' for motivating claims\",\"partial-equilibrium qualifier\",\"'we believe' in policy recommendation\"]},\"claim_evidence\":[{\"claim\":\"Passive pension investors are much less likely to hold equity outside the pension system than active investors.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Table II\"},{\"claim\":\"The calibrated model reproduces observed shares of passive and nonparticipating investors and their wealth and income profiles.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Figures 1-2\"},{\"claim\":\"Optimal DC equity share varies widely across default investors and with past stock returns.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 3-4, Table VI\"},{\"claim\":\"Full customization yields an average 1.5% welfare gain over 100-minus-age, and a rule of thumb attains 0.9%.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The rule of thumb reduces the share of opt-out investors by about 16.6 percentage points.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"}],\"structural_summary\":\"The introduction motivates the topic with the shift to DC plans and investor inertia, then states a dual empirical and theoretical contribution. It previews registry-data facts on how passive and active investors differ, model-based optimal allocations, a rule of thumb, and welfare magnitudes, and closes with robustness. The body describes the institution (Swedish pension system), then the data, then descriptive regressions, then the calibrated life-cycle model with endogenous opt-out, then counterfactual optimal-design and welfare analysis with an embedded robustness subsection. The conclusion restates the gains and encourages policymakers to adopt customized defaults.\",\"writing_profile_keywords\":[\"normative design\",\"life-cycle model\",\"calibration to registry data\",\"welfare certainty equivalent\",\"rule of thumb\",\"policy recommendation\",\"Swedish pension\"],\"confidence\":0.8,\"notes\":\"Many empty numbered section headings (idx 5-22) appear to be parsing artifacts from a matrix in the appendix-like model text. These were left unassigned, except idx 23, which covers the simulation method and model fit and was assigned to theory. No explicit roadmap appears in the visible intro excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "asset_pricing (life-cycle portfolio choice)", "primary_method": "calibration_simulation", "identification": {"estimand": "Optimal default-fund DC equity share and the welfare gain (certainty-equivalent retirement consumption) relative to a 100-minus-age default, within a calibrated life-cycle model.", "strategy": "Registry data documents how passive and active investors differ. A life-cycle model with endogenous opt-out and costly stock market participation is calibrated by matching data moments. Counterfactual default designs are then simulated.", "key_assumptions": ["Model structure (preferences, labor income process, return process) is correctly specified.", "Participation and opt-out cost heterogeneity is captured by the assumed joint distribution.", "Partial-equilibrium setting."], "threats_addressed": ["model misspecification of equity returns (low premium, skewness)", "suboptimal portfolio choice or frictions outside the pension system", "endogenous composition of default investors under alternative designs"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_data_or_measure", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "theory"}, {"idx": 23, "role": "theory"}, {"idx": 24, "role": "results"}, {"idx": 25, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["sensitivity_to_parameters", "alternative_specification", "subsample"]}, "tables": {"convention_notes": "Notes state the reporting currency (SEK) and the SEK/USD exchange rate, and define investor categories. Model tables report parameters, matched moments, and regressions run on simulated data, with simulation sizes (economies, investors) given.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["model-conditional phrasing (according to the model, our model suggests)", "'may' and 'seem' for motivating claims", "partial-equilibrium qualifier", "'we believe' in policy recommendation"]}, "claim_evidence": [{"claim": "Passive pension investors are much less likely to hold equity outside the pension system than active investors.", "evidence_type": "table", "evidence_ref": "Table I, Table II"}, {"claim": "The calibrated model reproduces observed shares of passive and nonparticipating investors and their wealth and income profiles.", "evidence_type": "table", "evidence_ref": "Table V, Figures 1-2"}, {"claim": "Optimal DC equity share varies widely across default investors and with past stock returns.", "evidence_type": "figure", "evidence_ref": "Figures 3-4, Table VI"}, {"claim": "Full customization yields an average 1.5% welfare gain over 100-minus-age, and a rule of thumb attains 0.9%.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The rule of thumb reduces the share of opt-out investors by about 16.6 percentage points.", "evidence_type": "table", "evidence_ref": "Table VII"}], "structural_summary": "The introduction motivates the topic with the shift to DC plans and investor inertia, then states a dual empirical and theoretical contribution. It previews registry-data facts on how passive and active investors differ, model-based optimal allocations, a rule of thumb, and welfare magnitudes, and closes with robustness. The body describes the institution (Swedish pension system), then the data, then descriptive regressions, then the calibrated life-cycle model with endogenous opt-out, then counterfactual optimal-design and welfare analysis with an embedded robustness subsection. The conclusion restates the gains and encourages policymakers to adopt customized defaults.", "writing_profile_keywords": ["normative design", "life-cycle model", "calibration to registry data", "welfare certainty equivalent", "rule of thumb", "policy recommendation", "Swedish pension"], "confidence": 0.8, "notes": "Many empty numbered section headings (idx 5-22) appear to be parsing artifacts from a matrix in the appendix-like model text. These were left unassigned, except idx 23, which covers the simulation method and model fit and was assigned to theory. No explicit roadmap appears in the visible intro excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.21, "usage": {"input": 8039, "output": 2065, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07425, "request_id": "req_011Cggq2K2KFt1hZGjCok99m", "cached": false, "ts": "2026-11-03T23:03:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/dfe5433dd5e48f558182c58508b2cd0c9a7cd35eb7bbcad1e409a04ea9e2e995.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"household_finance; peer effects in financial misconduct\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Effect of new coworkers' prior misconduct history on an advisor's probability of committing misconduct in the three years after a merger (coworker/peer influence)\",\"strategy\":\"Coworker changes induced by mergers of financial advisory firms. Merger-firm fixed effects in conditional logit and linear models compare branches of the same premerger firm that are combined with high- versus low-misconduct branches. Coworker misconduct is measured before the merger to avoid the reflection problem.\",\"key_assumptions\":[\"Conditional on a merger, firms do not select which branches or advisors to combine based on misconduct-related characteristics\",\"Merger-firm fixed effects absorb firm-level and time-specific endogeneity of mergers\",\"Premerger coworker misconduct is predetermined relative to the advisor's postmerger behavior\"],\"threats_addressed\":[\"selection of advisors into branches\",\"correlated branch environment\",\"reflection problem\",\"endogenous merger motives\",\"premerger trends\",\"geographic confounding\",\"complex products or productivity\",\"regulatory actions\",\"supervisor influence versus peer influence\",\"individual selection following mergers\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"subsample\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes report odds ratios from conditional logit models grouped by merger-firm, with linear-model columns alongside. Notes state the unit of observation (advisor or branch), the sample (477 mergers, 1999–2011), the three-year pre- and postmerger windows, and variable definitions.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'consistent with' for the mechanism interpretation\",\"'may occur' for candidate channels\",\"'attempt to' isolate peer versus supervisor effects\"]},\"claim_evidence\":[{\"claim\":\"Advisors are 37% more likely to commit misconduct when their merger-introduced coworkers have a misconduct history\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Coworker influence is asymmetric: misconduct spreads but there is no significant evidence that good conduct spreads\",\"evidence_type\":\"table\",\"evidence_ref\":\"Section II.B\"},{\"claim\":\"Placebo reassignment of branches yields odds ratios well below the actual estimate\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"No differential premerger trends in misconduct by introduced-branch misconduct status\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Selection bias tests do not undermine the identification\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The effect is stronger among demographically similar coworkers\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"Implied social multiplier of 1.59\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Conclusion\"},{\"claim\":\"Results are not driven by branch environment, monitoring, or supervision differences\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI\"}],\"structural_summary\":\"The introduction states the contagion finding up front and sketches two channels (social learning and social utility). It then gives a stylized merger example to explain identification and shows how timing of the coworker measure avoids the reflection problem. Results, including a magnitude and the asymmetry-based mechanism inference, are previewed next, followed by positioning against crime peer-effects, firm-level misconduct, and advisory-industry literatures. The body moves from data to the design and main results, then to a series of threat-specific sections: selection tests, similarity-based peer effects, confounders, and supervisors. The conclusion restates the magnitude and social multiplier and draws policy implications for penalties and disclosure.\",\"writing_profile_keywords\":[\"peer effects\",\"quasi-experiment via mergers\",\"merger-firm fixed effects\",\"conditional logit odds ratios\",\"Manski reflection problem\",\"threat-by-threat sections\",\"policy implications\",\"stylized illustrative example\"],\"confidence\":0.82,\"notes\":\"Section idx 3 has a garbled heading (an equation fragment); it contains the main results and placebo/trend subsections. Standard errors and significance conventions are not visible in the excerpts. Figure 1 uses cluster bootstrap placebo draws.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "household_finance; peer effects in financial misconduct", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Effect of new coworkers' prior misconduct history on an advisor's probability of committing misconduct in the three years after a merger (coworker/peer influence)", "strategy": "Coworker changes induced by mergers of financial advisory firms. Merger-firm fixed effects in conditional logit and linear models compare branches of the same premerger firm that are combined with high- versus low-misconduct branches. Coworker misconduct is measured before the merger to avoid the reflection problem.", "key_assumptions": ["Conditional on a merger, firms do not select which branches or advisors to combine based on misconduct-related characteristics", "Merger-firm fixed effects absorb firm-level and time-specific endogeneity of mergers", "Premerger coworker misconduct is predetermined relative to the advisor's postmerger behavior"], "threats_addressed": ["selection of advisors into branches", "correlated branch environment", "reflection problem", "endogenous merger motives", "premerger trends", "geographic confounding", "complex products or productivity", "regulatory actions", "supervisor influence versus peer influence", "individual selection following mergers"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_data_or_measure", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_specification", "subsample", "falsification", "other"]}, "tables": {"convention_notes": "Notes report odds ratios from conditional logit models grouped by merger-firm, with linear-model columns alongside. Notes state the unit of observation (advisor or branch), the sample (477 mergers, 1999–2011), the three-year pre- and postmerger windows, and variable definitions.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'consistent with' for the mechanism interpretation", "'may occur' for candidate channels", "'attempt to' isolate peer versus supervisor effects"]}, "claim_evidence": [{"claim": "Advisors are 37% more likely to commit misconduct when their merger-introduced coworkers have a misconduct history", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Coworker influence is asymmetric: misconduct spreads but there is no significant evidence that good conduct spreads", "evidence_type": "table", "evidence_ref": "Section II.B"}, {"claim": "Placebo reassignment of branches yields odds ratios well below the actual estimate", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "No differential premerger trends in misconduct by introduced-branch misconduct status", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Selection bias tests do not undermine the identification", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The effect is stronger among demographically similar coworkers", "evidence_type": "none_stated", "evidence_ref": "Section IV"}, {"claim": "Implied social multiplier of 1.59", "evidence_type": "none_stated", "evidence_ref": "Conclusion"}, {"claim": "Results are not driven by branch environment, monitoring, or supervision differences", "evidence_type": "table", "evidence_ref": "Tables V, VI"}], "structural_summary": "The introduction states the contagion finding up front and sketches two channels (social learning and social utility). It then gives a stylized merger example to explain identification and shows how timing of the coworker measure avoids the reflection problem. Results, including a magnitude and the asymmetry-based mechanism inference, are previewed next, followed by positioning against crime peer-effects, firm-level misconduct, and advisory-industry literatures. The body moves from data to the design and main results, then to a series of threat-specific sections: selection tests, similarity-based peer effects, confounders, and supervisors. The conclusion restates the magnitude and social multiplier and draws policy implications for penalties and disclosure.", "writing_profile_keywords": ["peer effects", "quasi-experiment via mergers", "merger-firm fixed effects", "conditional logit odds ratios", "Manski reflection problem", "threat-by-threat sections", "policy implications", "stylized illustrative example"], "confidence": 0.82, "notes": "Section idx 3 has a garbled heading (an equation fragment); it contains the main results and placebo/trend subsections. Standard errors and significance conventions are not visible in the excerpts. Figure 1 uses cluster bootstrap placebo draws."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.08, "usage": {"input": 7681, "output": 2413, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079778, "request_id": "req_011Cggpwzx26R6Ci7weNs6Tb", "cached": false, "ts": "2026-11-03T23:02:09+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e0bcf3fce9e0f322100133df1798c94ad459737b3d358018fa2cf2aabd2392ca.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Spot risk-neutral volatility, jump intensity and jump size distribution (left tail jump variation) extracted daily from short-dated options; predictive association of left tail measure with future excess returns\",\"strategy\":\"Seminonparametric 'structural calibration' using a new short-maturity asymptotic option pricing approximation with no-arbitrage constraints; daily estimation with confidence regions; predictive regressions; illustrative event windows around FOMC announcements\",\"key_assumptions\":[\"Arbitrage-free markets implying existence of a risk-neutral measure\",\"Volatility and jump intensity change little in expectation over short horizons\",\"Parametric restrictions only on the jump size distribution are adequate\"],\"threats_addressed\":[\"Noise or idiosyncratic pricing in short-dated options\",\"Liquidity/spread shifts around announcements\",\"Misspecification of volatility and jump intensity dynamics\",\"Noisy deep OTM quotes (data filters)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"other\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"identification_defense\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_method\",\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"},{\"idx\":13,\"role\":\"appendix\"},{\"idx\":14,\"role\":\"appendix\"},{\"idx\":15,\"role\":\"appendix\"},{\"idx\":16,\"role\":\"appendix\"},{\"idx\":17,\"role\":\"appendix\"},{\"idx\":18,\"role\":\"appendix\"},{\"idx\":19,\"role\":\"appendix\"},{\"idx\":20,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"appendix\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Few tables; descriptive table of contract shares by tenor and moneyness with sample period stated; predictive regression table reporting coefficients with Newey-West t-statistics in parentheses, lags matched to forecast horizon. Most evidence is carried by figures.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggestive of'\",\"'may have'\",\"'to the extent that'\",\"conditional framing ('if this is indeed a robust finding')\",\"explicit acknowledgement of potential concerns\"]},\"claim_evidence\":[{\"claim\":\"Trading in short-dated S&P 500 options has grown sharply and is skewed toward OTM contracts\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I; Figures 3-5\"},{\"claim\":\"Isolated shifts in the left risk-neutral jump tail occur and are visible in short-dated but not regular option prices\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Figure A.1\"},{\"claim\":\"FOMC announcements move priced tail risk differently from volatility\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2; Figure B.1\"},{\"claim\":\"Fitting short-dated options requires pronounced time variation in the jump size distribution\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 7, 9, 11; Appendix F-G\"},{\"claim\":\"Left jump tail variation is not tightly linked to diffusive volatility\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 12-14\"},{\"claim\":\"Left tail jump measure spills into pricing errors of standard parametric models on regular options\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 15; Appendix H\"},{\"claim\":\"Left jump tail measure helps predict future excess equity returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"}],\"structural_summary\":\"Opens with market-growth motivation for weekly options and asks why they matter, then previews a new short-maturity pricing approximation and seminonparametric 'structural calibration' contrasted with standard calibration and prior tail work. Illustrative figures (a 2012 tail shift, FOMC intraday reactions) motivate time-varying jump distributions, with noise and liquidity concerns addressed. Body covers data, setup, method, estimation, model comparison across jump specifications, characterization of time-varying jump risk, and asset pricing implications (spillovers to regular options, return predictability). Conclusion restates findings and proposes parametric modeling of tail dynamics as future work. Extensive appendices hold filters, pricing formulas and additional estimation results.\",\"writing_profile_keywords\":[\"option-implied tail risk\",\"seminonparametric estimation\",\"continuous-time finance\",\"figure-heavy evidence\",\"illustrative event cases\",\"no-arbitrage\",\"econometric inference\",\"model specification testing\"],\"confidence\":0.8,\"notes\":\"Primary method is labeled structural_estimation because estimation is option-based structural calibration; return predictability is secondary. Robustness appears mainly as appendix checks on data quality and alternative jump specifications.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "structural_estimation", "identification": {"estimand": "Spot risk-neutral volatility, jump intensity and jump size distribution (left tail jump variation) extracted daily from short-dated options; predictive association of left tail measure with future excess returns", "strategy": "Seminonparametric 'structural calibration' using a new short-maturity asymptotic option pricing approximation with no-arbitrage constraints; daily estimation with confidence regions; predictive regressions; illustrative event windows around FOMC announcements", "key_assumptions": ["Arbitrage-free markets implying existence of a risk-neutral measure", "Volatility and jump intensity change little in expectation over short horizons", "Parametric restrictions only on the jump size distribution are adequate"], "threats_addressed": ["Noise or idiosyncratic pricing in short-dated options", "Liquidity/spread shifts around announcements", "Misspecification of volatility and jump intensity dynamics", "Noisy deep OTM quotes (data filters)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "other"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "identification_defense"}, {"paragraph_index": 10, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "roadmap"}], "contribution_paragraph_index": 2, "contribution_types": ["new_method", "new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}, {"idx": 13, "role": "appendix"}, {"idx": 14, "role": "appendix"}, {"idx": 15, "role": "appendix"}, {"idx": 16, "role": "appendix"}, {"idx": 17, "role": "appendix"}, {"idx": 18, "role": "appendix"}, {"idx": 19, "role": "appendix"}, {"idx": 20, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": false}, "robustness": {"location": "appendix", "types": ["alternative_specification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Few tables; descriptive table of contract shares by tenor and moneyness with sample period stated; predictive regression table reporting coefficients with Newey-West t-statistics in parentheses, lags matched to forecast horizon. Most evidence is carried by figures.", "units_stated_in_notes": true, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggestive of'", "'may have'", "'to the extent that'", "conditional framing ('if this is indeed a robust finding')", "explicit acknowledgement of potential concerns"]}, "claim_evidence": [{"claim": "Trading in short-dated S&P 500 options has grown sharply and is skewed toward OTM contracts", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I; Figures 3-5"}, {"claim": "Isolated shifts in the left risk-neutral jump tail occur and are visible in short-dated but not regular option prices", "evidence_type": "figure", "evidence_ref": "Figure 1; Figure A.1"}, {"claim": "FOMC announcements move priced tail risk differently from volatility", "evidence_type": "figure", "evidence_ref": "Figure 2; Figure B.1"}, {"claim": "Fitting short-dated options requires pronounced time variation in the jump size distribution", "evidence_type": "figure", "evidence_ref": "Figures 7, 9, 11; Appendix F-G"}, {"claim": "Left jump tail variation is not tightly linked to diffusive volatility", "evidence_type": "figure", "evidence_ref": "Figures 12-14"}, {"claim": "Left tail jump measure spills into pricing errors of standard parametric models on regular options", "evidence_type": "figure", "evidence_ref": "Figure 15; Appendix H"}, {"claim": "Left jump tail measure helps predict future excess equity returns", "evidence_type": "table", "evidence_ref": "Table II"}], "structural_summary": "Opens with market-growth motivation for weekly options and asks why they matter, then previews a new short-maturity pricing approximation and seminonparametric 'structural calibration' contrasted with standard calibration and prior tail work. Illustrative figures (a 2012 tail shift, FOMC intraday reactions) motivate time-varying jump distributions, with noise and liquidity concerns addressed. Body covers data, setup, method, estimation, model comparison across jump specifications, characterization of time-varying jump risk, and asset pricing implications (spillovers to regular options, return predictability). Conclusion restates findings and proposes parametric modeling of tail dynamics as future work. Extensive appendices hold filters, pricing formulas and additional estimation results.", "writing_profile_keywords": ["option-implied tail risk", "seminonparametric estimation", "continuous-time finance", "figure-heavy evidence", "illustrative event cases", "no-arbitrage", "econometric inference", "model specification testing"], "confidence": 0.8, "notes": "Primary method is labeled structural_estimation because estimation is option-based structural calibration; return predictability is secondary. Robustness appears mainly as appendix checks on data quality and alternative jump specifications."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 21.31, "usage": {"input": 13739, "output": 2515, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.10605, "request_id": "req_011Cggpbc73PJo868918kDex", "cached": false, "ts": "2026-11-03T22:57:34+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e14a6041e27e1f7ba1575e3f32d9d015fb374b73e519443959f1bc653f4ae34b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between investor motives (intrinsic social preferences from a trust game, social signaling, return and risk expectations) and the probability and share of holding SRI equity funds\",\"strategy\":\"Probit and OLS cross-sectional regressions that link administrative brokerage holdings to survey answers and incentivized experimental measures, with a battery of controls\",\"key_assumptions\":[\"Second-mover behavior in an anonymous one-shot trust game cleanly measures intrinsic social preferences, free of reputation or image concerns\",\"Survey responses on expectations and signaling reflect true beliefs and behavior\"],\"threats_addressed\":[\"Strategic reputation or social image contamination of the preference measure, handled by an anonymous one-shot experiment\",\"Investors who regard SRI as a marketing trick (excluded in appendix)\",\"Inattentive experiment participants who failed practice questions (excluded in appendix)\",\"Inconsistent respondents (excluded in appendix)\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"},{\"idx\":13,\"role\":\"robustness\"},{\"idx\":14,\"role\":\"robustness\"},{\"idx\":15,\"role\":\"robustness\"},{\"idx\":16,\"role\":\"robustness\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"appendix\",\"types\":[\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes state whether each specification reports probit marginal effects or OLS coefficients and define dependent variables. They point to an appendix table of variable definitions, give robust standard errors in parentheses, and mark significance with stars at the 10/5/1% levels.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"robust\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggests\",\"indicates\",\"perhaps\",\"may\",\"explicit disclaimer that causal effects cannot be addressed\",\"discussion of generalizability beyond Dutch sample\"]},\"claim_evidence\":[{\"claim\":\"Investors with stronger intrinsic social preferences (trust game second-mover returns) are substantially more likely to hold SRI equity funds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"More frequent talking about investments (social signaling) is associated with SRI holding\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Table IV\"},{\"claim\":\"Most SRI investors expect SRI funds to underperform conventional funds\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Risk perceptions of SRI funds are unrelated to holding them\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2; Table III\"},{\"claim\":\"SRI investors donate more to charity, so SRI is not a substitute for donations\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Social preferences show few demographic covariates\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Dutch environmental attitudes resemble those in other countries, supporting generalizability\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"World Values Survey, ERNOP figures in conclusion\"}],\"structural_summary\":\"Opens with SRI growth framed as a finance puzzle and lists competing motives: social preferences, financial beliefs and signaling. It justifies an experimental measure of preferences, then previews results with magnitudes and links them to asset pricing via a taste-versus-beliefs argument, followed by a roadmap. Data and variable sections precede descriptive characteristics, the main regression analysis of motives, covariates of preferences and the charity-substitution test. The conclusion restates findings, draws asset-pricing implications and discusses generalizability. Appendix sample exclusions serve as robustness checks.\",\"writing_profile_keywords\":[\"administrative data linked to survey\",\"incentivized experiment\",\"trust game\",\"probit marginal effects\",\"social preferences\",\"social signaling\",\"household portfolio choice\",\"associational framing\"],\"confidence\":0.8,\"notes\":\"No explicit contribution paragraph is visible, so contribution_paragraph_index is set to -1. The causal-sounding wording ('determining', 'effect') is tempered by an explicit statement that causality cannot be addressed. Robustness checks are carried out as appendix sample exclusions.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "household_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between investor motives (intrinsic social preferences from a trust game, social signaling, return and risk expectations) and the probability and share of holding SRI equity funds", "strategy": "Probit and OLS cross-sectional regressions that link administrative brokerage holdings to survey answers and incentivized experimental measures, with a battery of controls", "key_assumptions": ["Second-mover behavior in an anonymous one-shot trust game cleanly measures intrinsic social preferences, free of reputation or image concerns", "Survey responses on expectations and signaling reflect true beliefs and behavior"], "threats_addressed": ["Strategic reputation or social image contamination of the preference measure, handled by an anonymous one-shot experiment", "Investors who regard SRI as a marketing trick (excluded in appendix)", "Inattentive experiment participants who failed practice questions (excluded in appendix)", "Inconsistent respondents (excluded in appendix)"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}, {"idx": 13, "role": "robustness"}, {"idx": 14, "role": "robustness"}, {"idx": 15, "role": "robustness"}, {"idx": 16, "role": "robustness"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "appendix", "types": ["subsample", "alternative_specification"]}, "tables": {"convention_notes": "Notes state whether each specification reports probit marginal effects or OLS coefficients and define dependent variables. They point to an appendix table of variable definitions, give robust standard errors in parentheses, and mark significance with stars at the 10/5/1% levels.", "units_stated_in_notes": false, "standard_errors_description": "robust"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggests", "indicates", "perhaps", "may", "explicit disclaimer that causal effects cannot be addressed", "discussion of generalizability beyond Dutch sample"]}, "claim_evidence": [{"claim": "Investors with stronger intrinsic social preferences (trust game second-mover returns) are substantially more likely to hold SRI equity funds", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "More frequent talking about investments (social signaling) is associated with SRI holding", "evidence_type": "table", "evidence_ref": "Table III; Table IV"}, {"claim": "Most SRI investors expect SRI funds to underperform conventional funds", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Risk perceptions of SRI funds are unrelated to holding them", "evidence_type": "figure", "evidence_ref": "Figure 2; Table III"}, {"claim": "SRI investors donate more to charity, so SRI is not a substitute for donations", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Social preferences show few demographic covariates", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Dutch environmental attitudes resemble those in other countries, supporting generalizability", "evidence_type": "descriptive_statistic", "evidence_ref": "World Values Survey, ERNOP figures in conclusion"}], "structural_summary": "Opens with SRI growth framed as a finance puzzle and lists competing motives: social preferences, financial beliefs and signaling. It justifies an experimental measure of preferences, then previews results with magnitudes and links them to asset pricing via a taste-versus-beliefs argument, followed by a roadmap. Data and variable sections precede descriptive characteristics, the main regression analysis of motives, covariates of preferences and the charity-substitution test. The conclusion restates findings, draws asset-pricing implications and discusses generalizability. Appendix sample exclusions serve as robustness checks.", "writing_profile_keywords": ["administrative data linked to survey", "incentivized experiment", "trust game", "probit marginal effects", "social preferences", "social signaling", "household portfolio choice", "associational framing"], "confidence": 0.8, "notes": "No explicit contribution paragraph is visible, so contribution_paragraph_index is set to -1. The causal-sounding wording ('determining', 'effect') is tempered by an explicit statement that causality cannot be addressed. Robustness checks are carried out as appendix sample exclusions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.5, "usage": {"input": 8190, "output": 2262, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078794, "request_id": "req_011CggpknVcEJitikcjo7wVj", "cached": false, "ts": "2026-11-03T22:59:36+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e1636c3eeaeaba77158b1e7df9bfddf70df78c48c82a5e900fe71401d07292e1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Causal effect of property value (mortgage debt), holding home equity fixed, and of home equity, holding property value fixed, on the stock share of liquid wealth and on stock participation\",\"strategy\":\"Three IV designs: (1) state house price indices at the current year and at the year of purchase; (2) national house prices interacted with state housing supply elasticity (Saiz 2010); (3) panel changes around home purchase, instrumented with the state price index at purchase and with household fixed effects\",\"key_assumptions\":[\"State house price variation is uncorrelated with other determinants of portfolio choice once fixed effects are included\",\"National price shocks interacted with supply elasticity affect portfolios only through housing\",\"Time-varying unobservables do not drive portfolio changes around home purchase\"],\"threats_addressed\":[\"omitted variable bias from local labor market conditions\",\"selection bias from endogenous timing of home purchase\",\"unobserved labor income correlated with property value\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"appendix\"},{\"idx\":13,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes list the fixed effects included (age, current year, purchase year, state) and controls such as a 10-piece liquid wealth spline. IV tables report first-stage OLS estimates alongside the second stage. The model table reports simulated optimal stock shares across parameter grids.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"acknowledges each design is not definitive alone\",\"stability across designs used as a robustness argument\",\"'may have' for broader implications\",\"'suggests' for interpretation\"]},\"claim_evidence\":[{\"claim\":\"Holding home equity fixed, higher property value lowers the stock share through illiquidity, risk exposure and lower lifetime wealth\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendix A\"},{\"claim\":\"A $10,000 rise in property value cuts the stock share by about 4%, while a $10,000 rise in home equity raises it by about 4.8%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Elasticities of the stock share are -0.2 with respect to mortgage debt and 0.3 with respect to home equity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Buying a house that costs $10,000 more lowers the stock share by about 5.2% in the year after purchase\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"State house price variation yields similar estimates in the first design\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Model calibrations imply sizable stock share responses that vary with mobility and the stock-housing return correlation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"}],\"structural_summary\":\"The introduction opens with a puzzle: theory predicts that housing crowds out risky assets, yet prior evidence finds no systematic link. The paper resolves this by separating property value from home equity in a stylized two-period model, then tests the model with three successively refined IV designs using SIPP data. The intro previews magnitudes and walks through the identification threats each design addresses. A short conclusion restates the elasticities and argues that stability across designs makes confounding unlikely. Appendices cover proofs, model extensions and numerical methods.\",\"writing_profile_keywords\":[\"puzzle resolution\",\"theory-guided empirics\",\"multiple research designs\",\"instrumental variables\",\"elasticity reporting\",\"household portfolios\"],\"confidence\":0.8,\"notes\":\"Section 5 is labeled as summary statistics but contains the main results subsections. The excerpts do not show how standard errors or significance are reported. The intro briefly mentions implications for welfare and policy, but the conclusion excerpt does not.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "asset_pricing", "primary_method": "instrumental_variables", "identification": {"estimand": "Causal effect of property value (mortgage debt), holding home equity fixed, and of home equity, holding property value fixed, on the stock share of liquid wealth and on stock participation", "strategy": "Three IV designs: (1) state house price indices at the current year and at the year of purchase; (2) national house prices interacted with state housing supply elasticity (Saiz 2010); (3) panel changes around home purchase, instrumented with the state price index at purchase and with household fixed effects", "key_assumptions": ["State house price variation is uncorrelated with other determinants of portfolio choice once fixed effects are included", "National price shocks interacted with supply elasticity affect portfolios only through housing", "Time-varying unobservables do not drive portfolio changes around home purchase"], "threats_addressed": ["omitted variable bias from local labor market conditions", "selection bias from endogenous timing of home purchase", "unobserved labor income correlated with property value"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_identification", "new_theory_or_model", "resolution_of_puzzle"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "appendix"}, {"idx": 13, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "Table notes list the fixed effects included (age, current year, purchase year, state) and controls such as a 10-piece liquid wealth spline. IV tables report first-stage OLS estimates alongside the second stage. The model table reports simulated optimal stock shares across parameter grids.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["acknowledges each design is not definitive alone", "stability across designs used as a robustness argument", "'may have' for broader implications", "'suggests' for interpretation"]}, "claim_evidence": [{"claim": "Holding home equity fixed, higher property value lowers the stock share through illiquidity, risk exposure and lower lifetime wealth", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendix A"}, {"claim": "A $10,000 rise in property value cuts the stock share by about 4%, while a $10,000 rise in home equity raises it by about 4.8%", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Elasticities of the stock share are -0.2 with respect to mortgage debt and 0.3 with respect to home equity", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Buying a house that costs $10,000 more lowers the stock share by about 5.2% in the year after purchase", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "State house price variation yields similar estimates in the first design", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Model calibrations imply sizable stock share responses that vary with mobility and the stock-housing return correlation", "evidence_type": "table", "evidence_ref": "Table I"}], "structural_summary": "The introduction opens with a puzzle: theory predicts that housing crowds out risky assets, yet prior evidence finds no systematic link. The paper resolves this by separating property value from home equity in a stylized two-period model, then tests the model with three successively refined IV designs using SIPP data. The intro previews magnitudes and walks through the identification threats each design addresses. A short conclusion restates the elasticities and argues that stability across designs makes confounding unlikely. Appendices cover proofs, model extensions and numerical methods.", "writing_profile_keywords": ["puzzle resolution", "theory-guided empirics", "multiple research designs", "instrumental variables", "elasticity reporting", "household portfolios"], "confidence": 0.8, "notes": "Section 5 is labeled as summary statistics but contains the main results subsections. The excerpts do not show how standard errors or significance are reported. The intro briefly mentions implications for welfare and policy, but the conclusion excerpt does not."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.58, "usage": {"input": 6950, "output": 2256, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073714, "request_id": "req_011CggpaJcYT4sHsYt5K6xW6", "cached": false, "ts": "2026-11-03T22:57:13+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e1d2d3958fbc0f4e81c15dcf4ddb2d8f3de6568557bd3cf3e40f1b4fa948d9ef.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of state antitakeover law coverage, specifically business combination laws, on firm outcomes, and whether prior estimates survive once legal and institutional context is controlled for\",\"strategy\":\"Re-estimates the staggered state-law difference-in-differences design. Compares a short regression with a long regression that adds controls for first-generation laws, other second-generation laws, court decisions, motivating (lobbying) firms and firm-level defenses. Includes an exact replication of Atanassov (2013) and simulations reported in the Internet Appendix.\",\"key_assumptions\":[\"Law adoption is exogenous for covered firms, except identifiable lobbying firms\",\"The added controls capture the relevant omitted legal context\",\"Takeover protection was low before business combination law adoption, which the paper argues is often violated\"],\"threats_addressed\":[\"omitted variables from other state laws\",\"preexisting firm-level defenses\",\"court decisions changing the legal regime\",\"endogenous adoption via lobbying firms\",\"selection through opt-outs and incorporation choice\",\"low test power\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"replication_or_correction\",\"new_data_or_measure\",\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"other\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"empirical_design\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"literature\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Three descriptive tables catalogue first- and second-generation laws and lobbying firms. Regression tables set short regressions against long regressions with added context controls. An appendix table lists 81 prior papers that use these laws. Notes define the variables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggesting\",\"tentatively identifying\",\"may best proceed inductively\",\"perhaps\",\"acknowledging that some prior results survive\"]},\"claim_evidence\":[{\"claim\":\"Most firms covered by business combination laws had earlier been covered by first-generation laws, so pre-period takeover protection was high\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I; Section II\"},{\"claim\":\"Adding legal-context controls makes the business combination law coefficient insignificant for four of seven outcomes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Business combination law effects are insignificant among firms with preexisting firm-level defenses\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Atanassov's innovation results disappear with context controls; poison pill laws instead relate negatively to patents\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Misspecified tests frequently produce spurious significant business combination law effects\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Internet Appendix simulations\"},{\"claim\":\"Laws were endogenous for identifiable lobbying firms\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Table III; Karpoff and Malatesta (1989), Gartman (2000)\"}],\"structural_summary\":\"Opens with endogeneity as a pervasive problem and the popularity of regulatory shocks as a remedy. Focuses on state antitakeover laws and lays out the standard identification construct. Long institutional sections show that other laws, firm defenses, court rulings and lobbying undermine that construct. These concerns are then recast as an omitted-variables problem, with new hand-collected data offered as the remedy. Replications contrast short and long regressions across prior studies, followed by an exact replication of an innovation paper. A separate section critiques a competing view on shadow poison pills. The paper closes with implications for future research and an argument for inductive inference.\",\"writing_profile_keywords\":[\"critique of identification\",\"replication\",\"institutional detail\",\"omitted variables\",\"legal context\",\"short vs long regression\",\"inductive inference\"],\"confidence\":0.8,\"notes\":\"Hybrid of methodological critique and replication; genre classified as replication_or_correction. Significance reporting conventions are not visible in the excerpts. Simulations are reported in the Internet Appendix.\"}", "json": {"genre": "replication_or_correction", "subfield": "corporate_finance", "subfield_secondary": "financial_econometrics", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of state antitakeover law coverage, specifically business combination laws, on firm outcomes, and whether prior estimates survive once legal and institutional context is controlled for", "strategy": "Re-estimates the staggered state-law difference-in-differences design. Compares a short regression with a long regression that adds controls for first-generation laws, other second-generation laws, court decisions, motivating (lobbying) firms and firm-level defenses. Includes an exact replication of Atanassov (2013) and simulations reported in the Internet Appendix.", "key_assumptions": ["Law adoption is exogenous for covered firms, except identifiable lobbying firms", "The added controls capture the relevant omitted legal context", "Takeover protection was low before business combination law adoption, which the paper argues is often violated"], "threats_addressed": ["omitted variables from other state laws", "preexisting firm-level defenses", "court decisions changing the legal regime", "endogenous adoption via lobbying firms", "selection through opt-outs and incorporation choice", "low test power"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}, {"paragraph_index": 7, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 2, "contribution_types": ["replication_or_correction", "new_data_or_measure", "new_evidence"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "other"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "empirical_design"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "literature"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Three descriptive tables catalogue first- and second-generation laws and lobbying firms. Regression tables set short regressions against long regressions with added context controls. An appendix table lists 81 prior papers that use these laws. Notes define the variables.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggesting", "tentatively identifying", "may best proceed inductively", "perhaps", "acknowledging that some prior results survive"]}, "claim_evidence": [{"claim": "Most firms covered by business combination laws had earlier been covered by first-generation laws, so pre-period takeover protection was high", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I; Section II"}, {"claim": "Adding legal-context controls makes the business combination law coefficient insignificant for four of seven outcomes", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Business combination law effects are insignificant among firms with preexisting firm-level defenses", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Atanassov's innovation results disappear with context controls; poison pill laws instead relate negatively to patents", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Misspecified tests frequently produce spurious significant business combination law effects", "evidence_type": "none_stated", "evidence_ref": "Internet Appendix simulations"}, {"claim": "Laws were endogenous for identifiable lobbying firms", "evidence_type": "prior_literature", "evidence_ref": "Table III; Karpoff and Malatesta (1989), Gartman (2000)"}], "structural_summary": "Opens with endogeneity as a pervasive problem and the popularity of regulatory shocks as a remedy. Focuses on state antitakeover laws and lays out the standard identification construct. Long institutional sections show that other laws, firm defenses, court rulings and lobbying undermine that construct. These concerns are then recast as an omitted-variables problem, with new hand-collected data offered as the remedy. Replications contrast short and long regressions across prior studies, followed by an exact replication of an innovation paper. A separate section critiques a competing view on shadow poison pills. The paper closes with implications for future research and an argument for inductive inference.", "writing_profile_keywords": ["critique of identification", "replication", "institutional detail", "omitted variables", "legal context", "short vs long regression", "inductive inference"], "confidence": 0.8, "notes": "Hybrid of methodological critique and replication; genre classified as replication_or_correction. Significance reporting conventions are not visible in the excerpts. Simulations are reported in the Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.16, "usage": {"input": 9962, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.083442, "request_id": "req_011Cggps5EMkbvL4oUNjSPRL", "cached": false, "ts": "2026-11-03T23:01:01+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e4a8e813ea4c83e311ef7b04396bb686d22003191b110c1edb8df2a5ddfde76b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Whether bank-specific CD funding dry-ups predict future bank performance conditional on public information, and whether funds shift toward high-quality banks under stress, as a test of adverse-selection versus informed-lender theories\",\"strategy\":\"Predictive panel regressions of future changes in ROA, impaired loans and size on dry-up dummies with current controls, plus event studies of stock returns, maturity shortening and rating downgrades, and probit models of reallocation\",\"key_assumptions\":[\"Future performance conditional on current public information proxies for quality unobserved by the market\",\"Impaired loans reflect loans extended before the dry-up and so are less prone to reverse causality\",\"CDs are unsecured and lenders are mostly money market funds, so asymmetric information concerns only issuer creditworthiness and liquidity hoarding by lenders is unlikely\"],\"threats_addressed\":[\"reverse causality from dry-ups to performance via fire sales or forgone lending\",\"self-fulfilling runs\",\"demand-driven switching to central bank funding\",\"heavy reliance on CD funding driving the results\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"subsample\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Notes open by naming the equation estimated and the dependent variable. Panels split partial and full dry-ups. Timing of observability (t-1 vs t) is spelled out, along with fixed effects and event windows. Descriptive tables report pooled means and quantiles, and an appendix table defines all variables with their data sources.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"suggesting\",\"we fail to find evidence\",\"cannot draw any definite conclusion\",\"unlikely to explain\"]},\"claim_evidence\":[{\"claim\":\"No market-wide freeze occurred in the European CD market from 2008 to 2014.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Section III.A\"},{\"claim\":\"Bank-specific dry-ups predict future declines in ROA, controlling for current performance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Dry-ups predict rising impaired loans, which helps address reverse causality.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Bank size and loan share stay stable after dry-ups, so fire sales are not indicated.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Negative abnormal stock returns precede dry-ups.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Maturity of new CD issues shortens in the months before dry-ups.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Short-term rating downgrades are followed by drops in CD funding.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"During stress, funds are reallocated toward banks with high future performance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"}],\"structural_summary\":\"The paper opens with the policy-relevant view that wholesale funding is fragile. It sets two classes of asymmetric-information theories against each other, which make opposite predictions about which banks lose funding, and derives testable predictions in a theory section. It then uses transaction-level CD data to document stable aggregate volume alongside bank-specific dry-ups. Dry-ups are shown to predict worse future performance, and reverse causality and run explanations are argued against. Further evidence on maturity shortening, downgrades, lack of reentry and reallocation supports informed-lender models. The conclusion bounds the regulatory and central-banking implications.\",\"writing_profile_keywords\":[\"theory-testing\",\"competing hypotheses\",\"transaction-level data\",\"predictive regressions\",\"event study\",\"reverse causality defense\",\"policy relevance\",\"banking fragility\"],\"confidence\":0.8,\"notes\":\"The empirical design is predictive rather than causal; the abstract explicitly separates predicting from causing. Standard-error treatment is not visible in the excerpts. Section V is labeled robustness by the rule, but its content is additional mechanism results, so it is annotated as results.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Whether bank-specific CD funding dry-ups predict future bank performance conditional on public information, and whether funds shift toward high-quality banks under stress, as a test of adverse-selection versus informed-lender theories", "strategy": "Predictive panel regressions of future changes in ROA, impaired loans and size on dry-up dummies with current controls, plus event studies of stock returns, maturity shortening and rating downgrades, and probit models of reallocation", "key_assumptions": ["Future performance conditional on current public information proxies for quality unobserved by the market", "Impaired loans reflect loans extended before the dry-up and so are less prone to reverse causality", "CDs are unsecured and lenders are mostly money market funds, so asymmetric information concerns only issuer creditworthiness and liquidity hoarding by lenders is unlikely"], "threats_addressed": ["reverse causality from dry-ups to performance via fire sales or forgone lending", "self-fulfilling runs", "demand-driven switching to central bank funding", "heavy reliance on CD funding driving the results"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 5}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_measure", "subsample", "falsification"]}, "tables": {"convention_notes": "Notes open by naming the equation estimated and the dependent variable. Panels split partial and full dry-ups. Timing of observability (t-1 vs t) is spelled out, along with fixed effects and event windows. Descriptive tables report pooled means and quantiles, and an appendix table defines all variables with their data sources.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "suggesting", "we fail to find evidence", "cannot draw any definite conclusion", "unlikely to explain"]}, "claim_evidence": [{"claim": "No market-wide freeze occurred in the European CD market from 2008 to 2014.", "evidence_type": "figure", "evidence_ref": "Figure 1; Section III.A"}, {"claim": "Bank-specific dry-ups predict future declines in ROA, controlling for current performance.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Dry-ups predict rising impaired loans, which helps address reverse causality.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Bank size and loan share stay stable after dry-ups, so fire sales are not indicated.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Negative abnormal stock returns precede dry-ups.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Maturity of new CD issues shortens in the months before dry-ups.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Short-term rating downgrades are followed by drops in CD funding.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "During stress, funds are reallocated toward banks with high future performance.", "evidence_type": "table", "evidence_ref": "Table XI"}], "structural_summary": "The paper opens with the policy-relevant view that wholesale funding is fragile. It sets two classes of asymmetric-information theories against each other, which make opposite predictions about which banks lose funding, and derives testable predictions in a theory section. It then uses transaction-level CD data to document stable aggregate volume alongside bank-specific dry-ups. Dry-ups are shown to predict worse future performance, and reverse causality and run explanations are argued against. Further evidence on maturity shortening, downgrades, lack of reentry and reallocation supports informed-lender models. The conclusion bounds the regulatory and central-banking implications.", "writing_profile_keywords": ["theory-testing", "competing hypotheses", "transaction-level data", "predictive regressions", "event study", "reverse causality defense", "policy relevance", "banking fragility"], "confidence": 0.8, "notes": "The empirical design is predictive rather than causal; the abstract explicitly separates predicting from causing. Standard-error treatment is not visible in the excerpts. Section V is labeled robustness by the rule, but its content is additional mechanism results, so it is annotated as results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.57, "usage": {"input": 8752, "output": 2321, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082222, "request_id": "req_011CggpqwzFcFqqL9gY1AXYW", "cached": false, "ts": "2026-11-03T23:00:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e66c86cf766fc4dedab8dcdffc66f6c5b9dc83933cbe0484e4bfa41b768cc4f7.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"household_finance; macro_finance\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Structural preference and shock-process parameters of a heterogeneous-household equilibrium model, together with the fit of aggregate and household cross-sectional moments; cross-sectional prices of household consumption risk\",\"strategy\":\"GMM estimation that jointly matches aggregate asset-market moments and cross-sectional moments of CEX household consumption growth; Fama-MacBeth two-pass regressions for the cross section of returns\",\"key_assumptions\":[\"Market incompleteness: households cannot trade claims on their wealth portfolios\",\"Autarchy equilibrium holds\",\"Aggregate consumption growth is i.i.d.\",\"A single countercyclical state variable drives the cross-sectional moments\",\"Household shocks are conditionally independent of the rest of the economy\",\"Log-linearization of the consumption-wealth ratio\"],\"threats_addressed\":[\"Measurement error in survey-based CEX consumption, handled by matching time-series averages of moments\",\"Sensitivity to sample period and data frequency\",\"Role of recursive versus CRRA preferences\",\"Role of skewness versus lognormal shocks\",\"Criticism of high-EIS models\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_sample_period\",\"alternative_specification\",\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Each estimation table reports parameter estimates alongside model-implied versus sample moments. Asymptotic Newey-West standard errors (two lags) appear in parentheses, and notes state the number of GMM moment restrictions. Fama-MacBeth tables report results by test-asset panel with adjusted R2. Notes specify the CEX tranche used and the sample period.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"can, in principle\",\"results suggest\",\"fits reasonably well\",\"acknowledgment that estimates are noisy due to measurement error\",\"at least as well as\"]},\"claim_evidence\":[{\"claim\":\"Cross-sectional skewness of household consumption growth is mostly negative and countercyclical, and moderately correlated with recessions\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Table II\"},{\"claim\":\"Moments of household consumption growth are persistent, with positive autocorrelations of the third moment and the fifth percentile\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"The estimated model matches household cross-sectional moments and asset-price moments with a low EIS\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Results are similar over the longer quarterly sample and at annual frequency\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V\"},{\"claim\":\"Recursive preferences are needed to address the equity premium and risk-free rate puzzles\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"Negative skewness of household shocks is pivotal for fitting the data\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Household consumption risk prices the cross section of portfolio returns at least as well as the Fama-French factors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X, XI\"},{\"claim\":\"Rare-disaster calibrations magnify annual consumption declines in an ad hoc way\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Constantinides (2008); Julliard and Ghosh (2012)\"}],\"structural_summary\":\"The introduction motivates the paper with labor-income evidence that the left skewness of idiosyncratic shocks is countercyclical. It positions uninsurable idiosyncratic risk as a paradigm alternative to long-run risks, habit, and disaster models, previews descriptive skewness evidence, and contrasts the approach with related literature. A parsimonious incomplete-market model with recursive preferences and a single state variable is then laid out. Data construction from CRSP and the CEX follows, along with descriptive household consumption statistics. The model is estimated by GMM jointly on aggregate and household moments. Variant estimations then probe sample period, data frequency, preferences, and the shock distribution. A Fama-MacBeth cross-sectional test is presented, and the conclusion restates the findings. Proofs and derivations are placed in appendices.\",\"writing_profile_keywords\":[\"structural GMM\",\"incomplete markets\",\"recursive preferences\",\"household heterogeneity\",\"model-fit tables\",\"paradigm positioning\",\"ablation of model ingredients\",\"Fama-MacBeth\"],\"confidence\":0.8,\"notes\":\"Section V functions as both robustness and ablation analysis. The introduction excerpt shows no explicit roadmap. Section III is descriptive evidence and is labeled as results here.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "household_finance; macro_finance", "primary_method": "structural_estimation", "identification": {"estimand": "Structural preference and shock-process parameters of a heterogeneous-household equilibrium model, together with the fit of aggregate and household cross-sectional moments; cross-sectional prices of household consumption risk", "strategy": "GMM estimation that jointly matches aggregate asset-market moments and cross-sectional moments of CEX household consumption growth; Fama-MacBeth two-pass regressions for the cross section of returns", "key_assumptions": ["Market incompleteness: households cannot trade claims on their wealth portfolios", "Autarchy equilibrium holds", "Aggregate consumption growth is i.i.d.", "A single countercyclical state variable drives the cross-sectional moments", "Household shocks are conditionally independent of the rest of the economy", "Log-linearization of the consumption-wealth ratio"], "threats_addressed": ["Measurement error in survey-based CEX consumption, handled by matching time-series averages of moments", "Sensitivity to sample period and data frequency", "Role of recursive versus CRRA preferences", "Role of skewness versus lognormal shocks", "Criticism of high-EIS models"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "related_literature_positioning"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_sample_period", "alternative_specification", "sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "Each estimation table reports parameter estimates alongside model-implied versus sample moments. Asymptotic Newey-West standard errors (two lags) appear in parentheses, and notes state the number of GMM moment restrictions. Fama-MacBeth tables report results by test-asset panel with adjusted R2. Notes specify the CEX tranche used and the sample period.", "units_stated_in_notes": false, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["can, in principle", "results suggest", "fits reasonably well", "acknowledgment that estimates are noisy due to measurement error", "at least as well as"]}, "claim_evidence": [{"claim": "Cross-sectional skewness of household consumption growth is mostly negative and countercyclical, and moderately correlated with recessions", "evidence_type": "figure", "evidence_ref": "Figure 1; Table II"}, {"claim": "Moments of household consumption growth are persistent, with positive autocorrelations of the third moment and the fifth percentile", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "The estimated model matches household cross-sectional moments and asset-price moments with a low EIS", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Results are similar over the longer quarterly sample and at annual frequency", "evidence_type": "table", "evidence_ref": "Tables IV, V"}, {"claim": "Recursive preferences are needed to address the equity premium and risk-free rate puzzles", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "Negative skewness of household shocks is pivotal for fitting the data", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Household consumption risk prices the cross section of portfolio returns at least as well as the Fama-French factors", "evidence_type": "table", "evidence_ref": "Tables X, XI"}, {"claim": "Rare-disaster calibrations magnify annual consumption declines in an ad hoc way", "evidence_type": "prior_literature", "evidence_ref": "Constantinides (2008); Julliard and Ghosh (2012)"}], "structural_summary": "The introduction motivates the paper with labor-income evidence that the left skewness of idiosyncratic shocks is countercyclical. It positions uninsurable idiosyncratic risk as a paradigm alternative to long-run risks, habit, and disaster models, previews descriptive skewness evidence, and contrasts the approach with related literature. A parsimonious incomplete-market model with recursive preferences and a single state variable is then laid out. Data construction from CRSP and the CEX follows, along with descriptive household consumption statistics. The model is estimated by GMM jointly on aggregate and household moments. Variant estimations then probe sample period, data frequency, preferences, and the shock distribution. A Fama-MacBeth cross-sectional test is presented, and the conclusion restates the findings. Proofs and derivations are placed in appendices.", "writing_profile_keywords": ["structural GMM", "incomplete markets", "recursive preferences", "household heterogeneity", "model-fit tables", "paradigm positioning", "ablation of model ingredients", "Fama-MacBeth"], "confidence": 0.8, "notes": "Section V functions as both robustness and ablation analysis. The introduction excerpt shows no explicit roadmap. Section III is descriptive evidence and is labeled as results here."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.56, "usage": {"input": 8958, "output": 2489, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086406, "request_id": "req_011CggpTnTALxqDYAnV6YtEi", "cached": false, "ts": "2026-11-03T22:55:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e7d7f9596c805eababe5f6d4b0fc39dc99306774b03b4544daef467c1856c114.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"asset_pricing (mutual fund performance)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in risk-adjusted (four-factor alpha) performance between commercial bank-affiliated and unaffiliated equity funds, and its relation to overweighting of the parent bank's lending-client stocks\",\"strategy\":\"Fund-quarter panel OLS with controls and fund fixed effects; difference-in-differences around divestitures of asset management divisions by banking groups; within-affiliated-fund calendar-time portfolio comparisons of client vs nonclient stock trades; placebo tests\",\"key_assumptions\":[\"Divestitures are exogenous to fund performance, e.g., driven by bank regulatory capital needs during the crisis\",\"Time-invariant unobserved fund heterogeneity is absorbed by fund fixed effects\",\"Pretreatment means are balanced between treated and control funds around divestitures\",\"Manager skill would show up equally in client and nonclient stock trades\"],\"threats_addressed\":[\"omitted variable bias\",\"reverse causality (past performance determines organizational form)\",\"lower manager skill at affiliated funds\",\"differences in affiliation type (insurance, investment bank, publicly traded parent)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":0},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"placebo\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes name the estimator (OLS, logit, probit, tobit, DiD), the dependent variable, and define key dummies. Variable definitions with data sources are collected in an appendix table. Summary statistics report mean-difference p-values. Clustering and SE details are not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"suggest\",\"we interpret these results as indicating\",\"support a causal interpretation\",\"do not support the skill hypothesis\"]},\"claim_evidence\":[{\"claim\":\"Bank-affiliated funds underperform unaffiliated funds by about 92 bp per year in four-factor alpha\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Underperformance is larger among affiliated funds that strongly overweight lending-client stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Underperformance varies across countries with lending division size relative to asset management\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"After divestiture, funds cut client stock holdings and their performance improves\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 2\"},{\"claim\":\"Overweighting funds underperform in client-stock trades but not nonclient trades, against a skill explanation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Affiliated fund holdings raise the bank's chance of becoming lead arranger on future loans\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Managers who overweight client stocks are less likely to be replaced\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Affiliated fund ownership is associated with lower voting dissent on executive pay\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Outside the U.S., flows to affiliated funds are insensitive to poor performance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"}],\"structural_summary\":\"The abstract opens with the question and international sample, states the headline magnitude, links heterogeneity to conflicts of interest, cites divestitures as causal support, and closes with a career-concerns interpretation. The introduction sets out industry facts and a lack of research, then poses three competing hypotheses (conflict of interest, information advantage, null). It previews endogeneity strategies (fixed effects, divestitures) and a skill test, then asks why such funds persist in equilibrium and previews benefits to the bank, manager and borrower plus investor inertia. The body runs: hypotheses, data, baseline and heterogeneity results, a dedicated endogeneity section, and an equilibrium section. The conclusion restates findings and identification.\",\"writing_profile_keywords\":[\"competing hypotheses\",\"conflict of interest\",\"international mutual fund panel\",\"quasi-natural experiment\",\"dedicated endogeneity section\",\"equilibrium rationale\",\"magnitude in abstract\"],\"confidence\":0.8,\"notes\":\"The endogeneity section mixes the divestiture DiD, calendar-time portfolio tests and placebo tests; it is coded as robustness, though it functions as the identification section. Results also include a 'Robustness Checks' subsection, and an Internet Appendix exists. No explicit contribution paragraph or roadmap is visible in the truncated introduction. How standard errors are computed is not visible.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "asset_pricing (mutual fund performance)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in risk-adjusted (four-factor alpha) performance between commercial bank-affiliated and unaffiliated equity funds, and its relation to overweighting of the parent bank's lending-client stocks", "strategy": "Fund-quarter panel OLS with controls and fund fixed effects; difference-in-differences around divestitures of asset management divisions by banking groups; within-affiliated-fund calendar-time portfolio comparisons of client vs nonclient stock trades; placebo tests", "key_assumptions": ["Divestitures are exogenous to fund performance, e.g., driven by bank regulatory capital needs during the crisis", "Time-invariant unobserved fund heterogeneity is absorbed by fund fixed effects", "Pretreatment means are balanced between treated and control funds around divestitures", "Manager skill would show up equally in client and nonclient stock trades"], "threats_addressed": ["omitted variable bias", "reverse causality (past performance determines organizational form)", "lower manager skill at affiliated funds", "differences in affiliation type (insurance, investment bank, publicly traded parent)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 0}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_measure", "placebo", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes name the estimator (OLS, logit, probit, tobit, DiD), the dependent variable, and define key dummies. Variable definitions with data sources are collected in an appendix table. Summary statistics report mean-difference p-values. Clustering and SE details are not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "suggest", "we interpret these results as indicating", "support a causal interpretation", "do not support the skill hypothesis"]}, "claim_evidence": [{"claim": "Bank-affiliated funds underperform unaffiliated funds by about 92 bp per year in four-factor alpha", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Underperformance is larger among affiliated funds that strongly overweight lending-client stocks", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Underperformance varies across countries with lending division size relative to asset management", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "After divestiture, funds cut client stock holdings and their performance improves", "evidence_type": "table", "evidence_ref": "Table VI; Figure 2"}, {"claim": "Overweighting funds underperform in client-stock trades but not nonclient trades, against a skill explanation", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Affiliated fund holdings raise the bank's chance of becoming lead arranger on future loans", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Managers who overweight client stocks are less likely to be replaced", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Affiliated fund ownership is associated with lower voting dissent on executive pay", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Outside the U.S., flows to affiliated funds are insensitive to poor performance", "evidence_type": "table", "evidence_ref": "Table XI"}], "structural_summary": "The abstract opens with the question and international sample, states the headline magnitude, links heterogeneity to conflicts of interest, cites divestitures as causal support, and closes with a career-concerns interpretation. The introduction sets out industry facts and a lack of research, then poses three competing hypotheses (conflict of interest, information advantage, null). It previews endogeneity strategies (fixed effects, divestitures) and a skill test, then asks why such funds persist in equilibrium and previews benefits to the bank, manager and borrower plus investor inertia. The body runs: hypotheses, data, baseline and heterogeneity results, a dedicated endogeneity section, and an equilibrium section. The conclusion restates findings and identification.", "writing_profile_keywords": ["competing hypotheses", "conflict of interest", "international mutual fund panel", "quasi-natural experiment", "dedicated endogeneity section", "equilibrium rationale", "magnitude in abstract"], "confidence": 0.8, "notes": "The endogeneity section mixes the divestiture DiD, calendar-time portfolio tests and placebo tests; it is coded as robustness, though it functions as the identification section. Results also include a 'Robustness Checks' subsection, and an Internet Appendix exists. No explicit contribution paragraph or roadmap is visible in the truncated introduction. How standard errors are computed is not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.05, "usage": {"input": 6819, "output": 2406, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07619, "request_id": "req_011Cggq4a1XoZiNRM3PY3qKy", "cached": false, "ts": "2026-11-03T23:03:38+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ec4e2b7d9b181c1ac74b0902ce64758fb5ad1b196e32d58304e5b1519fa848b8.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"corporate cash holdings / governance\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of financial portfolio size on the share of the portfolio held in risky assets; marginal value of risky versus safe financial assets\",\"strategy\":\"Panel regressions run by OLS and by 2SLS. Portfolio size is instrumented with unexpected operating cash flow innovations. Value tests regress stock returns and market-to-book on financial asset holdings, following the Faulkender–Wang approach.\",\"key_assumptions\":[\"Unexpected cash flow innovations shift portfolio size but affect portfolio composition only through size\",\"Fair value footnote disclosures accurately capture asset classes\"],\"threats_addressed\":[\"Joint determination of portfolio size and composition (endogeneity)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":2},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_measure\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes define variables and write out the regression equations, including the 2SLS first stage. OLS and 2SLS columns are reported side by side. Descriptive tables give aggregate value-weighted figures, with equal-weighted versions in an appendix. Appendix tables reproduce firm disclosures from Google, Apple and Intel as illustrations.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' framing of results against model predictions\",\"'suggest' for 2SLS estimates\",\"explicit sample limitation acknowledged in conclusion\"]},\"claim_evidence\":[{\"claim\":\"Risky assets make up about 40% of firms' financial portfolios and about 6% of book assets.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"The share held in risky assets rises from about 10% to 40% across portfolio-size quintiles, and allocation variation is higher in the top quintiles.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"A one-point increase in portfolio size raises the risky share by about 30 bp (2SLS estimate).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Permanently reinvested earnings relate positively to total financial assets but insignificantly to risky assets.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Poorly governed firms hold more risky assets.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Risky financial assets are valued at a 13–22% discount relative to safe assets.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Constrained firms optimally avoid risky assets; unconstrained firms are at best indifferent to them.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Internet Appendix\"},{\"claim\":\"There is no uniform tax incentive to invest in risky financial assets.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Appendix C\"}],\"structural_summary\":\"The paper opens with press anecdotes that challenge the assumption that corporate 'cash' is in fact cash. It introduces hand-collected fair value data made available by SFAS 157, then describes portfolio composition. A theory section sketches a precautionary-savings model extended to risky and illiquid assets, with the formal derivation in an internet appendix. Empirical tests follow: univariate sorts, then OLS/2SLS determinants (taxes, agency, CEO traits), then valuation and financing-source tests. The conclusion frames the findings as a largely unregulated shadow asset management industry.\",\"writing_profile_keywords\":[\"hand-collected data\",\"model-consistent predictions\",\"precautionary savings\",\"2SLS\",\"shadow hedge fund framing\",\"policy implications\",\"anecdotal motivation\"],\"confidence\":0.8,\"notes\":\"Theory is summarized in the main text and formally derived in an internet appendix. Standard error treatment is not visible in the excerpts. Section II is descriptive results.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "corporate_finance", "subfield_secondary": "corporate cash holdings / governance", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of financial portfolio size on the share of the portfolio held in risky assets; marginal value of risky versus safe financial assets", "strategy": "Panel regressions run by OLS and by 2SLS. Portfolio size is instrumented with unexpected operating cash flow innovations. Value tests regress stock returns and market-to-book on financial asset holdings, following the Faulkender–Wang approach.", "key_assumptions": ["Unexpected cash flow innovations shift portfolio size but affect portfolio composition only through size", "Fair value footnote disclosures accurately capture asset classes"], "threats_addressed": ["Joint determination of portfolio size and composition (endogeneity)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 2}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_data_or_measure", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_measure", "alternative_specification"]}, "tables": {"convention_notes": "Table notes define variables and write out the regression equations, including the 2SLS first stage. OLS and 2SLS columns are reported side by side. Descriptive tables give aggregate value-weighted figures, with equal-weighted versions in an appendix. Appendix tables reproduce firm disclosures from Google, Apple and Intel as illustrations.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' framing of results against model predictions", "'suggest' for 2SLS estimates", "explicit sample limitation acknowledged in conclusion"]}, "claim_evidence": [{"claim": "Risky assets make up about 40% of firms' financial portfolios and about 6% of book assets.", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "The share held in risky assets rises from about 10% to 40% across portfolio-size quintiles, and allocation variation is higher in the top quintiles.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "A one-point increase in portfolio size raises the risky share by about 30 bp (2SLS estimate).", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Permanently reinvested earnings relate positively to total financial assets but insignificantly to risky assets.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Poorly governed firms hold more risky assets.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Risky financial assets are valued at a 13–22% discount relative to safe assets.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Constrained firms optimally avoid risky assets; unconstrained firms are at best indifferent to them.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Internet Appendix"}, {"claim": "There is no uniform tax incentive to invest in risky financial assets.", "evidence_type": "equation_or_proof", "evidence_ref": "Appendix C"}], "structural_summary": "The paper opens with press anecdotes that challenge the assumption that corporate 'cash' is in fact cash. It introduces hand-collected fair value data made available by SFAS 157, then describes portfolio composition. A theory section sketches a precautionary-savings model extended to risky and illiquid assets, with the formal derivation in an internet appendix. Empirical tests follow: univariate sorts, then OLS/2SLS determinants (taxes, agency, CEO traits), then valuation and financing-source tests. The conclusion frames the findings as a largely unregulated shadow asset management industry.", "writing_profile_keywords": ["hand-collected data", "model-consistent predictions", "precautionary savings", "2SLS", "shadow hedge fund framing", "policy implications", "anecdotal motivation"], "confidence": 0.8, "notes": "Theory is summarized in the main text and formally derived in an internet appendix. Standard error treatment is not visible in the excerpts. Section II is descriptive results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.45, "usage": {"input": 8499, "output": 2273, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08025, "request_id": "req_011CggpXrUs8VJiNd5tJs5TJ", "cached": false, "ts": "2026-11-03T22:56:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f11c1d4f073c80c0f2f1450ea7b5e0911f465e979ecfdc82a3798ca76e58cd2b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of persistent shocks to the value of firms' real estate holdings, which relax financing constraints, on subsequent firm equity volatility, especially among R&D-intensive firms.\",\"strategy\":\"Local real estate price shocks are instrumented with the geographically determined availability of land, which shapes each region's exposure to national real estate cycles. These shocks are interacted with firms' preexisting (lagged) real estate share of total assets. A complementary financial crisis analysis relates the R&D volatility gap to firms' pre-crisis cash holdings and net leverage.\",\"key_assumptions\":[\"Land availability affects local real estate prices but has no other direct effect on firm equity volatility\",\"Preexisting real estate intensity does not proxy for other firm characteristics that drive volatility responses\",\"Local aggregate demand shocks do not drive the results\"],\"threats_addressed\":[\"local aggregate demand effects\",\"alternative channels via correlation of real estate ratio with firm characteristics (age, size, cash flow, Q)\",\"external validity (addressed via financial crisis analysis)\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the specification relative to a baseline equation or column, the sample (cross-referencing Table I panels), the unit of observation (firm-year or firm-month) and the outcome variables. Panels separate first-stage and other analyses.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'provides evidence that'\",\"'suggest'\",\"'can be an important determinant'\",\"'arguably'\",\"explicit acknowledgment of identification concerns\"]},\"claim_evidence\":[{\"claim\":\"Shocks to real estate values that relax financing constraints raise subsequent equity volatility.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Instrumented local real estate shocks are linked to land availability (first stage).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Balance sheet shocks raise investment and debt issuance modestly, lower market leverage, and do not raise cash flow volatility.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Effects are concentrated in the idiosyncratic volatility of R&D-intensive firms.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"No effect is found for mature or large firms, which are less reliant on growth options.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Results are not driven by local aggregate demand or alternative channels.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"},{\"claim\":\"The R&D volatility gap during the crisis varies with firms' pre-crisis cash and net leverage.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX, X\"},{\"claim\":\"Real estate shocks are persistent and the volatility increases are immediate.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Collateral constraints can raise equity volatility by enabling access to growth options.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I, Appendix A\"}],\"structural_summary\":\"The introduction motivates why idiosyncratic volatility matters and notes limited understanding of what determines it. It states the growth-options argument, names two identification challenges, then previews the stylized model, the IV design and the results, including mechanism and magnitudes. It positions the paper against cross-country volatility work and the financing-constraints asset pricing literature. The body proceeds through a theory section (proofs in an appendix), data, the main IV results with mechanism, subsample and identification subsections, and a crisis analysis for external validity. A brief conclusion restates the findings.\",\"writing_profile_keywords\":[\"financing constraints\",\"idiosyncratic volatility\",\"growth options\",\"collateral channel\",\"real estate shocks\",\"instrumental variables\",\"land availability instrument\",\"R&D intensity\",\"external validity\"],\"confidence\":0.8,\"notes\":\"Standard error clustering and the significance reporting style are not visible in the excerpts. Section III is labeled 'other' by the role rule but functions as the main results section, including identification checks.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "corporate_finance", "subfield_secondary": "asset_pricing", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of persistent shocks to the value of firms' real estate holdings, which relax financing constraints, on subsequent firm equity volatility, especially among R&D-intensive firms.", "strategy": "Local real estate price shocks are instrumented with the geographically determined availability of land, which shapes each region's exposure to national real estate cycles. These shocks are interacted with firms' preexisting (lagged) real estate share of total assets. A complementary financial crisis analysis relates the R&D volatility gap to firms' pre-crisis cash holdings and net leverage.", "key_assumptions": ["Land availability affects local real estate prices but has no other direct effect on firm equity volatility", "Preexisting real estate intensity does not proxy for other firm characteristics that drive volatility responses", "Local aggregate demand shocks do not drive the results"], "threats_addressed": ["local aggregate demand effects", "alternative channels via correlation of real estate ratio with firm characteristics (age, size, cash flow, Q)", "external validity (addressed via financial crisis analysis)"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["subsample", "falsification", "other"]}, "tables": {"convention_notes": "Table notes state the specification relative to a baseline equation or column, the sample (cross-referencing Table I panels), the unit of observation (firm-year or firm-month) and the outcome variables. Panels separate first-stage and other analyses.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'provides evidence that'", "'suggest'", "'can be an important determinant'", "'arguably'", "explicit acknowledgment of identification concerns"]}, "claim_evidence": [{"claim": "Shocks to real estate values that relax financing constraints raise subsequent equity volatility.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Instrumented local real estate shocks are linked to land availability (first stage).", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Balance sheet shocks raise investment and debt issuance modestly, lower market leverage, and do not raise cash flow volatility.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Effects are concentrated in the idiosyncratic volatility of R&D-intensive firms.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "No effect is found for mature or large firms, which are less reliant on growth options.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Results are not driven by local aggregate demand or alternative channels.", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}, {"claim": "The R&D volatility gap during the crisis varies with firms' pre-crisis cash and net leverage.", "evidence_type": "table", "evidence_ref": "Tables IX, X"}, {"claim": "Real estate shocks are persistent and the volatility increases are immediate.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Collateral constraints can raise equity volatility by enabling access to growth options.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I, Appendix A"}], "structural_summary": "The introduction motivates why idiosyncratic volatility matters and notes limited understanding of what determines it. It states the growth-options argument, names two identification challenges, then previews the stylized model, the IV design and the results, including mechanism and magnitudes. It positions the paper against cross-country volatility work and the financing-constraints asset pricing literature. The body proceeds through a theory section (proofs in an appendix), data, the main IV results with mechanism, subsample and identification subsections, and a crisis analysis for external validity. A brief conclusion restates the findings.", "writing_profile_keywords": ["financing constraints", "idiosyncratic volatility", "growth options", "collateral channel", "real estate shocks", "instrumental variables", "land availability instrument", "R&D intensity", "external validity"], "confidence": 0.8, "notes": "Standard error clustering and the significance reporting style are not visible in the excerpts. Section III is labeled 'other' by the role rule but functions as the main results section, including identification checks."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.93, "usage": {"input": 7219, "output": 2310, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07587, "request_id": "req_011CggpuvhjaJ3uS9CjuAwEo", "cached": false, "ts": "2026-11-03T23:01:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f1c72446d572216fa9007d378baf3a1adb70fae1aed7e889710e489edee7e2bf.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance (self-dealing/tunneling, securitization agency conflicts)\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Change in loan loss rates (and sale prices, liquidation volumes) for liquidated CMBS loans after ownership changes of four treated special servicers, relative to placebo servicers\",\"strategy\":\"Panel difference-in-differences that treats servicer ownership changes as shocks to firm affiliations, comparing treated with placebo special servicers, with servicer and month fixed effects; supplemented by a transaction-level case study of buyers for one servicer\",\"key_assumptions\":[\"Treated and placebo servicers face common market conditions\",\"Compositional differences in liquidated loans do not drive the effect\",\"Pre-change stockpiling of distressed debt is transitory and does not account for the post-period difference\"],\"threats_addressed\":[\"Unobserved market conditions around 2010\",\"Non-random assignment of loans across servicers / loan quality differences\",\"Stockpiling effect from liquidity crises that triggered the ownership changes\",\"Transitory confounders near the event date\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":3},\"magnitude\":{\"present\":true,\"sentence_index\":3},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_identification\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_sample_period\",\"subsample\",\"alternative_specification\",\"placebo\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state the estimator (OLS), the dependent-variable definition, and the key interaction regressor. Robustness tables are laid out row by row, each row a sample or specification variant. Standard errors are double-clustered by servicer and month with finite-sample adjustment. Significance is marked with stars at 10/5/1%.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' phrasing\",\"'potential' and 'potentially important'\",\"'unlikely to be explained away'\",\"'the weight of the evidence suggests'\",\"acknowledging mixed evidence\"]},\"claim_evidence\":[{\"claim\":\"Loans liquidated by treated servicers after ownership changes have about 8 p.p. higher loss rates than those of placebo servicers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The main effect survives dropping 12–18 months around the ownership changes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Predetermined and time-varying loan quality differences do not explain the effect\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figures 2-4\"},{\"claim\":\"Stockpiling explains at most 21% of the post-period loss difference\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Sale prices are 14% lower and liquidation volume rises 229%, while liquidation expenses do not rise, favoring steering over price discrimination\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The main effect is robust to sample period changes, propensity-score matching and aggregation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"}],\"structural_summary\":\"The introduction opens by stating that self-dealing is hard to measure. It then presents CMBS special servicers' ownership changes as a measurable setting and devotes several paragraphs to identification threats (market conditions, loan composition, stockpiling) and the robustness responses. It previews channel tests with magnitudes, positions the paper against the tunneling literature, and ends with RMBS and regulatory implications plus a roadmap. The body runs from institutional background with three self-dealing mechanisms, to data, to the DiD main results with balance and stockpiling checks, and then to a channel analysis with a case study and a discussion of offsetting vertical-integration benefits. The conclusion is hedged and adds policy points and future research directions.\",\"writing_profile_keywords\":[\"quasi-experimental\",\"placebo servicers\",\"threats-to-identification heavy intro\",\"bounding exercise\",\"case study complement\",\"channel discrimination\",\"policy implications\",\"hedged consistency language\"],\"confidence\":0.85,\"notes\":\"Single-author, first-person paper. The Background section (idx 1) is institutional rather than a literature review, so its role is coded 'other' despite the rule label. The figure captions do not show whether confidence intervals are plotted, so figure uncertainty is coded 'unclear'.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance (self-dealing/tunneling, securitization agency conflicts)", "primary_method": "difference_in_differences", "identification": {"estimand": "Change in loan loss rates (and sale prices, liquidation volumes) for liquidated CMBS loans after ownership changes of four treated special servicers, relative to placebo servicers", "strategy": "Panel difference-in-differences that treats servicer ownership changes as shocks to firm affiliations, comparing treated with placebo special servicers, with servicer and month fixed effects; supplemented by a transaction-level case study of buyers for one servicer", "key_assumptions": ["Treated and placebo servicers face common market conditions", "Compositional differences in liquidated loans do not drive the effect", "Pre-change stockpiling of distressed debt is transitory and does not account for the post-period difference"], "threats_addressed": ["Unobserved market conditions around 2010", "Non-random assignment of loans across servicers / loan quality differences", "Stockpiling effect from liquidity crises that triggered the ownership changes", "Transitory confounders near the event date"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": true, "sentence_index": 0}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 3}, "magnitude": {"present": true, "sentence_index": 3}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "policy_or_broader_implication"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 8, "contribution_types": ["new_evidence", "new_data_or_measure", "new_identification"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_sample_period", "subsample", "alternative_specification", "placebo", "other"]}, "tables": {"convention_notes": "Notes state the estimator (OLS), the dependent-variable definition, and the key interaction regressor. Robustness tables are laid out row by row, each row a sample or specification variant. Standard errors are double-clustered by servicer and month with finite-sample adjustment. Significance is marked with stars at 10/5/1%.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' phrasing", "'potential' and 'potentially important'", "'unlikely to be explained away'", "'the weight of the evidence suggests'", "acknowledging mixed evidence"]}, "claim_evidence": [{"claim": "Loans liquidated by treated servicers after ownership changes have about 8 p.p. higher loss rates than those of placebo servicers", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The main effect survives dropping 12–18 months around the ownership changes", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Predetermined and time-varying loan quality differences do not explain the effect", "evidence_type": "table", "evidence_ref": "Table IV; Figures 2-4"}, {"claim": "Stockpiling explains at most 21% of the post-period loss difference", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Sale prices are 14% lower and liquidation volume rises 229%, while liquidation expenses do not rise, favoring steering over price discrimination", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The main effect is robust to sample period changes, propensity-score matching and aggregation", "evidence_type": "table", "evidence_ref": "Table III"}], "structural_summary": "The introduction opens by stating that self-dealing is hard to measure. It then presents CMBS special servicers' ownership changes as a measurable setting and devotes several paragraphs to identification threats (market conditions, loan composition, stockpiling) and the robustness responses. It previews channel tests with magnitudes, positions the paper against the tunneling literature, and ends with RMBS and regulatory implications plus a roadmap. The body runs from institutional background with three self-dealing mechanisms, to data, to the DiD main results with balance and stockpiling checks, and then to a channel analysis with a case study and a discussion of offsetting vertical-integration benefits. The conclusion is hedged and adds policy points and future research directions.", "writing_profile_keywords": ["quasi-experimental", "placebo servicers", "threats-to-identification heavy intro", "bounding exercise", "case study complement", "channel discrimination", "policy implications", "hedged consistency language"], "confidence": 0.85, "notes": "Single-author, first-person paper. The Background section (idx 1) is institutional rather than a literature review, so its role is coded 'other' despite the rule label. The figure captions do not show whether confidence intervals are plotted, so figure uncertainty is coded 'unclear'."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.0, "usage": {"input": 7487, "output": 2254, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075822, "request_id": "req_011Cggq6b847mgjgDxAw9vKo", "cached": false, "ts": "2026-11-03T23:04:04+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f62fa2f7c3cd71c829f45ca98998fe6d00f992c8d22aaa86ee5e9bf9610f9988.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Differential effect of import penetration from China on firm outcomes (sales growth, ROA, capex, employment) as a function of the firm's ex ante R&D capital stock\",\"strategy\":\"Two instruments inside a difference-in-differences-style interaction design: China's import penetration in eight other high-income countries instruments U.S. import penetration, and tax-induced changes in the user cost of R&D, driven by staggered state R&D tax credits, instrument R&D stock. Firm fixed effects and industry-by-year fixed effects are included.\",\"key_assumptions\":[\"The surge in Chinese exports is driven by China-internal supply shocks common across high-income countries\",\"State R&D tax credit changes are exogenous to firm productivity, management quality and demand\",\"Both import penetration and R&D must be instrumented for the interaction coefficient to be unbiased, as the model shows\"],\"threats_addressed\":[\"Reverse causality from low U.S. productivity to higher imports\",\"Endogeneity of R&D to productivity, management quality and demand\",\"Industry-specific productivity and demand shocks, absorbed by industry-by-year fixed effects\",\"Import competition in intermediate input markets\",\"Alternative trade shock source: the 2000 NTR tariff change\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":true,\"sentence_index\":2},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"new_evidence\",\"new_identification\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the sample (Compustat manufacturing firms and years), the estimation model, the dependent variable and the fixed-effects structure for each column. Separate first-stage tables are reported for each instrument. Column pairs alternate between year fixed effects and industry-by-year fixed effects.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"described as 'consistent with' prior work or theory\",\"'provide evidence that' framing for the mechanism\",\"explicit caution about generalizing results\",\"acknowledgment that the difference-in-differences design cannot capture general equilibrium effects\"]},\"claim_evidence\":[{\"claim\":\"A one-standard-deviation rise in China import penetration lowers annual sales growth by about 1.8 percentage points on average.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Moving from the 25th to the 75th percentile of R&D stock halves the sales growth decline and offsets the ROA decline.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"The resilience effect of R&D operates through product differentiation rather than productivity.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"The R&D effect is stronger in industries where differentiation is more prevalent.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"High-R&D firms cut capital expenditures and employment much less in response to import shocks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XI, XII\"},{\"claim\":\"Imports from China surged in parallel in the U.S. and in other high-income countries, supporting the exogeneity of the trade instrument.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Table II\"},{\"claim\":\"Results hold with the NTR tariff gap instrument and with controls for import competition in input markets.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XIII, XIV\"},{\"claim\":\"Regressing performance on import penetration interacted with R&D is unbiased only if both variables are instrumented.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.B\"}],\"structural_summary\":\"The introduction opens with a widely held policy view (innovation shields firms from low-wage competition) and notes the scarce evidence behind it. It previews the results, uses a stylized model to motivate the identification strategy, defends the two instruments, and then previews magnitudes for performance, the mechanism and real effects. It ends by positioning the paper in the literature. The body proceeds from theoretical framework to data and instruments, then main results on performance, the product-differentiation mechanism, and capex and employment. A robustness section follows. The conclusion restates the findings, draws policy implications, bounds generality, and poses general-equilibrium questions for future work.\",\"writing_profile_keywords\":[\"model-guided identification\",\"double instrumentation\",\"interaction design\",\"mechanism tests\",\"economic magnitudes\",\"policy relevance\",\"explicit limitations\"],\"confidence\":0.85,\"notes\":\"Standard error type and significance notation are not visible in the table excerpts. Section idx 3 has a garbled heading but contains the R&D instrument and econometric specification subsections. The model for the mechanism is placed in the Internet Appendix.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "corporate_finance", "subfield_secondary": "international_finance", "primary_method": "instrumental_variables", "identification": {"estimand": "Differential effect of import penetration from China on firm outcomes (sales growth, ROA, capex, employment) as a function of the firm's ex ante R&D capital stock", "strategy": "Two instruments inside a difference-in-differences-style interaction design: China's import penetration in eight other high-income countries instruments U.S. import penetration, and tax-induced changes in the user cost of R&D, driven by staggered state R&D tax credits, instrument R&D stock. Firm fixed effects and industry-by-year fixed effects are included.", "key_assumptions": ["The surge in Chinese exports is driven by China-internal supply shocks common across high-income countries", "State R&D tax credit changes are exogenous to firm productivity, management quality and demand", "Both import penetration and R&D must be instrumented for the interaction coefficient to be unbiased, as the model shows"], "threats_addressed": ["Reverse causality from low U.S. productivity to higher imports", "Endogeneity of R&D to productivity, management quality and demand", "Industry-specific productivity and demand shocks, absorbed by industry-by-year fixed effects", "Import competition in intermediate input markets", "Alternative trade shock source: the 2000 NTR tariff change"], "causal_language_strength": 4, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": true, "sentence_index": 2}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}], "contribution_paragraph_index": 8, "contribution_types": ["new_evidence", "new_identification"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_measure", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes state the sample (Compustat manufacturing firms and years), the estimation model, the dependent variable and the fixed-effects structure for each column. Separate first-stage tables are reported for each instrument. Column pairs alternate between year fixed effects and industry-by-year fixed effects.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["described as 'consistent with' prior work or theory", "'provide evidence that' framing for the mechanism", "explicit caution about generalizing results", "acknowledgment that the difference-in-differences design cannot capture general equilibrium effects"]}, "claim_evidence": [{"claim": "A one-standard-deviation rise in China import penetration lowers annual sales growth by about 1.8 percentage points on average.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Moving from the 25th to the 75th percentile of R&D stock halves the sales growth decline and offsets the ROA decline.", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "The resilience effect of R&D operates through product differentiation rather than productivity.", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "The R&D effect is stronger in industries where differentiation is more prevalent.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "High-R&D firms cut capital expenditures and employment much less in response to import shocks.", "evidence_type": "table", "evidence_ref": "Tables XI, XII"}, {"claim": "Imports from China surged in parallel in the U.S. and in other high-income countries, supporting the exogeneity of the trade instrument.", "evidence_type": "figure", "evidence_ref": "Figure 1; Table II"}, {"claim": "Results hold with the NTR tariff gap instrument and with controls for import competition in input markets.", "evidence_type": "table", "evidence_ref": "Tables XIII, XIV"}, {"claim": "Regressing performance on import penetration interacted with R&D is unbiased only if both variables are instrumented.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.B"}], "structural_summary": "The introduction opens with a widely held policy view (innovation shields firms from low-wage competition) and notes the scarce evidence behind it. It previews the results, uses a stylized model to motivate the identification strategy, defends the two instruments, and then previews magnitudes for performance, the mechanism and real effects. It ends by positioning the paper in the literature. The body proceeds from theoretical framework to data and instruments, then main results on performance, the product-differentiation mechanism, and capex and employment. A robustness section follows. The conclusion restates the findings, draws policy implications, bounds generality, and poses general-equilibrium questions for future work.", "writing_profile_keywords": ["model-guided identification", "double instrumentation", "interaction design", "mechanism tests", "economic magnitudes", "policy relevance", "explicit limitations"], "confidence": 0.85, "notes": "Standard error type and significance notation are not visible in the table excerpts. Section idx 3 has a garbled heading but contains the R&D instrument and econometric specification subsections. The model for the mechanism is placed in the Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.68, "usage": {"input": 9464, "output": 2428, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08721, "request_id": "req_011Cggq33H4gx8GYV5J3xQSv", "cached": false, "ts": "2026-11-03T23:03:19+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/febf6433893047d4bfd4e6d3627dbc069a21051eed4a8d22f1b2488d27f54657.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model; comparative statics with respect to the news arrival frequency)\",\"strategy\":\"not applicable; equilibrium analysis of a dynamic limit order market in a DGP-style Poisson framework, with numerical comparative statics\",\"key_assumptions\":[\"Poisson arrival of unscheduled news\",\"stochastic investor reaction delays\",\"continuous idiosyncratic liquidity shocks\",\"aggregate preferences in steady state\",\"one-tick market with binding spread\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":4},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"No tables. Results appear as propositions, with proofs in the appendix and fuller versions in an Internet Appendix. Comparative statics and dynamics are shown in figures whose captions state parameter values and time units.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"first to my knowledge\",\"explicit acknowledgment that the model is stylized, with listed limitations in the extensions section\",\"conditional comparative-statics phrasing (holding price variability constant)\"]},\"claim_evidence\":[{\"claim\":\"More frequent news raises the adverse selection cost of limit orders, which lowers market depth.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Appendix proofs; Figure 5\"},{\"claim\":\"News arrival causes order flow and book imbalances and a volume spike, giving a positive covariance between volatility and volume.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6; Section III.E-F\"},{\"claim\":\"Holding unconditional price variability fixed, news frequency reduces the variability-volume covariance.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5; Section III.F\"},{\"claim\":\"Higher news frequency shortens the transition phase and raises the trades' share of order outflow.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5; Sections III.C-D\"}],\"structural_summary\":\"A theory paper. The introduction motivates the question with electronic limit order books and the winner's curse that unscheduled news creates. It then sketches the equilibrium mechanism across several paragraphs and closes with a contribution claim and literature positioning relative to DGP and earlier limit order models. The body sets up a model in a Poisson framework and solves the equilibrium phases backward (steady state, convergence, transition phase). It then derives cross-sectional and time-series empirical predictions through comparative statics in news frequency, discusses extensions and limitations, and gives a short summary conclusion. Proofs are in the appendix, with fuller versions in an Internet Appendix.\",\"writing_profile_keywords\":[\"equilibrium model\",\"limit order book\",\"comparative statics\",\"testable predictions\",\"mechanism-driven intro\",\"proofs in appendix\",\"stylized model caveats\"],\"confidence\":0.85,\"notes\":\"The empirical implications section contains model-derived predictions, not data analysis. The extensions section discusses limitations rather than robustness checks.\"}", "json": {"genre": "theoretical", "subfield": "market_microstructure", "subfield_secondary": "asset_pricing", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model; comparative statics with respect to the news arrival frequency)", "strategy": "not applicable; equilibrium analysis of a dynamic limit order market in a DGP-style Poisson framework, with numerical comparative statics", "key_assumptions": ["Poisson arrival of unscheduled news", "stochastic investor reaction delays", "continuous idiosyncratic liquidity shocks", "aggregate preferences in steady state", "one-tick market with binding spread"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": false, "sentence_index": -1}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 4}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}], "contribution_paragraph_index": 6, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "No tables. Results appear as propositions, with proofs in the appendix and fuller versions in an Internet Appendix. Comparative statics and dynamics are shown in figures whose captions state parameter values and time units.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["first to my knowledge", "explicit acknowledgment that the model is stylized, with listed limitations in the extensions section", "conditional comparative-statics phrasing (holding price variability constant)"]}, "claim_evidence": [{"claim": "More frequent news raises the adverse selection cost of limit orders, which lowers market depth.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Appendix proofs; Figure 5"}, {"claim": "News arrival causes order flow and book imbalances and a volume spike, giving a positive covariance between volatility and volume.", "evidence_type": "figure", "evidence_ref": "Figure 6; Section III.E-F"}, {"claim": "Holding unconditional price variability fixed, news frequency reduces the variability-volume covariance.", "evidence_type": "figure", "evidence_ref": "Figure 5; Section III.F"}, {"claim": "Higher news frequency shortens the transition phase and raises the trades' share of order outflow.", "evidence_type": "figure", "evidence_ref": "Figure 5; Sections III.C-D"}], "structural_summary": "A theory paper. The introduction motivates the question with electronic limit order books and the winner's curse that unscheduled news creates. It then sketches the equilibrium mechanism across several paragraphs and closes with a contribution claim and literature positioning relative to DGP and earlier limit order models. The body sets up a model in a Poisson framework and solves the equilibrium phases backward (steady state, convergence, transition phase). It then derives cross-sectional and time-series empirical predictions through comparative statics in news frequency, discusses extensions and limitations, and gives a short summary conclusion. Proofs are in the appendix, with fuller versions in an Internet Appendix.", "writing_profile_keywords": ["equilibrium model", "limit order book", "comparative statics", "testable predictions", "mechanism-driven intro", "proofs in appendix", "stylized model caveats"], "confidence": 0.85, "notes": "The empirical implications section contains model-derived predictions, not data analysis. The extensions section discusses limitations rather than robustness checks."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 13.77, "usage": {"input": 5154, "output": 1793, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.05727, "request_id": "req_011Cggq7LXkiiGuLEBGdphZE", "cached": false, "ts": "2026-11-03T23:04:10+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/annotations_table.csv +40 −0
@@ -0,0 +1,40 @@
1 +doc_id,schema_valid,genre,subfield,primary_method,causal_strength,uses_causal_claim,abs_question,abs_gap,abs_method,abs_results,abs_magnitude,abs_contribution,abs_question_idx,abs_gap_idx,abs_method_idx,abs_results_idx,abs_magnitude_idx,abs_contribution_idx,intro_moves_seq,intro_n_moves,contribution_para_idx,contribution_types,roadmap_llm,roadmap_regex,lit_positioning,econ_signif,stat_lang,magnitude_in_intro,robust_location,robust_types,se_desc,units_in_notes,fig_uncertainty,concl_bounded,concl_policy,concl_open,hedging,n_claims,claims_with_evidence,llm_confidence,abstract_n_sent,intro_n_paragraphs,intro_n_words,magnitude_regex_abstract,contribution_regex_intro,clustered_se_regex,stars_regex,role_agreement,cost_usd,model,validation_errors
2 +jof_12407,True,empirical,corporate_finance,instrumental_variables,4.0,True,False,False,True,True,False,True,-1.0,-1.0,1.0,0.0,-1.0,3.0,motivation|preview_of_approach|preview_of_results|identification_defense|preview_of_results|preview_of_results|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|roadmap,10.0,-1.0,new_evidence|new_data_or_measure|new_identification,True,True,separate_section,True,stars_and_tstats,True,mixed,placebo|alternative_measure|subsample|other,unclear,True,no_figures,False,False,False,low,7.0,7.0,0.8,4.0,6.0,855.0,False,False,False,False,1.0,0.06673,claude-opus-5-5,
3 +jof_12432,True,empirical,corporate_finance,reduced_form_regression,3.0,True,True,True,False,True,False,True,1.0,0.0,-1.0,2.0,-1.0,1.0,motivation|related_literature_positioning|literature_gap|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|identification_defense|robustness_mention|contribution_statement|policy_or_broader_implication|roadmap,12.0,11.0,new_evidence|new_identification,True,True,both,True,unclear,True,main_text_section,alternative_measure|placebo|alternative_specification|other,unclear,True,no_figures,False,False,True,moderate,7.0,7.0,0.8,5.0,14.0,1227.0,False,False,False,False,1.0,0.080506,claude-opus-5-5,
4 +jof_12434,True,empirical,corporate_finance,difference_in_differences,4.0,True,False,False,True,True,False,True,-1.0,-1.0,2.0,1.0,-1.0,0.0,motivation|preview_of_approach|identification_defense|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|robustness_mention|identification_defense|contribution_statement|related_literature_positioning|other|policy_or_broader_implication,13.0,7.0,new_evidence|new_identification,False,False,inside_introduction,True,unclear,True,mixed,placebo|alternative_specification|subsample|falsification,unclear,True,unclear,True,True,False,moderate,7.0,7.0,0.85,3.0,9.0,1304.0,False,True,False,False,0.75,0.079782,claude-opus-5-5,
5 +jof_12438,True,mixed_theory_empirical,asset_pricing,structural_estimation,2.0,False,True,False,True,True,False,True,0.0,-1.0,1.0,2.0,-1.0,0.0,motivation|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|literature_gap|contribution_statement|policy_or_broader_implication|roadmap,14.0,0.0,new_data_or_measure|resolution_of_puzzle|new_evidence|new_method,True,True,inside_introduction,True,unclear,True,within_results,alternative_measure|alternative_specification|subsample|alternative_sample_period,unclear,True,mixed,True,False,False,moderate,5.0,5.0,0.8,4.0,8.0,1290.0,False,True,False,False,1.0,0.082186,claude-opus-5-5,
6 +jof_12439,True,empirical,corporate_finance,difference_in_differences,4.0,True,True,False,True,True,True,True,0.0,-1.0,2.0,2.0,2.0,0.0,motivation|literature_gap|contribution_statement|preview_of_approach|preview_of_results|robustness_mention|identification_defense|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|policy_or_broader_implication|roadmap,12.0,0.0,new_evidence|new_identification,True,True,inside_introduction,True,unclear,True,mixed,placebo|alternative_specification|falsification,unclear,True,unclear,True,True,False,moderate,6.0,6.0,0.85,6.0,6.0,1025.0,True,True,False,False,0.5,0.066598,claude-opus-5-5,
7 +jof_12464,False,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,"'causal_language_strength' is a required property @ identification; 'uses_causal_claim' is a required property @ identification; [""Causal language strength rated 4: the abstract says booms 'form in response to' and the conclusion says the risk premium 'explains' waves, but the evidence is @ notes"
8 +jof_12465,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,3.0,True,True,False,True,True,False,False,0.0,-1.0,4.0,1.0,-1.0,-1.0,motivation|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|identification_defense|preview_of_results|robustness_mention|related_literature_positioning|related_literature_positioning|contribution_statement,11.0,8.0,new_evidence|new_data_or_measure|new_theory_or_model,False,False,inside_introduction,True,unclear,True,mixed,alternative_measure|alternative_sample_period|alternative_specification|subsample|other,unclear,True,no,True,False,False,moderate,6.0,6.0,0.8,5.0,9.0,1485.0,False,True,False,False,1.0,0.073646,claude-opus-5-5,
9 +jof_12466,True,empirical,asset_pricing,reduced_form_regression,3.0,True,False,False,True,True,False,True,-1.0,-1.0,0.0,0.0,-1.0,2.0,motivation|related_literature_positioning|literature_gap|mechanism_or_theory_sketch|identification_defense|preview_of_results|preview_of_results|preview_of_results|contribution_statement|roadmap,10.0,4.0,new_evidence|new_data_or_measure,True,True,both,True,unclear,True,mixed,alternative_measure|alternative_specification|subsample|other,unclear,True,no_figures,True,False,False,moderate,8.0,7.0,0.8,3.0,6.0,872.0,False,True,False,False,1.0,0.075834,claude-opus-5-5,
10 +jof_12467,True,empirical,financial_intermediation_banking,reduced_form_regression,3.0,True,True,False,True,True,False,True,0.0,-1.0,1.0,2.0,-1.0,0.0,motivation|policy_or_broader_implication|mechanism_or_theory_sketch|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_approach|preview_of_results|identification_defense|related_literature_positioning|contribution_statement|related_literature_positioning,11.0,4.0,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,within_results,subsample|alternative_measure|alternative_specification|other,unclear,True,unclear,True,False,True,moderate,6.0,6.0,0.8,5.0,6.0,1076.0,False,False,False,False,1.0,0.07403,claude-opus-5-5,
11 +jof_12468,True,empirical,entrepreneurial_finance_private_equity,difference_in_differences,4.0,True,False,False,True,True,False,True,-1.0,-1.0,1.0,2.0,-1.0,0.0,motivation|literature_gap|policy_or_broader_implication|preview_of_approach|preview_of_approach|preview_of_results|robustness_mention|preview_of_results|contribution_statement|related_literature_positioning,10.0,3.0,new_evidence|new_identification,False,False,inside_introduction,True,unclear,True,within_results,subsample|alternative_specification,unclear,False,yes,True,False,True,moderate,6.0,5.0,0.85,4.0,4.0,752.0,False,True,False,False,1.0,0.062382,claude-opus-5-5,
12 +jof_12469,True,empirical,household_finance,natural_experiment_event_study,4.0,True,True,False,True,True,False,True,0.0,-1.0,1.0,2.0,-1.0,1.0,tension_or_puzzle|literature_gap|contribution_statement|preview_of_approach|mechanism_or_theory_sketch|identification_defense|identification_defense|preview_of_approach|identification_defense|preview_of_results|preview_of_approach|preview_of_results|policy_or_broader_implication|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,18.0,1.0,new_evidence|new_identification|new_method,True,True,inside_introduction,True,stars_and_tstats,True,within_results,alternative_specification|alternative_measure|falsification|subsample|other,clustered,True,unclear,True,False,True,moderate,7.0,7.0,0.85,6.0,15.0,1483.0,False,True,False,False,0.75,0.07947,claude-opus-5-5,
13 +jof_12470,True,empirical,entrepreneurial_finance_private_equity,survey_or_experiment,4.0,True,True,False,True,True,False,True,0.0,-1.0,1.0,2.0,-1.0,4.0,motivation|literature_gap|contribution_statement|mechanism_or_theory_sketch|identification_defense|preview_of_approach|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|other|preview_of_results|related_literature_positioning|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,17.0,0.0,new_evidence|new_identification,True,True,inside_introduction,False,unclear,False,main_text_section,subsample|alternative_measure|alternative_specification|other,unclear,True,no,True,False,False,moderate,5.0,5.0,0.85,5.0,13.0,1529.0,False,True,False,False,0.6,0.076878,claude-opus-5-5,
14 +jof_12471,True,mixed_theory_empirical,asset_pricing,structural_estimation,2.0,False,True,False,True,True,False,True,0.0,-1.0,1.0,3.0,-1.0,0.0,motivation|related_literature_positioning|preview_of_approach|contribution_statement|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning|related_literature_positioning|contribution_statement|identification_defense|related_literature_positioning,11.0,1.0,new_evidence|new_theory_or_model,False,False,inside_introduction,True,mixed,True,main_text_section,alternative_sample_period|alternative_specification|sensitivity_to_parameters|other,newey_west,False,unclear,True,False,False,moderate,8.0,8.0,0.8,6.0,6.0,936.0,False,True,False,False,0.9,0.086406,claude-opus-5-5,
15 +jof_12472,True,empirical,market_microstructure,reduced_form_regression,2.0,False,True,False,True,True,True,True,0.0,-1.0,0.0,1.0,2.0,7.0,motivation|preview_of_approach|related_literature_positioning|preview_of_approach|other|preview_of_results|preview_of_results|identification_defense|contribution_statement|roadmap,10.0,4.0,new_evidence|new_data_or_measure,True,True,both,True,unclear,True,mixed,placebo|subsample|alternative_measure,unclear,True,unclear,True,False,True,moderate,7.0,7.0,0.82,8.0,5.0,779.0,True,True,False,False,1.0,0.069414,claude-opus-5-5,
16 +jof_12473,True,mixed_theory_empirical,household_finance,reduced_form_regression,2.0,False,True,False,True,True,False,True,0.0,-1.0,0.0,1.0,-1.0,3.0,motivation|literature_gap|preview_of_results|robustness_mention|identification_defense|contribution_statement|mechanism_or_theory_sketch|related_literature_positioning|roadmap,9.0,3.0,new_evidence|new_theory_or_model,True,True,both,True,unclear,True,mixed,subsample|alternative_measure|alternative_specification|other,unclear,False,unclear,True,False,True,moderate,6.0,6.0,0.8,4.0,7.0,1009.0,False,True,False,False,0.857,0.071746,claude-opus-5-5,
17 +jof_12482,True,empirical,household_finance,reduced_form_regression,3.0,True,True,False,False,True,False,True,0.0,-1.0,-1.0,2.0,-1.0,4.0,motivation|related_literature_positioning|contribution_statement|mechanism_or_theory_sketch|motivation|preview_of_approach|identification_defense|preview_of_results|preview_of_results|identification_defense|robustness_mention|policy_or_broader_implication,12.0,0.0,new_evidence,False,False,both,True,unclear,False,mixed,subsample|alternative_measure|alternative_specification|falsification|other,unclear,True,unclear,False,True,True,moderate,8.0,8.0,0.8,5.0,6.0,938.0,False,False,False,False,1.0,0.080942,claude-opus-5-5,
18 +jof_12483,True,empirical,entrepreneurial_finance_private_equity,difference_in_differences,3.0,True,True,False,False,True,False,False,0.0,-1.0,-1.0,1.0,-1.0,-1.0,motivation|related_literature_positioning|preview_of_approach|identification_defense|preview_of_results|identification_defense|preview_of_results|mechanism_or_theory_sketch|preview_of_results|contribution_statement,10.0,3.0,new_evidence,False,False,both,True,unclear,False,mixed,placebo|alternative_specification|alternative_measure,unclear,True,no_figures,True,False,False,moderate,7.0,7.0,0.8,6.0,4.0,860.0,False,True,False,False,1.0,0.067506,claude-opus-5-5,
19 +jof_12484,True,mixed_theory_empirical,household_finance,structural_estimation,2.0,False,False,False,True,True,False,True,-1.0,-1.0,1.0,0.0,-1.0,0.0,contribution_statement|preview_of_approach|related_literature_positioning|tension_or_puzzle|preview_of_approach|identification_defense|preview_of_results|literature_gap,8.0,0.0,new_evidence|new_data_or_measure|new_theory_or_model|resolution_of_puzzle,False,False,inside_introduction,True,unclear,True,internet_appendix,alternative_specification,unclear,True,unclear,True,False,False,moderate,5.0,4.0,0.8,2.0,4.0,752.0,False,True,False,False,1.0,0.071706,claude-opus-5-5,
20 +jof_12485,True,empirical,market_microstructure,natural_experiment_event_study,3.0,True,True,False,False,True,False,True,0.0,-1.0,-1.0,1.0,-1.0,3.0,motivation|related_literature_positioning|literature_gap|preview_of_results|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|contribution_statement,13.0,6.0,new_evidence,False,False,inside_introduction,True,unclear,True,mixed,alternative_specification|alternative_measure|subsample|other,unclear,False,no,False,True,False,moderate,7.0,7.0,0.8,6.0,7.0,1343.0,False,True,False,False,1.0,0.07733,claude-opus-5-5,
21 +jof_12486,True,mixed_theory_empirical,asset_pricing,structural_estimation,2.0,False,True,True,True,True,False,True,0.0,1.0,2.0,2.0,-1.0,3.0,motivation|tension_or_puzzle|preview_of_approach|contribution_statement|related_literature_positioning|preview_of_results|mechanism_or_theory_sketch|identification_defense|other|preview_of_results|identification_defense|preview_of_results|related_literature_positioning|roadmap,14.0,2.0,new_method|new_evidence|new_data_or_measure,True,True,inside_introduction,True,mixed,False,appendix,alternative_specification|alternative_measure|other,newey_west,True,mixed,True,False,True,moderate,7.0,7.0,0.8,4.0,12.0,2034.0,False,False,False,False,1.0,0.10605,claude-opus-5-5,
22 +jof_12487,True,empirical,corporate_finance,reduced_form_regression,2.0,False,False,False,True,True,False,False,-1.0,-1.0,0.0,1.0,-1.0,-1.0,motivation|tension_or_puzzle|related_literature_positioning|literature_gap|preview_of_approach|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|robustness_mention|roadmap,10.0,-1.0,new_evidence,True,True,inside_introduction,False,stars_and_tstats,False,within_results,alternative_specification|alternative_measure|placebo|other,unclear,True,no,True,False,False,moderate,7.0,7.0,0.8,3.0,6.0,1027.0,False,False,False,True,0.9,0.084146,claude-opus-5-5,
23 +jof_12488,True,mixed_theory_empirical,asset_pricing,structural_estimation,1.0,False,False,False,True,True,False,True,-1.0,-1.0,0.0,2.0,-1.0,1.0,motivation|literature_gap|contribution_statement|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|preview_of_results|roadmap,12.0,0.0,new_theory_or_model|new_evidence,True,True,inside_introduction,True,mixed,True,within_results,alternative_specification|subsample,mixed,True,unclear,False,False,True,low,7.0,6.0,0.85,3.0,9.0,1406.0,False,True,False,False,1.0,0.076602,claude-opus-5-5,
24 +jof_12489,False,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,Additional properties are not allowed ('figures_note_placeholder' was unexpected) @ figures
25 +jof_12490,True,mixed_theory_empirical,corporate_finance,instrumental_variables,3.0,True,True,False,True,True,True,True,2.0,-1.0,2.0,3.0,1.0,4.0,motivation|tension_or_puzzle|preview_of_approach|preview_of_approach|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|identification_defense|preview_of_results|preview_of_results|robustness_mention|contribution_statement|related_literature_positioning|policy_or_broader_implication,14.0,6.0,new_evidence|new_data_or_measure|new_theory_or_model,False,False,both,True,unclear,True,within_results,alternative_measure|alternative_specification,unclear,True,no,True,True,True,moderate,8.0,8.0,0.8,6.0,8.0,1027.0,True,False,False,False,1.0,0.08025,claude-opus-5-5,
26 +jof_12492,True,empirical,corporate_finance,instrumental_variables,3.0,True,True,False,True,True,True,False,0.0,-1.0,1.0,2.0,2.0,-1.0,motivation|motivation|literature_gap|tension_or_puzzle|contribution_statement|preview_of_approach|preview_of_approach|identification_defense|related_literature_positioning|related_literature_positioning|other,11.0,6.0,new_evidence|new_data_or_measure|new_identification,False,False,inside_introduction,True,unclear,False,mixed,alternative_measure|alternative_specification|subsample|other,unclear,True,unclear,True,True,False,moderate,9.0,9.0,0.8,6.0,12.0,1457.0,False,True,False,False,0.667,0.088078,claude-opus-5-5,
27 +jof_12493,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,1.0,False,False,False,True,True,False,True,-1.0,-1.0,1.0,0.0,-1.0,0.0,motivation|tension_or_puzzle|contribution_statement|mechanism_or_theory_sketch|preview_of_approach|motivation|motivation|related_literature_positioning|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,14.0,1.0,new_theory_or_model,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,False,False,moderate,5.0,5.0,0.85,5.0,9.0,1642.0,False,True,False,False,1.0,0.06903,claude-opus-5-5,
28 +jof_12494,False,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,Additional properties are not allowed ('figures_note_placeholder_removed' was unexpected) @ figures
29 +jof_12495,True,empirical,entrepreneurial_finance_private_equity,instrumental_variables,4.0,True,True,True,True,True,False,False,1.0,0.0,1.0,2.0,-1.0,-1.0,motivation|literature_gap|motivation|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|mechanism_or_theory_sketch|robustness_mention|related_literature_positioning|roadmap,11.0,1.0,new_evidence|new_identification,True,True,inside_introduction,True,unclear,False,mixed,alternative_measure|alternative_specification|subsample|alternative_sample_period|other,unclear,True,no,True,False,True,moderate,8.0,8.0,0.8,5.0,7.0,1049.0,False,True,False,False,1.0,0.072966,claude-opus-5-5,
30 +jof_12496,True,empirical,market_microstructure,reduced_form_regression,3.0,True,False,False,False,True,False,True,-1.0,-1.0,-1.0,2.0,-1.0,3.0,motivation|tension_or_puzzle|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement|preview_of_approach|preview_of_results|preview_of_results|policy_or_broader_implication|roadmap,10.0,2.0,new_evidence,True,True,inside_introduction,True,unclear,False,none_seen,,unclear,True,unclear,True,True,False,moderate,7.0,7.0,0.8,5.0,6.0,800.0,False,True,False,False,1.0,0.073094,claude-opus-5-5,
31 +jof_12498,True,empirical,market_microstructure,reduced_form_regression,2.0,False,True,False,True,True,False,False,0.0,-1.0,2.0,2.0,-1.0,-1.0,motivation|motivation|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|related_literature_positioning|preview_of_results|contribution_statement|roadmap,10.0,45.0,new_evidence,True,True,inside_introduction,True,unclear,True,none_seen,,unclear,True,unclear,True,True,False,moderate,6.0,6.0,0.72,6.0,46.0,1247.0,False,True,False,False,0.5,0.062926,claude-opus-5-5,
32 +jof_12499,False,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,Additional properties are not allowed ('convention_notes_placeholder_removed' was unexpected) @ figures
33 +jof_12500,True,mixed_theory_empirical,household_finance,instrumental_variables,4.0,True,True,False,True,True,True,True,0.0,-1.0,1.0,2.0,3.0,0.0,motivation|tension_or_puzzle|contribution_statement|mechanism_or_theory_sketch|preview_of_results|identification_defense|preview_of_approach|preview_of_results|preview_of_approach|identification_defense|preview_of_results|policy_or_broader_implication|roadmap,13.0,1.0,new_evidence|new_identification|new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|subsample|sensitivity_to_parameters|other,unclear,False,no,True,False,False,moderate,6.0,6.0,0.8,4.0,7.0,1045.0,True,True,False,False,0.923,0.073714,claude-opus-5-5,
34 +jof_12501,True,mixed_theory_empirical,asset_pricing,structural_estimation,2.0,False,True,False,True,True,False,True,0.0,-1.0,0.0,1.0,-1.0,4.0,motivation|preview_of_results|preview_of_results|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|related_literature_positioning|preview_of_results|policy_or_broader_implication|preview_of_results|related_literature_positioning|literature_gap|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,17.0,9.0,new_evidence|new_theory_or_model,True,True,inside_introduction,True,confidence_intervals,True,appendix,alternative_specification,none,True,yes,True,False,False,moderate,6.0,6.0,0.8,6.0,11.0,1501.0,False,True,False,False,1.0,0.08133,claude-opus-5-5,
35 +jof_12507,True,empirical,market_microstructure,regression_discontinuity,4.0,True,True,False,True,True,False,True,0.0,-1.0,2.0,3.0,-1.0,1.0,motivation|tension_or_puzzle|preview_of_approach|related_literature_positioning|preview_of_results|identification_defense|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|literature_gap|roadmap,11.0,0.0,new_evidence|new_identification,True,True,inside_introduction,True,unclear,False,within_results,alternative_specification|alternative_measure|placebo|sensitivity_to_parameters,unclear,True,unclear,True,True,False,moderate,7.0,7.0,0.85,5.0,5.0,831.0,False,False,False,False,0.8,0.066786,claude-opus-5-5,
36 +jof_12509,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2.0,False,False,False,True,True,False,False,-1.0,-1.0,0.0,2.0,-1.0,-1.0,motivation|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,13.0,6.0,new_evidence|new_theory_or_model,True,True,inside_introduction,True,unclear,True,main_text_section,alternative_measure|alternative_specification|alternative_sample_period,unclear,False,no,True,False,False,moderate,7.0,7.0,0.85,8.0,10.0,1171.0,False,False,False,False,1.0,0.076466,claude-opus-5-5,
37 +jof_12514,True,empirical,household_finance,reduced_form_regression,3.0,True,False,False,True,True,True,False,-1.0,-1.0,0.0,1.0,4.0,-1.0,motivation|literature_gap|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|other|identification_defense|preview_of_results|contribution_statement|roadmap,11.0,4.0,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,within_results,other,unclear,True,mixed,False,False,False,moderate,7.0,7.0,0.8,5.0,6.0,797.0,True,True,False,False,1.0,0.068798,claude-opus-5-5,
38 +jof_12530,True,survey_or_address,financial_econometrics,other,2.0,False,True,True,True,True,False,True,4.0,3.0,5.0,1.0,-1.0,6.0,motivation|tension_or_puzzle|mechanism_or_theory_sketch|related_literature_positioning|policy_or_broader_implication|roadmap,6.0,2.0,new_method|synthesis,True,False,both,False,mixed,True,none_seen,,none,True,no,True,True,True,moderate,7.0,7.0,0.8,16.0,3.0,600.0,True,True,False,False,0.75,0.072698,claude-opus-5-5,
39 +jof_12589,True,empirical,financial_intermediation_banking,instrumental_variables,4.0,True,True,False,True,True,True,False,0.0,-1.0,0.0,1.0,1.0,-1.0,motivation|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_approach|identification_defense|preview_of_results|identification_defense|preview_of_approach|preview_of_results|preview_of_results|policy_or_broader_implication|contribution_statement|related_literature_positioning|roadmap,14.0,6.0,new_evidence|new_data_or_measure|new_identification|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,placebo|alternative_specification|other,unclear,False,unclear,False,False,False,moderate,6.0,6.0,0.8,4.0,9.0,1408.0,True,True,False,False,1.0,0.09826,claude-opus-5-5,
40 +jof_12590,True,theoretical,corporate_finance,theoretical_model_analytical,1.0,False,False,False,True,True,False,True,-1.0,-1.0,0.0,3.0,-1.0,4.0,motivation|tension_or_puzzle|related_literature_positioning|preview_of_approach|preview_of_approach|preview_of_approach|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|roadmap,13.0,0.0,new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,False,unclear,False,main_text_section,other,none,False,no,True,False,True,moderate,6.0,6.0,0.88,5.0,8.0,1553.0,False,False,False,False,1.0,0.078502,claude-opus-5-5,
added data/annotations/annotations_table.parquet +0 −0

Binary file not shown.

added data/annotations/api_ledger.jsonl +146 −0
@@ -0,0 +1,146 @@
1 +{"ts": "2026-11-03T22:54:20+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7870, "output": 2347, "cache_read": 0, "cache_write": 3968}, "cost_usd": 0.09826, "latency_s": 19.23, "request_id": "req_011CggpMRmJ56eBjjbFYjXy9", "stop_reason": "end_turn", "cache_key": "b98c63f6848a3834d63cb0a3918eaf1b485d7875ac981e68b8ce1b12b0c351da", "code_version": "feb6c9e-dirty", "doc_id": "jof_12589"}
2 +{"ts": "2026-11-03T22:54:38+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8022, "output": 2281, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078502, "latency_s": 18.05, "request_id": "req_011CggpNrJwyPzEQPVYETQou", "stop_reason": "end_turn", "cache_key": "2ac63d13c067c9a6e4963e037323128c54af00d32a50f15c3f61663d475157ab", "code_version": "feb6c9e-dirty", "doc_id": "jof_12590"}
3 +{"ts": "2026-11-03T22:55:23+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6624, "output": 1972, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06673, "latency_s": 17.04, "request_id": "req_011CggpSEVX1d38giR4Rjxbb", "stop_reason": "end_turn", "cache_key": "114afe74e671995a8c1aa4b9fac91a3759b8afabe1235e38bf4f6a7c9493979e", "code_version": "feb6c9e-dirty", "doc_id": "jof_12407"}
4 +{"ts": "2026-11-03T22:55:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8177, "output": 2314, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079782, "latency_s": 19.35, "request_id": "req_011CggpSEWmGLpptcvkQCqXk", "stop_reason": "end_turn", "cache_key": "05028c35361907c71442407bc74ba778a9f8f64668709441e4700ac646c215e0", "code_version": "feb6c9e-dirty", "doc_id": "jof_12434"}
5 +{"ts": "2026-11-03T22:55:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7744, "output": 2385, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07947, "latency_s": 19.78, "request_id": "req_011CggpSEW2QrqobxsXayFPw", "stop_reason": "end_turn", "cache_key": "2ef52b123465f7d1de149601106d2aaf2ea333486f2c9d1d69510c7e5a24d054", "code_version": "feb6c9e-dirty", "doc_id": "jof_12469"}
6 +{"ts": "2026-11-03T22:55:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7888, "output": 2408, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080506, "latency_s": 20.99, "request_id": "req_011CggpSEWGEgMr5GvRdYip3", "stop_reason": "end_turn", "cache_key": "9116af05cca6c59a9864a9965c8298b9413155362f97a718c4c537eb5da94e81", "code_version": "feb6c9e-dirty", "doc_id": "jof_12432"}
7 +{"ts": "2026-11-03T22:55:39+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6306, "output": 2029, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066598, "latency_s": 16.64, "request_id": "req_011CggpTW7QiEoBdCikcmxkN", "stop_reason": "end_turn", "cache_key": "99d41201d3e5b0bbd488490b868926d7ebc47018290597924f1059567ffdd399", "code_version": "feb6c9e-dirty", "doc_id": "jof_12439"}
8 +{"ts": "2026-11-03T22:55:42+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8108, "output": 2021, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073646, "latency_s": 17.23, "request_id": "req_011CggpTfUy1ngLBvHDm2Qvu", "stop_reason": "end_turn", "cache_key": "2931d45123c629803b10a2b8482a17ed8ce80e804cd9c05565644bda35c9fc81", "code_version": "feb6c9e-dirty", "doc_id": "jof_12465"}
9 +{"ts": "2026-11-03T22:55:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8616, "output": 2568, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086618, "latency_s": 20.49, "request_id": "req_011CggpThPZKLuuR58E9Q9sk", "stop_reason": "end_turn", "cache_key": "a68b40a852f093ad27958461df974eefe8b948153e6d853e108e6ee44543ee14", "code_version": "feb6c9e-dirty", "doc_id": "jof_12464"}
10 +{"ts": "2026-11-03T22:55:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8958, "output": 2489, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086406, "latency_s": 19.56, "request_id": "req_011CggpTnTALxqDYAnV6YtEi", "stop_reason": "end_turn", "cache_key": "e66c86cf766fc4dedab8dcdffc66f6c5b9dc83933cbe0484e4bfa41b768cc4f7", "code_version": "feb6c9e-dirty", "doc_id": "jof_12471"}
11 +{"ts": "2026-11-03T22:55:57+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6460, "output": 2139, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069414, "latency_s": 17.76, "request_id": "req_011CggpUj8WQTmW1Q6F8gkaQ", "stop_reason": "end_turn", "cache_key": "c712d6a87b2a4e82767caf965f1f50746e4ceec0b45e066e4fbb40cbe9ee805f", "code_version": "feb6c9e-dirty", "doc_id": "jof_12472"}
12 +{"ts": "2026-11-03T22:56:00+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9163, "output": 2237, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082186, "latency_s": 17.92, "request_id": "req_011CggpUwX99UMZ6CSdXUqQv", "stop_reason": "end_turn", "cache_key": "9aef7f4d5fab154669e45ab4728b30a9de73c2ae1f9acdcd14cd940f611fa2b3", "code_version": "feb6c9e-dirty", "doc_id": "jof_12438"}
13 +{"ts": "2026-11-03T22:56:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7843, "output": 1979, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071746, "latency_s": 16.85, "request_id": "req_011CggpVD8FAv98BkAEmNqQN", "stop_reason": "end_turn", "cache_key": "12564936cfa421bf29098a6256c4be0cdf436047fe7f6272a9d491e8f31af073", "code_version": "feb6c9e-dirty", "doc_id": "jof_12473"}
14 +{"ts": "2026-11-03T22:56:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5832, "output": 1913, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062382, "latency_s": 16.88, "request_id": "req_011CggpVEKSPA7VTNM8ZPije", "stop_reason": "end_turn", "cache_key": "84971cd34ee70f51d1d784652edb2486e4348059191347f18e56b095f941a3bd", "code_version": "feb6c9e-dirty", "doc_id": "jof_12468"}
15 +{"ts": "2026-11-03T22:56:16+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7796, "output": 2245, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076878, "latency_s": 18.38, "request_id": "req_011CggpW3Hwckb3yVmdXQ6tj", "stop_reason": "end_turn", "cache_key": "c06ea2a7f8eb82e1a09e257e73230929468ddc19a71a51f067676055dc86772b", "code_version": "feb6c9e-dirty", "doc_id": "jof_12470"}
16 +{"ts": "2026-11-03T22:56:18+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6083, "output": 2119, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067506, "latency_s": 17.36, "request_id": "req_011CggpWGRxWPcC9iNnzmoAU", "stop_reason": "end_turn", "cache_key": "27f19f86ee71c7a8c4a76df91bc799e2b62bc67a354b269fd599a55a629dd275", "code_version": "feb6c9e-dirty", "doc_id": "jof_12483"}
17 +{"ts": "2026-11-03T22:56:21+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7754, "output": 2111, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07403, "latency_s": 18.16, "request_id": "req_011CggpWUprHSiSyMixMJ3cu", "stop_reason": "end_turn", "cache_key": "68d6bae21147f0d83f71c9bca92374523d2d8af9bddc95f1aaa39b1c0f359551", "code_version": "feb6c9e-dirty", "doc_id": "jof_12467"}
18 +{"ts": "2026-11-03T22:56:22+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8442, "output": 2319, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080942, "latency_s": 18.89, "request_id": "req_011CggpWTUj8goewU5fubnP1", "stop_reason": "end_turn", "cache_key": "db59df98a24e3e097eed12234df17027d1e3ffed50ad9453c97d169888559b15", "code_version": "feb6c9e-dirty", "doc_id": "jof_12482"}
19 +{"ts": "2026-11-03T22:56:35+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7618, "output": 2022, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071706, "latency_s": 17.17, "request_id": "req_011CggpXYsCnTrsijZKtwQJX", "stop_reason": "end_turn", "cache_key": "717c143d09808825d1abd93d9666ae7578fadcbcb4797fdb9520c8fd7244b203", "code_version": "feb6c9e-dirty", "doc_id": "jof_12484"}
20 +{"ts": "2026-11-03T22:56:36+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7657, "output": 2259, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076602, "latency_s": 20.64, "request_id": "req_011CggpXR9ceQyG8aAaTdHLQ", "stop_reason": "end_turn", "cache_key": "91406d7ea71f4af36f093c845e488c680ac089006700f5254bce5b1f76854275", "code_version": "feb6c9e-dirty", "doc_id": "jof_12488"}
21 +{"ts": "2026-11-03T22:56:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7330, "output": 2286, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075834, "latency_s": 18.76, "request_id": "req_011CggpXpfyceYJzHT1Gh13g", "stop_reason": "end_turn", "cache_key": "7dbd22b2e37326f592e271c10deb1c4c7e38d9fc11816feaa681d9a362ccdcf4", "code_version": "feb6c9e-dirty", "doc_id": "jof_12466"}
22 +{"ts": "2026-11-03T22:56:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8499, "output": 2273, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08025, "latency_s": 18.45, "request_id": "req_011CggpXrUs8VJiNd5tJs5TJ", "stop_reason": "end_turn", "cache_key": "ec4e2b7d9b181c1ac74b0902ce64758fb5ad1b196e32d58304e5b1519fa848b8", "code_version": "feb6c9e-dirty", "doc_id": "jof_12490"}
23 +{"ts": "2026-11-03T22:56:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7036, "output": 2229, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073518, "latency_s": 18.05, "request_id": "req_011CggpYvguM4oRi2dVp599B", "stop_reason": "end_turn", "cache_key": "b3d4ec80ca625e673836f522671ade7a9b9a1bf8b2b34accfed4d18a16f7bfa6", "code_version": "feb6c9e-dirty", "doc_id": "jof_12489"}
24 +{"ts": "2026-11-03T22:56:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8623, "output": 2443, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084146, "latency_s": 20.05, "request_id": "req_011CggpYpd1yJoXawyuvzD99", "stop_reason": "end_turn", "cache_key": "ba6e81c8ca490d18b286e9587c704cb0f1674e4a734bb0e3564dcce389193f04", "code_version": "feb6c9e-dirty", "doc_id": "jof_12487"}
25 +{"ts": "2026-11-03T22:56:59+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6067, "output": 2216, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069382, "latency_s": 18.07, "request_id": "req_011CggpZDf7Ywsc1Kb3AhxNU", "stop_reason": "end_turn", "cache_key": "6c16d6e4b905758791709941f97f8c9c578b541e63f152ff29bbee0ab842c3d9", "code_version": "feb6c9e-dirty", "doc_id": "jof_12494"}
26 +{"ts": "2026-11-03T22:57:01+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9001, "output": 2564, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088078, "latency_s": 20.52, "request_id": "req_011CggpZDKXBkaGYQ6qHe2x2", "stop_reason": "end_turn", "cache_key": "469c06408ebb74919053bfe6cc5d25c76d4ce1b1613ad7dffcdd7ab3c577a116", "code_version": "feb6c9e-dirty", "doc_id": "jof_12492"}
27 +{"ts": "2026-11-03T22:57:11+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8070, "output": 2001, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073094, "latency_s": 16.33, "request_id": "req_011CggpaGDvrU3R5jfFgrJGt", "stop_reason": "end_turn", "cache_key": "391304afeba68f9701a99c579179912ab8baa4c9f9ba1d9d23d97a283811872c", "code_version": "feb6c9e-dirty", "doc_id": "jof_12496"}
28 +{"ts": "2026-11-03T22:57:13+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6950, "output": 2256, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073714, "latency_s": 17.58, "request_id": "req_011CggpaJcYT4sHsYt5K6xW6", "stop_reason": "end_turn", "cache_key": "e1636c3eeaeaba77158b1e7df9bfddf70df78c48c82a5e900fe71401d07292e1", "code_version": "feb6c9e-dirty", "doc_id": "jof_12500"}
29 +{"ts": "2026-11-03T22:57:17+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6884, "output": 2035, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06903, "latency_s": 16.04, "request_id": "req_011CggpajDPXTVamw2kLieaY", "stop_reason": "end_turn", "cache_key": "17b3a5e1159d8f3551100a95a70b527a120b13b0f02af56db8da2d5812363c2d", "code_version": "feb6c9e-dirty", "doc_id": "jof_12493"}
30 +{"ts": "2026-11-03T22:57:21+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8219, "output": 2639, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08645, "latency_s": 22.64, "request_id": "req_011CggpaZC8nFHh4jFxnj4se", "stop_reason": "end_turn", "cache_key": "68573fb6287686843e7ef5441b110b3ba2ecb45f4855c727763d828684bbac10", "code_version": "feb6c9e-dirty", "doc_id": "jof_12499"}
31 +{"ts": "2026-11-03T22:57:29+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7584, "output": 2310, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07733, "latency_s": 18.23, "request_id": "req_011CggpbUMxCwXjHoWuYjC65", "stop_reason": "end_turn", "cache_key": "80b4001d204f9b31d629d29ab3efed964154a654f9940d1e9516a6f452f32ef8", "code_version": "feb6c9e-dirty", "doc_id": "jof_12485"}
32 +{"ts": "2026-11-03T22:57:33+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6018, "output": 1903, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062926, "latency_s": 16.29, "request_id": "req_011CggpbvDTrBjcqYfxnxnAv", "stop_reason": "end_turn", "cache_key": "13439cad7752a6bfe1e44cee2d43deedf217ab4cb6b42e014d47e97b88dcd43a", "code_version": "feb6c9e-dirty", "doc_id": "jof_12498"}
33 +{"ts": "2026-11-03T22:57:34+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 13739, "output": 2515, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.10605, "latency_s": 21.31, "request_id": "req_011Cggpbc73PJo868918kDex", "stop_reason": "end_turn", "cache_key": "e0bcf3fce9e0f322100133df1798c94ad459737b3d358018fa2cf2aabd2392ca", "code_version": "feb6c9e-dirty", "doc_id": "jof_12486"}
34 +{"ts": "2026-11-03T22:57:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7343, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072966, "latency_s": 18.63, "request_id": "req_011CggpcETKLpNq2RhMDyQ4L", "stop_reason": "end_turn", "cache_key": "3fd3f626f36bea4e9e417df114ee8e28a9afe2330b92c8e4bbe3e0222ba92706", "code_version": "feb6c9e-dirty", "doc_id": "jof_12495"}
35 +{"ts": "2026-11-03T22:57:47+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7476, "output": 2100, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072698, "latency_s": 17.36, "request_id": "req_011CggpcpmpCGxdqD1gXthDB", "stop_reason": "end_turn", "cache_key": "96812bbe77ead8b9ba355452945bee90075875e35e4f0968b86a2a0e2aa91748", "code_version": "feb6c9e-dirty", "doc_id": "jof_12530"}
36 +{"ts": "2026-11-03T22:57:52+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6266, "output": 2147, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068798, "latency_s": 18.07, "request_id": "req_011Cggpd87721PchFpgzdAHV", "stop_reason": "end_turn", "cache_key": "a6304ba055c6f043a96eef5d545d710d2cf2c770eeb03b8b61a03f43b131cb2a", "code_version": "feb6c9e-dirty", "doc_id": "jof_12514"}
37 +{"ts": "2026-11-03T22:57:53+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7288, "output": 2326, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076466, "latency_s": 18.69, "request_id": "req_011CggpdBTWWQeqXj48BGve5", "stop_reason": "end_turn", "cache_key": "0d21fdb91550bc31a988c9d97b023eefcdfc335590edcad60539edb23bb26b89", "code_version": "feb6c9e-dirty", "doc_id": "jof_12509"}
38 +{"ts": "2026-11-03T22:58:00+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8419, "output": 2343, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08133, "latency_s": 20.34, "request_id": "req_011CggpdcTRn5zCxq5bG1H3D", "stop_reason": "end_turn", "cache_key": "5ed40cbf05976698685318b11ae9b53e62e099eb8f9c5b0a6a929a8a03bbfd79", "code_version": "feb6c9e-dirty", "doc_id": "jof_12501"}
39 +{"ts": "2026-11-03T22:58:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6438, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066786, "latency_s": 16.71, "request_id": "req_011Cggpe79LKkBbdTs8VTKky", "stop_reason": "end_turn", "cache_key": "2854eaa79139a46dfe374dc1308865dae423ae87c0efaa8256b21ea2182acace", "code_version": "feb6c9e-dirty", "doc_id": "jof_12507"}
40 +{"ts": "2026-11-03T22:58:08+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5877, "output": 2016, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064622, "latency_s": 16.77, "request_id": "req_011CggpeTfcrxHG5drcG6sTs", "stop_reason": "end_turn", "cache_key": "1f5b570c5679d291d7b01413a321df38a39bce66444c619b7ab81f687e0b4c08", "code_version": "feb6c9e-dirty", "doc_id": "jof_12513"}
41 +{"ts": "2026-11-03T22:58:10+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8026, "output": 2170, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076298, "latency_s": 17.32, "request_id": "req_011CggpeZe3Vsp9duoyQgftp", "stop_reason": "end_turn", "cache_key": "93ff9eb2c8766125b7caa3faa8bd577d003be9ced11f5f9843c630d421eb6304", "code_version": "feb6c9e-dirty", "doc_id": "jof_12506"}
42 +{"ts": "2026-11-03T22:58:18+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7027, "output": 2049, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069882, "latency_s": 17.34, "request_id": "req_011Cggpf7iLfvmEhFxGrLnum", "stop_reason": "end_turn", "cache_key": "2d0f038836fbd5ddea5aba9f28832c2a6766c0547be9fe244ce48fe407c7c9d2", "code_version": "feb6c9e-dirty", "doc_id": "jof_12511"}
43 +{"ts": "2026-11-03T22:58:23+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9752, "output": 2470, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.089202, "latency_s": 19.95, "request_id": "req_011CggpfM1YjYEn5V5xt9y6b", "stop_reason": "end_turn", "cache_key": "27406ebc0ae1376a90cbf92731e906e708ab2bd36e265ede7c2b912382ec6c2e", "code_version": "feb6c9e-dirty", "doc_id": "jof_12512"}
44 +{"ts": "2026-11-03T22:58:26+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5540, "output": 2121, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065374, "latency_s": 17.81, "request_id": "req_011CggpfhfGLy8SvPtJsyyAY", "stop_reason": "end_turn", "cache_key": "a9967e4cd6bc9065fc8d350988734904461d72b9a4bed0778a3dc780cc04b88e", "code_version": "feb6c9e-dirty", "doc_id": "jof_12505"}
45 +{"ts": "2026-11-03T22:58:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7084, "output": 2132, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07177, "latency_s": 16.91, "request_id": "req_011Cggpfr2oUQttk2HfTG2E9", "stop_reason": "end_turn", "cache_key": "68e686251a536e60fc5256267b033c0f879ca8fab6d203a8a04392209d1bb6a2", "code_version": "feb6c9e-dirty", "doc_id": "jof_12497"}
46 +{"ts": "2026-11-03T22:58:38+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9141, "output": 2343, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084218, "latency_s": 20.28, "request_id": "req_011CggpgQ3daNZ4LAGW4zepz", "stop_reason": "end_turn", "cache_key": "35a082abb20dc96de5483bff59de8242cd9ecfd5396cca14ef9e17b712a4b99e", "code_version": "feb6c9e-dirty", "doc_id": "jof_12520"}
47 +{"ts": "2026-11-03T22:58:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7283, "output": 1962, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069166, "latency_s": 16.1, "request_id": "req_011CggpgpN7qPop4HAoqdN13", "stop_reason": "end_turn", "cache_key": "39e75956c9d05070cbfb1c8c8b982077aea761cdf6b07254f63ae685a9c27cab", "code_version": "feb6c9e-dirty", "doc_id": "jof_12519"}
48 +{"ts": "2026-11-03T22:58:41+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6278, "output": 1937, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064646, "latency_s": 15.11, "request_id": "req_011Cggph274eHbEbFRmDdQqx", "stop_reason": "end_turn", "cache_key": "2ed2f326d2baff68a61418c25f88130c95eba899ef53f2413788a3c2180d1679", "code_version": "feb6c9e-dirty", "doc_id": "jof_12525"}
49 +{"ts": "2026-11-03T22:58:45+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5930, "output": 2034, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065194, "latency_s": 17.13, "request_id": "req_011Cggph8QKmdoupfMwz4MyG", "stop_reason": "end_turn", "cache_key": "3b1a72f736281efa4fdc86c2ad1760beb84c5c5cae0c8ceef5c9f9ad3ac48667", "code_version": "feb6c9e-dirty", "doc_id": "jof_12524"}
50 +{"ts": "2026-11-03T22:58:57+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9197, "output": 2236, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082302, "latency_s": 18.93, "request_id": "req_011Cggphu79gT5xDcCNv15Fo", "stop_reason": "end_turn", "cache_key": "bf8953e88697fe7709378c0602f3c5411b7328d76ac77b5b377f7565a9fdec7a", "code_version": "feb6c9e-dirty", "doc_id": "jof_12508"}
51 +{"ts": "2026-11-03T22:58:58+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 3773, "output": 1852, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.052926, "latency_s": 16.72, "request_id": "req_011Cggpi94nVZTmVD7qfL9Fi", "stop_reason": "end_turn", "cache_key": "902f017174cc08dac668fe95224dd85483e898a2974b1f799c961db6145a4690", "code_version": "feb6c9e-dirty", "doc_id": "jof_12526"}
52 +{"ts": "2026-11-03T22:58:59+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8938, "output": 2405, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084646, "latency_s": 18.86, "request_id": "req_011Cggpi1Y7G1FaQpkW64PC5", "stop_reason": "end_turn", "cache_key": "3c274e8ab145f499cb1251d8a4d1d9206b3d84219d620d040fc2e678af702234", "code_version": "feb6c9e-dirty", "doc_id": "jof_12504"}
53 +{"ts": "2026-11-03T22:59:01+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9999, "output": 2165, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08409, "latency_s": 16.61, "request_id": "req_011CggpiNCK2no8ULgMj8aPY", "stop_reason": "end_turn", "cache_key": "1d6c0149aff0d147ced718c9f83c97460f2e17f64f4bb2c79946fbbab4ddbfef", "code_version": "feb6c9e-dirty", "doc_id": "jof_12521"}
54 +{"ts": "2026-11-03T22:59:16+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7763, "output": 2158, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075006, "latency_s": 18.59, "request_id": "req_011CggpjJVqsfs4uEEvQbLBR", "stop_reason": "end_turn", "cache_key": "46c8de1094f213281ef592eaa709b90a556509f4530eaba3217f210c07b8490d", "code_version": "feb6c9e-dirty", "doc_id": "jof_12518"}
55 +{"ts": "2026-11-03T22:59:17+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7041, "output": 2274, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074438, "latency_s": 18.73, "request_id": "req_011CggpjPkMd3HEbAgrGHxnf", "stop_reason": "end_turn", "cache_key": "77504260fc5c8bf98949cb11272bb50f1bb60d5e4b0dc9c1d0fa08e7160779e4", "code_version": "feb6c9e-dirty", "doc_id": "jof_12522"}
56 +{"ts": "2026-11-03T22:59:18+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8517, "output": 2384, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082542, "latency_s": 19.06, "request_id": "req_011CggpjQVG8CmPPRnu6HP6T", "stop_reason": "end_turn", "cache_key": "64039def73ed5d504824a573cd2581c61f6f2786dc7428ad1c6f748c042a8ddd", "code_version": "feb6c9e-dirty", "doc_id": "jof_12540"}
57 +{"ts": "2026-11-03T22:59:19+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7965, "output": 2234, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077334, "latency_s": 18.09, "request_id": "req_011CggpjbeyRwnQVvhd3o8eJ", "stop_reason": "end_turn", "cache_key": "3805d6b5d9706c0378663a49bf01dc0cb28fa500fe0d6394be94b3ac0a7c1dd2", "code_version": "feb6c9e-dirty", "doc_id": "jof_12523"}
58 +{"ts": "2026-11-03T22:59:34+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6854, "output": 2242, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07305, "latency_s": 18.35, "request_id": "req_011CggpkgFa27VHncb9jkV9D", "stop_reason": "end_turn", "cache_key": "4d6992a34ecf27cb2dd16f7d4bd5776a49f56ce78fe67ed5b805c16b0ec7325d", "code_version": "feb6c9e-dirty", "doc_id": "jof_12542"}
59 +{"ts": "2026-11-03T22:59:35+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7030, "output": 2096, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070834, "latency_s": 17.45, "request_id": "req_011CggpkpE3CkeqHV3TSvy3N", "stop_reason": "end_turn", "cache_key": "16924f337f3237e02506879130ce92a08a596be46796efc91d3cd1c165bd3531", "code_version": "feb6c9e-dirty", "doc_id": "jof_12545"}
60 +{"ts": "2026-11-03T22:59:35+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6074, "output": 1826, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06161, "latency_s": 15.81, "request_id": "req_011CggpkwVcEBK8aSNfuickp", "stop_reason": "end_turn", "cache_key": "b9f37a3f6aa5677023a96b0ecd19ea3352ae0b5fff1d7d25fe7bfe2599468a09", "code_version": "feb6c9e-dirty", "doc_id": "jof_12541"}
61 +{"ts": "2026-11-03T22:59:36+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8190, "output": 2262, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078794, "latency_s": 18.5, "request_id": "req_011CggpknVcEJitikcjo7wVj", "stop_reason": "end_turn", "cache_key": "e14a6041e27e1f7ba1575e3f32d9d015fb374b73e519443959f1bc653f4ae34b", "code_version": "feb6c9e-dirty", "doc_id": "jof_12547"}
62 +{"ts": "2026-11-03T22:59:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 2692, "output": 1536, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.042282, "latency_s": 12.33, "request_id": "req_011Cggpn8ywscmTMrAR9b42d", "stop_reason": "end_turn", "cache_key": "2a9715297db6415cdec3167e8b5c80019034e68d34d40a70aade6795089c2425", "code_version": "feb6c9e-dirty", "doc_id": "jof_12548"}
63 +{"ts": "2026-11-03T22:59:51+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6638, "output": 1929, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065926, "latency_s": 15.79, "request_id": "req_011Cggpn76LRKtVDJEp4cDgQ", "stop_reason": "end_turn", "cache_key": "a2a48a79b5b1b0b53cfad15b34fe527af4950abac3bb4ab7db032802af034d16", "code_version": "feb6c9e-dirty", "doc_id": "jof_12544"}
64 +{"ts": "2026-11-03T22:59:51+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6779, "output": 1998, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06787, "latency_s": 15.99, "request_id": "req_011Cggpn756kc6isVapqRaMf", "stop_reason": "end_turn", "cache_key": "af98ad51ec92a67a895b0b99ac3b5a5420693f2dc0bd7b43346b2e3180fe645a", "code_version": "feb6c9e-dirty", "doc_id": "jof_12510"}
65 +{"ts": "2026-11-03T22:59:51+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6759, "output": 2113, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07009, "latency_s": 16.99, "request_id": "req_011Cggpn37DmwWz9ecfjEuPU", "stop_reason": "end_turn", "cache_key": "27565031974f4a9b3c05db909c11391e43347b0c0fbba4db2e65ea2bf710b220", "code_version": "feb6c9e-dirty", "doc_id": "jof_12546"}
66 +{"ts": "2026-11-03T23:00:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6832, "output": 1800, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064122, "latency_s": 14.85, "request_id": "req_011Cggpo3gyrkHX2tNkBYVUv", "stop_reason": "end_turn", "cache_key": "daf786ca42c98c39ee19b47aade036d419cab2e757a13cea265dca52e24e06db", "code_version": "feb6c9e-dirty", "doc_id": "jof_12588"}
67 +{"ts": "2026-11-03T23:00:07+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5670, "output": 1936, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062194, "latency_s": 15.56, "request_id": "req_011CggpoHhLdfXjkQEptAiWg", "stop_reason": "end_turn", "cache_key": "61f586c8a687efe4fee98bebfe95e190e46201bb0653111cf819f341cf15119c", "code_version": "feb6c9e-dirty", "doc_id": "jof_12584"}
68 +{"ts": "2026-11-03T23:00:08+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5653, "output": 1982, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.063046, "latency_s": 17.23, "request_id": "req_011CggpoJAs4f1ifBATuiYyU", "stop_reason": "end_turn", "cache_key": "095173e62259d89082fb0fc424aa1fea9f7ca3ae9a728e332260a18c01487542", "code_version": "feb6c9e-dirty", "doc_id": "jof_12586"}
69 +{"ts": "2026-11-03T23:00:11+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7666, "output": 2450, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080458, "latency_s": 19.54, "request_id": "req_011CggpoHp3AX4XaoHuxAaUo", "stop_reason": "end_turn", "cache_key": "36006ac1366745a5882d383986a6868311c1ee20dd95c876668765a0f9831720", "code_version": "feb6c9e-dirty", "doc_id": "jof_12585"}
70 +{"ts": "2026-11-03T23:00:22+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7341, "output": 2304, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076238, "latency_s": 18.58, "request_id": "req_011Cggpp9jdXgcMqcfjA3nWQ", "stop_reason": "end_turn", "cache_key": "dce5b7fab6d1c8a0aec5d7f020178a76036a1e3c4577a40932b4497f39edb4e7", "code_version": "feb6c9e-dirty", "doc_id": "jof_12539"}
71 +{"ts": "2026-11-03T23:00:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6777, "output": 2179, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071482, "latency_s": 17.53, "request_id": "req_011CggppSbA8GCouf3nWJZXN", "stop_reason": "end_turn", "cache_key": "62fb6cad517007f71f64d2578aaae8b557476a9f41e25095045b05892277067b", "code_version": "feb6c9e-dirty", "doc_id": "jof_12587"}
72 +{"ts": "2026-11-03T23:00:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7420, "output": 2013, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070734, "latency_s": 16.14, "request_id": "req_011CggppjiYreP4Y1q26AeVm", "stop_reason": "end_turn", "cache_key": "8461f53eeaacbf7c7cec9c17fb0cb2c73bde1ce5c15213403740f88c6ab09eda", "code_version": "feb6c9e-dirty", "doc_id": "jof_12591"}
73 +{"ts": "2026-11-03T23:00:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7726, "output": 2176, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075218, "latency_s": 18.7, "request_id": "req_011CggppYqhVccuPdcDQBtVp", "stop_reason": "end_turn", "cache_key": "5d32141da98f4fd3c0f4cde643e0084d36d59122f6583ef1d3260996099da9a3", "code_version": "feb6c9e-dirty", "doc_id": "jof_12583"}
74 +{"ts": "2026-11-03T23:00:43+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9828, "output": 2396, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088026, "latency_s": 20.97, "request_id": "req_011CggpqXTsTjn69nRgraWhz", "stop_reason": "end_turn", "cache_key": "cbddda01fd301f3269b4d754fd6444f236420becbe9ca7313a26033256f265fa", "code_version": "feb6c9e-dirty", "doc_id": "jof_12593"}
75 +{"ts": "2026-11-03T23:00:44+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7759, "output": 2274, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07731, "latency_s": 19.86, "request_id": "req_011Cggpqjn3o2jc1TEjkgPfH", "stop_reason": "end_turn", "cache_key": "436975d688e36a4927227083acbcc05fdd7dc111cc4d6dd4a0540f6c5afa3be8", "code_version": "feb6c9e-dirty", "doc_id": "jof_12602"}
76 +{"ts": "2026-11-03T23:00:45+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4680, "output": 2057, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.060654, "latency_s": 17.54, "request_id": "req_011CggpqvzEnvVgrfFzbnuD9", "stop_reason": "end_turn", "cache_key": "4d59b6136497e94ba7bb0ee520742fd86af2f0712b475227fdc47d8f2e27ffb5", "code_version": "feb6c9e-dirty", "doc_id": "jof_12606"}
77 +{"ts": "2026-11-03T23:00:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8752, "output": 2321, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082222, "latency_s": 20.57, "request_id": "req_011CggpqwzFcFqqL9gY1AXYW", "stop_reason": "end_turn", "cache_key": "e4a8e813ea4c83e311ef7b04396bb686d22003191b110c1edb8df2a5ddfde76b", "code_version": "feb6c9e-dirty", "doc_id": "jof_12592"}
78 +{"ts": "2026-11-03T23:01:01+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9962, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.083442, "latency_s": 18.16, "request_id": "req_011Cggps5EMkbvL4oUNjSPRL", "stop_reason": "end_turn", "cache_key": "e1d2d3958fbc0f4e81c15dcf4ddb2d8f3de6568557bd3cf3e40f1b4fa948d9ef", "code_version": "feb6c9e-dirty", "doc_id": "jof_12600"}
79 +{"ts": "2026-11-03T23:01:04+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8447, "output": 2312, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080822, "latency_s": 19.0, "request_id": "req_011CggpsD3emDcfqvmpDHv9r", "stop_reason": "end_turn", "cache_key": "3e64809cbee0355608db222ccf008d046d22a1372570424d03c4df29b8a89216", "code_version": "feb6c9e-dirty", "doc_id": "jof_12607"}
80 +{"ts": "2026-11-03T23:01:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6821, "output": 2166, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071398, "latency_s": 18.19, "request_id": "req_011CggpsTJQAJL9BoHfCssfc", "stop_reason": "end_turn", "cache_key": "dc5af954604d2857fcc8d236f78d1544e524c5ecd48ccdd576d52bf3629f2562", "code_version": "feb6c9e-dirty", "doc_id": "jof_12604"}
81 +{"ts": "2026-11-03T23:01:08+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4934, "output": 2500, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07053, "latency_s": 22.63, "request_id": "req_011CggpsEr3sggcL996niFU6", "stop_reason": "end_turn", "cache_key": "466048fb6368fbae0427d80a3e781a24f911ddcd695a480cf08b68afb46908e6", "code_version": "feb6c9e-dirty", "doc_id": "jof_12601"}
82 +{"ts": "2026-11-03T23:01:19+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6023, "output": 2133, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067546, "latency_s": 17.66, "request_id": "req_011CggptREu5Xmii87CayiFh", "stop_reason": "end_turn", "cache_key": "cfec39edd59c95fd48cf93d0b7d85b9690cd34c2716e0dfee0e096f149a176b0", "code_version": "feb6c9e-dirty", "doc_id": "jof_12605"}
83 +{"ts": "2026-11-03T23:01:21+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7800, "output": 2164, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075274, "latency_s": 17.73, "request_id": "req_011CggptcbXCJjzwwWnSpLeS", "stop_reason": "end_turn", "cache_key": "4e650c20bf4f14df88172d81773935f49bc255f2cfb1b93c66d0938e740ba439", "code_version": "feb6c9e-dirty", "doc_id": "jof_12603"}
84 +{"ts": "2026-11-03T23:01:23+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6525, "output": 2098, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068854, "latency_s": 17.12, "request_id": "req_011CggptoPfLDa5p1KWqJVr5", "stop_reason": "end_turn", "cache_key": "2d0b9356ab798fd506afbacc97cf09f75d9ee809d88fd47d46e8ec1ba60238ce", "code_version": "feb6c9e-dirty", "doc_id": "jof_12614"}
85 +{"ts": "2026-11-03T23:01:35+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8895, "output": 3213, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.100634, "latency_s": 27.11, "request_id": "req_011CggptuewogAJ5vrMxMAZo", "stop_reason": "end_turn", "cache_key": "580b3f76810d041a434a6172e92c5a01ce123770c612789fc07d6efdd31e6d27", "code_version": "feb6c9e-dirty", "doc_id": "jof_12615"}
86 +{"ts": "2026-11-03T23:01:36+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5071, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061318, "latency_s": 17.37, "request_id": "req_011Cggpuj5yY8TLUtu2MwPiY", "stop_reason": "end_turn", "cache_key": "471a4f9935dd471327c89504592ec0b66c247ab73d335f7db72174cf8bf2ce4f", "code_version": "feb6c9e-dirty", "doc_id": "jof_12619"}
87 +{"ts": "2026-11-03T23:01:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7219, "output": 2310, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07587, "latency_s": 18.93, "request_id": "req_011CggpuvhjaJ3uS9CjuAwEo", "stop_reason": "end_turn", "cache_key": "f11c1d4f073c80c0f2f1450ea7b5e0911f465e979ecfdc82a3798ca76e58cd2b", "code_version": "feb6c9e-dirty", "doc_id": "jof_12610"}
88 +{"ts": "2026-11-03T23:01:42+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8690, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080594, "latency_s": 18.56, "request_id": "req_011Cggpv4srUfcHvY4htXQxh", "stop_reason": "end_turn", "cache_key": "66c6bbacc5f66e6f4cb33ad83597fdaa1613c45f9a30f38af24e6753a519090f", "code_version": "feb6c9e-dirty", "doc_id": "jof_12612"}
89 +{"ts": "2026-11-03T23:01:49+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5855, "output": 1877, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061754, "latency_s": 14.69, "request_id": "req_011CggpvurA5PMW2cneVnGTZ", "stop_reason": "end_turn", "cache_key": "c8b9faf575e0b5f39d5591adc1f31b80b881e65ce1670ed5b18a9b8e662ef37d", "code_version": "feb6c9e-dirty", "doc_id": "jof_12618"}
90 +{"ts": "2026-11-03T23:01:54+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6992, "output": 2286, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074482, "latency_s": 17.89, "request_id": "req_011Cggpw1bwUb5qw4g8FPHvq", "stop_reason": "end_turn", "cache_key": "52300d61a9177b256b0aeb37eecdde7a2ae40f2a7ed0c3f7907985b1930b8752", "code_version": "feb6c9e-dirty", "doc_id": "jof_12616"}
91 +{"ts": "2026-11-03T23:01:58+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6420, "output": 2070, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067874, "latency_s": 17.74, "request_id": "req_011CggpwKxzWwis9vkDqr1f2", "stop_reason": "end_turn", "cache_key": "740e13c7b987eff9e95784627fc42bc9325408b587b2032e295f09d8a4766656", "code_version": "feb6c9e-dirty", "doc_id": "jof_12609"}
92 +{"ts": "2026-11-03T23:02:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8479, "output": 2508, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08487, "latency_s": 20.98, "request_id": "req_011CggpwSa757mg6NkeWrKJB", "stop_reason": "end_turn", "cache_key": "2ac62d87ff3e8905a031288ca362560fded28341510bdec0fb24fcb37c5e9ef7", "code_version": "feb6c9e-dirty", "doc_id": "jof_12617"}
93 +{"ts": "2026-11-03T23:02:09+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7681, "output": 2413, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079778, "latency_s": 19.08, "request_id": "req_011Cggpwzx26R6Ci7weNs6Tb", "stop_reason": "end_turn", "cache_key": "dfe5433dd5e48f558182c58508b2cd0c9a7cd35eb7bbcad1e409a04ea9e2e995", "code_version": "feb6c9e-dirty", "doc_id": "jof_12613"}
94 +{"ts": "2026-11-03T23:02:13+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8256, "output": 2356, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080938, "latency_s": 18.73, "request_id": "req_011CggpxLktKSEqCHawdPPA2", "stop_reason": "end_turn", "cache_key": "36fc61e29d7e4edf9e3003a1f3c5103056deae144898354b4439fe22222d4cfc", "code_version": "feb6c9e-dirty", "doc_id": "jof_12620"}
95 +{"ts": "2026-11-03T23:02:18+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8143, "output": 2327, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079906, "latency_s": 19.33, "request_id": "req_011Cggpxe6S8PS4aD3CqPySw", "stop_reason": "end_turn", "cache_key": "dbcf558280519460269319f8f1da733a7a4d0d33d3adf3c7fd00ac4d762e0dc5", "code_version": "feb6c9e-dirty", "doc_id": "jof_12611"}
96 +{"ts": "2026-11-03T23:02:22+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8882, "output": 2329, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082902, "latency_s": 19.19, "request_id": "req_011CggpxzyHNGegZam58fwRe", "stop_reason": "end_turn", "cache_key": "199908f8bc7d399578f8ab034e7a91453de4a21bbada406029f92625c441cdf5", "code_version": "feb6c9e-dirty", "doc_id": "jof_12710"}
97 +{"ts": "2026-11-03T23:02:31+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8489, "output": 2452, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08379, "latency_s": 21.89, "request_id": "req_011CggpyQvT4x5LbudWjdLoi", "stop_reason": "end_turn", "cache_key": "0743830246f87c71fda4767d01bb822d368b32f489aa559d5f1133b9f76e5d92", "code_version": "feb6c9e-dirty", "doc_id": "jof_12698"}
98 +{"ts": "2026-11-03T23:02:31+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8269, "output": 2130, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07647, "latency_s": 17.77, "request_id": "req_011CggpyixtzBwxJLpmxpdtR", "stop_reason": "end_turn", "cache_key": "85b14a267230438b12476adf4ab3309d3ac8a4c789d9eefb045ab8e1b481e2f9", "code_version": "feb6c9e-dirty", "doc_id": "jof_12685"}
99 +{"ts": "2026-11-03T23:02:39+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6183, "output": 2057, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066666, "latency_s": 16.48, "request_id": "req_011CggpzR1SNXEjmjWPEQNby", "stop_reason": "end_turn", "cache_key": "327b0caf6a62f7e9717830edf8818e7babae68d31b67ee75dc638badcb08d4b3", "code_version": "feb6c9e-dirty", "doc_id": "jof_12696"}
100 +{"ts": "2026-11-03T23:02:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8055, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076454, "latency_s": 22.18, "request_id": "req_011Cggpz56MnZDUbEpiDsuin", "stop_reason": "end_turn", "cache_key": "2ecb08379d79c8d07ae8ef2a950b8ce9fa0999fa753c1294e0bc0928da10bf8b", "code_version": "feb6c9e-dirty", "doc_id": "jof_12694"}
101 +{"ts": "2026-11-03T23:02:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5439, "output": 2009, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06273, "latency_s": 15.28, "request_id": "req_011Cggq12oV1W83oJgBkdHSU", "stop_reason": "end_turn", "cache_key": "299bc406441467932eec9d28b244c56d8a29bfb606a3ca9458fcf52895b3918d", "code_version": "feb6c9e-dirty", "doc_id": "jof_12689"}
102 +{"ts": "2026-11-03T23:02:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4574, "output": 2084, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06077, "latency_s": 17.2, "request_id": "req_011Cggq13955aR1X3ngVFmCv", "stop_reason": "end_turn", "cache_key": "45b8805e39601d8e791639f18bb37d2764747b025124b270d1682df9ee3fac7d", "code_version": "feb6c9e-dirty", "doc_id": "jof_12688"}
103 +{"ts": "2026-11-03T23:02:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6159, "output": 2065, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06673, "latency_s": 17.55, "request_id": "req_011Cggq1dsNFCpXhUZj8U3gC", "stop_reason": "end_turn", "cache_key": "48aa94471af6f752f948cdf51239c83b1aaccc0db9e317c80b9bf6fba8314e20", "code_version": "feb6c9e-dirty", "doc_id": "jof_12684"}
104 +{"ts": "2026-11-03T23:02:58+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8748, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080826, "latency_s": 17.96, "request_id": "req_011Cggq1iF3NHiLxmnSX3QLq", "stop_reason": "end_turn", "cache_key": "666773bb31a6d893186453ada85a9af560ecc96a7315afb19f041902a5b33be4", "code_version": "feb6c9e-dirty", "doc_id": "jof_12687"}
105 +{"ts": "2026-11-03T23:03:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6803, "output": 1938, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066766, "latency_s": 17.1, "request_id": "req_011Cggq2Af3QB3oknBMjycR1", "stop_reason": "end_turn", "cache_key": "700cd6eede478c4dcb495bcf44e8246b7ba9a3e783316432ff0f92f05548bfc3", "code_version": "feb6c9e-dirty", "doc_id": "jof_12695"}
106 +{"ts": "2026-11-03T23:03:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8039, "output": 2065, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07425, "latency_s": 18.21, "request_id": "req_011Cggq2K2KFt1hZGjCok99m", "stop_reason": "end_turn", "cache_key": "deedd6eb76f35d9eb54d790f4fbb57831883419abb05f25854aabc4d167f5aad", "code_version": "feb6c9e-dirty", "doc_id": "jof_12697"}
107 +{"ts": "2026-11-03T23:03:14+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8184, "output": 2331, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08015, "latency_s": 17.53, "request_id": "req_011Cggq2wa1WmUbXuGjYDRrA", "stop_reason": "end_turn", "cache_key": "65bbba998d9417f425d090d4cac61dfb184a8afedda834e213a9a651dae062bc", "code_version": "feb6c9e-dirty", "doc_id": "jof_12718"}
108 +{"ts": "2026-11-03T23:03:19+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9464, "output": 2428, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08721, "latency_s": 20.68, "request_id": "req_011Cggq33H4gx8GYV5J3xQSv", "stop_reason": "end_turn", "cache_key": "f62fa2f7c3cd71c829f45ca98998fe6d00f992c8d22aaa86ee5e9bf9610f9988", "code_version": "feb6c9e-dirty", "doc_id": "jof_12691"}
109 +{"ts": "2026-11-03T23:03:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8535, "output": 2236, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079654, "latency_s": 19.04, "request_id": "req_011Cggq3fD3bwhUtKUdbbWbB", "stop_reason": "end_turn", "cache_key": "87edcde20968ed5b8e8eb341e154d9b6d9c1c7812d0387498764153fc0bc1ee8", "code_version": "feb6c9e-dirty", "doc_id": "jof_12704"}
110 +{"ts": "2026-11-03T23:03:31+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7788, "output": 2293, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077806, "latency_s": 27.36, "request_id": "req_011Cggq3Rx51rEycaWHgmR1J", "stop_reason": "end_turn", "cache_key": "cbc338de4c33690ff77323b03f16135d822e0098cfc993e635dc010d794e9cf3", "code_version": "feb6c9e-dirty", "doc_id": "jof_12706"}
111 +{"ts": "2026-11-03T23:03:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7669, "output": 2227, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07601, "latency_s": 18.12, "request_id": "req_011Cggq4EWGXCbkSqozjtKs9", "stop_reason": "end_turn", "cache_key": "2e48830ce6dbd2ff98c14df5c17e5620b4b1e6e237a19c949c5dd799152353e3", "code_version": "feb6c9e-dirty", "doc_id": "jof_12700"}
112 +{"ts": "2026-11-03T23:03:38+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6819, "output": 2406, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07619, "latency_s": 19.05, "request_id": "req_011Cggq4a1XoZiNRM3PY3qKy", "stop_reason": "end_turn", "cache_key": "e7d7f9596c805eababe5f6d4b0fc39dc99306774b03b4544daef467c1856c114", "code_version": "feb6c9e-dirty", "doc_id": "jof_12702"}
113 +{"ts": "2026-11-03T23:03:43+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7815, "output": 2165, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075354, "latency_s": 17.73, "request_id": "req_011Cggq54qsG6DMdXyTrK64P", "stop_reason": "end_turn", "cache_key": "0558df49948e0ed9c1bd18fffe9215b398c1b0470c0ca43e425fda8c1b98cb7f", "code_version": "feb6c9e-dirty", "doc_id": "jof_12699"}
114 +{"ts": "2026-11-03T23:03:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7582, "output": 1993, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070982, "latency_s": 15.38, "request_id": "req_011Cggq5TJEKhimziquDM3Kf", "stop_reason": "end_turn", "cache_key": "c3c47e8ed7f92d8e4c908e7c6374f831a6098b8a382a9ff882eafd6050b4b094", "code_version": "feb6c9e-dirty", "doc_id": "jof_12701"}
115 +{"ts": "2026-11-03T23:03:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7310, "output": 2014, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070314, "latency_s": 16.12, "request_id": "req_011Cggq5aKuhSCAdwZ8XFe4J", "stop_reason": "end_turn", "cache_key": "d76871d030d2d80c7ef4d99f74d2b9ee0f49ee8cfc6d02bb8218e2b399a98bb4", "code_version": "feb6c9e-dirty", "doc_id": "jof_12703"}
116 +{"ts": "2026-11-03T23:03:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6793, "output": 2217, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072306, "latency_s": 18.34, "request_id": "req_011Cggq5yonQUcEbrsPLyUME", "stop_reason": "end_turn", "cache_key": "6f086478c29ee5b996c862f7898193f143fbe6fea089a9457b97c5ac6543aff9", "code_version": "feb6c9e-dirty", "doc_id": "jof_12705"}
117 +{"ts": "2026-11-03T23:04:04+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7487, "output": 2254, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075822, "latency_s": 18.0, "request_id": "req_011Cggq6b847mgjgDxAw9vKo", "stop_reason": "end_turn", "cache_key": "f1c72446d572216fa9007d378baf3a1adb70fae1aed7e889710e489edee7e2bf", "code_version": "feb6c9e-dirty", "doc_id": "jof_12690"}
118 +{"ts": "2026-11-03T23:04:04+00:00", "purpose": "write:E_full_engine:upgrade", "prompt_id": "condition_D_profiles_structured", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "high", "usage": {"input": 15046, "output": 6000, "cache_read": 0, "cache_write": 0}, "cost_usd": 0.180184, "latency_s": 51.13, "request_id": "req_011Cggq4BHHhAfSnM64Uyi96", "stop_reason": "max_tokens", "cache_key": "450b88722a3f826918f6dc17039c091f938fa5ecaf502cbaeba681ec359f7e47", "code_version": "feb6c9e-dirty", "section": "introduction", "paper": "/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/finance/wp41_uqo"}
119 +{"ts": "2026-11-03T23:04:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7466, "output": 2066, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071978, "latency_s": 17.21, "request_id": "req_011Cggq6marPSEsMNDeTpav9", "stop_reason": "end_turn", "cache_key": "03432647aa61462a9220c698add3503dbded4858f9b51fc851331e5ae0b90656", "code_version": "feb6c9e-dirty", "doc_id": "jof_12727"}
120 +{"ts": "2026-11-03T23:04:08+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6092, "output": 2478, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074722, "latency_s": 24.38, "request_id": "req_011Cggq6P2HdxqAktsPBKu7J", "stop_reason": "end_turn", "cache_key": "73e80a718367c106243d0c33ec94c0360d0187638afc2b4fe339a0152cb46153", "code_version": "feb6c9e-dirty", "doc_id": "jof_12692"}
121 +{"ts": "2026-11-03T23:04:10+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5154, "output": 1793, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.05727, "latency_s": 13.77, "request_id": "req_011Cggq7LXkiiGuLEBGdphZE", "stop_reason": "end_turn", "cache_key": "febf6433893047d4bfd4e6d3627dbc069a21051eed4a8d22f1b2488d27f54657", "code_version": "feb6c9e-dirty", "doc_id": "jof_12717"}
122 +{"ts": "2026-11-03T23:04:22+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6688, "output": 2040, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068346, "latency_s": 16.67, "request_id": "req_011Cggq83S8BAPiaB23GCfXq", "stop_reason": "end_turn", "cache_key": "cadc21897c580856f56100a89ea1be9199dbc415f8c7cba43aba876407109318", "code_version": "feb6c9e-dirty", "doc_id": "jof_12721"}
123 +{"ts": "2026-11-03T23:04:23+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5966, "output": 2162, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067898, "latency_s": 18.95, "request_id": "req_011Cggq7vhJrD6RGCMKiVQKQ", "stop_reason": "end_turn", "cache_key": "9fb6d560ab8a1055eb85907ce2ed3d1e4f70014e62a69e4182bd2805e6b940a7", "code_version": "feb6c9e-dirty", "doc_id": "jof_12719"}
124 +{"ts": "2026-11-03T23:04:26+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6963, "output": 2203, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072706, "latency_s": 18.59, "request_id": "req_011Cggq8BZX8HcirHFuhjYh9", "stop_reason": "end_turn", "cache_key": "814265f77bcdec2ff2f329c07d96eefbdebfcf63c2ad007e45397355f6197975", "code_version": "feb6c9e-dirty", "doc_id": "jof_12725"}
125 +{"ts": "2026-11-03T23:04:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6026, "output": 2077, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066438, "latency_s": 17.25, "request_id": "req_011Cggq8MiTsSuWqgnAftuMh", "stop_reason": "end_turn", "cache_key": "754321de1a80c2a20a1d3c8305d744299a04b63fcbadb6df448f3cd8d1905201", "code_version": "feb6c9e-dirty", "doc_id": "jof_12720"}
126 +{"ts": "2026-11-03T23:04:42+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7600, "output": 2388, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078954, "latency_s": 18.84, "request_id": "req_011Cggq9KmAJMWbHJ6ftMwtH", "stop_reason": "end_turn", "cache_key": "338e0e5b8572193f6da6b0a692302dddf9abc9f02363eaafb2b444740849361f", "code_version": "feb6c9e-dirty", "doc_id": "jof_12726"}
127 +{"ts": "2026-11-03T23:04:44+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7557, "output": 2191, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074842, "latency_s": 16.9, "request_id": "req_011Cggq9dktfbhb85ekoojAD", "stop_reason": "end_turn", "cache_key": "6b4c77f8b3ffbdceef7b6b71e1fb6cff33a9479bfee127ccc881cc1f8c89f2cf", "code_version": "feb6c9e-dirty", "doc_id": "jof_12723"}
128 +{"ts": "2026-11-03T23:04:47+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8169, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07691, "latency_s": 24.17, "request_id": "req_011Cggq9H7BDxjvv2Ve7dStn", "stop_reason": "end_turn", "cache_key": "c583cea899d2dcd81ac3456c675850c66b02633461c2d7b95aa287898b8588c6", "code_version": "feb6c9e-dirty", "doc_id": "jof_12722"}
129 +{"ts": "2026-11-03T23:04:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7318, "output": 3355, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.097166, "latency_s": 28.36, "request_id": "req_011Cggq9ZRCDpHtQmLfbSoSD", "stop_reason": "end_turn", "cache_key": "8930267bbf8a8193a4525d0abe3b5938c92ae998360d0a3a65f6b6fa919d694f", "code_version": "feb6c9e-dirty", "doc_id": "jof_12724"}
130 +{"ts": "2026-11-03T23:04:57+00:00", "purpose": "referee:E_full_engine:upgrade", "prompt_id": "condition_E_specialist_referee", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "high", "usage": {"input": 16301, "output": 6000, "cache_read": 0, "cache_write": 0}, "cost_usd": 0.185204, "latency_s": 52.21, "request_id": "req_011Cggq7yAeePxHxBsKTKLkC", "stop_reason": "max_tokens", "cache_key": "238efa48f8f10e1f654dd868d263fcdafae12a4e5d489bd16e06d4f05015b38c", "code_version": "feb6c9e-dirty", "section": "introduction", "paper": "/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/finance/wp41_uqo"}
131 +{"ts": "2026-11-03T23:05:02+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7909, "output": 2157, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07557, "latency_s": 17.29, "request_id": "req_011CggqAtLWt3NMW7h3UgJS2", "stop_reason": "end_turn", "cache_key": "052b21227ee450698be9622092c52eff7ace73e07efbfe0d00b67e8aefb8928a", "code_version": "feb6c9e-dirty", "doc_id": "jof_12732"}
132 +{"ts": "2026-11-03T23:05:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8775, "output": 2443, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084754, "latency_s": 20.75, "request_id": "req_011CggqAifM61etHqPSKysKs", "stop_reason": "end_turn", "cache_key": "91a7b3a82e114ac7f442bc63c10ac9f63e8457cc63cbaf27394c19b6c3b0bf05", "code_version": "feb6c9e-dirty", "doc_id": "jof_12739"}
133 +{"ts": "2026-11-03T23:05:05+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6589, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07219, "latency_s": 18.39, "request_id": "req_011CggqB4XSsXu4eySxF7xSi", "stop_reason": "end_turn", "cache_key": "756e184688f33b4b0fa370af648383ea52c9875e0022d861c7c6fc218056d579", "code_version": "feb6c9e-dirty", "doc_id": "jof_12738"}
134 +{"ts": "2026-11-03T23:05:13+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8040, "output": 2252, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077994, "latency_s": 18.33, "request_id": "req_011CggqBesg1TV1RPRbtWdBg", "stop_reason": "end_turn", "cache_key": "b4476c02a9753adea77f974a6ac5f7fd6d6d556be8eb7b0b54cd6337c3a5d02b", "code_version": "feb6c9e-dirty", "doc_id": "jof_12730"}
135 +{"ts": "2026-11-03T23:05:20+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8140, "output": 2318, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079714, "latency_s": 18.67, "request_id": "req_011CggqCAZMoNQNqy4vSvwLL", "stop_reason": "end_turn", "cache_key": "6c3e6bdd674ba37c6af180ae0b666d6247dddeac6623dd32c00bab519f0feb18", "code_version": "feb6c9e-dirty", "doc_id": "jof_12737"}
136 +{"ts": "2026-11-03T23:05:22+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9507, "output": 2029, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079402, "latency_s": 17.04, "request_id": "req_011CggqCRXGhZ1DhUmaMhmCE", "stop_reason": "end_turn", "cache_key": "39c4a0cadfd55619b3d575999c04db8991a4ccf19b2b62595a2aa7cc5233ebe1", "code_version": "feb6c9e-dirty", "doc_id": "jof_12731"}
137 +{"ts": "2026-11-03T23:05:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8820, "output": 2304, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082154, "latency_s": 18.44, "request_id": "req_011CggqD1ZPsxXcVhjHqJ5YE", "stop_reason": "end_turn", "cache_key": "449f407999a6a4ce5ec43bbf42efda2cfe53558d735ade566282c7906bacc6c1", "code_version": "feb6c9e-dirty", "doc_id": "jof_12733"}
138 +{"ts": "2026-11-03T23:05:33+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4725, "output": 1922, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.058134, "latency_s": 30.16, "request_id": "req_011CggqCFm9Tn1hBEYYb3Bbo", "stop_reason": "end_turn", "cache_key": "a390f5144faf928d8992516699729c0deeaa1426c380fe8de73fed958abc745f", "code_version": "feb6c9e-dirty", "doc_id": "jof_12729"}
139 +{"ts": "2026-11-03T23:05:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6769, "output": 3182, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.09151, "latency_s": 27.29, "request_id": "req_011CggqDYm8FTsHzqcoMy92w", "stop_reason": "end_turn", "cache_key": "7e76c79e10c38abac5883cccbc15304a911b3db307b394c68156f9a08f746458", "code_version": "feb6c9e-dirty", "doc_id": "jof_12736"}
140 +{"ts": "2026-11-03T23:05:51+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7469, "output": 2316, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07699, "latency_s": 19.24, "request_id": "req_011CggqEP9CqRshJLfZq9a2i", "stop_reason": "end_turn", "cache_key": "8b459b7ea7ad1ee09aa4bb129b2368013603ce147a83b1c985ecdb3ded528cd6", "code_version": "feb6c9e-dirty", "doc_id": "jof_12735"}
141 +{"ts": "2026-11-03T23:05:53+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9337, "output": 2419, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086522, "latency_s": 19.98, "request_id": "req_011CggqEUyCaGmKuyx6hjE3n", "stop_reason": "end_turn", "cache_key": "73fdb4bc2200559df23a5e9f188d43b390bc22dfc72c3e714dea4027fe95a7ce", "code_version": "feb6c9e-dirty", "doc_id": "jof_12728"}
142 +{"ts": "2026-11-03T23:05:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8261, "output": 2357, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080978, "latency_s": 33.41, "request_id": "req_011CggqDgmaR5zMGopWMED9N", "stop_reason": "end_turn", "cache_key": "9f2a363b55a34ce4f3517f2702249ee2cd5b2e30b0361849ea4034ae5ee84858", "code_version": "feb6c9e-dirty", "doc_id": "jof_12740"}
143 +{"ts": "2026-11-03T23:06:07+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 11201, "output": 2497, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.095538, "latency_s": 19.59, "request_id": "req_011CggqFZsej3JzRp4mSAUVi", "stop_reason": "end_turn", "cache_key": "4499d259c9d3016f28ffc4d8b519489d5f773161841d89a88ca6bc62ae631ea8", "code_version": "feb6c9e-dirty", "doc_id": "jof_12747"}
144 +{"ts": "2026-11-03T23:06:09+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8134, "output": 2145, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07623, "latency_s": 17.69, "request_id": "req_011CggqFpKpAmYDmVkifPVwV", "stop_reason": "end_turn", "cache_key": "8b377d5e496a28154da743125af2cdb1fddf102d8d553b9e6ba5d716293dead6", "code_version": "feb6c9e-dirty", "doc_id": "jof_12751"}
145 +{"ts": "2026-11-03T23:06:11+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7914, "output": 2258, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07761, "latency_s": 17.72, "request_id": "req_011CggqFyJK4gmDMYqXZQF5D", "stop_reason": "end_turn", "cache_key": "4d9a97cf0c9f923308ec4265b5b9b472b9a89d20971eecb99fefd7af347a3fcc", "code_version": "feb6c9e-dirty", "doc_id": "jof_12749"}
146 +{"ts": "2026-11-03T23:06:13+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5901, "output": 2185, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068098, "latency_s": 17.07, "request_id": "req_011CggqGABeYE9wTYTzhNLup", "stop_reason": "end_turn", "cache_key": "c8da58e67a3b5da6893b6169200fa3158c7dd7d578d4b05c5146947dc10f9a4e", "code_version": "feb6c9e-dirty", "doc_id": "jof_12734"}
added data/annotations/documents/jof_12407.json +267 −0
@@ -0,0 +1,267 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12407",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:23+00:00",
9 + "source_sha256": "9fd05d6953387ae8c54afe8d8da512039663ccbaae7b4fdea5537f05e603ba72",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20359,
12 + "usage": {
13 + "input": 6624,
14 + "output": 1972,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06673,
19 + "request_id": "req_011CggpSEVX1d38giR4Rjxbb",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "international_finance",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Causal effect of the local same-ethnicity resident population share near firm headquarters on the firm's export/import ratio with that country",
27 + "strategy": "Fixed-effects panel regressions of trade ratios on Connected Population, followed by IV using the WWII Japanese internment camp locations as exogenous shocks to local Japanese population; placebo tests on other Asian populations",
28 + "key_assumptions": [
29 + "Internment camp locations affect current firm trade with Japan only through local Japanese population",
30 + "Camp siting was unrelated to factors driving later firm trade with Japan"
31 + ],
32 + "threats_addressed": [
33 + "Reverse causality: firms bringing in foreign populations when planning trade",
34 + "Omitted common factors driving co-location of ethnic groups and trading firms",
35 + "Familiarity bias versus value-creating information channel"
36 + ],
37 + "causal_language_strength": 4,
38 + "uses_causal_claim": true
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": false,
43 + "sentence_index": -1
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": true,
51 + "sentence_index": 1
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 0
56 + },
57 + "magnitude": {
58 + "present": false,
59 + "sentence_index": -1
60 + },
61 + "contribution": {
62 + "present": true,
63 + "sentence_index": 3
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 1,
74 + "move": "preview_of_approach"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "preview_of_results"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "identification_defense"
83 + },
84 + {
85 + "paragraph_index": 2,
86 + "move": "preview_of_results"
87 + },
88 + {
89 + "paragraph_index": 3,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 4,
94 + "move": "mechanism_or_theory_sketch"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 5,
102 + "move": "mechanism_or_theory_sketch"
103 + },
104 + {
105 + "paragraph_index": 5,
106 + "move": "roadmap"
107 + }
108 + ],
109 + "contribution_paragraph_index": -1,
110 + "contribution_types": [
111 + "new_evidence",
112 + "new_data_or_measure",
113 + "new_identification"
114 + ],
115 + "roadmap_present": true,
116 + "literature_positioning_location": "separate_section"
117 + },
118 + "section_roles": [
119 + {
120 + "idx": 0,
121 + "role": "introduction"
122 + },
123 + {
124 + "idx": 1,
125 + "role": "literature"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "data"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "empirical_design"
138 + },
139 + {
140 + "idx": 5,
141 + "role": "results"
142 + },
143 + {
144 + "idx": 6,
145 + "role": "results"
146 + },
147 + {
148 + "idx": 7,
149 + "role": "robustness"
150 + },
151 + {
152 + "idx": 8,
153 + "role": "conclusion"
154 + }
155 + ],
156 + "results_presentation": {
157 + "economic_significance_discussed": true,
158 + "statistical_significance_language": "stars_and_tstats",
159 + "magnitude_statements_in_intro": true
160 + },
161 + "robustness": {
162 + "location": "mixed",
163 + "types": [
164 + "placebo",
165 + "alternative_measure",
166 + "subsample",
167 + "other"
168 + ]
169 + },
170 + "tables": {
171 + "convention_notes": "Table notes write out the regression equation, define dependent and key variables, and list the fixed effects. IV tables report first and second stages in separate panels. Descriptive tables give dollar values in billions.",
172 + "units_stated_in_notes": true,
173 + "standard_errors_description": "unclear"
174 + },
175 + "figures": {
176 + "uncertainty_shown": "no_figures"
177 + },
178 + "conclusion": {
179 + "bounded_claims": false,
180 + "policy_implications": false,
181 + "open_questions": false,
182 + "restates_contribution": true
183 + },
184 + "hedging": {
185 + "level": "low",
186 + "typical_devices": [
187 + "plausibly exogenous",
188 + "consistent with",
189 + "point to"
190 + ]
191 + },
192 + "claim_evidence": [
193 + {
194 + "claim": "Firms trade more with countries whose ethnic group is well represented near headquarters",
195 + "evidence_type": "table",
196 + "evidence_ref": "Table III"
197 + },
198 + {
199 + "claim": "Internment camps caused persistent local Japanese population shifts that raise trade with Japan",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table V"
202 + },
203 + {
204 + "claim": "Placebo checks support the instrument; no effect on other Asian populations",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table VI"
207 + },
208 + {
209 + "claim": "Strategic exporters see about 11% higher future profitability than nonstrategic exporters",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table VII"
212 + },
213 + {
214 + "claim": "Connected networks raise cross-border M&A and foreign segment sales",
215 + "evidence_type": "table",
216 + "evidence_ref": "Tables VIII, IX"
217 + },
218 + {
219 + "claim": "Network effects are stronger with low tariffs and differentiated goods; connected trades get better prices",
220 + "evidence_type": "table",
221 + "evidence_ref": "Tables X, XI"
222 + },
223 + {
224 + "claim": "Connected board members are associated with 68% more exports to the connected country",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table III (board member variable)"
227 + }
228 + ],
229 + "structural_summary": "Opens with the global-markets motivation and introduces local resident networks. The intro previews the correlational trade results, flags the endogeneity concern and the internment-camp instrument, then previews value effects, M&A and segment extensions, the board channel and mechanism tests, and closes with a roadmap. The body runs: literature section, data, baseline fixed-effects results, a long IV section (natural experiment, placebos, corroboration, pre-war firms), real effects, other corporate decisions, and mechanism tests. A short conclusion restates the contribution.",
230 + "writing_profile_keywords": [
231 + "natural experiment",
232 + "instrumental variables",
233 + "novel trade data",
234 + "ethnic networks",
235 + "mechanism tests",
236 + "economic magnitudes in intro"
237 + ],
238 + "confidence": 0.8,
239 + "notes": "Standard error treatment is not visible in the excerpts. Robustness evidence appears in the Internet Appendix (state-level measure), in a placebo subsection, and in the mechanism section, which is why location is marked mixed."
240 + },
241 + "deterministic_checks": {
242 + "roadmap_regex": true,
243 + "contribution_regex_in_intro": false,
244 + "magnitude_in_abstract": false,
245 + "abstract_n_sentences": 4,
246 + "intro_n_paragraphs": 6,
247 + "intro_n_words": 855,
248 + "causal_words_in_abstract": 2,
249 + "hedge_words_in_abstract": 2,
250 + "stars_mentioned": false,
251 + "clustered_se_mentioned": false,
252 + "internet_appendix_mentioned": true,
253 + "section_roles_rule": {
254 + "0": "introduction",
255 + "1": "literature",
256 + "2": "data",
257 + "3": "other",
258 + "4": "other",
259 + "5": "other",
260 + "6": "other",
261 + "7": "robustness",
262 + "8": "conclusion"
263 + }
264 + },
265 + "schema_valid": true,
266 + "code_version": "feb6c9e-dirty"
267 +}
\ No newline at end of file
added data/annotations/documents/jof_12432.json +271 −0
@@ -0,0 +1,271 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12432",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:27+00:00",
9 + "source_sha256": "ca0ebf3d253fbf0194f7d3a3f16b844b78695593229c84582679f39f735f66cf",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22627,
12 + "usage": {
13 + "input": 7888,
14 + "output": 2408,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.080506,
19 + "request_id": "req_011CggpSEWGEgMr5GvRdYip3",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Effect of the intensity of a CEO's early-life exposure to fatal natural disasters (none vs. medium vs. extreme fatality) on corporate risk-taking policies and outcomes.",
27 + "strategy": "Panel OLS and linear probability models using early-life disaster exposure inferred from birth county and year. Specifications include firm, year, CEO birth-year and CEO birth-state fixed effects and control for the county's historical disaster incidence. Supplemented by a difference-in-differences comparison with neighboring unaffected counties, a placebo with random reassignment of birth county, and an analysis of exogenous CEO turnovers.",
28 + "key_assumptions": [
29 + "Disaster exposure during the formative years is exogenous to later corporate outcomes, conditional on fixed effects and historical county risk.",
30 + "Birth county proxies for where the CEO lived during childhood.",
31 + "Disaster experience does not affect selection into becoming a CEO, nor CEO-firm matching in ways that drive results."
32 + ],
33 + "threats_addressed": [
34 + "cohort effects",
35 + "growing up in a high-risk county vs. actually experiencing a disaster",
36 + "measurement error in childhood residence",
37 + "omitted variables",
38 + "CEO-firm matching",
39 + "spurious correlation (placebo)"
40 + ],
41 + "causal_language_strength": 3,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 1
48 + },
49 + "gap": {
50 + "present": true,
51 + "sentence_index": 0
52 + },
53 + "method": {
54 + "present": false,
55 + "sentence_index": -1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 1
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 5,
78 + "move": "related_literature_positioning"
79 + },
80 + {
81 + "paragraph_index": 5,
82 + "move": "literature_gap"
83 + },
84 + {
85 + "paragraph_index": 6,
86 + "move": "preview_of_approach"
87 + },
88 + {
89 + "paragraph_index": 7,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 8,
94 + "move": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 9,
98 + "move": "mechanism_or_theory_sketch"
99 + },
100 + {
101 + "paragraph_index": 10,
102 + "move": "identification_defense"
103 + },
104 + {
105 + "paragraph_index": 10,
106 + "move": "robustness_mention"
107 + },
108 + {
109 + "paragraph_index": 11,
110 + "move": "contribution_statement"
111 + },
112 + {
113 + "paragraph_index": 12,
114 + "move": "policy_or_broader_implication"
115 + },
116 + {
117 + "paragraph_index": 13,
118 + "move": "roadmap"
119 + }
120 + ],
121 + "contribution_paragraph_index": 11,
122 + "contribution_types": [
123 + "new_evidence",
124 + "new_identification"
125 + ],
126 + "roadmap_present": true,
127 + "literature_positioning_location": "both"
128 + },
129 + "section_roles": [
130 + {
131 + "idx": 0,
132 + "role": "introduction"
133 + },
134 + {
135 + "idx": 1,
136 + "role": "literature"
137 + },
138 + {
139 + "idx": 2,
140 + "role": "data"
141 + },
142 + {
143 + "idx": 3,
144 + "role": "results"
145 + },
146 + {
147 + "idx": 4,
148 + "role": "robustness"
149 + },
150 + {
151 + "idx": 5,
152 + "role": "conclusion"
153 + },
154 + {
155 + "idx": 6,
156 + "role": "appendix"
157 + }
158 + ],
159 + "results_presentation": {
160 + "economic_significance_discussed": true,
161 + "statistical_significance_language": "unclear",
162 + "magnitude_statements_in_intro": true
163 + },
164 + "robustness": {
165 + "location": "main_text_section",
166 + "types": [
167 + "alternative_measure",
168 + "placebo",
169 + "alternative_specification",
170 + "other"
171 + ]
172 + },
173 + "tables": {
174 + "convention_notes": "Notes state the estimator (OLS or LPM), the dependent variable by column, and the fixed-effect set (firm, year, CEO birth year, CEO birth state). Descriptive tables come first, followed by policy-by-policy regression tables, then robustness tables (placebo, DiD-style treated vs. control, exogenous turnovers). The excerpts do not show how standard errors are computed.",
175 + "units_stated_in_notes": true,
176 + "standard_errors_description": "unclear"
177 + },
178 + "figures": {
179 + "uncertainty_shown": "no_figures"
180 + },
181 + "conclusion": {
182 + "bounded_claims": false,
183 + "policy_implications": false,
184 + "open_questions": true,
185 + "restates_contribution": true
186 + },
187 + "hedging": {
188 + "level": "moderate",
189 + "typical_devices": [
190 + "'results support the view/hypothesis'",
191 + "'appear to be more cautious'",
192 + "'suggests that our results may have broader implications'",
193 + "'all else equal' framing"
194 + ]
195 + },
196 + "claim_evidence": [
197 + {
198 + "claim": "Inverse U-shaped relation between the intensity of CEO fatal disaster exposure and firm leverage and cash policy.",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table IV"
201 + },
202 + {
203 + "claim": "Leverage differences reflect active financing choices: moderate-disaster CEOs fill financing deficits with debt.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table V"
206 + },
207 + {
208 + "claim": "Moderate-disaster CEOs face higher interest costs and loan spreads, lower ratings and more bankruptcy; extreme-disaster CEOs show the reverse.",
209 + "evidence_type": "table",
210 + "evidence_ref": "Tables VI, VII"
211 + },
212 + {
213 + "claim": "Acquisition frequency, payment method, diversification and announcement returns vary nonmonotonically with disaster intensity.",
214 + "evidence_type": "table",
215 + "evidence_ref": "Tables VIII, IX"
216 + },
217 + {
218 + "claim": "Moderate-disaster CEOs run firms with higher total and idiosyncratic volatility.",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table X"
221 + },
222 + {
223 + "claim": "CEOs' childhood disaster exposure resembles that of the general U.S. population, suggesting no selection into CEO role.",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table II"
226 + },
227 + {
228 + "claim": "Results survive random reassignment placebo, treated vs. neighboring control counties, and exogenous turnovers.",
229 + "evidence_type": "table",
230 + "evidence_ref": "Tables XI, XII, XIII"
231 + }
232 + ],
233 + "structural_summary": "The paper opens with an epigraph and a literature-based observation that prior work assumes a monotonic link between experience and managerial risk-taking, then poses a dosage (intensity) hypothesis. The introduction previews the measure, policy-by-policy results with magnitudes, and the interpretation (desensitization vs. caution). It then gives an identification defense (fixed effects, historical-risk control, DiD), contribution claims, broader implications and a roadmap. A short motivation section grounds the hypothesis in the literature, followed by data construction, main results across financing, credit, acquisitions and volatility, and a dedicated robustness section. The conclusion restates the findings and points to investor-behavior research.",
234 + "writing_profile_keywords": [
235 + "managerial style",
236 + "early-life experience",
237 + "nonmonotonic effect",
238 + "natural disasters",
239 + "fixed effects",
240 + "dosage metaphor",
241 + "epigraph opening",
242 + "policy-by-policy results"
243 + ],
244 + "confidence": 0.8,
245 + "notes": "The Motivation section (I) functions as literature-based hypothesis development; it is coded as literature. The excerpts do not show standard errors or how significance is reported. The introduction's first five paragraph entries are fragments of the epigraph, so the substantive intro text begins at index 5."
246 + },
247 + "deterministic_checks": {
248 + "roadmap_regex": true,
249 + "contribution_regex_in_intro": false,
250 + "magnitude_in_abstract": false,
251 + "abstract_n_sentences": 5,
252 + "intro_n_paragraphs": 14,
253 + "intro_n_words": 1227,
254 + "causal_words_in_abstract": 0,
255 + "hedge_words_in_abstract": 0,
256 + "stars_mentioned": false,
257 + "clustered_se_mentioned": false,
258 + "internet_appendix_mentioned": false,
259 + "section_roles_rule": {
260 + "0": "introduction",
261 + "1": "other",
262 + "2": "data",
263 + "3": "results",
264 + "4": "robustness",
265 + "5": "conclusion",
266 + "6": "appendix"
267 + }
268 + },
269 + "schema_valid": true,
270 + "code_version": "feb6c9e-dirty"
271 +}
\ No newline at end of file
added data/annotations/documents/jof_12434.json +275 −0
@@ -0,0 +1,275 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12434",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:25+00:00",
9 + "source_sha256": "3c97b721b594e1f26313be6488f6bd5ca338f72d9cedd1032f3c0d3f01c3de37",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24257,
12 + "usage": {
13 + "input": 8177,
14 + "output": 2314,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.079782,
19 + "request_id": "req_011CggpSEWmGLpptcvkQCqXk",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "international_finance",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Average effect of treatment on the treated (ATT) of a sovereign downgrade on investment, debt/equity issuance and bond yields for firms rated at or above the sovereign ceiling (bound firms), relative to matched nonbound firms",
27 + "strategy": "Abadie-Imbens bias-corrected matching combined with difference-in-differences. Bound firms are treated and matched nonbound firms from the same country and year serve as controls. The source of variation is the sovereign ceiling policy of rating agencies. Linear DiD regressions are used as an alternative.",
28 + "key_assumptions": [
29 + "Nonbound firms have similar but slightly lower credit quality, so fundamentals shocks affect both groups similarly or affect controls more",
30 + "Asymmetric corporate downgrades arise from the ceiling rule rather than from changes in fundamentals",
31 + "Exposure to government is similar across treated and control firms"
32 + ],
33 + "threats_addressed": [
34 + "common macro shocks and spillovers",
35 + "changes in fundamentals and credit risk",
36 + "differential government exposure and government support or ownership",
37 + "recessions and crisis periods unrelated to downgrades (placebo)",
38 + "credit demand versus supply"
39 + ],
40 + "causal_language_strength": 4,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": false,
46 + "sentence_index": -1
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 2
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 1
59 + },
60 + "magnitude": {
61 + "present": false,
62 + "sentence_index": -1
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 0
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 1,
77 + "move": "preview_of_approach"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "identification_defense"
82 + },
83 + {
84 + "paragraph_index": 2,
85 + "move": "mechanism_or_theory_sketch"
86 + },
87 + {
88 + "paragraph_index": 3,
89 + "move": "mechanism_or_theory_sketch"
90 + },
91 + {
92 + "paragraph_index": 4,
93 + "move": "preview_of_approach"
94 + },
95 + {
96 + "paragraph_index": 5,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 5,
101 + "move": "robustness_mention"
102 + },
103 + {
104 + "paragraph_index": 6,
105 + "move": "identification_defense"
106 + },
107 + {
108 + "paragraph_index": 7,
109 + "move": "contribution_statement"
110 + },
111 + {
112 + "paragraph_index": 7,
113 + "move": "related_literature_positioning"
114 + },
115 + {
116 + "paragraph_index": 8,
117 + "move": "other"
118 + },
119 + {
120 + "paragraph_index": 8,
121 + "move": "policy_or_broader_implication"
122 + }
123 + ],
124 + "contribution_paragraph_index": 7,
125 + "contribution_types": [
126 + "new_evidence",
127 + "new_identification"
128 + ],
129 + "roadmap_present": false,
130 + "literature_positioning_location": "inside_introduction"
131 + },
132 + "section_roles": [
133 + {
134 + "idx": 0,
135 + "role": "introduction"
136 + },
137 + {
138 + "idx": 1,
139 + "role": "empirical_design"
140 + },
141 + {
142 + "idx": 2,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 3,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 4,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 5,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 6,
159 + "role": "conclusion"
160 + }
161 + ],
162 + "results_presentation": {
163 + "economic_significance_discussed": true,
164 + "statistical_significance_language": "unclear",
165 + "magnitude_statements_in_intro": true
166 + },
167 + "robustness": {
168 + "location": "mixed",
169 + "types": [
170 + "placebo",
171 + "alternative_specification",
172 + "subsample",
173 + "falsification"
174 + ]
175 + },
176 + "tables": {
177 + "convention_notes": "Notes define the treated group (rating at or above sovereign in year before downgrade), define each variable as a ratio to lagged assets or PP&E, and describe the sample source and period. Most tables report DiD matching estimators, with linear regression tables for the alternative and government-exposure tests. Inference details are not visible in the excerpts.",
178 + "units_stated_in_notes": true,
179 + "standard_errors_description": "unclear"
180 + },
181 + "figures": {
182 + "uncertainty_shown": "unclear"
183 + },
184 + "conclusion": {
185 + "bounded_claims": true,
186 + "policy_implications": true,
187 + "open_questions": false,
188 + "restates_contribution": true
189 + },
190 + "hedging": {
191 + "level": "moderate",
192 + "typical_devices": [
193 + "'likely to be due to'",
194 + "'cannot completely rule out'",
195 + "explicit limitations paragraph on quasi-natural experiments",
196 + "'interpreted with caution due to small sample size and lack of statistical power'",
197 + "'consistent with'",
198 + "'can have'"
199 + ]
200 + },
201 + "claim_evidence": [
202 + {
203 + "claim": "Following sovereign downgrades, bound firms are more likely to be downgraded than nonbound firms",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table III; Figures 2-3"
206 + },
207 + {
208 + "claim": "Treated firms cut investment by about 8.9 pp of capital (ATT), relative to matched controls",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table IV; Figure 4"
211 + },
212 + {
213 + "claim": "Treated firms reduce net debt issuance more than controls (ATT -5.5 pp), with some increase in equity issuance",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table VIII; Figure 5"
216 + },
217 + {
218 + "claim": "Bond yields of bound firms rise about 34 bp more over 3 months and 61 bp over 6 months",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table X; Figure 6"
221 + },
222 + {
223 + "claim": "Effects do not appear in recessions without downgrades or in other placebo windows",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table V"
226 + },
227 + {
228 + "claim": "Exposure to government appears similar across treated and control firms",
229 + "evidence_type": "table",
230 + "evidence_ref": "Tables VI-VII"
231 + },
232 + {
233 + "claim": "Effects are stronger when downgrades move sovereigns across Basel II risk-weight bins",
234 + "evidence_type": "table",
235 + "evidence_ref": "Table XI"
236 + }
237 + ],
238 + "structural_summary": "The introduction opens with recent sovereign downgrades and poses a direct question. It then presents the ceiling-based identification and defends it on credit-quality grounds. A regulatory mechanism is sketched, followed by a preview of the matching estimator and quantified results, a government-exposure caveat, three contributions positioned against the ratings literature, and an explicit limitations paragraph. The body describes the institutional background and matching design, then sequential outcome sections: ratings, investment and financing (with placebo, linear and robustness subsections), and cost of debt. A mechanism section uses Basel II bins. The conclusion restates findings and draws a public debt management implication.",
239 + "writing_profile_keywords": [
240 + "quasi-natural experiment",
241 + "sovereign ceiling",
242 + "matching estimator",
243 + "difference-in-differences",
244 + "credit supply",
245 + "real effects",
246 + "explicit limitations"
247 + ],
248 + "confidence": 0.85,
249 + "notes": "The intro paragraph on government exposure tests appears partially truncated. Significance reporting conventions are not visible in the table notes. Section V is labelled robustness by the rule but is actually a mechanism or results section."
250 + },
251 + "deterministic_checks": {
252 + "roadmap_regex": false,
253 + "contribution_regex_in_intro": true,
254 + "magnitude_in_abstract": false,
255 + "abstract_n_sentences": 3,
256 + "intro_n_paragraphs": 9,
257 + "intro_n_words": 1304,
258 + "causal_words_in_abstract": 1,
259 + "hedge_words_in_abstract": 0,
260 + "stars_mentioned": false,
261 + "clustered_se_mentioned": false,
262 + "internet_appendix_mentioned": false,
263 + "section_roles_rule": {
264 + "0": "introduction",
265 + "1": "empirical_design",
266 + "2": "other",
267 + "3": "other",
268 + "4": "other",
269 + "5": "robustness",
270 + "6": "conclusion"
271 + }
272 + },
273 + "schema_valid": true,
274 + "code_version": "feb6c9e-dirty"
275 +}
\ No newline at end of file
added data/annotations/documents/jof_12438.json +293 −0
@@ -0,0 +1,293 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12438",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:00+00:00",
9 + "source_sha256": "32e6e6c3e69ad8623c8908d22be92d5edf3a54024def41a76f6aea51d38ae2ba",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 25333,
12 + "usage": {
13 + "input": 9163,
14 + "output": 2237,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.082186,
19 + "request_id": "req_011CggpUwX99UMZ6CSdXUqQv",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Coefficient of relative risk aversion implied by consumption covariances with stock returns (GMM Euler equations), and cross-sectional consumption risk premia",
27 + "strategy": "A Kalman-filter model of NIPA consumption estimation is used to reverse (unfilter) reported NIPA consumption. Stock returns are time-aggregated to address time-aggregation bias. The resulting series is used in GMM Euler-equation estimation and Fama-MacBeth cross-sectional tests. A simulation experiment validates the procedure.",
28 + "key_assumptions": [
29 + "BEA estimation process can be approximated by an optimal (Kalman) filter",
30 + "Measurement error in observable consumption is uncorrelated with stock returns",
31 + "Calibrated filter parameters (signal-to-noise, consumption volatility) are appropriate"
32 + ],
33 + "threats_addressed": [
34 + "time-aggregation bias",
35 + "heterogeneity between nondurables and services",
36 + "choice of consumption volatility measure in the unfilter",
37 + "conditioning information and long-horizon returns"
38 + ],
39 + "causal_language_strength": 2,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 2
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 0
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "mechanism_or_theory_sketch"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_results"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "related_literature_positioning"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "related_literature_positioning"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "related_literature_positioning"
105 + },
106 + {
107 + "paragraph_index": 6,
108 + "move": "related_literature_positioning"
109 + },
110 + {
111 + "paragraph_index": 7,
112 + "move": "literature_gap"
113 + },
114 + {
115 + "paragraph_index": 7,
116 + "move": "contribution_statement"
117 + },
118 + {
119 + "paragraph_index": 7,
120 + "move": "policy_or_broader_implication"
121 + },
122 + {
123 + "paragraph_index": 7,
124 + "move": "roadmap"
125 + }
126 + ],
127 + "contribution_paragraph_index": 0,
128 + "contribution_types": [
129 + "new_data_or_measure",
130 + "resolution_of_puzzle",
131 + "new_evidence",
132 + "new_method"
133 + ],
134 + "roadmap_present": true,
135 + "literature_positioning_location": "inside_introduction"
136 + },
137 + "section_roles": [
138 + {
139 + "idx": 0,
140 + "role": "introduction"
141 + },
142 + {
143 + "idx": 1,
144 + "role": "theory"
145 + },
146 + {
147 + "idx": 2,
148 + "role": "results"
149 + },
150 + {
151 + "idx": 3,
152 + "role": "results"
153 + },
154 + {
155 + "idx": 4,
156 + "role": "theory"
157 + },
158 + {
159 + "idx": 5,
160 + "role": "conclusion"
161 + },
162 + {
163 + "idx": 6,
164 + "role": "appendix"
165 + },
166 + {
167 + "idx": 7,
168 + "role": "appendix"
169 + },
170 + {
171 + "idx": 8,
172 + "role": "appendix"
173 + },
174 + {
175 + "idx": 9,
176 + "role": "appendix"
177 + },
178 + {
179 + "idx": 10,
180 + "role": "appendix"
181 + },
182 + {
183 + "idx": 11,
184 + "role": "appendix"
185 + }
186 + ],
187 + "results_presentation": {
188 + "economic_significance_discussed": true,
189 + "statistical_significance_language": "unclear",
190 + "magnitude_statements_in_intro": true
191 + },
192 + "robustness": {
193 + "location": "within_results",
194 + "types": [
195 + "alternative_measure",
196 + "alternative_specification",
197 + "subsample",
198 + "alternative_sample_period"
199 + ]
200 + },
201 + "tables": {
202 + "convention_notes": "Notes define moment restrictions and estimators and fix the discount factor. They state sample periods (full sample and postwar). Some tables scale moments (×100). Simulation tables report medians of 10,000 simulations.",
203 + "units_stated_in_notes": true,
204 + "standard_errors_description": "unclear"
205 + },
206 + "figures": {
207 + "uncertainty_shown": "mixed"
208 + },
209 + "conclusion": {
210 + "bounded_claims": true,
211 + "policy_implications": false,
212 + "open_questions": false,
213 + "restates_contribution": true
214 + },
215 + "hedging": {
216 + "level": "moderate",
217 + "typical_devices": [
218 + "'might be attributed, at least in part'",
219 + "'may underestimate/overstate'",
220 + "'provides a first attempt'",
221 + "explicit acknowledgment of failures (profitability premium, return predictability)"
222 + ]
223 + },
224 + "claim_evidence": [
225 + {
226 + "claim": "Unfiltered NIPA consumption explains the equity premium with a reasonable relative risk aversion",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table III"
229 + },
230 + {
231 + "claim": "Unfiltering recovers properties of observable consumption in a simulated economy",
232 + "evidence_type": "figure",
233 + "evidence_ref": "Table I; Figure 1"
234 + },
235 + {
236 + "claim": "Unfiltered consumption prices size, value, and investment portfolios with low pricing errors but fails for profitability",
237 + "evidence_type": "table",
238 + "evidence_ref": "Tables VIII, IX"
239 + },
240 + {
241 + "claim": "The filter model explains differences between NIPA nondurables and services",
242 + "evidence_type": "table",
243 + "evidence_ref": "Table IV"
244 + },
245 + {
246 + "claim": "Calibrating to unfiltered consumption changes the implications of habit and long-run risk models",
247 + "evidence_type": "table",
248 + "evidence_ref": "Table X; Figure 5"
249 + }
250 + ],
251 + "structural_summary": "The introduction frames the paper around the puzzle of why garbage beats NIPA consumption in the consumption CAPM. It proposes statistical filtering by NIPA as the explanation and enumerates findings, including failures. Literature positioning (real estate unsmoothing, time aggregation) closes the introduction with a roadmap. Section I builds a Kalman-filter model, inverts it, and validates it by simulation. Section II compares consumption measures and estimates risk aversion by GMM, with several robustness variants. Section III runs cross-sectional tests. Section IV recalibrates habit and long-run risk models. A short conclusion restates the contribution and acknowledges the unexplained return predictability.",
252 + "writing_profile_keywords": [
253 + "measurement-error explanation",
254 + "puzzle resolution",
255 + "Kalman unfiltering",
256 + "GMM Euler equations",
257 + "simulation validation",
258 + "model calibration implications",
259 + "enumerated findings"
260 + ],
261 + "confidence": 0.8,
262 + "notes": "Primary method is a hybrid of filter modeling, calibration/simulation, and GMM estimation. Structural_estimation was chosen as the closest label. The excerpts do not show how statistical significance is reported."
263 + },
264 + "deterministic_checks": {
265 + "roadmap_regex": true,
266 + "contribution_regex_in_intro": true,
267 + "magnitude_in_abstract": false,
268 + "abstract_n_sentences": 4,
269 + "intro_n_paragraphs": 8,
270 + "intro_n_words": 1290,
271 + "causal_words_in_abstract": 0,
272 + "hedge_words_in_abstract": 1,
273 + "stars_mentioned": false,
274 + "clustered_se_mentioned": false,
275 + "internet_appendix_mentioned": false,
276 + "section_roles_rule": {
277 + "0": "introduction",
278 + "1": "other",
279 + "2": "other",
280 + "3": "other",
281 + "4": "theory",
282 + "5": "conclusion",
283 + "6": "appendix",
284 + "7": "appendix",
285 + "8": "appendix",
286 + "9": "appendix",
287 + "10": "appendix",
288 + "11": "appendix"
289 + }
290 + },
291 + "schema_valid": true,
292 + "code_version": "feb6c9e-dirty"
293 +}
\ No newline at end of file
added data/annotations/documents/jof_12439.json +265 −0
@@ -0,0 +1,265 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12439",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:39+00:00",
9 + "source_sha256": "16a5dc8018daa19a4073dbd046dc496777080d561ce5fd9913d6d6da5bc5bb6d",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 18993,
12 + "usage": {
13 + "input": 6306,
14 + "output": 2029,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.066598,
19 + "request_id": "req_011CggpTW7QiEoBdCikcmxkN",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "organizational economics / trade credit and supply chains",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of restricting trade credit maturity to 30 days on the likelihood of trade with affected suppliers, on vertical integration, and on total procurement",
27 + "strategy": "Within-product difference-in-differences comparing suppliers just below and just above a revenue cutoff (UF 100,000), before and after the Agreement, with product-time and firm-product fixed effects. Also exploits product-level exposure and time-varying firm fixed effects.",
28 + "key_assumptions": [
29 + "Parallel trends between Treated and Control suppliers in the absence of the Agreement",
30 + "No manipulation of the revenue cutoff",
31 + "Cutoff determination not confounded by contemporaneous legislation",
32 + "Enforcement of the Agreement"
33 + ],
34 + "threats_addressed": [
35 + "Differential trends between small and large firms",
36 + "Differential exit of smaller firms",
37 + "Manipulation around the threshold",
38 + "Contemporaneous legislation"
39 + ],
40 + "causal_language_strength": 4,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": true,
46 + "sentence_index": 0
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 2
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 2
59 + },
60 + "magnitude": {
61 + "present": true,
62 + "sentence_index": 2
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 0
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "literature_gap"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "contribution_statement"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 2,
93 + "move": "robustness_mention"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "identification_defense"
98 + },
99 + {
100 + "paragraph_index": 3,
101 + "move": "mechanism_or_theory_sketch"
102 + },
103 + {
104 + "paragraph_index": 3,
105 + "move": "preview_of_results"
106 + },
107 + {
108 + "paragraph_index": 4,
109 + "move": "related_literature_positioning"
110 + },
111 + {
112 + "paragraph_index": 4,
113 + "move": "policy_or_broader_implication"
114 + },
115 + {
116 + "paragraph_index": 5,
117 + "move": "roadmap"
118 + }
119 + ],
120 + "contribution_paragraph_index": 0,
121 + "contribution_types": [
122 + "new_evidence",
123 + "new_identification"
124 + ],
125 + "roadmap_present": true,
126 + "literature_positioning_location": "inside_introduction"
127 + },
128 + "section_roles": [
129 + {
130 + "idx": 0,
131 + "role": "introduction"
132 + },
133 + {
134 + "idx": 1,
135 + "role": "data"
136 + },
137 + {
138 + "idx": 2,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 3,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 4,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 5,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 6,
155 + "role": "conclusion"
156 + }
157 + ],
158 + "results_presentation": {
159 + "economic_significance_discussed": true,
160 + "statistical_significance_language": "unclear",
161 + "magnitude_statements_in_intro": true
162 + },
163 + "robustness": {
164 + "location": "mixed",
165 + "types": [
166 + "placebo",
167 + "alternative_specification",
168 + "falsification"
169 + ]
170 + },
171 + "tables": {
172 + "convention_notes": "Table notes state the estimating equation with its fixed effects and define the Treated/Control groups and exposure measures. Summary statistics compare the two groups in the preperiod. The excerpts do not show how standard errors are computed.",
173 + "units_stated_in_notes": true,
174 + "standard_errors_description": "unclear"
175 + },
176 + "figures": {
177 + "uncertainty_shown": "unclear"
178 + },
179 + "conclusion": {
180 + "bounded_claims": true,
181 + "policy_implications": true,
182 + "open_questions": false,
183 + "restates_contribution": true
184 + },
185 + "hedging": {
186 + "level": "moderate",
187 + "typical_devices": [
188 + "consistent with",
189 + "suggests",
190 + "likely",
191 + "we interpret this as",
192 + "most consistent with",
193 + "in this setting"
194 + ]
195 + },
196 + "claim_evidence": [
197 + {
198 + "claim": "The trade credit maturity restriction lowers the probability that a Treated supplier trades a product with the retailer by 11%",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table II; Figure 2"
201 + },
202 + {
203 + "claim": "The retailer becomes more likely to source from its own subsidiaries for highly exposed products",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table IV; Figure 3"
206 + },
207 + {
208 + "claim": "Total procurement falls for products mostly bought from affected suppliers",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table IV"
211 + },
212 + {
213 + "claim": "Results are not driven by differential trends or differential exit of small firms",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table III; Figure 1"
216 + },
217 + {
218 + "claim": "Exclusive or high-share relationships mitigate the negative effects",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table V"
221 + },
222 + {
223 + "claim": "Effects concentrate in durable/nonperishable goods and non-factoring firms, consistent with trade credit as a quality guarantee",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table VI"
226 + }
227 + ],
228 + "structural_summary": "The introduction frames the question in organizational economics, notes that prior evidence is limited to case studies and industry-level work, and claims first causal evidence. It previews a regulatory difference-in-differences design and three margins of adjustment with magnitudes, then robustness checks, relational-contract heterogeneity and the mechanism. It closes by contrasting the results with French evidence and giving a roadmap. The body describes the setting and data, then presents the trade-margin results with graphical evidence and robustness. Later sections cover vertical integration and procurement, exclusivity as a mitigating factor, and tests of competing trade credit theories. The conclusion restates the three margins and notes welfare implications.",
229 + "writing_profile_keywords": [
230 + "regulatory natural experiment",
231 + "within-product DiD",
232 + "trade credit",
233 + "vertical integration",
234 + "mechanism discrimination",
235 + "proprietary data",
236 + "first causal evidence"
237 + ],
238 + "confidence": 0.85,
239 + "notes": "The section role rules were overridden: Section I is a setting/data section and Section V is a mechanism-testing results section. The excerpts do not show how standard errors are reported."
240 + },
241 + "deterministic_checks": {
242 + "roadmap_regex": true,
243 + "contribution_regex_in_intro": true,
244 + "magnitude_in_abstract": true,
245 + "abstract_n_sentences": 6,
246 + "intro_n_paragraphs": 6,
247 + "intro_n_words": 1025,
248 + "causal_words_in_abstract": 0,
249 + "hedge_words_in_abstract": 1,
250 + "stars_mentioned": false,
251 + "clustered_se_mentioned": false,
252 + "internet_appendix_mentioned": false,
253 + "section_roles_rule": {
254 + "0": "introduction",
255 + "1": "theory",
256 + "2": "other",
257 + "3": "other",
258 + "4": "other",
259 + "5": "robustness",
260 + "6": "conclusion"
261 + }
262 + },
263 + "schema_valid": true,
264 + "code_version": "feb6c9e-dirty"
265 +}
\ No newline at end of file
added data/annotations/documents/jof_12464.json +319 −0
@@ -0,0 +1,319 @@
1 +{
2 + "validation_errors": [
3 + "'causal_language_strength' is a required property @ identification",
4 + "'uses_causal_claim' is a required property @ identification",
5 + "[\"Causal language strength rated 4: the abstract says booms 'form in response to' and the conclusion says the risk premium 'explains' waves, but the evidence is @ notes"
6 + ],
7 + "doc_id": "jof_12464",
8 + "ontology_version": "1.0",
9 + "prompt_version": "1.0",
10 + "model": "claude-opus-5-5",
11 + "effort": "medium",
12 + "annotated_at": "2026-11-03T22:55:46+00:00",
13 + "source_sha256": "2ca10cacc14b69a439286f84226831d7e5a8bfe7a7b838caf66f391b5f6c08a8",
14 + "parser": "pymupdf-jof-heuristics-1.3",
15 + "payload_chars": 24477,
16 + "usage": {
17 + "input": 8616,
18 + "output": 2568,
19 + "cache_read": 3968,
20 + "cache_write": 0
21 + },
22 + "cost_usd": 0.086618,
23 + "request_id": "req_011CggpThPZKLuuR58E9Q9sk",
24 + "annotation": {
25 + "genre": "mixed_theory_empirical",
26 + "subfield": "entrepreneurial_finance_private_equity",
27 + "subfield_secondary": "corporate_finance; asset_pricing (time-varying discount rates)",
28 + "primary_method": "reduced_form_regression",
29 + "identification": {
30 + "estimand": "Association between the predicted aggregate equity risk premium and buyout activity at the aggregate and firm level, plus cross-sectional differences in that sensitivity",
31 + "strategy": "Time-series regressions of buyout volume and value on predicted risk premium (OLS predictive and VAR/Campbell-Shiller estimates) versus credit factors. Firm panel deal-likelihood regressions with firm or time fixed effects. Interactions of the risk premium with firm risk, agency proxies and exit liquidity to test model-specific predictions.",
32 + "key_assumptions": [
33 + "Predicted excess return from D/P, cay and the T-bill rate (or a VAR) captures the aggregate risk premium",
34 + "Credit-spread measures adequately proxy for credit-market explanations",
35 + "Cross-sectional predictions (beta versus idiosyncratic risk) separate the risk-premium mechanism from debt-capacity stories"
36 + ],
37 + "threats_addressed": [
38 + "Credit market conditions as an alternative explanation",
39 + "Debt-equity mispricing alternative",
40 + "Persistence of regressors (Newey-West SEs)",
41 + "Seasonality (quarter dummies)",
42 + "Firm heterogeneity (firm fixed effects)",
43 + "Time-varying confounds in cross-section (time fixed effects)",
44 + "Expected growth as a separate valuation component"
45 + ]
46 + },
47 + "abstract_moves": {
48 + "question": {
49 + "present": true,
50 + "sentence_index": 0
51 + },
52 + "gap": {
53 + "present": false,
54 + "sentence_index": -1
55 + },
56 + "method": {
57 + "present": true,
58 + "sentence_index": 3
59 + },
60 + "results": {
61 + "present": true,
62 + "sentence_index": 1
63 + },
64 + "magnitude": {
65 + "present": false,
66 + "sentence_index": -1
67 + },
68 + "contribution": {
69 + "present": true,
70 + "sentence_index": 5
71 + }
72 + },
73 + "introduction": {
74 + "moves": [
75 + {
76 + "paragraph_index": 0,
77 + "move": "motivation"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "tension_or_puzzle"
82 + },
83 + {
84 + "paragraph_index": 0,
85 + "move": "related_literature_positioning"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "mechanism_or_theory_sketch"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "preview_of_approach"
98 + },
99 + {
100 + "paragraph_index": 2,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 2,
105 + "move": "robustness_mention"
106 + },
107 + {
108 + "paragraph_index": 3,
109 + "move": "identification_defense"
110 + },
111 + {
112 + "paragraph_index": 4,
113 + "move": "preview_of_results"
114 + },
115 + {
116 + "paragraph_index": 4,
117 + "move": "identification_defense"
118 + },
119 + {
120 + "paragraph_index": 5,
121 + "move": "preview_of_results"
122 + },
123 + {
124 + "paragraph_index": 5,
125 + "move": "robustness_mention"
126 + },
127 + {
128 + "paragraph_index": 6,
129 + "move": "contribution_statement"
130 + },
131 + {
132 + "paragraph_index": 6,
133 + "move": "related_literature_positioning"
134 + },
135 + {
136 + "paragraph_index": 7,
137 + "move": "roadmap"
138 + }
139 + ],
140 + "contribution_paragraph_index": 6,
141 + "contribution_types": [
142 + "new_evidence",
143 + "new_theory_or_model"
144 + ],
145 + "roadmap_present": true,
146 + "literature_positioning_location": "inside_introduction"
147 + },
148 + "section_roles": [
149 + {
150 + "idx": 0,
151 + "role": "introduction"
152 + },
153 + {
154 + "idx": 1,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 2,
159 + "role": "results"
160 + },
161 + {
162 + "idx": 3,
163 + "role": "theory"
164 + },
165 + {
166 + "idx": 4,
167 + "role": "theory"
168 + },
169 + {
170 + "idx": 5,
171 + "role": "results"
172 + },
173 + {
174 + "idx": 6,
175 + "role": "results"
176 + },
177 + {
178 + "idx": 7,
179 + "role": "results"
180 + },
181 + {
182 + "idx": 8,
183 + "role": "results"
184 + },
185 + {
186 + "idx": 9,
187 + "role": "conclusion"
188 + },
189 + {
190 + "idx": 10,
191 + "role": "appendix"
192 + },
193 + {
194 + "idx": 11,
195 + "role": "appendix"
196 + }
197 + ],
198 + "results_presentation": {
199 + "economic_significance_discussed": true,
200 + "statistical_significance_language": "mixed",
201 + "magnitude_statements_in_intro": true
202 + },
203 + "robustness": {
204 + "location": "within_results",
205 + "types": [
206 + "alternative_measure",
207 + "alternative_specification"
208 + ]
209 + },
210 + "tables": {
211 + "convention_notes": "Notes give the sample period (1982Q4–2011Q4) and define variables, including the risk premium estimators. Dollar values are deflated to 2010 dollars and activity is scaled in basis points of public firms. Aggregate tables report summary statistics first, then coefficient tables with panels or columns for alternative dependent variables.",
212 + "units_stated_in_notes": true,
213 + "standard_errors_description": "mixed"
214 + },
215 + "figures": {
216 + "uncertainty_shown": "no"
217 + },
218 + "conclusion": {
219 + "bounded_claims": false,
220 + "policy_implications": false,
221 + "open_questions": true,
222 + "restates_contribution": true
223 + },
224 + "hedging": {
225 + "level": "low",
226 + "typical_devices": [
227 + "'consistent with' phrasing",
228 + "'broadly match' for extension results",
229 + "acknowledging the simplified view of fund participants"
230 + ]
231 + },
232 + "claim_evidence": [
233 + {
234 + "claim": "Buyout activity falls when the aggregate risk premium is high and rises when it is low",
235 + "evidence_type": "figure",
236 + "evidence_ref": "Figure 1"
237 + },
238 + {
239 + "claim": "Risk premium explains about 30% of activity variation versus up to 10% for credit factors",
240 + "evidence_type": "table",
241 + "evidence_ref": "Table II"
242 + },
243 + {
244 + "claim": "Expected market growth is positively related to buyout activity",
245 + "evidence_type": "table",
246 + "evidence_ref": "Table IV"
247 + },
248 + {
249 + "claim": "Firms with high beta or idiosyncratic volatility are less likely buyout targets",
250 + "evidence_type": "table",
251 + "evidence_ref": "Table VII"
252 + },
253 + {
254 + "claim": "Target beta distribution shifts lower in high risk-premium periods while idiosyncratic risk does not",
255 + "evidence_type": "figure",
256 + "evidence_ref": "Figure 3; Table VIII"
257 + },
258 + {
259 + "claim": "Poor-governance and free-cash-flow firms are more sensitive, and more liquid industries less sensitive, to the risk premium",
260 + "evidence_type": "table",
261 + "evidence_ref": "Table IX"
262 + },
263 + {
264 + "claim": "Low risk premium raises the NPV of performance gains and lowers the cost of GP incentive risk",
265 + "evidence_type": "equation_or_proof",
266 + "evidence_ref": "Section II model; Appendix A–B"
267 + },
268 + {
269 + "claim": "M&A and IPO activity also co-move with the risk premium",
270 + "evidence_type": "table",
271 + "evidence_ref": "Table X"
272 + }
273 + ],
274 + "structural_summary": "Opens with large swings in buyout activity and offers a discount-rate explanation, set against credit-centric literature. Aggregate time-series evidence comes first (risk premium versus credit spreads, then a valuation decomposition). A two-period agency model follows, with performance and illiquidity channels. Firm-panel tests check model-specific cross-sectional predictions (beta versus idiosyncratic risk, agency proxies, exit liquidity) to rule out debt-capacity stories. The paper extends the analysis to M&A and IPOs and closes with heterogeneity caveats and a research agenda.",
275 + "writing_profile_keywords": [
276 + "discount-rate channel",
277 + "contrast with credit-market view",
278 + "model-derived cross-sectional predictions",
279 + "evidence-first then theory",
280 + "confident tone",
281 + "unified explanation"
282 + ],
283 + "confidence": 0.8,
284 + "notes": [
285 + "Causal language strength rated 4: the abstract says booms 'form in response to' and the conclusion says the risk premium 'explains' waves, but the evidence is regression-based with no exogenous variation.",
286 + "Standard errors are Newey-West in the aggregate regressions and two-way clustered by firm and quarter in the panel; the schema value 'mixed' covers both.",
287 + "Several 'sections' are actually equation fragments produced by parsing."
288 + ]
289 + },
290 + "deterministic_checks": {
291 + "roadmap_regex": false,
292 + "contribution_regex_in_intro": true,
293 + "magnitude_in_abstract": false,
294 + "abstract_n_sentences": 8,
295 + "intro_n_paragraphs": 8,
296 + "intro_n_words": 1160,
297 + "causal_words_in_abstract": 0,
298 + "hedge_words_in_abstract": 0,
299 + "stars_mentioned": false,
300 + "clustered_se_mentioned": false,
301 + "internet_appendix_mentioned": false,
302 + "section_roles_rule": {
303 + "0": "introduction",
304 + "1": "other",
305 + "2": "other",
306 + "3": "theory",
307 + "4": "other",
308 + "5": "other",
309 + "6": "other",
310 + "7": "other",
311 + "8": "other",
312 + "9": "conclusion",
313 + "10": "appendix",
314 + "11": "appendix"
315 + }
316 + },
317 + "schema_valid": false,
318 + "code_version": "feb6c9e-dirty"
319 +}
\ No newline at end of file
added data/annotations/documents/jof_12465.json +262 −0
@@ -0,0 +1,262 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12465",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:42+00:00",
9 + "source_sha256": "eb147af174fd904332d336776060a7b154281096a288c2a18053495227dc17cb",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24381,
12 + "usage": {
13 + "input": 8108,
14 + "output": 2021,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.073646,
19 + "request_id": "req_011CggpTfUy1ngLBvHDm2Qvu",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Differential sensitivity of local factor prices (wages, real estate) to aggregate GDP shocks by MSA local beta, and the cross-sectional relation between local beta and firms' industry-adjusted returns and conditional betas",
27 + "strategy": "Panel regressions that interact aggregate shocks with an industry-composition-weighted local beta; firm-level panel regressions and local-beta portfolio sorts within industry; supported by a calibrated production-based equilibrium model",
28 + "key_assumptions": [
29 + "No local factor mobility across markets at business-cycle frequency",
30 + "Firm location choice treated as given",
31 + "Headquarters location proxies for where the firm operates"
32 + ],
33 + "threats_addressed": [
34 + "labor mobility across areas",
35 + "headquarters location as a noisy proxy for operations (geographically focused firms)",
36 + "outlier firms driving results",
37 + "residual correlation structure",
38 + "sample period dependence",
39 + "measurement of local beta"
40 + ],
41 + "causal_language_strength": 3,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 4
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 1
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": false,
67 + "sentence_index": -1
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "mechanism_or_theory_sketch"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 2,
86 + "move": "preview_of_approach"
87 + },
88 + {
89 + "paragraph_index": 3,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 4,
94 + "move": "identification_defense"
95 + },
96 + {
97 + "paragraph_index": 5,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 6,
102 + "move": "robustness_mention"
103 + },
104 + {
105 + "paragraph_index": 7,
106 + "move": "related_literature_positioning"
107 + },
108 + {
109 + "paragraph_index": 8,
110 + "move": "related_literature_positioning"
111 + },
112 + {
113 + "paragraph_index": 8,
114 + "move": "contribution_statement"
115 + }
116 + ],
117 + "contribution_paragraph_index": 8,
118 + "contribution_types": [
119 + "new_evidence",
120 + "new_data_or_measure",
121 + "new_theory_or_model"
122 + ],
123 + "roadmap_present": false,
124 + "literature_positioning_location": "inside_introduction"
125 + },
126 + "section_roles": [
127 + {
128 + "idx": 0,
129 + "role": "introduction"
130 + },
131 + {
132 + "idx": 1,
133 + "role": "theory"
134 + },
135 + {
136 + "idx": 2,
137 + "role": "data"
138 + },
139 + {
140 + "idx": 3,
141 + "role": "results"
142 + },
143 + {
144 + "idx": 4,
145 + "role": "theory"
146 + },
147 + {
148 + "idx": 5,
149 + "role": "conclusion"
150 + },
151 + {
152 + "idx": 6,
153 + "role": "appendix"
154 + }
155 + ],
156 + "results_presentation": {
157 + "economic_significance_discussed": true,
158 + "statistical_significance_language": "unclear",
159 + "magnitude_statements_in_intro": true
160 + },
161 + "robustness": {
162 + "location": "mixed",
163 + "types": [
164 + "alternative_measure",
165 + "alternative_sample_period",
166 + "alternative_specification",
167 + "subsample",
168 + "other"
169 + ]
170 + },
171 + "tables": {
172 + "convention_notes": "Notes define the key variable (local beta) and refer back to its construction in an earlier table. Empirical tables are panel regressions of factor prices, betas and returns on local beta; one table reports portfolio sorts with three-factor alphas. Model tables give parameter values and the model-implied counterparts of the empirical regressions. Appendix tables list data sets and headquarters employment shares.",
173 + "units_stated_in_notes": true,
174 + "standard_errors_description": "unclear"
175 + },
176 + "figures": {
177 + "uncertainty_shown": "no"
178 + },
179 + "conclusion": {
180 + "bounded_claims": true,
181 + "policy_implications": false,
182 + "open_questions": false,
183 + "restates_contribution": true
184 + },
185 + "hedging": {
186 + "level": "moderate",
187 + "typical_devices": [
188 + "consistent with the view",
189 + "likely to",
190 + "suggest",
191 + "explicit discussion of simplifying assumptions"
192 + ]
193 + },
194 + "claim_evidence": [
195 + {
196 + "claim": "Wages respond more strongly to aggregate shocks in high local beta MSAs",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table III"
199 + },
200 + {
201 + "claim": "Housing and commercial real estate returns are more procyclical in high local beta areas",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table IV"
204 + },
205 + {
206 + "claim": "Firms in higher local beta areas have lower conditional equity betas",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table V"
209 + },
210 + {
211 + "claim": "Firms in higher local beta areas earn lower future returns, especially when real estate holdings are low",
212 + "evidence_type": "table",
213 + "evidence_ref": "Tables VI, VII, VIII"
214 + },
215 + {
216 + "claim": "Calibrated model reproduces procyclical factor prices and lower firm risk in high-beta areas",
217 + "evidence_type": "table",
218 + "evidence_ref": "Tables X, XI"
219 + },
220 + {
221 + "claim": "Labor mobility across counties is low at annual frequency",
222 + "evidence_type": "descriptive_statistic",
223 + "evidence_ref": "Intro paragraph 4, footnote 2"
224 + }
225 + ],
226 + "structural_summary": "The introduction moves from a stylized fact about immobile local factors to a mechanism illustrated by an example, then previews the model and results. It defends its simplifying assumptions, lists extensions and robustness checks, and positions the paper in the literature. A simplified labor-only model builds intuition before the data section constructs the local beta. The empirical section first tests factor prices, then firm risk and returns. A full three-input production model is calibrated to match the empirical patterns. The conclusion restates the findings without discussing policy.",
227 + "writing_profile_keywords": [
228 + "local beta",
229 + "production-based asset pricing",
230 + "model-then-evidence-then-full-model",
231 + "natural hedge mechanism",
232 + "panel regressions",
233 + "calibration"
234 + ],
235 + "confidence": 0.8,
236 + "notes": "Standard-error treatment is not visible; the introduction mentions checking various residual correlation assumptions. The model in Section IV is calibrated, not estimated."
237 + },
238 + "deterministic_checks": {
239 + "roadmap_regex": false,
240 + "contribution_regex_in_intro": true,
241 + "magnitude_in_abstract": false,
242 + "abstract_n_sentences": 5,
243 + "intro_n_paragraphs": 9,
244 + "intro_n_words": 1485,
245 + "causal_words_in_abstract": 0,
246 + "hedge_words_in_abstract": 0,
247 + "stars_mentioned": false,
248 + "clustered_se_mentioned": false,
249 + "internet_appendix_mentioned": true,
250 + "section_roles_rule": {
251 + "0": "introduction",
252 + "1": "theory",
253 + "2": "data",
254 + "3": "other",
255 + "4": "theory",
256 + "5": "conclusion",
257 + "6": "appendix"
258 + }
259 + },
260 + "schema_valid": true,
261 + "code_version": "feb6c9e-dirty"
262 +}
\ No newline at end of file
added data/annotations/documents/jof_12466.json +283 −0
@@ -0,0 +1,283 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12466",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:40+00:00",
9 + "source_sha256": "9d49fe27275082c2c93138bc92803f500bb95706ef4d018a546ff234d3522e92",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21840,
12 + "usage": {
13 + "input": 7330,
14 + "output": 2286,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.075834,
19 + "request_id": "req_011CggpXpfyceYJzHT1Gh13g",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "analyst research / information intermediaries (capital markets)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Difference in forecast accuracy, all-star likelihood, and market reactions between analysts with industry-related preanalyst experience and other analysts; DiD change in information asymmetry and prices after losing an expert versus a non-expert analyst",
27 + "strategy": "OLS and logit regressions with controls for analyst characteristics. Difference-in-differences around plausibly exogenous coverage terminations caused by brokerage house closures. Reg FD split used to probe the private-information channel.",
28 + "key_assumptions": [
29 + "Brokerage closures are exogenous to covered firms' information environment",
30 + "Controls capture other analyst skill attributes (experience, portfolio complexity, broker prestige)",
31 + "Self-selection of analysts into related industries does not drive results"
32 + ],
33 + "threats_addressed": [
34 + "private information via social connections (Reg FD test)",
35 + "sample selection/self-selection",
36 + "industry classification choice",
37 + "alternative performance measures and fixed effects",
38 + "analyst effort"
39 + ],
40 + "causal_language_strength": 3,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": false,
46 + "sentence_index": -1
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 0
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 0
59 + },
60 + "magnitude": {
61 + "present": false,
62 + "sentence_index": -1
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 2
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "related_literature_positioning"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "literature_gap"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "mechanism_or_theory_sketch"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "identification_defense"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "preview_of_results"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 3,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 4,
105 + "move": "contribution_statement"
106 + },
107 + {
108 + "paragraph_index": 5,
109 + "move": "roadmap"
110 + }
111 + ],
112 + "contribution_paragraph_index": 4,
113 + "contribution_types": [
114 + "new_evidence",
115 + "new_data_or_measure"
116 + ],
117 + "roadmap_present": true,
118 + "literature_positioning_location": "both"
119 + },
120 + "section_roles": [
121 + {
122 + "idx": 0,
123 + "role": "introduction"
124 + },
125 + {
126 + "idx": 1,
127 + "role": "literature"
128 + },
129 + {
130 + "idx": 2,
131 + "role": "data"
132 + },
133 + {
134 + "idx": 3,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 4,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 5,
143 + "role": "robustness"
144 + },
145 + {
146 + "idx": 6,
147 + "role": "conclusion"
148 + },
149 + {
150 + "idx": 7,
151 + "role": "appendix"
152 + },
153 + {
154 + "idx": 8,
155 + "role": "appendix"
156 + },
157 + {
158 + "idx": 9,
159 + "role": "appendix"
160 + }
161 + ],
162 + "results_presentation": {
163 + "economic_significance_discussed": true,
164 + "statistical_significance_language": "unclear",
165 + "magnitude_statements_in_intro": true
166 + },
167 + "robustness": {
168 + "location": "mixed",
169 + "types": [
170 + "alternative_measure",
171 + "alternative_specification",
172 + "subsample",
173 + "other"
174 + ]
175 + },
176 + "tables": {
177 + "convention_notes": "Table notes name the estimator (OLS, logistic, DiD means) and give dependent variable definitions in detail, pointing to Appendix C for other variables. Data sources and the sample period are stated. Notes are truncated, so the reporting of standard errors cannot be seen.",
178 + "units_stated_in_notes": true,
179 + "standard_errors_description": "unclear"
180 + },
181 + "figures": {
182 + "uncertainty_shown": "no_figures"
183 + },
184 + "conclusion": {
185 + "bounded_claims": true,
186 + "policy_implications": false,
187 + "open_questions": false,
188 + "restates_contribution": true
189 + },
190 + "hedging": {
191 + "level": "moderate",
192 + "typical_devices": [
193 + "plausibly exogenous",
194 + "consistent with",
195 + "cannot completely rule out",
196 + "unlikely to be a primary explanation",
197 + "suggesting"
198 + ]
199 + },
200 + "claim_evidence": [
201 + {
202 + "claim": "Analysts with preanalyst experience in related industries forecast more accurately; the effect holds only for related-industry firms.",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table II"
205 + },
206 + {
207 + "claim": "Odds of all-star status are 75% higher for related-experienced analysts than for unrelated-experienced analysts.",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table III"
210 + },
211 + {
212 + "claim": "Forecast revisions by related-experienced analysts trigger about 0.43–0.45% stronger abnormal reactions than those of inexperienced analysts.",
213 + "evidence_type": "table",
214 + "evidence_ref": "Table IV"
215 + },
216 + {
217 + "claim": "Losing an industry-expert analyst through broker closure raises information asymmetry more than losing other analysts.",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table VI"
220 + },
221 + {
222 + "claim": "The 3-day CAR difference-in-differences is 0.77% lower when an expert analyst is lost, and it does not reverse.",
223 + "evidence_type": "table",
224 + "evidence_ref": "Table VII"
225 + },
226 + {
227 + "claim": "The industry-experience effect persists after Reg FD, so a private-information channel is unlikely to be the main driver.",
228 + "evidence_type": "none_stated",
229 + "evidence_ref": "Introduction paragraph 1"
230 + },
231 + {
232 + "claim": "Results are robust to GICS classification and to alternative accuracy measures.",
233 + "evidence_type": "table",
234 + "evidence_ref": "Tables VIII, IX"
235 + },
236 + {
237 + "claim": "Recommendations of expert analysts have higher investment value.",
238 + "evidence_type": "table",
239 + "evidence_ref": "Table X"
240 + }
241 + ],
242 + "structural_summary": "The paper opens by citing practitioner surveys that rank industry knowledge as the most important analyst trait, then notes the lack of systematic evidence on it. It introduces hand-collected biographical data and previews results in sequence: forecast accuracy, a Reg FD channel test, all-star career outcomes, market reactions to revisions, and DiD tests on broker-closure terminations. A contribution paragraph and a roadmap close the introduction. The body then runs through a literature section, data, main results, analyst allocation and real effects, and robustness (largely in the Internet Appendix). The conclusion restates the findings.",
243 + "writing_profile_keywords": [
244 + "sell-side analysts",
245 + "hand-collected data",
246 + "practitioner motivation",
247 + "industry expertise",
248 + "difference-in-differences",
249 + "broker closures",
250 + "Internet Appendix robustness",
251 + "sequential results preview"
252 + ],
253 + "confidence": 0.8,
254 + "notes": "Paragraphs of the introduction appear to be partially missing between paragraphs 0 and 1, where the data and accuracy results are presumably introduced. The form of significance reporting is not visible in the excerpts. The Reg FD finding is stated in the introduction without a table reference, so its evidence type is recorded as none_stated."
255 + },
256 + "deterministic_checks": {
257 + "roadmap_regex": true,
258 + "contribution_regex_in_intro": true,
259 + "magnitude_in_abstract": false,
260 + "abstract_n_sentences": 3,
261 + "intro_n_paragraphs": 6,
262 + "intro_n_words": 872,
263 + "causal_words_in_abstract": 2,
264 + "hedge_words_in_abstract": 2,
265 + "stars_mentioned": false,
266 + "clustered_se_mentioned": false,
267 + "internet_appendix_mentioned": true,
268 + "section_roles_rule": {
269 + "0": "introduction",
270 + "1": "other",
271 + "2": "data",
272 + "3": "other",
273 + "4": "other",
274 + "5": "robustness",
275 + "6": "conclusion",
276 + "7": "appendix",
277 + "8": "appendix",
278 + "9": "appendix"
279 + }
280 + },
281 + "schema_valid": true,
282 + "code_version": "feb6c9e-dirty"
283 +}
\ No newline at end of file
added data/annotations/documents/jof_12467.json +255 −0
@@ -0,0 +1,255 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12467",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:21+00:00",
9 + "source_sha256": "4093c4ee9f55187820d8b17add241413334bf8a03bcc902064bce6113f166283",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21989,
12 + "usage": {
13 + "input": 7754,
14 + "output": 2111,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07403,
19 + "request_id": "req_011CggpWUprHSiSyMixMJ3cu",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "macro_finance (monetary policy transmission)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Effect of the short-term policy rate (federal funds rate) on the ex ante risk rating of newly issued bank loans, and how that effect varies with bank capitalization",
27 + "strategy": "Loan-level panel regressions of internal loan risk ratings on the federal funds rate interacted with bank capital, with loan, bank, regional and macro controls. Endogeneity is addressed by using new loans, Taylor rule residuals, regional business-cycle asynchrony, and small-bank and state exclusions.",
28 + "key_assumptions": [
29 + "Policy rate is plausibly exogenous to individual bank lending risk at loan origination",
30 + "Internal risk ratings comparably measure ex ante loan risk across banks",
31 + "Controls capture demand-side and economic-condition confounders"
32 + ],
33 + "threats_addressed": [
34 + "Reverse causality of monetary policy responding to bank risk-taking",
35 + "Changes in economic conditions",
36 + "Securitization and loan sales",
37 + "Loans under preexisting commitment",
38 + "Large-bank influence on policy",
39 + "Liquidity versus capital as the relevant bank characteristic"
40 + ],
41 + "causal_language_strength": 3,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 0
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "policy_or_broader_implication"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "mechanism_or_theory_sketch"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "mechanism_or_theory_sketch"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "tension_or_puzzle"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "preview_of_approach"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "identification_defense"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "related_literature_positioning"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "contribution_statement"
111 + },
112 + {
113 + "paragraph_index": 5,
114 + "move": "related_literature_positioning"
115 + }
116 + ],
117 + "contribution_paragraph_index": 4,
118 + "contribution_types": [
119 + "new_evidence",
120 + "new_data_or_measure"
121 + ],
122 + "roadmap_present": false,
123 + "literature_positioning_location": "inside_introduction"
124 + },
125 + "section_roles": [
126 + {
127 + "idx": 0,
128 + "role": "introduction"
129 + },
130 + {
131 + "idx": 1,
132 + "role": "empirical_design"
133 + },
134 + {
135 + "idx": 2,
136 + "role": "data"
137 + },
138 + {
139 + "idx": 3,
140 + "role": "results"
141 + },
142 + {
143 + "idx": 4,
144 + "role": "conclusion"
145 + },
146 + {
147 + "idx": 5,
148 + "role": "appendix"
149 + }
150 + ],
151 + "results_presentation": {
152 + "economic_significance_discussed": true,
153 + "statistical_significance_language": "unclear",
154 + "magnitude_statements_in_intro": true
155 + },
156 + "robustness": {
157 + "location": "within_results",
158 + "types": [
159 + "subsample",
160 + "alternative_measure",
161 + "alternative_specification",
162 + "other"
163 + ]
164 + },
165 + "tables": {
166 + "convention_notes": "Table notes state the sample period (1997Q2–2011Q4), define the dependent variable as the STBL internal risk rating on a 1–5 scale, list the data sources for the controls, and describe sample restrictions column by column.",
167 + "units_stated_in_notes": true,
168 + "standard_errors_description": "unclear"
169 + },
170 + "figures": {
171 + "uncertainty_shown": "unclear"
172 + },
173 + "conclusion": {
174 + "bounded_claims": true,
175 + "policy_implications": false,
176 + "open_questions": true,
177 + "restates_contribution": true
178 + },
179 + "hedging": {
180 + "level": "moderate",
181 + "typical_devices": [
182 + "'associated with' wording alongside causal claims",
183 + "'consistent with' a theoretical channel",
184 + "scope qualifier limiting results to the new-loan margin",
185 + "noting difficulty of quantifying the channel versus establishing it"
186 + ]
187 + },
188 + "claim_evidence": [
189 + {
190 + "claim": "Loan risk ratings rise when the federal funds rate falls; a one-SD rate decrease raises ratings by about 0.11",
191 + "evidence_type": "table",
192 + "evidence_ref": "Table IV"
193 + },
194 + {
195 + "claim": "The negative rate–risk relationship is weaker for poorly capitalized banks, consistent with risk-shifting",
196 + "evidence_type": "table",
197 + "evidence_ref": "Table IV"
198 + },
199 + {
200 + "claim": "Average loan risk rating co-moves negatively with the target federal funds rate over time",
201 + "evidence_type": "figure",
202 + "evidence_ref": "Figure 1"
203 + },
204 + {
205 + "claim": "Results are robust to Taylor rule residuals and to controls for economic conditions",
206 + "evidence_type": "table",
207 + "evidence_ref": "Table VIII"
208 + },
209 + {
210 + "claim": "Theory gives ambiguous predictions depending on maturity mismatch and leverage",
211 + "evidence_type": "equation_or_proof",
212 + "evidence_ref": "Appendix A"
213 + },
214 + {
215 + "claim": "Capital, not liquidity, drives the heterogeneity",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table VI"
218 + }
219 + ],
220 + "structural_summary": "The introduction motivates the topic with post-crisis policy debate and lays out competing theoretical channels (portfolio, risk-shifting, search for yield), concluding that the sign is an empirical question. It previews confidential loan-level data, main results with magnitudes, and an endogeneity defense, then positions the paper against prior risk measures and Jimenez et al. The body follows a short methodology section, a data section, and a long results section whose subsections are mostly robustness checks. The conclusion restates magnitudes and contrasts the findings with Spain, and a simple theoretical model sits in an appendix.",
221 + "writing_profile_keywords": [
222 + "loan-level confidential data",
223 + "interaction with bank capital",
224 + "risk-taking channel",
225 + "competing theoretical predictions",
226 + "extensive robustness subsections",
227 + "policy-relevant motivation"
228 + ],
229 + "confidence": 0.8,
230 + "notes": "Standard-error and significance conventions are not visible in the table-note excerpts. The abstract mentions regional business-cycle synchrony and financial distress, which correspond to the Table X analyses."
231 + },
232 + "deterministic_checks": {
233 + "roadmap_regex": false,
234 + "contribution_regex_in_intro": false,
235 + "magnitude_in_abstract": false,
236 + "abstract_n_sentences": 5,
237 + "intro_n_paragraphs": 6,
238 + "intro_n_words": 1076,
239 + "causal_words_in_abstract": 0,
240 + "hedge_words_in_abstract": 0,
241 + "stars_mentioned": false,
242 + "clustered_se_mentioned": false,
243 + "internet_appendix_mentioned": false,
244 + "section_roles_rule": {
245 + "0": "introduction",
246 + "1": "empirical_design",
247 + "2": "data",
248 + "3": "results",
249 + "4": "conclusion",
250 + "5": "appendix"
251 + }
252 + },
253 + "schema_valid": true,
254 + "code_version": "feb6c9e-dirty"
255 +}
\ No newline at end of file
added data/annotations/documents/jof_12468.json +247 −0
@@ -0,0 +1,247 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12468",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:03+00:00",
9 + "source_sha256": "cc1f315e36e8a23c68702ecd3dd7139fbb6b8fa8e0cf385e9aeecb1ca542f2e0",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 17358,
12 + "usage": {
13 + "input": 5832,
14 + "output": 1913,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.062382,
19 + "request_id": "req_011CggpVEKSPA7VTNM8ZPije",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "entrepreneurial_finance_private_equity",
23 + "subfield_secondary": "corporate_finance",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of an increase in housing collateral value on the probability of starting a firm and on postentry size, debt, growth and survival",
27 + "strategy": "Difference-in-differences-like comparison: local house-price appreciation interacted with homeownership. Treated homeowners are compared with renters, and with homeowners carrying an outstanding mortgage (who cannot take a second mortgage in France), within region-by-year fixed effects. A regional-level analysis interacts price growth with the homeownership share.",
28 + "key_assumptions": [
29 + "Local demand shocks affect owners and controls (renters, mortgaged owners) in the same region similarly",
30 + "French institutions prevent home equity withdrawal for mortgaged owners",
31 + "House-price variation is not correlated with owner-specific entrepreneurial opportunities"
32 + ],
33 + "threats_addressed": [
34 + "local economic or demand shocks",
35 + "wealth versus liquidity effects (full versus partial owners)",
36 + "individual characteristics",
37 + "crowding out at the regional level"
38 + ],
39 + "causal_language_strength": 4,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 2
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 0
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "literature_gap"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "policy_or_broader_implication"
81 + },
82 + {
83 + "paragraph_index": 0,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "preview_of_approach"
89 + },
90 + {
91 + "paragraph_index": 1,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 1,
96 + "move": "robustness_mention"
97 + },
98 + {
99 + "paragraph_index": 2,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 3,
104 + "move": "contribution_statement"
105 + },
106 + {
107 + "paragraph_index": 3,
108 + "move": "related_literature_positioning"
109 + }
110 + ],
111 + "contribution_paragraph_index": 3,
112 + "contribution_types": [
113 + "new_evidence",
114 + "new_identification"
115 + ],
116 + "roadmap_present": false,
117 + "literature_positioning_location": "inside_introduction"
118 + },
119 + "section_roles": [
120 + {
121 + "idx": 0,
122 + "role": "introduction"
123 + },
124 + {
125 + "idx": 1,
126 + "role": "other"
127 + },
128 + {
129 + "idx": 2,
130 + "role": "results"
131 + },
132 + {
133 + "idx": 3,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 4,
138 + "role": "results"
139 + },
140 + {
141 + "idx": 5,
142 + "role": "conclusion"
143 + }
144 + ],
145 + "results_presentation": {
146 + "economic_significance_discussed": true,
147 + "statistical_significance_language": "unclear",
148 + "magnitude_statements_in_intro": true
149 + },
150 + "robustness": {
151 + "location": "within_results",
152 + "types": [
153 + "subsample",
154 + "alternative_specification"
155 + ]
156 + },
157 + "tables": {
158 + "convention_notes": "Table notes describe the regression model (linear probability or OLS), the interaction term between house-price growth and owner status, the fixed effects and controls, and the sample period. Summary statistics tables precede regression tables for each analysis.",
159 + "units_stated_in_notes": false,
160 + "standard_errors_description": "unclear"
161 + },
162 + "figures": {
163 + "uncertainty_shown": "yes"
164 + },
165 + "conclusion": {
166 + "bounded_claims": true,
167 + "policy_implications": false,
168 + "open_questions": true,
169 + "restates_contribution": true
170 + },
171 + "hedging": {
172 + "level": "moderate",
173 + "typical_devices": [
174 + "consistent with",
175 + "may, in turn",
176 + "suggests",
177 + "improving identification"
178 + ]
179 + },
180 + "claim_evidence": [
181 + {
182 + "claim": "Collateral value increases raise homeowners' probability of entering entrepreneurship relative to renters and mortgaged owners",
183 + "evidence_type": "table",
184 + "evidence_ref": "Tables II, III"
185 + },
186 + {
187 + "claim": "Treated entrepreneurs start larger firms and use more debt at creation",
188 + "evidence_type": "table",
189 + "evidence_ref": "Table V"
190 + },
191 + {
192 + "claim": "Size differences persist through 2006",
193 + "evidence_type": "figure",
194 + "evidence_ref": "Table VI, Figure 2"
195 + },
196 + {
197 + "claim": "Effects are stronger in industries with more prevalent credit constraints",
198 + "evidence_type": "table",
199 + "evidence_ref": "Table VII"
200 + },
201 + {
202 + "claim": "Regional firm creation is more sensitive to house prices where the homeownership share is higher",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table VIII"
205 + },
206 + {
207 + "claim": "Moving from the 25th to the 75th percentile of house-price growth raises firm assets by about 13%",
208 + "evidence_type": "none_stated",
209 + "evidence_ref": "Intro paragraph 1"
210 + }
211 + ],
212 + "structural_summary": "The introduction opens with the claim of the paper, notes that the existing wealth–entrepreneurship correlation is ambiguous, and stresses its policy relevance. It then sets out a three-part design (extensive margin, intensive margin, aggregate) with results previewed for each part, and ends with literature positioning. An institutional section on French mortgages justifies the control groups. Three empirical sections follow a parallel structure: data, specification, identification, results and sample splits. A short conclusion restates the contribution, contrasts it with Mian and Sufi, and leaves quantification of the relative channels for future research.",
213 + "writing_profile_keywords": [
214 + "institutional setting as identification",
215 + "parallel section structure",
216 + "extensive and intensive margin",
217 + "administrative data",
218 + "policy motivation",
219 + "interaction design"
220 + ],
221 + "confidence": 0.85,
222 + "notes": "The excerpts give no detail on standard errors or significance notation. Robustness checks appear within results via sample splits and controls."
223 + },
224 + "deterministic_checks": {
225 + "roadmap_regex": false,
226 + "contribution_regex_in_intro": true,
227 + "magnitude_in_abstract": false,
228 + "abstract_n_sentences": 4,
229 + "intro_n_paragraphs": 4,
230 + "intro_n_words": 752,
231 + "causal_words_in_abstract": 1,
232 + "hedge_words_in_abstract": 0,
233 + "stars_mentioned": false,
234 + "clustered_se_mentioned": false,
235 + "internet_appendix_mentioned": true,
236 + "section_roles_rule": {
237 + "0": "introduction",
238 + "1": "other",
239 + "2": "other",
240 + "3": "other",
241 + "4": "other",
242 + "5": "conclusion"
243 + }
244 + },
245 + "schema_valid": true,
246 + "code_version": "feb6c9e-dirty"
247 +}
\ No newline at end of file
added data/annotations/documents/jof_12469.json +287 −0
@@ -0,0 +1,287 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12469",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:25+00:00",
9 + "source_sha256": "a3f17f97e3aea7a3a86094562189444cf1f29fd622c82c3ca19718bff41d800a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24431,
12 + "usage": {
13 + "input": 7744,
14 + "output": 2385,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07947,
19 + "request_id": "req_011CggpSEW2QrqobxsXayFPw",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "natural_experiment_event_study",
25 + "identification": {
26 + "estimand": "Effect of depression-era local (region-occupation) labor market deterioration on long-run risky asset investment more than a decade later, plus the part not working through income, employment or wealth",
27 + "strategy": "Cross-sectional variation in region-occupation unemployment changes during the 1991-93 Finnish depression, conditional on region and occupation fixed effects and predepression controls; secondhand experiences of neighbors and family isolate non-wealth channels; matching as an alternative estimator",
28 + "key_assumptions": [
29 + "Conditional independence: exposure to local labor market distress is unrelated to unobserved worker characteristics given fixed effects and predepression controls",
30 + "Secondhand experiences do not affect the individual's wealth, income or employment",
31 + "Depression shocks (Soviet collapse, twin crisis) were unexpected"
32 + ],
33 + "threats_addressed": [
34 + "Omitted variables / selection into occupations and regions",
35 + "Wealth, income and employment channels as bad controls",
36 + "Measurement error from recall bias in survey data",
37 + "Sensitivity to the labor market definition"
38 + ],
39 + "causal_language_strength": 4,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 2
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "tension_or_puzzle"
73 + },
74 + {
75 + "paragraph_index": 1,
76 + "move": "literature_gap"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "contribution_statement"
81 + },
82 + {
83 + "paragraph_index": 2,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "identification_defense"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "identification_defense"
97 + },
98 + {
99 + "paragraph_index": 5,
100 + "move": "preview_of_approach"
101 + },
102 + {
103 + "paragraph_index": 6,
104 + "move": "identification_defense"
105 + },
106 + {
107 + "paragraph_index": 7,
108 + "move": "preview_of_results"
109 + },
110 + {
111 + "paragraph_index": 8,
112 + "move": "preview_of_approach"
113 + },
114 + {
115 + "paragraph_index": 9,
116 + "move": "preview_of_results"
117 + },
118 + {
119 + "paragraph_index": 10,
120 + "move": "policy_or_broader_implication"
121 + },
122 + {
123 + "paragraph_index": 10,
124 + "move": "related_literature_positioning"
125 + },
126 + {
127 + "paragraph_index": 11,
128 + "move": "related_literature_positioning"
129 + },
130 + {
131 + "paragraph_index": 12,
132 + "move": "related_literature_positioning"
133 + },
134 + {
135 + "paragraph_index": 13,
136 + "move": "related_literature_positioning"
137 + },
138 + {
139 + "paragraph_index": 14,
140 + "move": "roadmap"
141 + }
142 + ],
143 + "contribution_paragraph_index": 1,
144 + "contribution_types": [
145 + "new_evidence",
146 + "new_identification",
147 + "new_method"
148 + ],
149 + "roadmap_present": true,
150 + "literature_positioning_location": "inside_introduction"
151 + },
152 + "section_roles": [
153 + {
154 + "idx": 0,
155 + "role": "introduction"
156 + },
157 + {
158 + "idx": 1,
159 + "role": "empirical_design"
160 + },
161 + {
162 + "idx": 2,
163 + "role": "results"
164 + },
165 + {
166 + "idx": 3,
167 + "role": "results"
168 + },
169 + {
170 + "idx": 4,
171 + "role": "conclusion"
172 + }
173 + ],
174 + "results_presentation": {
175 + "economic_significance_discussed": true,
176 + "statistical_significance_language": "stars_and_tstats",
177 + "magnitude_statements_in_intro": true
178 + },
179 + "robustness": {
180 + "location": "within_results",
181 + "types": [
182 + "alternative_specification",
183 + "alternative_measure",
184 + "falsification",
185 + "subsample",
186 + "other"
187 + ]
188 + },
189 + "tables": {
190 + "convention_notes": "Notes report coefficients with t-values and spell out the dependent variable, control sets and sample. Income and wealth are in 2005 CPI-adjusted terms. Later tables cross-reference the control sets of earlier columns. Table I pairs a falsification test with descriptive statistics.",
191 + "units_stated_in_notes": true,
192 + "standard_errors_description": "clustered"
193 + },
194 + "figures": {
195 + "uncertainty_shown": "unclear"
196 + },
197 + "conclusion": {
198 + "bounded_claims": true,
199 + "policy_implications": false,
200 + "open_questions": true,
201 + "restates_contribution": true
202 + },
203 + "hedging": {
204 + "level": "moderate",
205 + "typical_devices": [
206 + "plausibly exogenous",
207 + "suggest",
208 + "likely",
209 + "unlikely to explain",
210 + "interpret as lower bound",
211 + "candidate explanations"
212 + ]
213 + },
214 + "claim_evidence": [
215 + {
216 + "claim": "A one-SD worse labor market experience lowers stock market participation by about 2.8-3.6 pp, against a 22% base rate",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table II"
219 + },
220 + {
221 + "claim": "Depression-era labor conditions do not predict predepression stock market participation (falsification)",
222 + "evidence_type": "figure",
223 + "evidence_ref": "Figure 3; Table I"
224 + },
225 + {
226 + "claim": "Labor market conditions affect later income, unemployment and net worth",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table III"
229 + },
230 + {
231 + "claim": "Results hold with family fixed effects, parental controls and alternative labor market definitions",
232 + "evidence_type": "table",
233 + "evidence_ref": "Table IV"
234 + },
235 + {
236 + "claim": "Matching adversely affected workers to less affected ones yields similar results",
237 + "evidence_type": "table",
238 + "evidence_ref": "Table V"
239 + },
240 + {
241 + "claim": "Adverse experiences also reduce holdings of fixed income funds, balanced funds and derivatives",
242 + "evidence_type": "table",
243 + "evidence_ref": "Table VI"
244 + },
245 + {
246 + "claim": "Neighbors' and family members' adverse experiences reduce risky investment by 0.4-0.6 pp, so at least 11% of the effect is not via wealth or income",
247 + "evidence_type": "table",
248 + "evidence_ref": "Tables VII, VIII"
249 + }
250 + ],
251 + "structural_summary": "The introduction frames portfolio heterogeneity as a puzzle that twin studies leave open. It then organizes the design around three identification challenges, each answered by a feature of the Finnish depression setting, and previews effect sizes with clustered t-values. Four literature strands follow, then a roadmap. Section I covers the timeline, register data and a conditional independence defense. Section II gives baseline firsthand-experience regressions followed by robustness checks: controls, labor market definitions, regional analyses, matching and alternative assets. Section III uses secondhand experiences of neighbors and family to separate preference and belief channels from wealth channels. The conclusion restates the contribution, bounds the mechanism to preferences, beliefs or trust, and suggests extensions.",
252 + "writing_profile_keywords": [
253 + "natural experiment",
254 + "register data",
255 + "identification challenges enumerated",
256 + "falsification test",
257 + "secondhand experiences",
258 + "lower-bound interpretation",
259 + "clustered t-values",
260 + "mechanism isolation"
261 + ],
262 + "confidence": 0.85,
263 + "notes": "Section II carries the main results; the 'other' role assigned by the rule was reinterpreted as results. Section III was labeled robustness by the rule but serves as channel analysis, so it is classified as results. Whether figures show uncertainty cannot be determined from the captions."
264 + },
265 + "deterministic_checks": {
266 + "roadmap_regex": true,
267 + "contribution_regex_in_intro": true,
268 + "magnitude_in_abstract": false,
269 + "abstract_n_sentences": 6,
270 + "intro_n_paragraphs": 15,
271 + "intro_n_words": 1483,
272 + "causal_words_in_abstract": 3,
273 + "hedge_words_in_abstract": 1,
274 + "stars_mentioned": false,
275 + "clustered_se_mentioned": false,
276 + "internet_appendix_mentioned": false,
277 + "section_roles_rule": {
278 + "0": "introduction",
279 + "1": "empirical_design",
280 + "2": "other",
281 + "3": "robustness",
282 + "4": "conclusion"
283 + }
284 + },
285 + "schema_valid": true,
286 + "code_version": "feb6c9e-dirty"
287 +}
\ No newline at end of file
added data/annotations/documents/jof_12470.json +291 −0
@@ -0,0 +1,291 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12470",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:16+00:00",
9 + "source_sha256": "6b3aa29ef900c7fdb1d4cd7740fc23be08efe0f8fba7adc741966c44783bde07",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24571,
12 + "usage": {
13 + "input": 7796,
14 + "output": 2245,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.076878,
19 + "request_id": "req_011CggpW3Hwckb3yVmdXQ6tj",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "entrepreneurial_finance_private_equity",
23 + "subfield_secondary": "corporate_finance",
24 + "primary_method": "survey_or_experiment",
25 + "identification": {
26 + "estimand": "Effect on investor click-through (interest) of disclosing each information category (team, traction, current investors) about a fund-raising start-up",
27 + "strategy": "Randomized field experiment: among the categories that pass the platform's disclosure threshold, which ones appear in featured emails is randomized; linear regressions of clicks on category indicators with start-up fixed effects, using opened emails only",
28 + "key_assumptions": [
29 + "Randomization of displayed categories is independent of investor and start-up characteristics",
30 + "Investors read a missing category as signaling low quality",
31 + "Click-through captures investor interest at the initial screening stage"
32 + ],
33 + "threats_addressed": [
34 + "Omitted unobservable start-up characteristics (start-up fixed effects plus randomization)",
35 + "Investors not seeing the email (sample restricted to opened emails)",
36 + "External validity of the 21 experiment firms (compared with the broad AngelList sample)",
37 + "Ordering of information categories",
38 + "Differences in signal-to-noise ratios across categories",
39 + "Interpretation of click rates as interest"
40 + ],
41 + "causal_language_strength": 4,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 4
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "literature_gap"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "contribution_statement"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "mechanism_or_theory_sketch"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "identification_defense"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "preview_of_approach"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "preview_of_approach"
99 + },
100 + {
101 + "paragraph_index": 5,
102 + "move": "preview_of_results"
103 + },
104 + {
105 + "paragraph_index": 6,
106 + "move": "mechanism_or_theory_sketch"
107 + },
108 + {
109 + "paragraph_index": 6,
110 + "move": "preview_of_results"
111 + },
112 + {
113 + "paragraph_index": 7,
114 + "move": "other"
115 + },
116 + {
117 + "paragraph_index": 8,
118 + "move": "preview_of_results"
119 + },
120 + {
121 + "paragraph_index": 9,
122 + "move": "related_literature_positioning"
123 + },
124 + {
125 + "paragraph_index": 10,
126 + "move": "related_literature_positioning"
127 + },
128 + {
129 + "paragraph_index": 10,
130 + "move": "contribution_statement"
131 + },
132 + {
133 + "paragraph_index": 11,
134 + "move": "related_literature_positioning"
135 + },
136 + {
137 + "paragraph_index": 12,
138 + "move": "roadmap"
139 + }
140 + ],
141 + "contribution_paragraph_index": 0,
142 + "contribution_types": [
143 + "new_evidence",
144 + "new_identification"
145 + ],
146 + "roadmap_present": true,
147 + "literature_positioning_location": "inside_introduction"
148 + },
149 + "section_roles": [
150 + {
151 + "idx": 0,
152 + "role": "introduction"
153 + },
154 + {
155 + "idx": 1,
156 + "role": "other"
157 + },
158 + {
159 + "idx": 2,
160 + "role": "empirical_design"
161 + },
162 + {
163 + "idx": 3,
164 + "role": "data"
165 + },
166 + {
167 + "idx": 4,
168 + "role": "results"
169 + },
170 + {
171 + "idx": 5,
172 + "role": "results"
173 + },
174 + {
175 + "idx": 6,
176 + "role": "results"
177 + },
178 + {
179 + "idx": 7,
180 + "role": "robustness"
181 + },
182 + {
183 + "idx": 8,
184 + "role": "conclusion"
185 + }
186 + ],
187 + "results_presentation": {
188 + "economic_significance_discussed": false,
189 + "statistical_significance_language": "unclear",
190 + "magnitude_statements_in_intro": false
191 + },
192 + "robustness": {
193 + "location": "main_text_section",
194 + "types": [
195 + "subsample",
196 + "alternative_measure",
197 + "alternative_specification",
198 + "other"
199 + ]
200 + },
201 + "tables": {
202 + "convention_notes": "Table notes define the binary click outcome and the information-category indicators and state that only opened emails are used. Descriptive tables cover emails, start-ups and investors, and a final table compares experiment firms with the broad platform sample. No coefficient values or standard-error details are visible in the excerpts.",
203 + "units_stated_in_notes": true,
204 + "standard_errors_description": "unclear"
205 + },
206 + "figures": {
207 + "uncertainty_shown": "no"
208 + },
209 + "conclusion": {
210 + "bounded_claims": true,
211 + "policy_implications": false,
212 + "open_questions": false,
213 + "restates_contribution": true
214 + },
215 + "hedging": {
216 + "level": "moderate",
217 + "typical_devices": [
218 + "suggestive evidence",
219 + "indirect evidence",
220 + "explicit statement of limitations (long-run outcomes not observed)",
221 + "results consistent with an existing model",
222 + "disclaimer that nonhuman assets may still matter"
223 + ]
224 + },
225 + "claim_evidence": [
226 + {
227 + "claim": "Investors' click rates respond strongly to founding-team information but not significantly to traction or current-investor information",
228 + "evidence_type": "table",
229 + "evidence_ref": "Table IV"
230 + },
231 + {
232 + "claim": "Investors specialized in the start-up's sector react as strongly to team information, indicating that operational ability matters beyond signaling",
233 + "evidence_type": "table",
234 + "evidence_ref": "Table VI"
235 + },
236 + {
237 + "claim": "More experienced and successful investors react only to team information, indirectly supporting team-based selection as a viable strategy",
238 + "evidence_type": "table",
239 + "evidence_ref": "Tables VII, IX"
240 + },
241 + {
242 + "claim": "Experiment start-ups are compared with the broad AngelList population to assess external validity",
243 + "evidence_type": "table",
244 + "evidence_ref": "Table X"
245 + },
246 + {
247 + "claim": "Successful early-stage investing reflects skill, as shown by persistence of VC returns",
248 + "evidence_type": "prior_literature",
249 + "evidence_ref": "Kaplan and Schoar (2005) et al."
250 + }
251 + ],
252 + "structural_summary": "The introduction motivates the question of how early-stage investors select start-ups and notes scarce evidence. It frames three competing hypotheses drawn from theories of the firm and herding, explains the omitted-variable and selected-sample problems, and introduces a randomized email experiment on AngelList. The body describes the platform, the experimental design and summary statistics, then reports main click-rate regressions. Two follow-up sections test mechanisms (signal versus operational capability) and the rationality of team-based selection. A discussion section addresses robustness and external validity, and a short conclusion restates the findings and links them to Rajan (2012).",
253 + "writing_profile_keywords": [
254 + "randomized field experiment",
255 + "competing hypotheses framing",
256 + "online platform data",
257 + "mechanism exploration",
258 + "explicit limitations",
259 + "first experimental evidence claim",
260 + "entrepreneurial finance"
261 + ],
262 + "confidence": 0.85,
263 + "notes": "The section role labeled 'conclusion' for Summary Statistics has been corrected to 'data'. The Discussion section serves as the robustness section. Standard-error treatment and significance notation are not visible in the excerpts."
264 + },
265 + "deterministic_checks": {
266 + "roadmap_regex": true,
267 + "contribution_regex_in_intro": true,
268 + "magnitude_in_abstract": false,
269 + "abstract_n_sentences": 5,
270 + "intro_n_paragraphs": 13,
271 + "intro_n_words": 1529,
272 + "causal_words_in_abstract": 0,
273 + "hedge_words_in_abstract": 1,
274 + "stars_mentioned": false,
275 + "clustered_se_mentioned": false,
276 + "internet_appendix_mentioned": false,
277 + "section_roles_rule": {
278 + "0": "introduction",
279 + "1": "other",
280 + "2": "other",
281 + "3": "conclusion",
282 + "4": "results",
283 + "5": "other",
284 + "6": "other",
285 + "7": "conclusion",
286 + "8": "conclusion"
287 + }
288 + },
289 + "schema_valid": true,
290 + "code_version": "feb6c9e-dirty"
291 +}
\ No newline at end of file
added data/annotations/documents/jof_12471.json +300 −0
@@ -0,0 +1,300 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12471",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:46+00:00",
9 + "source_sha256": "472bd685e66434a2558b39b37ac206974f1bfe7003beef332e75c88f47f5858a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 25431,
12 + "usage": {
13 + "input": 8958,
14 + "output": 2489,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.086406,
19 + "request_id": "req_011CggpTnTALxqDYAnV6YtEi",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "household_finance; macro_finance",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Structural preference and shock-process parameters of a heterogeneous-household equilibrium model, together with the fit of aggregate and household cross-sectional moments; cross-sectional prices of household consumption risk",
27 + "strategy": "GMM estimation that jointly matches aggregate asset-market moments and cross-sectional moments of CEX household consumption growth; Fama-MacBeth two-pass regressions for the cross section of returns",
28 + "key_assumptions": [
29 + "Market incompleteness: households cannot trade claims on their wealth portfolios",
30 + "Autarchy equilibrium holds",
31 + "Aggregate consumption growth is i.i.d.",
32 + "A single countercyclical state variable drives the cross-sectional moments",
33 + "Household shocks are conditionally independent of the rest of the economy",
34 + "Log-linearization of the consumption-wealth ratio"
35 + ],
36 + "threats_addressed": [
37 + "Measurement error in survey-based CEX consumption, handled by matching time-series averages of moments",
38 + "Sensitivity to sample period and data frequency",
39 + "Role of recursive versus CRRA preferences",
40 + "Role of skewness versus lognormal shocks",
41 + "Criticism of high-EIS models"
42 + ],
43 + "causal_language_strength": 2,
44 + "uses_causal_claim": false
45 + },
46 + "abstract_moves": {
47 + "question": {
48 + "present": true,
49 + "sentence_index": 0
50 + },
51 + "gap": {
52 + "present": false,
53 + "sentence_index": -1
54 + },
55 + "method": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "results": {
60 + "present": true,
61 + "sentence_index": 3
62 + },
63 + "magnitude": {
64 + "present": false,
65 + "sentence_index": -1
66 + },
67 + "contribution": {
68 + "present": true,
69 + "sentence_index": 0
70 + }
71 + },
72 + "introduction": {
73 + "moves": [
74 + {
75 + "paragraph_index": 0,
76 + "move": "motivation"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "related_literature_positioning"
81 + },
82 + {
83 + "paragraph_index": 0,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "contribution_statement"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "mechanism_or_theory_sketch"
97 + },
98 + {
99 + "paragraph_index": 2,
100 + "move": "related_literature_positioning"
101 + },
102 + {
103 + "paragraph_index": 3,
104 + "move": "related_literature_positioning"
105 + },
106 + {
107 + "paragraph_index": 3,
108 + "move": "contribution_statement"
109 + },
110 + {
111 + "paragraph_index": 4,
112 + "move": "identification_defense"
113 + },
114 + {
115 + "paragraph_index": 5,
116 + "move": "related_literature_positioning"
117 + }
118 + ],
119 + "contribution_paragraph_index": 1,
120 + "contribution_types": [
121 + "new_evidence",
122 + "new_theory_or_model"
123 + ],
124 + "roadmap_present": false,
125 + "literature_positioning_location": "inside_introduction"
126 + },
127 + "section_roles": [
128 + {
129 + "idx": 0,
130 + "role": "introduction"
131 + },
132 + {
133 + "idx": 1,
134 + "role": "theory"
135 + },
136 + {
137 + "idx": 2,
138 + "role": "data"
139 + },
140 + {
141 + "idx": 3,
142 + "role": "results"
143 + },
144 + {
145 + "idx": 4,
146 + "role": "empirical_design"
147 + },
148 + {
149 + "idx": 5,
150 + "role": "robustness"
151 + },
152 + {
153 + "idx": 6,
154 + "role": "results"
155 + },
156 + {
157 + "idx": 7,
158 + "role": "conclusion"
159 + },
160 + {
161 + "idx": 8,
162 + "role": "appendix"
163 + },
164 + {
165 + "idx": 9,
166 + "role": "appendix"
167 + },
168 + {
169 + "idx": 10,
170 + "role": "appendix"
171 + },
172 + {
173 + "idx": 11,
174 + "role": "appendix"
175 + }
176 + ],
177 + "results_presentation": {
178 + "economic_significance_discussed": true,
179 + "statistical_significance_language": "mixed",
180 + "magnitude_statements_in_intro": true
181 + },
182 + "robustness": {
183 + "location": "main_text_section",
184 + "types": [
185 + "alternative_sample_period",
186 + "alternative_specification",
187 + "sensitivity_to_parameters",
188 + "other"
189 + ]
190 + },
191 + "tables": {
192 + "convention_notes": "Each estimation table reports parameter estimates alongside model-implied versus sample moments. Asymptotic Newey-West standard errors (two lags) appear in parentheses, and notes state the number of GMM moment restrictions. Fama-MacBeth tables report results by test-asset panel with adjusted R2. Notes specify the CEX tranche used and the sample period.",
193 + "units_stated_in_notes": false,
194 + "standard_errors_description": "newey_west"
195 + },
196 + "figures": {
197 + "uncertainty_shown": "unclear"
198 + },
199 + "conclusion": {
200 + "bounded_claims": true,
201 + "policy_implications": false,
202 + "open_questions": false,
203 + "restates_contribution": true
204 + },
205 + "hedging": {
206 + "level": "moderate",
207 + "typical_devices": [
208 + "can, in principle",
209 + "results suggest",
210 + "fits reasonably well",
211 + "acknowledgment that estimates are noisy due to measurement error",
212 + "at least as well as"
213 + ]
214 + },
215 + "claim_evidence": [
216 + {
217 + "claim": "Cross-sectional skewness of household consumption growth is mostly negative and countercyclical, and moderately correlated with recessions",
218 + "evidence_type": "figure",
219 + "evidence_ref": "Figure 1; Table II"
220 + },
221 + {
222 + "claim": "Moments of household consumption growth are persistent, with positive autocorrelations of the third moment and the fifth percentile",
223 + "evidence_type": "table",
224 + "evidence_ref": "Table I"
225 + },
226 + {
227 + "claim": "The estimated model matches household cross-sectional moments and asset-price moments with a low EIS",
228 + "evidence_type": "table",
229 + "evidence_ref": "Table III"
230 + },
231 + {
232 + "claim": "Results are similar over the longer quarterly sample and at annual frequency",
233 + "evidence_type": "table",
234 + "evidence_ref": "Tables IV, V"
235 + },
236 + {
237 + "claim": "Recursive preferences are needed to address the equity premium and risk-free rate puzzles",
238 + "evidence_type": "table",
239 + "evidence_ref": "Tables VI, VII"
240 + },
241 + {
242 + "claim": "Negative skewness of household shocks is pivotal for fitting the data",
243 + "evidence_type": "table",
244 + "evidence_ref": "Table VIII"
245 + },
246 + {
247 + "claim": "Household consumption risk prices the cross section of portfolio returns at least as well as the Fama-French factors",
248 + "evidence_type": "table",
249 + "evidence_ref": "Tables X, XI"
250 + },
251 + {
252 + "claim": "Rare-disaster calibrations magnify annual consumption declines in an ad hoc way",
253 + "evidence_type": "prior_literature",
254 + "evidence_ref": "Constantinides (2008); Julliard and Ghosh (2012)"
255 + }
256 + ],
257 + "structural_summary": "The introduction motivates the paper with labor-income evidence that the left skewness of idiosyncratic shocks is countercyclical. It positions uninsurable idiosyncratic risk as a paradigm alternative to long-run risks, habit, and disaster models, previews descriptive skewness evidence, and contrasts the approach with related literature. A parsimonious incomplete-market model with recursive preferences and a single state variable is then laid out. Data construction from CRSP and the CEX follows, along with descriptive household consumption statistics. The model is estimated by GMM jointly on aggregate and household moments. Variant estimations then probe sample period, data frequency, preferences, and the shock distribution. A Fama-MacBeth cross-sectional test is presented, and the conclusion restates the findings. Proofs and derivations are placed in appendices.",
258 + "writing_profile_keywords": [
259 + "structural GMM",
260 + "incomplete markets",
261 + "recursive preferences",
262 + "household heterogeneity",
263 + "model-fit tables",
264 + "paradigm positioning",
265 + "ablation of model ingredients",
266 + "Fama-MacBeth"
267 + ],
268 + "confidence": 0.8,
269 + "notes": "Section V functions as both robustness and ablation analysis. The introduction excerpt shows no explicit roadmap. Section III is descriptive evidence and is labeled as results here."
270 + },
271 + "deterministic_checks": {
272 + "roadmap_regex": false,
273 + "contribution_regex_in_intro": true,
274 + "magnitude_in_abstract": false,
275 + "abstract_n_sentences": 6,
276 + "intro_n_paragraphs": 6,
277 + "intro_n_words": 936,
278 + "causal_words_in_abstract": 0,
279 + "hedge_words_in_abstract": 0,
280 + "stars_mentioned": false,
281 + "clustered_se_mentioned": false,
282 + "internet_appendix_mentioned": false,
283 + "section_roles_rule": {
284 + "0": "introduction",
285 + "1": "theory",
286 + "2": "data",
287 + "3": "other",
288 + "4": "empirical_design",
289 + "5": "results",
290 + "6": "other",
291 + "7": "conclusion",
292 + "8": "appendix",
293 + "9": "appendix",
294 + "10": "appendix",
295 + "11": "appendix"
296 + }
297 + },
298 + "schema_valid": true,
299 + "code_version": "feb6c9e-dirty"
300 +}
\ No newline at end of file
added data/annotations/documents/jof_12472.json +253 −0
@@ -0,0 +1,253 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12472",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:55:57+00:00",
9 + "source_sha256": "433811aa3e514577e71252136a2cd02482802ae8735846591970292a99f43597",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19451,
12 + "usage": {
13 + "input": 6460,
14 + "output": 2139,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.069414,
19 + "request_id": "req_011CggpUj8WQTmW1Q6F8gkaQ",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "asset_pricing (ex-dividend pricing, institutional investor skill)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Abnormal profits of institutional dividend capture trades after transaction costs, and the cross-sectional relation between those profits and prior trade execution skill",
27 + "strategy": "Transaction-level institutional data (Abel Noser) used to compute realized ex-day returns net of commissions; portfolio sorts by abnormal buy intensity and by skill deciles; regressions of buy intensity on event and trader characteristics; comparison against placebo non-ex-days and non-dividend trades",
28 + "key_assumptions": [
29 + "Prior execution-skill measures (AIPV) capture persistent trader ability rather than stock-picking skill",
30 + "Placebo non-ex-days are a valid comparison for ex-days",
31 + "Benchmark-adjusted abnormal returns (DGTW) correctly remove risk-related returns"
32 + ],
33 + "threats_addressed": [
34 + "Skill-profit relation could be generic rather than ex-day specific (placebo non-ex-day test)",
35 + "Profits could come from stock-picking rather than execution (stock-picking skill and undervaluation checks)",
36 + "Transaction costs eliminating average ex-day returns (actual execution prices and commissions used)"
37 + ],
38 + "causal_language_strength": 2,
39 + "uses_causal_claim": false
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 0
45 + },
46 + "gap": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 0
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 1
57 + },
58 + "magnitude": {
59 + "present": true,
60 + "sentence_index": 2
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 7
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "preview_of_approach"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "related_literature_positioning"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "other"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 3,
99 + "move": "identification_defense"
100 + },
101 + {
102 + "paragraph_index": 4,
103 + "move": "contribution_statement"
104 + },
105 + {
106 + "paragraph_index": 4,
107 + "move": "roadmap"
108 + }
109 + ],
110 + "contribution_paragraph_index": 4,
111 + "contribution_types": [
112 + "new_evidence",
113 + "new_data_or_measure"
114 + ],
115 + "roadmap_present": true,
116 + "literature_positioning_location": "both"
117 + },
118 + "section_roles": [
119 + {
120 + "idx": 0,
121 + "role": "introduction"
122 + },
123 + {
124 + "idx": 1,
125 + "role": "literature"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "data"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "results"
138 + },
139 + {
140 + "idx": 5,
141 + "role": "conclusion"
142 + }
143 + ],
144 + "results_presentation": {
145 + "economic_significance_discussed": true,
146 + "statistical_significance_language": "unclear",
147 + "magnitude_statements_in_intro": true
148 + },
149 + "robustness": {
150 + "location": "mixed",
151 + "types": [
152 + "placebo",
153 + "subsample",
154 + "alternative_measure"
155 + ]
156 + },
157 + "tables": {
158 + "convention_notes": "Table notes define variables and benchmarks in detail: abnormal volume formulas, VWAP execution prices before and after commissions, DGTW-adjusted holding-period returns, and quintile/decile sort construction. Results are shown as panels of means by sort and as regressions. Standard-error treatment is not visible in the excerpts.",
159 + "units_stated_in_notes": true,
160 + "standard_errors_description": "unclear"
161 + },
162 + "figures": {
163 + "uncertainty_shown": "unclear"
164 + },
165 + "conclusion": {
166 + "bounded_claims": true,
167 + "policy_implications": false,
168 + "open_questions": true,
169 + "restates_contribution": true
170 + },
171 + "hedging": {
172 + "level": "moderate",
173 + "typical_devices": [
174 + "results suggest",
175 + "one potential interpretation is",
176 + "may be as important",
177 + "evidence suggests"
178 + ]
179 + },
180 + "claim_evidence": [
181 + {
182 + "claim": "Average ex-day returns are not significantly positive once actual execution prices and commissions are incorporated",
183 + "evidence_type": "table",
184 + "evidence_ref": "Table III"
185 + },
186 + {
187 + "claim": "Dividend capture buys earn significantly positive post-commission abnormal returns, higher than other buys",
188 + "evidence_type": "table",
189 + "evidence_ref": "Table VIII"
190 + },
191 + {
192 + "claim": "Dividend capture trades are under 6% of buys but contribute about 15% of abnormal returns",
193 + "evidence_type": "table",
194 + "evidence_ref": "Table VIII"
195 + },
196 + {
197 + "claim": "Institutional dividend capture buying intensity persists for at least two subsequent quarters",
198 + "evidence_type": "table",
199 + "evidence_ref": "Table VI; Figure 1"
200 + },
201 + {
202 + "claim": "Profitability is higher for ex-day events with high abnormal cum-day buy intensity",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table V"
205 + },
206 + {
207 + "claim": "High- vs low-skill institutions differ by about 40 bps in dividend capture profitability, with no relation on placebo days",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table IX"
210 + },
211 + {
212 + "claim": "Buy intensity is related to yield, idiosyncratic risk, transaction costs and lagged ex-day returns",
213 + "evidence_type": "table",
214 + "evidence_ref": "Table VII"
215 + }
216 + ],
217 + "structural_summary": "The paper opens by citing evidence that execution skill earns profits and asks whether skilled institutions benefit uniformly or target specific opportunities. It proposes dividend capture as the setting and states three research questions. Results are previewed with magnitudes, followed by a placebo-based defense and contributions to two literatures. A short theory/literature section precedes the data and descriptive statistics, which show that net ex-day returns vanish after costs. Two results sections follow: profitability and persistence at the client-manager level, then the link to trader skill with cross-sectional characteristics. The conclusion is structured around the three research questions and closes with a future-research direction.",
218 + "writing_profile_keywords": [
219 + "proprietary transaction data",
220 + "question-driven structure",
221 + "placebo test",
222 + "portfolio sorts",
223 + "transaction costs",
224 + "institutional skill",
225 + "economic magnitudes"
226 + ],
227 + "confidence": 0.82,
228 + "notes": "Robustness material appears within the results and in an Internet Appendix. Significance reporting conventions are not visible in the excerpts. A single equation appears in Section IV."
229 + },
230 + "deterministic_checks": {
231 + "roadmap_regex": true,
232 + "contribution_regex_in_intro": true,
233 + "magnitude_in_abstract": true,
234 + "abstract_n_sentences": 8,
235 + "intro_n_paragraphs": 5,
236 + "intro_n_words": 779,
237 + "causal_words_in_abstract": 0,
238 + "hedge_words_in_abstract": 1,
239 + "stars_mentioned": false,
240 + "clustered_se_mentioned": false,
241 + "internet_appendix_mentioned": false,
242 + "section_roles_rule": {
243 + "0": "introduction",
244 + "1": "literature",
245 + "2": "data",
246 + "3": "other",
247 + "4": "other",
248 + "5": "conclusion"
249 + }
250 + },
251 + "schema_valid": true,
252 + "code_version": "feb6c9e-dirty"
253 +}
\ No newline at end of file
added data/annotations/documents/jof_12473.json +268 −0
@@ -0,0 +1,268 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12473",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:03+00:00",
9 + "source_sha256": "b7e376f397246444f24d03a82ab6bb142573b397c44e9833cada7b2391ae09e7",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22870,
12 + "usage": {
13 + "input": 7843,
14 + "output": 1979,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.071746,
19 + "request_id": "req_011CggpVD8FAv98BkAEmNqQN",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Conditional associations between household portfolio value loadings and investor characteristics (age, wealth, leverage, human capital, labor income risk, industry macroeconomic exposure)",
27 + "strategy": "Pooled panel regressions of portfolio value loadings on household characteristics with year, industry and county fixed effects, using Swedish administrative data. Cohort plots and industry sorts complement the regressions. A twin subsample controls for latent family-level fixed effects.",
28 + "key_assumptions": [
29 + "Industry sensitivity to national income growth proxies systematic labor income risk",
30 + "Twin fixed effects absorb latent background and risk attitudes",
31 + "Observed characteristics are not proxies for other stock characteristics"
32 + ],
33 + "threats_addressed": [
34 + "popular widely held stocks driving results",
35 + "investor experience",
36 + "stock characteristics other than value loading (dividend yield, size, firm age, skewness, taxes, professional proximity)",
37 + "latent investor fixed effects (twins)",
38 + "passive versus active changes in loadings"
39 + ],
40 + "causal_language_strength": 2,
41 + "uses_causal_claim": false
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": true,
46 + "sentence_index": 0
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 0
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 1
59 + },
60 + "magnitude": {
61 + "present": false,
62 + "sentence_index": -1
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 3
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "literature_gap"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "preview_of_results"
82 + },
83 + {
84 + "paragraph_index": 2,
85 + "move": "robustness_mention"
86 + },
87 + {
88 + "paragraph_index": 2,
89 + "move": "identification_defense"
90 + },
91 + {
92 + "paragraph_index": 3,
93 + "move": "contribution_statement"
94 + },
95 + {
96 + "paragraph_index": 4,
97 + "move": "mechanism_or_theory_sketch"
98 + },
99 + {
100 + "paragraph_index": 5,
101 + "move": "related_literature_positioning"
102 + },
103 + {
104 + "paragraph_index": 6,
105 + "move": "roadmap"
106 + }
107 + ],
108 + "contribution_paragraph_index": 3,
109 + "contribution_types": [
110 + "new_evidence",
111 + "new_theory_or_model"
112 + ],
113 + "roadmap_present": true,
114 + "literature_positioning_location": "both"
115 + },
116 + "section_roles": [
117 + {
118 + "idx": 0,
119 + "role": "introduction"
120 + },
121 + {
122 + "idx": 1,
123 + "role": "data"
124 + },
125 + {
126 + "idx": 2,
127 + "role": "results"
128 + },
129 + {
130 + "idx": 3,
131 + "role": "results"
132 + },
133 + {
134 + "idx": 4,
135 + "role": "theory"
136 + },
137 + {
138 + "idx": 5,
139 + "role": "theory"
140 + },
141 + {
142 + "idx": 6,
143 + "role": "theory"
144 + },
145 + {
146 + "idx": 7,
147 + "role": "robustness"
148 + },
149 + {
150 + "idx": 8,
151 + "role": "conclusion"
152 + },
153 + {
154 + "idx": 9,
155 + "role": "appendix"
156 + }
157 + ],
158 + "results_presentation": {
159 + "economic_significance_discussed": true,
160 + "statistical_significance_language": "unclear",
161 + "magnitude_statements_in_intro": true
162 + },
163 + "robustness": {
164 + "location": "mixed",
165 + "types": [
166 + "subsample",
167 + "alternative_measure",
168 + "alternative_specification",
169 + "other"
170 + ]
171 + },
172 + "tables": {
173 + "convention_notes": "Table notes describe the sample, the portfolio level (risky, stock or fund), the fixed effects included and the reference date (end of 2003). A dedicated table translates life-cycle changes into economic significance. A variable definitions table appears in the appendix.",
174 + "units_stated_in_notes": false,
175 + "standard_errors_description": "unclear"
176 + },
177 + "figures": {
178 + "uncertainty_shown": "unclear"
179 + },
180 + "conclusion": {
181 + "bounded_claims": true,
182 + "policy_implications": false,
183 + "open_questions": true,
184 + "restates_contribution": true
185 + },
186 + "hedging": {
187 + "level": "moderate",
188 + "typical_devices": [
189 + "consistent with",
190 + "seem evident",
191 + "appear remarkably consistent",
192 + "to the best of our knowledge",
193 + "unlikely to be due to"
194 + ]
195 + },
196 + "claim_evidence": [
197 + {
198 + "claim": "Households shift from growth to value as they age (the value ladder)",
199 + "evidence_type": "figure",
200 + "evidence_ref": "Figure 2"
201 + },
202 + {
203 + "claim": "Value loading rises with age and balance sheet strength and falls with human capital and income risk",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table III"
206 + },
207 + {
208 + "claim": "A single national income factor explains on average 88% of time-series variation in industry per-capita income",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table VIII"
211 + },
212 + {
213 + "claim": "Households in highly cyclical industries hold lower value loadings, especially when young",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table IX; Figure 3"
216 + },
217 + {
218 + "claim": "Results are not driven by the most widely held popular stocks",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table X"
221 + },
222 + {
223 + "claim": "Households actively rebalance their value loadings",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table VII"
226 + }
227 + ],
228 + "structural_summary": "The paper frames the value premium puzzle as a contest between risk-based and behavioral explanations and argues that holdings data can help discriminate between them. It then describes Swedish administrative data and the construction of local factors. The core findings follow: life-cycle and balance-sheet determinants of portfolio value tilts, followed by labor income risk across industries. An equilibrium model relates the evidence to risk-based theories, with a shorter section on sentiment explanations. Robustness checks, mostly in the Internet Appendix, precede a conclusion that restates the findings and suggests modeling extensions.",
229 + "writing_profile_keywords": [
230 + "administrative panel",
231 + "household portfolios",
232 + "value premium",
233 + "life cycle",
234 + "hedging demand",
235 + "consistency with theory",
236 + "Internet Appendix heavy"
237 + ],
238 + "confidence": 0.8,
239 + "notes": "Section idx 5 is a parsing artifact (equation fragment) within the model section. The equilibrium model is derived in the Internet Appendix. Standard error type is not visible in the excerpts."
240 + },
241 + "deterministic_checks": {
242 + "roadmap_regex": true,
243 + "contribution_regex_in_intro": true,
244 + "magnitude_in_abstract": false,
245 + "abstract_n_sentences": 4,
246 + "intro_n_paragraphs": 7,
247 + "intro_n_words": 1009,
248 + "causal_words_in_abstract": 0,
249 + "hedge_words_in_abstract": 1,
250 + "stars_mentioned": false,
251 + "clustered_se_mentioned": false,
252 + "internet_appendix_mentioned": true,
253 + "section_roles_rule": {
254 + "0": "introduction",
255 + "1": "data",
256 + "2": "other",
257 + "3": "other",
258 + "4": "theory",
259 + "5": "other",
260 + "6": "theory",
261 + "7": "empirical_design",
262 + "8": "conclusion",
263 + "9": "appendix"
264 + }
265 + },
266 + "schema_valid": true,
267 + "code_version": "feb6c9e-dirty"
268 +}
\ No newline at end of file
added data/annotations/documents/jof_12482.json +291 −0
@@ -0,0 +1,291 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12482",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:22+00:00",
9 + "source_sha256": "7128b3f254dfc38112d4db048add41e1574f2cc611580fba0579249dd703e83d",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 25837,
12 + "usage": {
13 + "input": 8442,
14 + "output": 2319,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.080942,
19 + "request_id": "req_011CggpWTUj8goewU5fubnP1",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "corporate_finance (debt overhang theory applied to households); real estate",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Difference in quarterly housing investment (home improvement/maintenance spending, unscheduled principal payments) associated with negative equity or estimated mortgage default risk",
27 + "strategy": "Cross-sectional regressions of spending on a negative-equity indicator or estimated default risk, with controls and state-year fixed effects. Supplemented by a propensity-score matched sample, subsamples of financially unconstrained households, a within-household comparison across durable categories, household fixed effects (changes over time and across properties of multiproperty owners), and recourse vs. nonrecourse state contrasts.",
28 + "key_assumptions": [
29 + "Conditional on observables, negative equity is not correlated with unobserved preferences for home investment",
30 + "Negative equity and estimated default risk proxy for perceived default risk",
31 + "High-income or high-asset households are not financially constrained",
32 + "Durables retained in default share the same household-level confounders as home improvements"
33 + ],
34 + "threats_addressed": [
35 + "financial/liquidity constraints",
36 + "omitted household characteristics (household fixed effects)",
37 + "general spending decline or low expected income",
38 + "intertemporal substitution of past improvements",
39 + "selection on observables (matching)"
40 + ],
41 + "causal_language_strength": 3,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": false,
55 + "sentence_index": -1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 4
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "related_literature_positioning"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "contribution_statement"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "mechanism_or_theory_sketch"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "motivation"
91 + },
92 + {
93 + "paragraph_index": 1,
94 + "move": "preview_of_approach"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "identification_defense"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "preview_of_results"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "preview_of_results"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "identification_defense"
111 + },
112 + {
113 + "paragraph_index": 5,
114 + "move": "robustness_mention"
115 + },
116 + {
117 + "paragraph_index": 5,
118 + "move": "policy_or_broader_implication"
119 + }
120 + ],
121 + "contribution_paragraph_index": 0,
122 + "contribution_types": [
123 + "new_evidence"
124 + ],
125 + "roadmap_present": false,
126 + "literature_positioning_location": "both"
127 + },
128 + "section_roles": [
129 + {
130 + "idx": 0,
131 + "role": "introduction"
132 + },
133 + {
134 + "idx": 1,
135 + "role": "theory"
136 + },
137 + {
138 + "idx": 2,
139 + "role": "data"
140 + },
141 + {
142 + "idx": 3,
143 + "role": "empirical_design"
144 + },
145 + {
146 + "idx": 4,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 5,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 6,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 7,
159 + "role": "results"
160 + },
161 + {
162 + "idx": 8,
163 + "role": "robustness"
164 + },
165 + {
166 + "idx": 9,
167 + "role": "conclusion"
168 + }
169 + ],
170 + "results_presentation": {
171 + "economic_significance_discussed": true,
172 + "statistical_significance_language": "unclear",
173 + "magnitude_statements_in_intro": false
174 + },
175 + "robustness": {
176 + "location": "mixed",
177 + "types": [
178 + "subsample",
179 + "alternative_measure",
180 + "alternative_specification",
181 + "falsification",
182 + "other"
183 + ]
184 + },
185 + "tables": {
186 + "convention_notes": "Table notes describe the sample and period, the dependent and key independent variables, and the fixed effects used. Mean spending is reported for the comparison group so coefficients can be read as percentages. Matching procedures are described in the notes. Standard-error treatment is not visible in the excerpts.",
187 + "units_stated_in_notes": true,
188 + "standard_errors_description": "unclear"
189 + },
190 + "figures": {
191 + "uncertainty_shown": "unclear"
192 + },
193 + "conclusion": {
194 + "bounded_claims": false,
195 + "policy_implications": true,
196 + "open_questions": true,
197 + "restates_contribution": true
198 + },
199 + "hedging": {
200 + "level": "moderate",
201 + "typical_devices": [
202 + "'consistent with' language",
203 + "acknowledging that results do not rule out alternatives",
204 + "noting low power of fixed-effects tests",
205 + "'mixed' evidence admission",
206 + "'may create' deadweight loss"
207 + ]
208 + },
209 + "claim_evidence": [
210 + {
211 + "claim": "Negative-equity homeowners cut improvement/maintenance spending and unscheduled principal payments by roughly 30%",
212 + "evidence_type": "table",
213 + "evidence_ref": "Tables III, IV"
214 + },
215 + {
216 + "claim": "Housing investment declines with a continuous estimate of default risk",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table V; Figure 2"
219 + },
220 + {
221 + "claim": "Spending falls with loan-to-value in raw and residualized data",
222 + "evidence_type": "figure",
223 + "evidence_ref": "Figure 3"
224 + },
225 + {
226 + "claim": "Financially unconstrained households also cut housing investment when underwater",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table VII"
229 + },
230 + {
231 + "claim": "Spending on durables retained in default is not reduced by negative-equity households",
232 + "evidence_type": "table",
233 + "evidence_ref": "Table VIII"
234 + },
235 + {
236 + "claim": "Wealthy underwater homeowners cut investment more in nonrecourse states",
237 + "evidence_type": "table",
238 + "evidence_ref": "Table X"
239 + },
240 + {
241 + "claim": "Household fixed-effects estimates broadly confirm improvement results; principal payment evidence mixed",
242 + "evidence_type": "table",
243 + "evidence_ref": "Table IX"
244 + },
245 + {
246 + "claim": "Results robust in a propensity-matched sample",
247 + "evidence_type": "table",
248 + "evidence_ref": "Table VI"
249 + }
250 + ],
251 + "structural_summary": "The paper opens by framing household behavior through the classic corporate and public debt overhang theory. It then motivates the question with post-2006 negative-equity statistics and previews a baseline finding from consumer expenditure survey data. It acknowledges a competing explanation, financial constraints, and presents a sequence of discriminating tests: unconstrained subsamples, recourse laws, cross-durable comparisons, and household fixed effects. Brief theory and data sections are followed by a research-design section, several results sections each addressing one alternative, and an additional-tests section referring to the internet appendix. The conclusion restates the findings, gives an aggregate magnitude, and suggests future work on social costs.",
252 + "writing_profile_keywords": [
253 + "debt overhang",
254 + "household finance",
255 + "alternative-hypothesis elimination",
256 + "sequential falsification tests",
257 + "survey data",
258 + "negative equity",
259 + "economic magnitudes"
260 + ],
261 + "confidence": 0.8,
262 + "notes": "How statistical significance is reported is not visible in the excerpts. The intro excerpt is truncated, so the main magnitude preview may be in the missing text; the 30% figure appears in the conclusion. Robustness appears both in the main text and in the internet appendix."
263 + },
264 + "deterministic_checks": {
265 + "roadmap_regex": false,
266 + "contribution_regex_in_intro": false,
267 + "magnitude_in_abstract": false,
268 + "abstract_n_sentences": 5,
269 + "intro_n_paragraphs": 6,
270 + "intro_n_words": 938,
271 + "causal_words_in_abstract": 0,
272 + "hedge_words_in_abstract": 2,
273 + "stars_mentioned": false,
274 + "clustered_se_mentioned": false,
275 + "internet_appendix_mentioned": true,
276 + "section_roles_rule": {
277 + "0": "introduction",
278 + "1": "theory",
279 + "2": "data",
280 + "3": "empirical_design",
281 + "4": "other",
282 + "5": "other",
283 + "6": "other",
284 + "7": "other",
285 + "8": "robustness",
286 + "9": "conclusion"
287 + }
288 + },
289 + "schema_valid": true,
290 + "code_version": "feb6c9e-dirty"
291 +}
\ No newline at end of file
added data/annotations/documents/jof_12483.json +265 −0
@@ -0,0 +1,265 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12483",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:18+00:00",
9 + "source_sha256": "a5a3d2a5b3778bcd2d8e0ff3e9129ae2ae928e57091d3faf5f10bd9304799b11",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19429,
12 + "usage": {
13 + "input": 6083,
14 + "output": 2119,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.067506,
19 + "request_id": "req_011CggpWGRxWPcC9iNnzmoAU",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "entrepreneurial_finance_private_equity",
23 + "subfield_secondary": "corporate_finance (political connections, labor)",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Difference in post-buyout establishment-level employment growth between targets of politically connected and nonconnected private equity firms",
27 + "strategy": "Pre/post establishment panel regressions comparing connected vs nonconnected PE targets, supplemented by propensity score matching on buyout year/state/type and establishment traits, and placebo/DDD comparisons with failed buyouts withdrawn for presumably exogenous reasons",
28 + "key_assumptions": [
29 + "Matching on observables removes relevant differences between connected and nonconnected PE targets",
30 + "Failed buyouts (deals cancelled by the target or a third party, or presumably positive-NPV deals) form a valid counterfactual for successful buyouts",
31 + "Political connection classification from keyword searches is accurate"
32 + ],
33 + "threats_addressed": [
34 + "selection of targets by connected PE firms",
35 + "nonrandom assignment to treatment and control groups (local market conditions, buyout type, PE firm size)",
36 + "endogeneity of the deal withdrawal decision",
37 + "missed job creation at newly opened establishments"
38 + ],
39 + "causal_language_strength": 3,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": false,
53 + "sentence_index": -1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": false,
65 + "sentence_index": -1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "related_literature_positioning"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "preview_of_approach"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "identification_defense"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "identification_defense"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "mechanism_or_theory_sketch"
101 + },
102 + {
103 + "paragraph_index": 3,
104 + "move": "preview_of_results"
105 + },
106 + {
107 + "paragraph_index": 3,
108 + "move": "contribution_statement"
109 + }
110 + ],
111 + "contribution_paragraph_index": 3,
112 + "contribution_types": [
113 + "new_evidence"
114 + ],
115 + "roadmap_present": false,
116 + "literature_positioning_location": "both"
117 + },
118 + "section_roles": [
119 + {
120 + "idx": 0,
121 + "role": "introduction"
122 + },
123 + {
124 + "idx": 1,
125 + "role": "literature"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "data"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "results"
138 + },
139 + {
140 + "idx": 5,
141 + "role": "robustness"
142 + },
143 + {
144 + "idx": 6,
145 + "role": "conclusion"
146 + },
147 + {
148 + "idx": 7,
149 + "role": "appendix"
150 + }
151 + ],
152 + "results_presentation": {
153 + "economic_significance_discussed": true,
154 + "statistical_significance_language": "unclear",
155 + "magnitude_statements_in_intro": false
156 + },
157 + "robustness": {
158 + "location": "mixed",
159 + "types": [
160 + "placebo",
161 + "alternative_specification",
162 + "alternative_measure"
163 + ]
164 + },
165 + "tables": {
166 + "convention_notes": "Table notes define the dependent variable as an explicit log-difference formula and define indicator variables. They specify the estimator (panel OLS, conditional logit, negative binomial with fixed effects) and the matching procedure. Univariate pre/post means come first, followed by the regression tables.",
167 + "units_stated_in_notes": true,
168 + "standard_errors_description": "unclear"
169 + },
170 + "figures": {
171 + "uncertainty_shown": "no_figures"
172 + },
173 + "conclusion": {
174 + "bounded_claims": true,
175 + "policy_implications": false,
176 + "open_questions": false,
177 + "restates_contribution": true
178 + },
179 + "hedging": {
180 + "level": "moderate",
181 + "typical_devices": [
182 + "consistent with",
183 + "presumably exogenous",
184 + "mitigate rather than eliminate concerns",
185 + "only suggestive of a causal relation",
186 + "appears to"
187 + ]
188 + },
189 + "claim_evidence": [
190 + {
191 + "claim": "Establishments of politically connected PE targets have higher post-buyout employment growth than those of nonconnected PE targets",
192 + "evidence_type": "table",
193 + "evidence_ref": "Tables I, II"
194 + },
195 + {
196 + "claim": "The employment growth difference persists against propensity-score-matched peers",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table III"
199 + },
200 + {
201 + "claim": "Failed buyouts show no systematic job creation; DDD estimates favor successful connected buyouts",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table IV"
204 + },
205 + {
206 + "claim": "Employment increases are larger in election years and in more corrupt states",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table V"
209 + },
210 + {
211 + "claim": "State employment changes are positively associated with incumbent reelection",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table VI"
214 + },
215 + {
216 + "claim": "Targets of connected PE firms are more likely to receive government contracts and grants",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table VII"
219 + },
220 + {
221 + "claim": "Results are not driven by differences in establishment openings or closures, or by using employment levels instead of growth",
222 + "evidence_type": "table",
223 + "evidence_ref": "Tables VIII, IX"
224 + }
225 + ],
226 + "structural_summary": "The introduction opens with a public controversy anecdote, then links it to the recent literature on PE and employment. It previews the data, followed by a sequence of identification concerns, each paired with a design (matching, failed-buyout placebo/DDD). It next poses a 'why' question and lists alternative explanations before testing the exchange-of-favors story with three pieces of evidence. A separate literature section follows. The results are split into an employment dynamics section and a mechanism section, then come a short robustness section and a conclusion that explicitly limits causal interpretation.",
227 + "writing_profile_keywords": [
228 + "political connections",
229 + "private equity buyouts",
230 + "establishment-level employment",
231 + "selection concerns",
232 + "propensity score matching",
233 + "placebo failed deals",
234 + "exchange of favors",
235 + "cautious causal language"
236 + ],
237 + "confidence": 0.8,
238 + "notes": "No figures. The way standard errors are computed is not visible in the excerpts. The abstract mentions robustness to selection tests but gives no method detail. Primary method is borderline between DiD and reduced-form panel regression with matching; DiD was chosen because of the pre/post comparison across groups plus the DDD."
239 + },
240 + "deterministic_checks": {
241 + "roadmap_regex": false,
242 + "contribution_regex_in_intro": true,
243 + "magnitude_in_abstract": false,
244 + "abstract_n_sentences": 6,
245 + "intro_n_paragraphs": 4,
246 + "intro_n_words": 860,
247 + "causal_words_in_abstract": 0,
248 + "hedge_words_in_abstract": 1,
249 + "stars_mentioned": false,
250 + "clustered_se_mentioned": false,
251 + "internet_appendix_mentioned": true,
252 + "section_roles_rule": {
253 + "0": "introduction",
254 + "1": "literature",
255 + "2": "data",
256 + "3": "other",
257 + "4": "other",
258 + "5": "robustness",
259 + "6": "conclusion",
260 + "7": "appendix"
261 + }
262 + },
263 + "schema_valid": true,
264 + "code_version": "feb6c9e-dirty"
265 +}
\ No newline at end of file
added data/annotations/documents/jof_12484.json +268 −0
@@ -0,0 +1,268 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12484",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:35+00:00",
9 + "source_sha256": "a96f8b9bc8e4d7736047d9b68b260101eb9073f2d214be77ef91bfc5868e30ed",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22377,
12 + "usage": {
13 + "input": 7618,
14 + "output": 2022,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.071706,
19 + "request_id": "req_011CggpXYsCnTrsijZKtwQJX",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Age profiles of stock market participation, conditional risky share and financial wealth, plus structural preference and cost parameters (risk aversion, participation cost, crash probability)",
27 + "strategy": "Heckman selection model to handle endogenous participation, cohort effects separated with identification restrictions (Deaton-Paxson zero-sum time effects, GDP-trend proxies), and moment-matching structural estimation of an extended life-cycle model",
28 + "key_assumptions": [
29 + "Exclusion and selection restrictions in the Heckman model",
30 + "Time/age/cohort identification restrictions such as zero-sum time effects",
31 + "Administrative tax records are free of measurement error with little tax evasion",
32 + "Model specification: CRRA preferences, labor income process, crash probability"
33 + ],
34 + "threats_addressed": [
35 + "endogeneity of participation",
36 + "separation of age, time and cohort effects",
37 + "measurement error and underreporting",
38 + "wealth-mortality correlation",
39 + "attrition"
40 + ],
41 + "causal_language_strength": 2,
42 + "uses_causal_claim": false
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 0
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 0
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "contribution_statement"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "preview_of_approach"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "related_literature_positioning"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "tension_or_puzzle"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "preview_of_approach"
91 + },
92 + {
93 + "paragraph_index": 2,
94 + "move": "identification_defense"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "literature_gap"
103 + }
104 + ],
105 + "contribution_paragraph_index": 0,
106 + "contribution_types": [
107 + "new_evidence",
108 + "new_data_or_measure",
109 + "new_theory_or_model",
110 + "resolution_of_puzzle"
111 + ],
112 + "roadmap_present": false,
113 + "literature_positioning_location": "inside_introduction"
114 + },
115 + "section_roles": [
116 + {
117 + "idx": 0,
118 + "role": "introduction"
119 + },
120 + {
121 + "idx": 1,
122 + "role": "data"
123 + },
124 + {
125 + "idx": 2,
126 + "role": "empirical_design"
127 + },
128 + {
129 + "idx": 3,
130 + "role": "theory"
131 + },
132 + {
133 + "idx": 4,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 5,
138 + "role": "conclusion"
139 + },
140 + {
141 + "idx": 6,
142 + "role": "appendix"
143 + },
144 + {
145 + "idx": 7,
146 + "role": "appendix"
147 + },
148 + {
149 + "idx": 8,
150 + "role": "appendix"
151 + },
152 + {
153 + "idx": 9,
154 + "role": "appendix"
155 + },
156 + {
157 + "idx": 10,
158 + "role": "appendix"
159 + },
160 + {
161 + "idx": 11,
162 + "role": "appendix"
163 + }
164 + ],
165 + "results_presentation": {
166 + "economic_significance_discussed": true,
167 + "statistical_significance_language": "unclear",
168 + "magnitude_statements_in_intro": true
169 + },
170 + "robustness": {
171 + "location": "internet_appendix",
172 + "types": [
173 + "alternative_specification"
174 + ]
175 + },
176 + "tables": {
177 + "convention_notes": "Notes define variables and samples and state that monetary values are in 1995 USD. The parameterization table lists each value with its source. Some tables report t-values in parentheses. The structural estimation table describes the targeted moments.",
178 + "units_stated_in_notes": true,
179 + "standard_errors_description": "unclear"
180 + },
181 + "figures": {
182 + "uncertainty_shown": "unclear"
183 + },
184 + "conclusion": {
185 + "bounded_claims": true,
186 + "policy_implications": false,
187 + "open_questions": false,
188 + "restates_contribution": true
189 + },
190 + "hedging": {
191 + "level": "moderate",
192 + "typical_devices": [
193 + "we argue that this is likely",
194 + "seem to depart",
195 + "suggests",
196 + "consistent with model predictions"
197 + ]
198 + },
199 + "claim_evidence": [
200 + {
201 + "claim": "Households reduce the conditional risky share by just under 1 pp per year from about age 45 until retirement.",
202 + "evidence_type": "figure",
203 + "evidence_ref": "Figure 5; Table III"
204 + },
205 + {
206 + "claim": "Participation rises to about 60% by age 45 and households exit the stock market after retirement.",
207 + "evidence_type": "figure",
208 + "evidence_ref": "Figures 1, 2, 5, 6"
209 + },
210 + {
211 + "claim": "Tax-registry data minimize measurement error and tax evasion is unlikely.",
212 + "evidence_type": "none_stated",
213 + "evidence_ref": "Section I"
214 + },
215 + {
216 + "claim": "Best-fitting parameters combine high risk aversion, a small participation cost and a crash probability in line with Norwegian history.",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table V; Figures 11-12; Appendix C"
219 + },
220 + {
221 + "claim": "Existing calibrated models predict rebalancing toward safer portfolios with age but not stock market exit.",
222 + "evidence_type": "prior_literature",
223 + "evidence_ref": "Merton 1971; Cocco, Gomes, and Maenhout 2005"
224 + }
225 + ],
226 + "structural_summary": "The paper opens by claiming new stylized facts and a model that fits them. It then sets out the life-cycle theory prediction and its mismatch with micro data, and presents the Norwegian registry data as the remedy. Results are previewed with magnitudes. Descriptive cohort evidence comes next, followed by an econometric treatment of selection and cohort effects. An extended life-cycle model with a participation cost and crash risk follows, then policy functions, simulations and structural estimation. The conclusion argues the data are consistent with model predictions.",
227 + "writing_profile_keywords": [
228 + "stylized facts",
229 + "administrative data",
230 + "life-cycle model",
231 + "structural estimation",
232 + "Heckman selection",
233 + "cohort effects",
234 + "double adjustment"
235 + ],
236 + "confidence": 0.8,
237 + "notes": "The introduction and conclusion are truncated in the excerpts. No robustness section is visible; the internet appendix is referenced for estimation details, so the robustness fields are tentative. Significance conventions are not clearly stated beyond t-values in Table A.I."
238 + },
239 + "deterministic_checks": {
240 + "roadmap_regex": false,
241 + "contribution_regex_in_intro": true,
242 + "magnitude_in_abstract": false,
243 + "abstract_n_sentences": 2,
244 + "intro_n_paragraphs": 4,
245 + "intro_n_words": 752,
246 + "causal_words_in_abstract": 0,
247 + "hedge_words_in_abstract": 0,
248 + "stars_mentioned": false,
249 + "clustered_se_mentioned": false,
250 + "internet_appendix_mentioned": true,
251 + "section_roles_rule": {
252 + "0": "introduction",
253 + "1": "data",
254 + "2": "empirical_design",
255 + "3": "theory",
256 + "4": "results",
257 + "5": "conclusion",
258 + "6": "appendix",
259 + "7": "appendix",
260 + "8": "appendix",
261 + "9": "appendix",
262 + "10": "appendix",
263 + "11": "appendix"
264 + }
265 + },
266 + "schema_valid": true,
267 + "code_version": "feb6c9e-dirty"
268 +}
\ No newline at end of file
added data/annotations/documents/jof_12485.json +276 −0
@@ -0,0 +1,276 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12485",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:29+00:00",
9 + "source_sha256": "1f1aa0ac42f7f938bc0ac08f78c5eae5b9e17a4dc685c9c0d907e2cf79c513bf",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23721,
12 + "usage": {
13 + "input": 7584,
14 + "output": 2310,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07733,
19 + "request_id": "req_011CggpbUMxCwXjHoWuYjC65",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "information intermediaries / sell-side analysts",
24 + "primary_method": "natural_experiment_event_study",
25 + "identification": {
26 + "estimand": "Effect of industry-level drops in the number of analysts on the forecast accuracy, bias, effort and informativeness of the remaining analysts covering that industry, incremental to firm-level coverage effects",
27 + "strategy": "Uses analyst drops caused by brokerage house mergers and closures as plausibly exogenous shocks to industry-level analyst coverage, estimated with industry panel regressions; complemented by firm-level OLS and difference-in-differences regressions extending Hong and Kacperczyk (2010), and by analyses of all coverage changes",
28 + "key_assumptions": [
29 + "Analyst drops from brokerage mergers and closures are unrelated to the industry's underlying information environment",
30 + "Industry-level competition is the channel through which coverage changes affect other analysts"
31 + ],
32 + "threats_addressed": [
33 + "endogeneity of analyst coverage changes",
34 + "external validity limits of drop-only tests (addressed by examining all changes, including increases)",
35 + "confounding of industry-level effects with firm-level coverage effects"
36 + ],
37 + "causal_language_strength": 3,
38 + "uses_causal_claim": true
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": true,
43 + "sentence_index": 0
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "magnitude": {
58 + "present": false,
59 + "sentence_index": -1
60 + },
61 + "contribution": {
62 + "present": true,
63 + "sentence_index": 3
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "related_literature_positioning"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "literature_gap"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 2,
86 + "move": "preview_of_approach"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "mechanism_or_theory_sketch"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 5,
102 + "move": "mechanism_or_theory_sketch"
103 + },
104 + {
105 + "paragraph_index": 5,
106 + "move": "preview_of_results"
107 + },
108 + {
109 + "paragraph_index": 6,
110 + "move": "preview_of_approach"
111 + },
112 + {
113 + "paragraph_index": 6,
114 + "move": "preview_of_results"
115 + },
116 + {
117 + "paragraph_index": 6,
118 + "move": "contribution_statement"
119 + }
120 + ],
121 + "contribution_paragraph_index": 6,
122 + "contribution_types": [
123 + "new_evidence"
124 + ],
125 + "roadmap_present": false,
126 + "literature_positioning_location": "inside_introduction"
127 + },
128 + "section_roles": [
129 + {
130 + "idx": 0,
131 + "role": "introduction"
132 + },
133 + {
134 + "idx": 1,
135 + "role": "data"
136 + },
137 + {
138 + "idx": 2,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 3,
143 + "role": "robustness"
144 + },
145 + {
146 + "idx": 4,
147 + "role": "conclusion"
148 + },
149 + {
150 + "idx": 5,
151 + "role": "appendix"
152 + },
153 + {
154 + "idx": 6,
155 + "role": "appendix"
156 + },
157 + {
158 + "idx": 7,
159 + "role": "appendix"
160 + }
161 + ],
162 + "results_presentation": {
163 + "economic_significance_discussed": true,
164 + "statistical_significance_language": "unclear",
165 + "magnitude_statements_in_intro": true
166 + },
167 + "robustness": {
168 + "location": "mixed",
169 + "types": [
170 + "alternative_specification",
171 + "alternative_measure",
172 + "subsample",
173 + "other"
174 + ]
175 + },
176 + "tables": {
177 + "convention_notes": "Panel-regression tables in which notes define the dependent variables (e.g., absolute and signed forecast error as proxies for accuracy and bias), the treatment variable (analyst drops from mergers and closures) and interaction indicators. Notes cross-reference variable definitions in Appendix A and earlier tables. Several tables are split into panels.",
178 + "units_stated_in_notes": false,
179 + "standard_errors_description": "unclear"
180 + },
181 + "figures": {
182 + "uncertainty_shown": "no"
183 + },
184 + "conclusion": {
185 + "bounded_claims": false,
186 + "policy_implications": true,
187 + "open_questions": false,
188 + "restates_contribution": true
189 + },
190 + "hedging": {
191 + "level": "moderate",
192 + "typical_devices": [
193 + "our findings suggest",
194 + "plausibly exogenous",
195 + "consistent with",
196 + "we argue",
197 + "can potentially",
198 + "appears to"
199 + ]
200 + },
201 + "claim_evidence": [
202 + {
203 + "claim": "Number of analysts in an industry rises with industry IPO activity and trading volume changes; large brokers are most sensitive to IPOs",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table I"
206 + },
207 + {
208 + "claim": "Analyst headcount grew moderately over 1990-2010, with small brokers increasing their share",
209 + "evidence_type": "figure",
210 + "evidence_ref": "Figure 1"
211 + },
212 + {
213 + "claim": "Industry analyst drops from broker mergers/closures reduce remaining analysts' forecast accuracy and increase optimism bias",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table II"
216 + },
217 + {
218 + "claim": "Effects are stronger in industries with weaker information environments and when All-Star analysts are dropped",
219 + "evidence_type": "table",
220 + "evidence_ref": "Tables III-V"
221 + },
222 + {
223 + "claim": "After drops, remaining analysts exert less effort and forecast dispersion rises",
224 + "evidence_type": "table",
225 + "evidence_ref": "Tables VI-VII"
226 + },
227 + {
228 + "claim": "Analyst report informativeness falls about 0.28% per analyst lost, while earnings announcement informativeness rises about 0.42%",
229 + "evidence_type": "table",
230 + "evidence_ref": "Table VIII"
231 + },
232 + {
233 + "claim": "Industry-level coverage effects are incremental to firm-level effects",
234 + "evidence_type": "table",
235 + "evidence_ref": "Tables IX-X"
236 + }
237 + ],
238 + "structural_summary": "The introduction motivates the role of analysts and identifies a gap: no aggregate, industry-level view of the analyst industry exists. It previews descriptive evidence on how analyst scope varies over time and its economic correlates, then sketches a competition and spillover mechanism. It previews results from merger- and closure-induced analyst drops on forecast quality, effort, convergence and price informativeness, and closes with firm- versus industry-level comparisons and contribution claims. In the body, Section I gives descriptive time-series evidence. Section II presents the quasi-experimental results with mechanism tests. Section III extends to all coverage changes and firm-level DID. The conclusion restates the findings and draws implications for the literature and regulation.",
239 + "writing_profile_keywords": [
240 + "sell-side analysts",
241 + "industry competition",
242 + "spillover effects",
243 + "brokerage mergers",
244 + "quasi-natural experiment",
245 + "forecast accuracy",
246 + "information dissemination"
247 + ],
248 + "confidence": 0.8,
249 + "notes": "Section I is labeled 'other' in the source but functions as data and descriptive analysis. The source says Section II's first paragraph refers to changes 'documented in Section II'; this appears to be a numbering mismatch. The excerpts do not state how standard errors are computed. The internet appendix is mentioned."
250 + },
251 + "deterministic_checks": {
252 + "roadmap_regex": false,
253 + "contribution_regex_in_intro": true,
254 + "magnitude_in_abstract": false,
255 + "abstract_n_sentences": 6,
256 + "intro_n_paragraphs": 7,
257 + "intro_n_words": 1343,
258 + "causal_words_in_abstract": 0,
259 + "hedge_words_in_abstract": 1,
260 + "stars_mentioned": false,
261 + "clustered_se_mentioned": false,
262 + "internet_appendix_mentioned": true,
263 + "section_roles_rule": {
264 + "0": "introduction",
265 + "1": "other",
266 + "2": "other",
267 + "3": "robustness",
268 + "4": "conclusion",
269 + "5": "appendix",
270 + "6": "appendix",
271 + "7": "appendix"
272 + }
273 + },
274 + "schema_valid": true,
275 + "code_version": "feb6c9e-dirty"
276 +}
\ No newline at end of file
added data/annotations/documents/jof_12486.json +348 −0
@@ -0,0 +1,348 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12486",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:34+00:00",
9 + "source_sha256": "8f27b1d0a56f4b177536000961d2b4cb6db9acee80efff444c4afd7a0f4498dd",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 40542,
12 + "usage": {
13 + "input": 13739,
14 + "output": 2515,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.10605,
19 + "request_id": "req_011Cggpbc73PJo868918kDex",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "financial_econometrics",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Spot risk-neutral volatility, jump intensity and jump size distribution (left tail jump variation) extracted daily from short-dated options; predictive association of left tail measure with future excess returns",
27 + "strategy": "Seminonparametric 'structural calibration' using a new short-maturity asymptotic option pricing approximation with no-arbitrage constraints; daily estimation with confidence regions; predictive regressions; illustrative event windows around FOMC announcements",
28 + "key_assumptions": [
29 + "Arbitrage-free markets implying existence of a risk-neutral measure",
30 + "Volatility and jump intensity change little in expectation over short horizons",
31 + "Parametric restrictions only on the jump size distribution are adequate"
32 + ],
33 + "threats_addressed": [
34 + "Noise or idiosyncratic pricing in short-dated options",
35 + "Liquidity/spread shifts around announcements",
36 + "Misspecification of volatility and jump intensity dynamics",
37 + "Noisy deep OTM quotes (data filters)"
38 + ],
39 + "causal_language_strength": 2,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": true,
49 + "sentence_index": 1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 2
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 2
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 3
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_approach"
81 + },
82 + {
83 + "paragraph_index": 2,
84 + "move": "contribution_statement"
85 + },
86 + {
87 + "paragraph_index": 3,
88 + "move": "related_literature_positioning"
89 + },
90 + {
91 + "paragraph_index": 4,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 5,
96 + "move": "mechanism_or_theory_sketch"
97 + },
98 + {
99 + "paragraph_index": 6,
100 + "move": "identification_defense"
101 + },
102 + {
103 + "paragraph_index": 7,
104 + "move": "other"
105 + },
106 + {
107 + "paragraph_index": 8,
108 + "move": "preview_of_results"
109 + },
110 + {
111 + "paragraph_index": 9,
112 + "move": "identification_defense"
113 + },
114 + {
115 + "paragraph_index": 10,
116 + "move": "preview_of_results"
117 + },
118 + {
119 + "paragraph_index": 10,
120 + "move": "related_literature_positioning"
121 + },
122 + {
123 + "paragraph_index": 11,
124 + "move": "roadmap"
125 + }
126 + ],
127 + "contribution_paragraph_index": 2,
128 + "contribution_types": [
129 + "new_method",
130 + "new_evidence",
131 + "new_data_or_measure"
132 + ],
133 + "roadmap_present": true,
134 + "literature_positioning_location": "inside_introduction"
135 + },
136 + "section_roles": [
137 + {
138 + "idx": 0,
139 + "role": "introduction"
140 + },
141 + {
142 + "idx": 1,
143 + "role": "data"
144 + },
145 + {
146 + "idx": 2,
147 + "role": "theory"
148 + },
149 + {
150 + "idx": 3,
151 + "role": "theory"
152 + },
153 + {
154 + "idx": 4,
155 + "role": "empirical_design"
156 + },
157 + {
158 + "idx": 5,
159 + "role": "results"
160 + },
161 + {
162 + "idx": 6,
163 + "role": "results"
164 + },
165 + {
166 + "idx": 7,
167 + "role": "results"
168 + },
169 + {
170 + "idx": 8,
171 + "role": "conclusion"
172 + },
173 + {
174 + "idx": 9,
175 + "role": "appendix"
176 + },
177 + {
178 + "idx": 10,
179 + "role": "appendix"
180 + },
181 + {
182 + "idx": 11,
183 + "role": "appendix"
184 + },
185 + {
186 + "idx": 12,
187 + "role": "appendix"
188 + },
189 + {
190 + "idx": 13,
191 + "role": "appendix"
192 + },
193 + {
194 + "idx": 14,
195 + "role": "appendix"
196 + },
197 + {
198 + "idx": 15,
199 + "role": "appendix"
200 + },
201 + {
202 + "idx": 16,
203 + "role": "appendix"
204 + },
205 + {
206 + "idx": 17,
207 + "role": "appendix"
208 + },
209 + {
210 + "idx": 18,
211 + "role": "appendix"
212 + },
213 + {
214 + "idx": 19,
215 + "role": "appendix"
216 + },
217 + {
218 + "idx": 20,
219 + "role": "appendix"
220 + }
221 + ],
222 + "results_presentation": {
223 + "economic_significance_discussed": true,
224 + "statistical_significance_language": "mixed",
225 + "magnitude_statements_in_intro": false
226 + },
227 + "robustness": {
228 + "location": "appendix",
229 + "types": [
230 + "alternative_specification",
231 + "alternative_measure",
232 + "other"
233 + ]
234 + },
235 + "tables": {
236 + "convention_notes": "Few tables; descriptive table of contract shares by tenor and moneyness with sample period stated; predictive regression table reporting coefficients with Newey-West t-statistics in parentheses, lags matched to forecast horizon. Most evidence is carried by figures.",
237 + "units_stated_in_notes": true,
238 + "standard_errors_description": "newey_west"
239 + },
240 + "figures": {
241 + "uncertainty_shown": "mixed"
242 + },
243 + "conclusion": {
244 + "bounded_claims": true,
245 + "policy_implications": false,
246 + "open_questions": true,
247 + "restates_contribution": true
248 + },
249 + "hedging": {
250 + "level": "moderate",
251 + "typical_devices": [
252 + "'suggestive of'",
253 + "'may have'",
254 + "'to the extent that'",
255 + "conditional framing ('if this is indeed a robust finding')",
256 + "explicit acknowledgement of potential concerns"
257 + ]
258 + },
259 + "claim_evidence": [
260 + {
261 + "claim": "Trading in short-dated S&P 500 options has grown sharply and is skewed toward OTM contracts",
262 + "evidence_type": "descriptive_statistic",
263 + "evidence_ref": "Table I; Figures 3-5"
264 + },
265 + {
266 + "claim": "Isolated shifts in the left risk-neutral jump tail occur and are visible in short-dated but not regular option prices",
267 + "evidence_type": "figure",
268 + "evidence_ref": "Figure 1; Figure A.1"
269 + },
270 + {
271 + "claim": "FOMC announcements move priced tail risk differently from volatility",
272 + "evidence_type": "figure",
273 + "evidence_ref": "Figure 2; Figure B.1"
274 + },
275 + {
276 + "claim": "Fitting short-dated options requires pronounced time variation in the jump size distribution",
277 + "evidence_type": "figure",
278 + "evidence_ref": "Figures 7, 9, 11; Appendix F-G"
279 + },
280 + {
281 + "claim": "Left jump tail variation is not tightly linked to diffusive volatility",
282 + "evidence_type": "figure",
283 + "evidence_ref": "Figures 12-14"
284 + },
285 + {
286 + "claim": "Left tail jump measure spills into pricing errors of standard parametric models on regular options",
287 + "evidence_type": "figure",
288 + "evidence_ref": "Figure 15; Appendix H"
289 + },
290 + {
291 + "claim": "Left jump tail measure helps predict future excess equity returns",
292 + "evidence_type": "table",
293 + "evidence_ref": "Table II"
294 + }
295 + ],
296 + "structural_summary": "Opens with market-growth motivation for weekly options and asks why they matter, then previews a new short-maturity pricing approximation and seminonparametric 'structural calibration' contrasted with standard calibration and prior tail work. Illustrative figures (a 2012 tail shift, FOMC intraday reactions) motivate time-varying jump distributions, with noise and liquidity concerns addressed. Body covers data, setup, method, estimation, model comparison across jump specifications, characterization of time-varying jump risk, and asset pricing implications (spillovers to regular options, return predictability). Conclusion restates findings and proposes parametric modeling of tail dynamics as future work. Extensive appendices hold filters, pricing formulas and additional estimation results.",
297 + "writing_profile_keywords": [
298 + "option-implied tail risk",
299 + "seminonparametric estimation",
300 + "continuous-time finance",
301 + "figure-heavy evidence",
302 + "illustrative event cases",
303 + "no-arbitrage",
304 + "econometric inference",
305 + "model specification testing"
306 + ],
307 + "confidence": 0.8,
308 + "notes": "Primary method is labeled structural_estimation because estimation is option-based structural calibration; return predictability is secondary. Robustness appears mainly as appendix checks on data quality and alternative jump specifications."
309 + },
310 + "deterministic_checks": {
311 + "roadmap_regex": true,
312 + "contribution_regex_in_intro": false,
313 + "magnitude_in_abstract": false,
314 + "abstract_n_sentences": 4,
315 + "intro_n_paragraphs": 12,
316 + "intro_n_words": 2034,
317 + "causal_words_in_abstract": 0,
318 + "hedge_words_in_abstract": 0,
319 + "stars_mentioned": false,
320 + "clustered_se_mentioned": false,
321 + "internet_appendix_mentioned": false,
322 + "section_roles_rule": {
323 + "0": "introduction",
324 + "1": "other",
325 + "2": "theory",
326 + "3": "theory",
327 + "4": "empirical_design",
328 + "5": "results",
329 + "6": "other",
330 + "7": "other",
331 + "8": "conclusion",
332 + "9": "appendix",
333 + "10": "appendix",
334 + "11": "appendix",
335 + "12": "appendix",
336 + "13": "appendix",
337 + "14": "appendix",
338 + "15": "appendix",
339 + "16": "appendix",
340 + "17": "appendix",
341 + "18": "appendix",
342 + "19": "appendix",
343 + "20": "appendix"
344 + }
345 + },
346 + "schema_valid": true,
347 + "code_version": "feb6c9e-dirty"
348 +}
\ No newline at end of file
added data/annotations/documents/jof_12487.json +279 −0
@@ -0,0 +1,279 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12487",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:55+00:00",
9 + "source_sha256": "e327f42795508f12d6b95430d6b8b5b4ca898bc0415cc5d8e6367b94fadd7467",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 25069,
12 + "usage": {
13 + "input": 8623,
14 + "output": 2443,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.084146,
19 + "request_id": "req_011CggpYpd1yJoXawyuvzD99",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "international_finance (law and finance, CSR/ESG)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Association between a country's legal origin (common vs. civil, by family) and firm-level CSR ratings, plus the differential responsiveness of CSR to demand shocks by legal origin.",
27 + "strategy": "Cross-country panel regressions (pooled OLS, random-effects GLS, random-effects ordered probit) of CSR ratings on legal origin dummies with year and industry fixed effects and country- and firm-level controls. Quasi-natural experiments (Chinese milk scandal, Indian Ocean earthquake, Deepwater Horizon) estimated as difference-in-differences with country fixed effects, plus placebo tests and mechanism tests.",
28 + "key_assumptions": [
29 + "Legal origin is predetermined and does not merely proxy for omitted country characteristics once controls are included",
30 + "Scandals and disasters shift CSR demand exogenously and comparably across legal origins",
31 + "Placebo industries are unaffected by the shocks"
32 + ],
33 + "threats_addressed": [
34 + "alternative explanations: doing good by doing well, ownership, political institutions, globalization, culture",
35 + "country-level omitted channels, handled with country fixed effects in the shock experiments",
36 + "changes in market shares or consumer demand driving the responsiveness results",
37 + "choice of CSR measure or data provider"
38 + ],
39 + "causal_language_strength": 2,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": false,
65 + "sentence_index": -1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "related_literature_positioning"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "literature_gap"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "preview_of_approach"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "mechanism_or_theory_sketch"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "preview_of_approach"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "robustness_mention"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "roadmap"
109 + }
110 + ],
111 + "contribution_paragraph_index": -1,
112 + "contribution_types": [
113 + "new_evidence"
114 + ],
115 + "roadmap_present": true,
116 + "literature_positioning_location": "inside_introduction"
117 + },
118 + "section_roles": [
119 + {
120 + "idx": 0,
121 + "role": "introduction"
122 + },
123 + {
124 + "idx": 1,
125 + "role": "theory"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "data"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "results"
138 + },
139 + {
140 + "idx": 5,
141 + "role": "results"
142 + },
143 + {
144 + "idx": 6,
145 + "role": "conclusion"
146 + },
147 + {
148 + "idx": 7,
149 + "role": "appendix"
150 + },
151 + {
152 + "idx": 8,
153 + "role": "appendix"
154 + },
155 + {
156 + "idx": 9,
157 + "role": "appendix"
158 + },
159 + {
160 + "idx": 10,
161 + "role": "appendix"
162 + }
163 + ],
164 + "results_presentation": {
165 + "economic_significance_discussed": false,
166 + "statistical_significance_language": "stars_and_tstats",
167 + "magnitude_statements_in_intro": false
168 + },
169 + "robustness": {
170 + "location": "within_results",
171 + "types": [
172 + "alternative_specification",
173 + "alternative_measure",
174 + "placebo",
175 + "other"
176 + ]
177 + },
178 + "tables": {
179 + "convention_notes": "Tables open with descriptive and nonparametric comparisons by legal origin, followed by the main regression table across multiple estimators. Later robustness tables are framed as repeating one benchmark model with alternative controls, dependent variables or samples. Notes state the estimation method and fixed effects and use a 10/5/1% star convention. Descriptive tables report standard deviations in brackets.",
180 + "units_stated_in_notes": true,
181 + "standard_errors_description": "unclear"
182 + },
183 + "figures": {
184 + "uncertainty_shown": "no"
185 + },
186 + "conclusion": {
187 + "bounded_claims": true,
188 + "policy_implications": false,
189 + "open_questions": false,
190 + "restates_contribution": true
191 + },
192 + "hedging": {
193 + "level": "moderate",
194 + "typical_devices": [
195 + "'appears to be' and 'suggests' attached to findings",
196 + "framing results as correlations or associations",
197 + "explicit caveat that higher CSR does not imply higher total social responsibility",
198 + "'may reflect' alternative interpretations such as constraints rather than managerial objectives"
199 + ]
200 + },
201 + "claim_evidence": [
202 + {
203 + "claim": "Firms from common law countries have lower CSR ratings than civil law firms, and Scandinavian-origin firms score highest.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Tables II-IV"
206 + },
207 + {
208 + "claim": "Legal origin outperforms ownership, political institutions, culture and firm financials as an explanation of CSR.",
209 + "evidence_type": "table",
210 + "evidence_ref": "Tables IV-V"
211 + },
212 + {
213 + "claim": "The legal origin result holds across CSR subindices and alternative ESG data providers.",
214 + "evidence_type": "table",
215 + "evidence_ref": "Tables VI-VII"
216 + },
217 + {
218 + "claim": "Civil law firms respond more strongly in CSR to scandals and disasters.",
219 + "evidence_type": "table",
220 + "evidence_ref": "Tables VIII-IX"
221 + },
222 + {
223 + "claim": "The greater responsiveness of civil law firms is not explained by changes in market shares.",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table X"
226 + },
227 + {
228 + "claim": "Civil law firms face lower shareholder litigation risk and more regulation, which serve as channels.",
229 + "evidence_type": "table",
230 + "evidence_ref": "Table XI"
231 + },
232 + {
233 + "claim": "Countries with higher sustainability ratings tend to be civil law countries.",
234 + "evidence_type": "figure",
235 + "evidence_ref": "Figure 1"
236 + }
237 + ],
238 + "structural_summary": "The paper opens with the tension between shareholder primacy and widespread CSR activity, then argues that the doing good/doing well literature cannot explain cross-country variation. It proposes legal origin as the explanation and grounds this in law-and-finance theory in a dedicated section. Global CSR data feed descriptive and nonparametric comparisons, followed by panel regressions using several estimators. Robustness is organized as tests of alternative theories, alternative dependent variables and alternative samples. The paper then adds difference-in-differences shock experiments with placebo tests and a market-share check, and closes with tests of mechanisms. The conclusion restates macro- and micro-level contributions and caveats against welfare interpretations.",
239 + "writing_profile_keywords": [
240 + "law and finance",
241 + "cross-country panel",
242 + "horse race against alternative explanations",
243 + "quasi-natural experiments",
244 + "mechanism tests",
245 + "multiple data sources",
246 + "associational framing"
247 + ],
248 + "confidence": 0.8,
249 + "notes": "The section titled Economic Mechanisms was labeled robustness by the rule; I assigned it the results role because it directly tests channels. The excerpts do not state how standard errors are computed or clustered, so that field is unclear. Causal wording is mostly associational ('strongly correlated', 'predictor'), although the shock experiments are framed as quasi-natural experiments; I therefore rated causal language strength at 2."
250 + },
251 + "deterministic_checks": {
252 + "roadmap_regex": true,
253 + "contribution_regex_in_intro": false,
254 + "magnitude_in_abstract": false,
255 + "abstract_n_sentences": 3,
256 + "intro_n_paragraphs": 6,
257 + "intro_n_words": 1027,
258 + "causal_words_in_abstract": 0,
259 + "hedge_words_in_abstract": 0,
260 + "stars_mentioned": true,
261 + "clustered_se_mentioned": false,
262 + "internet_appendix_mentioned": false,
263 + "section_roles_rule": {
264 + "0": "introduction",
265 + "1": "other",
266 + "2": "data",
267 + "3": "results",
268 + "4": "results",
269 + "5": "robustness",
270 + "6": "conclusion",
271 + "7": "appendix",
272 + "8": "appendix",
273 + "9": "appendix",
274 + "10": "appendix"
275 + }
276 + },
277 + "schema_valid": true,
278 + "code_version": "feb6c9e-dirty"
279 +}
\ No newline at end of file
added data/annotations/documents/jof_12488.json +269 −0
@@ -0,0 +1,269 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12488",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:36+00:00",
9 + "source_sha256": "bb9f0855c72da6b7717b229327ca3ed8eb01047f2deb7681b72377405b5a1831",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22201,
12 + "usage": {
13 + "input": 7657,
14 + "output": 2259,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.076602,
19 + "request_id": "req_011CggpXR9ceQyG8aAaTdHLQ",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "term structure modeling / fixed income derivatives",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Structural parameters of a linear-rational term structure model, plus model fit to swap rates and swaption implied volatilities",
27 + "strategy": "Quasi-maximum likelihood with the Kalman filter on a weekly panel of swaps and swaptions, followed by simulation-based checks of the model-implied dynamics against the data",
28 + "key_assumptions": [
29 + "Factor process has a linear drift and the state price density is linear in the factors",
30 + "Three term structure factors with one to three USV factors",
31 + "Quasi-ML approximation with the Kalman filter is adequate"
32 + ],
33 + "threats_addressed": [
34 + "Model misspecification (compared across numbers of USV factors)",
35 + "Fit near the zero lower bound",
36 + "Benchmarking against an exponential-affine model"
37 + ],
38 + "causal_language_strength": 1,
39 + "uses_causal_claim": false
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": false,
44 + "sentence_index": -1
45 + },
46 + "gap": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 0
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 2
57 + },
58 + "magnitude": {
59 + "present": false,
60 + "sentence_index": -1
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 1
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "literature_gap"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "contribution_statement"
80 + },
81 + {
82 + "paragraph_index": 2,
83 + "move": "mechanism_or_theory_sketch"
84 + },
85 + {
86 + "paragraph_index": 3,
87 + "move": "preview_of_approach"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 4,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 5,
99 + "move": "preview_of_results"
100 + },
101 + {
102 + "paragraph_index": 6,
103 + "move": "related_literature_positioning"
104 + },
105 + {
106 + "paragraph_index": 7,
107 + "move": "related_literature_positioning"
108 + },
109 + {
110 + "paragraph_index": 8,
111 + "move": "preview_of_results"
112 + },
113 + {
114 + "paragraph_index": 8,
115 + "move": "roadmap"
116 + }
117 + ],
118 + "contribution_paragraph_index": 0,
119 + "contribution_types": [
120 + "new_theory_or_model",
121 + "new_evidence"
122 + ],
123 + "roadmap_present": true,
124 + "literature_positioning_location": "inside_introduction"
125 + },
126 + "section_roles": [
127 + {
128 + "idx": 0,
129 + "role": "introduction"
130 + },
131 + {
132 + "idx": 1,
133 + "role": "theory"
134 + },
135 + {
136 + "idx": 2,
137 + "role": "theory"
138 + },
139 + {
140 + "idx": 3,
141 + "role": "theory"
142 + },
143 + {
144 + "idx": 4,
145 + "role": "results"
146 + },
147 + {
148 + "idx": 5,
149 + "role": "conclusion"
150 + },
151 + {
152 + "idx": 6,
153 + "role": "appendix"
154 + },
155 + {
156 + "idx": 7,
157 + "role": "appendix"
158 + }
159 + ],
160 + "results_presentation": {
161 + "economic_significance_discussed": true,
162 + "statistical_significance_language": "mixed",
163 + "magnitude_statements_in_intro": true
164 + },
165 + "robustness": {
166 + "location": "within_results",
167 + "types": [
168 + "alternative_specification",
169 + "subsample"
170 + ]
171 + },
172 + "tables": {
173 + "convention_notes": "Notes define variables, units (basis points, percent, annualized) and data periods. Parameter estimates are reported with asymptotic standard errors in parentheses and regression slopes with t-statistics in parentheses. Data and model-simulated panels are shown side by side. One table compares frameworks qualitatively.",
174 + "units_stated_in_notes": true,
175 + "standard_errors_description": "mixed"
176 + },
177 + "figures": {
178 + "uncertainty_shown": "unclear"
179 + },
180 + "conclusion": {
181 + "bounded_claims": false,
182 + "policy_implications": false,
183 + "open_questions": true,
184 + "restates_contribution": true
185 + },
186 + "hedging": {
187 + "level": "low",
188 + "typical_devices": [
189 + "'arguably' when describing the dominant literature",
190 + "qualified statements such as 'a very good fit' and 'many features'",
191 + "explicit statement of the model's focus or scope"
192 + ]
193 + },
194 + "claim_evidence": [
195 + {
196 + "claim": "The linear-rational framework guarantees nonnegative short rates, accommodates USV, and gives semi-analytical swaption prices.",
197 + "evidence_type": "equation_or_proof",
198 + "evidence_ref": "Section I theorems; Appendix A proofs"
199 + },
200 + {
201 + "claim": "No exponential-affine model can jointly ensure nonnegative rates and USV, or price swaptions semi-analytically.",
202 + "evidence_type": "prior_literature",
203 + "evidence_ref": "Duffie and Kan (1996), Dai and Singleton (2000), Joslin (2014); Table I"
204 + },
205 + {
206 + "claim": "The preferred specification with three USV factors fits swaps and swaptions well, including near the ZLB.",
207 + "evidence_type": "table",
208 + "evidence_ref": "Tables II, III; Figure 2"
209 + },
210 + {
211 + "claim": "Volatility becomes compressed and level-dependent near the ZLB, with R2 around 0.5, but shows little level-dependence at moderate rates.",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table IV; Figures 1, 7"
214 + },
215 + {
216 + "claim": "Near the ZLB the model generates asymmetric short-rate distributions and shifts the first PC from a level toward a slope factor.",
217 + "evidence_type": "figure",
218 + "evidence_ref": "Figures 5, 6"
219 + },
220 + {
221 + "claim": "The model captures unconditional and conditional swap risk premium features.",
222 + "evidence_type": "table",
223 + "evidence_ref": "Tables V, VI; Figure 8"
224 + },
225 + {
226 + "claim": "On swaps alone, LRSQ pricing performance is virtually identical to an exponential-affine square-root model.",
227 + "evidence_type": "none_stated",
228 + "evidence_ref": "Section IV.J"
229 + }
230 + ],
231 + "structural_summary": "Opens by motivating the problem that low rates create for existing term structure models and states a twofold contribution: a new model class and an empirical analysis. It then sketches the state-price-density construction and previews the estimation design and the simulated results on term structure, volatility and risk premium dynamics near the ZLB. Positioning against the LG and exponential-affine frameworks follows, then a roadmap. The body develops the theory (framework, LRSQ model, permanent-component extension) before a long empirical section covering estimation, specification comparison, ZLB dynamics, risk premia and an affine benchmark. Proofs are in the appendix, and the conclusion restates the contribution and lists extensions.",
232 + "writing_profile_keywords": [
233 + "model class introduction",
234 + "tractability advantages enumerated",
235 + "Kalman filter quasi-ML",
236 + "simulation-based model evaluation",
237 + "zero lower bound",
238 + "unspanned stochastic volatility",
239 + "benchmark comparison"
240 + ],
241 + "confidence": 0.85,
242 + "notes": "Model-introduction paper with substantial estimation, so the genre is mixed. Causal identification is not applicable. Section III's role was mapped to theory."
243 + },
244 + "deterministic_checks": {
245 + "roadmap_regex": true,
246 + "contribution_regex_in_intro": true,
247 + "magnitude_in_abstract": false,
248 + "abstract_n_sentences": 3,
249 + "intro_n_paragraphs": 9,
250 + "intro_n_words": 1406,
251 + "causal_words_in_abstract": 0,
252 + "hedge_words_in_abstract": 0,
253 + "stars_mentioned": false,
254 + "clustered_se_mentioned": false,
255 + "internet_appendix_mentioned": true,
256 + "section_roles_rule": {
257 + "0": "introduction",
258 + "1": "theory",
259 + "2": "theory",
260 + "3": "other",
261 + "4": "other",
262 + "5": "conclusion",
263 + "6": "appendix",
264 + "7": "appendix"
265 + }
266 + },
267 + "schema_valid": true,
268 + "code_version": "feb6c9e-dirty"
269 +}
\ No newline at end of file
added data/annotations/documents/jof_12489.json +258 −0
@@ -0,0 +1,258 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('figures_note_placeholder' was unexpected) @ figures"
4 + ],
5 + "doc_id": "jof_12489",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:56:55+00:00",
11 + "source_sha256": "891df134f35f992e0ff749ce882850fd37996ca9fdfbf8c8260046e8aaf503c0",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 20682,
14 + "usage": {
15 + "input": 7036,
16 + "output": 2229,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.073518,
21 + "request_id": "req_011CggpYvguM4oRi2dVp599B",
22 + "annotation": {
23 + "genre": "mixed_theory_empirical",
24 + "subfield": "household_finance",
25 + "subfield_secondary": "macro_finance",
26 + "primary_method": "structural_estimation",
27 + "identification": {
28 + "estimand": "Model-implied reverse mortgage take-up rates and welfare gains, and how they respond to counterfactual changes in risks, bequest motives, contract terms and Great Recession shocks",
29 + "strategy": "Life-cycle retirement model with uninsurable health, mortality, medical expense and house price risk. It is calibrated by two-stage estimation (Gourinchas–Parker style) on HRS data; RMLs are added afterward and counterfactual experiments are run.",
30 + "key_assumptions": [
31 + "Model structure, including preferences, bequest motive and risk processes, describes retiree behavior",
32 + "No aggregate uncertainty in the baseline model",
33 + "Parameters estimated without RMLs remain valid once RMLs are introduced",
34 + "Retirees aged 65+ with no labor supply decision"
35 + ],
36 + "threats_addressed": [
37 + "Model fit checked against untargeted moments: distribution of debt, portfolio shares, and age distribution of RML borrowers",
38 + "Sensitivity analysis reported in the Internet Appendix",
39 + "Additional experiments with aggregate house price dynamics"
40 + ],
41 + "causal_language_strength": 3,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": true,
51 + "sentence_index": 1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 2
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 4
60 + },
61 + "magnitude": {
62 + "present": true,
63 + "sentence_index": 3
64 + },
65 + "contribution": {
66 + "present": false,
67 + "sentence_index": -1
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "literature_gap"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "mechanism_or_theory_sketch"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "tension_or_puzzle"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "preview_of_approach"
91 + },
92 + {
93 + "paragraph_index": 2,
94 + "move": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "policy_or_broader_implication"
103 + },
104 + {
105 + "paragraph_index": 3,
106 + "move": "contribution_statement"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "related_literature_positioning"
111 + },
112 + {
113 + "paragraph_index": 5,
114 + "move": "related_literature_positioning"
115 + },
116 + {
117 + "paragraph_index": 6,
118 + "move": "roadmap"
119 + }
120 + ],
121 + "contribution_paragraph_index": 3,
122 + "contribution_types": [
123 + "new_theory_or_model",
124 + "new_evidence"
125 + ],
126 + "roadmap_present": true,
127 + "literature_positioning_location": "inside_introduction"
128 + },
129 + "section_roles": [
130 + {
131 + "idx": 0,
132 + "role": "introduction"
133 + },
134 + {
135 + "idx": 1,
136 + "role": "other"
137 + },
138 + {
139 + "idx": 2,
140 + "role": "theory"
141 + },
142 + {
143 + "idx": 3,
144 + "role": "empirical_design"
145 + },
146 + {
147 + "idx": 4,
148 + "role": "results"
149 + },
150 + {
151 + "idx": 5,
152 + "role": "conclusion"
153 + }
154 + ],
155 + "results_presentation": {
156 + "economic_significance_discussed": true,
157 + "statistical_significance_language": "unclear",
158 + "magnitude_statements_in_intro": true
159 + },
160 + "robustness": {
161 + "location": "internet_appendix",
162 + "types": [
163 + "sensitivity_to_parameters"
164 + ]
165 + },
166 + "tables": {
167 + "convention_notes": "Tables report the calibrated parameters and model-generated take-up rates and welfare gains under baseline and counterfactual scenarios. Notes define each column and the welfare measure (a one-time payment at age 65 equating expected lifetime utility), cross-reference Table II's notes, and state parameter frequency (annual vs. biennial). No standard errors are shown.",
168 + "units_stated_in_notes": true,
169 + "standard_errors_description": "none"
170 + },
171 + "figures": {
172 + "uncertainty_shown": "unclear",
173 + "figures_note_placeholder": null
174 + },
175 + "conclusion": {
176 + "bounded_claims": true,
177 + "policy_implications": true,
178 + "open_questions": true,
179 + "restates_contribution": true
180 + },
181 + "hedging": {
182 + "level": "moderate",
183 + "typical_devices": [
184 + "'according to the model' / 'our model indicates' attribution",
185 + "modal verbs such as 'may' and 'is likely to'",
186 + "conditioning on the existing RML contract",
187 + "acknowledging that aggregate house price dynamics are absent from the model"
188 + ]
189 + },
190 + "claim_evidence": [
191 + {
192 + "claim": "Average welfare gain from RMLs is small per homeowner but larger for borrowers, low-income households and the oldest households",
193 + "evidence_type": "table",
194 + "evidence_ref": "Table II"
195 + },
196 + {
197 + "claim": "Bequest motives and uncertainty about health and expenses dampen RML demand; without bequest motives, take-up rises sharply",
198 + "evidence_type": "table",
199 + "evidence_ref": "Table III"
200 + },
201 + {
202 + "claim": "Eliminating up-front costs, or making RMLs recourse loans, roughly triples demand",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table IV"
205 + },
206 + {
207 + "claim": "The Great Recession lowers aggregate take-up in the short run but triples demand among the oldest, lowest-income households",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table V"
210 + },
211 + {
212 + "claim": "The RML take-up share among older homeowners is low but has grown over time",
213 + "evidence_type": "figure",
214 + "evidence_ref": "Figure 1"
215 + },
216 + {
217 + "claim": "The calibrated model matches asset age profiles, the debt distribution and the age distribution of RML borrowers",
218 + "evidence_type": "figure",
219 + "evidence_ref": "Figures 2-5, 7"
220 + }
221 + ],
222 + "structural_summary": "The introduction opens with the product, the gap in research, and the low-take-up puzzle, then poses four research questions. It previews heterogeneous welfare gains, counterfactual findings and Great Recession effects, and closes with contributions and positioning against the literatures on retirement saving and mortgage choice. The body proceeds through: institutional overview; a structural life-cycle model covering renters, owners, and owners with an RML; two-stage calibration with checks of fit on untargeted moments; and results built on counterfactual experiments. The conclusion restates the findings, notes policy trade-offs and a model limitation, and points to future work.",
223 + "writing_profile_keywords": [
224 + "quantitative structural model",
225 + "life-cycle retirement",
226 + "counterfactual experiments",
227 + "welfare gains",
228 + "policy reform evaluation",
229 + "model fit validation",
230 + "heterogeneity across income and age"
231 + ],
232 + "confidence": 0.85,
233 + "notes": "Calibration uses a two-stage estimation in the style of Gourinchas–Parker. It was labeled structural_estimation, though the paper calls itself calibrated. Section I is an institutional overview. The 0.89% model take-up rate differs from the 1.9% data figure."
234 + },
235 + "deterministic_checks": {
236 + "roadmap_regex": true,
237 + "contribution_regex_in_intro": true,
238 + "magnitude_in_abstract": true,
239 + "abstract_n_sentences": 6,
240 + "intro_n_paragraphs": 7,
241 + "intro_n_words": 1105,
242 + "causal_words_in_abstract": 0,
243 + "hedge_words_in_abstract": 0,
244 + "stars_mentioned": false,
245 + "clustered_se_mentioned": false,
246 + "internet_appendix_mentioned": true,
247 + "section_roles_rule": {
248 + "0": "introduction",
249 + "1": "other",
250 + "2": "theory",
251 + "3": "other",
252 + "4": "results",
253 + "5": "conclusion"
254 + }
255 + },
256 + "schema_valid": false,
257 + "code_version": "feb6c9e-dirty"
258 +}
\ No newline at end of file
added data/annotations/documents/jof_12490.json +295 −0
@@ -0,0 +1,295 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12490",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:56:40+00:00",
9 + "source_sha256": "22332f923950a346e53845513bb3c898a690075826c1ff3f9c38cbd237058120",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24514,
12 + "usage": {
13 + "input": 8499,
14 + "output": 2273,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.08025,
19 + "request_id": "req_011CggpXrUs8VJiNd5tJs5TJ",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "corporate cash holdings / governance",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Effect of financial portfolio size on the share of the portfolio held in risky assets; marginal value of risky versus safe financial assets",
27 + "strategy": "Panel regressions run by OLS and by 2SLS. Portfolio size is instrumented with unexpected operating cash flow innovations. Value tests regress stock returns and market-to-book on financial asset holdings, following the Faulkender–Wang approach.",
28 + "key_assumptions": [
29 + "Unexpected cash flow innovations shift portfolio size but affect portfolio composition only through size",
30 + "Fair value footnote disclosures accurately capture asset classes"
31 + ],
32 + "threats_addressed": [
33 + "Joint determination of portfolio size and composition (endogeneity)"
34 + ],
35 + "causal_language_strength": 3,
36 + "uses_causal_claim": true
37 + },
38 + "abstract_moves": {
39 + "question": {
40 + "present": true,
41 + "sentence_index": 2
42 + },
43 + "gap": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "method": {
48 + "present": true,
49 + "sentence_index": 2
50 + },
51 + "results": {
52 + "present": true,
53 + "sentence_index": 3
54 + },
55 + "magnitude": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "contribution": {
60 + "present": true,
61 + "sentence_index": 4
62 + }
63 + },
64 + "introduction": {
65 + "moves": [
66 + {
67 + "paragraph_index": 0,
68 + "move": "motivation"
69 + },
70 + {
71 + "paragraph_index": 0,
72 + "move": "tension_or_puzzle"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "preview_of_approach"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_approach"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "identification_defense"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "preview_of_results"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "robustness_mention"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "contribution_statement"
113 + },
114 + {
115 + "paragraph_index": 6,
116 + "move": "related_literature_positioning"
117 + },
118 + {
119 + "paragraph_index": 7,
120 + "move": "policy_or_broader_implication"
121 + }
122 + ],
123 + "contribution_paragraph_index": 6,
124 + "contribution_types": [
125 + "new_evidence",
126 + "new_data_or_measure",
127 + "new_theory_or_model"
128 + ],
129 + "roadmap_present": false,
130 + "literature_positioning_location": "both"
131 + },
132 + "section_roles": [
133 + {
134 + "idx": 0,
135 + "role": "introduction"
136 + },
137 + {
138 + "idx": 1,
139 + "role": "data"
140 + },
141 + {
142 + "idx": 2,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 3,
147 + "role": "theory"
148 + },
149 + {
150 + "idx": 4,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 5,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 6,
159 + "role": "conclusion"
160 + },
161 + {
162 + "idx": 7,
163 + "role": "appendix"
164 + },
165 + {
166 + "idx": 8,
167 + "role": "appendix"
168 + },
169 + {
170 + "idx": 9,
171 + "role": "appendix"
172 + },
173 + {
174 + "idx": 10,
175 + "role": "appendix"
176 + }
177 + ],
178 + "results_presentation": {
179 + "economic_significance_discussed": true,
180 + "statistical_significance_language": "unclear",
181 + "magnitude_statements_in_intro": true
182 + },
183 + "robustness": {
184 + "location": "within_results",
185 + "types": [
186 + "alternative_measure",
187 + "alternative_specification"
188 + ]
189 + },
190 + "tables": {
191 + "convention_notes": "Table notes define variables and write out the regression equations, including the 2SLS first stage. OLS and 2SLS columns are reported side by side. Descriptive tables give aggregate value-weighted figures, with equal-weighted versions in an appendix. Appendix tables reproduce firm disclosures from Google, Apple and Intel as illustrations.",
192 + "units_stated_in_notes": true,
193 + "standard_errors_description": "unclear"
194 + },
195 + "figures": {
196 + "uncertainty_shown": "no"
197 + },
198 + "conclusion": {
199 + "bounded_claims": true,
200 + "policy_implications": true,
201 + "open_questions": true,
202 + "restates_contribution": true
203 + },
204 + "hedging": {
205 + "level": "moderate",
206 + "typical_devices": [
207 + "'consistent with' framing of results against model predictions",
208 + "'suggest' for 2SLS estimates",
209 + "explicit sample limitation acknowledged in conclusion"
210 + ]
211 + },
212 + "claim_evidence": [
213 + {
214 + "claim": "Risky assets make up about 40% of firms' financial portfolios and about 6% of book assets.",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table I"
217 + },
218 + {
219 + "claim": "The share held in risky assets rises from about 10% to 40% across portfolio-size quintiles, and allocation variation is higher in the top quintiles.",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table II"
222 + },
223 + {
224 + "claim": "A one-point increase in portfolio size raises the risky share by about 30 bp (2SLS estimate).",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table III"
227 + },
228 + {
229 + "claim": "Permanently reinvested earnings relate positively to total financial assets but insignificantly to risky assets.",
230 + "evidence_type": "table",
231 + "evidence_ref": "Table IV"
232 + },
233 + {
234 + "claim": "Poorly governed firms hold more risky assets.",
235 + "evidence_type": "table",
236 + "evidence_ref": "Table V"
237 + },
238 + {
239 + "claim": "Risky financial assets are valued at a 13–22% discount relative to safe assets.",
240 + "evidence_type": "table",
241 + "evidence_ref": "Table VII"
242 + },
243 + {
244 + "claim": "Constrained firms optimally avoid risky assets; unconstrained firms are at best indifferent to them.",
245 + "evidence_type": "equation_or_proof",
246 + "evidence_ref": "Section III; Internet Appendix"
247 + },
248 + {
249 + "claim": "There is no uniform tax incentive to invest in risky financial assets.",
250 + "evidence_type": "equation_or_proof",
251 + "evidence_ref": "Appendix C"
252 + }
253 + ],
254 + "structural_summary": "The paper opens with press anecdotes that challenge the assumption that corporate 'cash' is in fact cash. It introduces hand-collected fair value data made available by SFAS 157, then describes portfolio composition. A theory section sketches a precautionary-savings model extended to risky and illiquid assets, with the formal derivation in an internet appendix. Empirical tests follow: univariate sorts, then OLS/2SLS determinants (taxes, agency, CEO traits), then valuation and financing-source tests. The conclusion frames the findings as a largely unregulated shadow asset management industry.",
255 + "writing_profile_keywords": [
256 + "hand-collected data",
257 + "model-consistent predictions",
258 + "precautionary savings",
259 + "2SLS",
260 + "shadow hedge fund framing",
261 + "policy implications",
262 + "anecdotal motivation"
263 + ],
264 + "confidence": 0.8,
265 + "notes": "Theory is summarized in the main text and formally derived in an internet appendix. Standard error treatment is not visible in the excerpts. Section II is descriptive results."
266 + },
267 + "deterministic_checks": {
268 + "roadmap_regex": false,
269 + "contribution_regex_in_intro": false,
270 + "magnitude_in_abstract": true,
271 + "abstract_n_sentences": 6,
272 + "intro_n_paragraphs": 8,
273 + "intro_n_words": 1027,
274 + "causal_words_in_abstract": 0,
275 + "hedge_words_in_abstract": 1,
276 + "stars_mentioned": false,
277 + "clustered_se_mentioned": false,
278 + "internet_appendix_mentioned": true,
279 + "section_roles_rule": {
280 + "0": "introduction",
281 + "1": "data",
282 + "2": "other",
283 + "3": "theory",
284 + "4": "other",
285 + "5": "other",
286 + "6": "conclusion",
287 + "7": "appendix",
288 + "8": "appendix",
289 + "9": "appendix",
290 + "10": "appendix"
291 + }
292 + },
293 + "schema_valid": true,
294 + "code_version": "feb6c9e-dirty"
295 +}
\ No newline at end of file
added data/annotations/documents/jof_12492.json +282 −0
@@ -0,0 +1,282 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12492",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:01+00:00",
9 + "source_sha256": "90490da86b17d09e74b3f92843e603754efcafb27d151e2026d42411ac64daeb",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 27298,
12 + "usage": {
13 + "input": 9001,
14 + "output": 2564,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.088078,
19 + "request_id": "req_011CggpZDKXBkaGYQ6qHe2x2",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "financial_intermediation_banking; small business credit / government loan guarantee evaluation",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Average effect on the treated of SBA loan amount (per $1 million) on firm employment growth, measured as post-loan minus pre-loan average employment, for growing small firms",
27 + "strategy": "Longitudinal matching (exact matching plus kernel propensity scores on employment history) combined with treatment-control group fixed-effects regressions. IVs built from the lagged local presence or share of branches of Preferred Lender Program banks, with PLP activity defined outside the local market.",
28 + "key_assumptions": [
29 + "Bank entry into a county is not based on potential growth of SBA borrowers there",
30 + "A bank's choice to become a PLP lender is not based on local SBA-recipient growth potential",
31 + "For the branch-share IV, PLP banks do not gain local market share based on SBA-recipient growth potential",
32 + "Small businesses rely on local lenders",
33 + "For matching/OLS: no correlated time-varying unobservables"
34 + ],
35 + "threats_addressed": [
36 + "Selection into loan receipt (positive or negative)",
37 + "Pre-loan growth trends of recipients",
38 + "Correlation of SBA availability with local financial development and growth",
39 + "Firm exit/survivorship",
40 + "Displacement of competing firms (nontradables)",
41 + "Time-varying idiosyncratic demand, productivity or cost shocks"
42 + ],
43 + "causal_language_strength": 3,
44 + "uses_causal_claim": true
45 + },
46 + "abstract_moves": {
47 + "question": {
48 + "present": true,
49 + "sentence_index": 0
50 + },
51 + "gap": {
52 + "present": false,
53 + "sentence_index": -1
54 + },
55 + "method": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "results": {
60 + "present": true,
61 + "sentence_index": 2
62 + },
63 + "magnitude": {
64 + "present": true,
65 + "sentence_index": 2
66 + },
67 + "contribution": {
68 + "present": false,
69 + "sentence_index": -1
70 + }
71 + },
72 + "introduction": {
73 + "moves": [
74 + {
75 + "paragraph_index": 0,
76 + "move": "motivation"
77 + },
78 + {
79 + "paragraph_index": 5,
80 + "move": "motivation"
81 + },
82 + {
83 + "paragraph_index": 5,
84 + "move": "literature_gap"
85 + },
86 + {
87 + "paragraph_index": 5,
88 + "move": "tension_or_puzzle"
89 + },
90 + {
91 + "paragraph_index": 6,
92 + "move": "contribution_statement"
93 + },
94 + {
95 + "paragraph_index": 6,
96 + "move": "preview_of_approach"
97 + },
98 + {
99 + "paragraph_index": 7,
100 + "move": "preview_of_approach"
101 + },
102 + {
103 + "paragraph_index": 8,
104 + "move": "identification_defense"
105 + },
106 + {
107 + "paragraph_index": 9,
108 + "move": "related_literature_positioning"
109 + },
110 + {
111 + "paragraph_index": 10,
112 + "move": "related_literature_positioning"
113 + },
114 + {
115 + "paragraph_index": 11,
116 + "move": "other"
117 + }
118 + ],
119 + "contribution_paragraph_index": 6,
120 + "contribution_types": [
121 + "new_evidence",
122 + "new_data_or_measure",
123 + "new_identification"
124 + ],
125 + "roadmap_present": false,
126 + "literature_positioning_location": "inside_introduction"
127 + },
128 + "section_roles": [
129 + {
130 + "idx": 0,
131 + "role": "introduction"
132 + },
133 + {
134 + "idx": 1,
135 + "role": "other"
136 + },
137 + {
138 + "idx": 2,
139 + "role": "data"
140 + },
141 + {
142 + "idx": 3,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 4,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 5,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 6,
155 + "role": "conclusion"
156 + }
157 + ],
158 + "results_presentation": {
159 + "economic_significance_discussed": true,
160 + "statistical_significance_language": "unclear",
161 + "magnitude_statements_in_intro": false
162 + },
163 + "robustness": {
164 + "location": "mixed",
165 + "types": [
166 + "alternative_measure",
167 + "alternative_specification",
168 + "subsample",
169 + "other"
170 + ]
171 + },
172 + "tables": {
173 + "convention_notes": "Table notes define the dependent variable as the change in average employment, t+1..t+3 versus t-3..t-1. They state that loan amount is in millions of 2010 dollars, list the county-level controls, and distinguish OLS from IV and scaled from unscaled specifications. Summary-statistics tables come first, then OLS, IV, alternative-IV and heterogeneity tables.",
174 + "units_stated_in_notes": true,
175 + "standard_errors_description": "unclear"
176 + },
177 + "figures": {
178 + "uncertainty_shown": "unclear"
179 + },
180 + "conclusion": {
181 + "bounded_claims": true,
182 + "policy_implications": true,
183 + "open_questions": false,
184 + "restates_contribution": true
185 + },
186 + "hedging": {
187 + "level": "moderate",
188 + "typical_devices": [
189 + "explicit caveats paragraph on non-random allocation and external validity",
190 + "'arguably exogenous' framing of instruments",
191 + "'if these estimates are correct' conditional",
192 + "lower-bound interpretation",
193 + "'preliminary investigation' and 'rough estimates' qualifiers",
194 + "'suggesting' when linking results to credit constraints"
195 + ]
196 + },
197 + "claim_evidence": [
198 + {
199 + "claim": "Each $1 million of SBA loans raises recipient employment by about 3–3.5 jobs over three post-loan years",
200 + "evidence_type": "table",
201 + "evidence_ref": "Tables IV, VI, VII"
202 + },
203 + {
204 + "claim": "Effects are larger for younger and larger firms",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table VIII"
207 + },
208 + {
209 + "claim": "No systematic difference between 7(a) and 504 programs or over the business cycle",
210 + "evidence_type": "table",
211 + "evidence_ref": "Tables IX, X"
212 + },
213 + {
214 + "claim": "Effects are smaller where local conventional credit is more plentiful",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table XI"
217 + },
218 + {
219 + "claim": "Displacement of competing firms is around 10% of the total effect in nontradables",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table XII"
222 + },
223 + {
224 + "claim": "Results are similar when exited firms are assigned zero employment",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table XIII"
227 + },
228 + {
229 + "claim": "Taxpayer cost per job is roughly $21,000–$25,000, low relative to other government programs",
230 + "evidence_type": "prior_literature",
231 + "evidence_ref": "Section V.B; Neumark (2013) comparison in conclusion"
232 + },
233 + {
234 + "claim": "PLP lenders account for a large share of SBA loans, especially larger loans",
235 + "evidence_type": "figure",
236 + "evidence_ref": "Figure 1"
237 + },
238 + {
239 + "claim": "Only 3.2% of SBA loans occur in counties with no PLP branch in the prior year",
240 + "evidence_type": "descriptive_statistic",
241 + "evidence_ref": "Introduction"
242 + }
243 + ],
244 + "structural_summary": "The paper is a policy-evaluation empirical study. The introduction opens with a historical epigraph and the size of the SBA program, then notes that the employment effect is theoretically ambiguous. It previews linked administrative-census data and the matching method, and motivates instruments from lender institutional features while stating the identifying assumptions. Literature positioning follows, then an explicit caveats paragraph on external validity. The body describes the program institutions and the data, presents matching/OLS results, then the IV approach, and finally heterogeneity, displacement, exit and cost-per-job extensions. The conclusion restates the estimates, bounds them as a lower bound, and draws a cost-effectiveness policy implication.",
245 + "writing_profile_keywords": [
246 + "program evaluation",
247 + "linked administrative data",
248 + "matching estimator",
249 + "instrumental variables",
250 + "explicit caveats",
251 + "cost per job",
252 + "heterogeneous effects",
253 + "policy relevance"
254 + ],
255 + "confidence": 0.8,
256 + "notes": "The introduction-paragraph indices are fragmented: paragraphs 0–4 are pieces of the epigraph, and paragraph 5 contains the opening motivation, theoretical ambiguity and empirical difficulty. Section IV is labeled 'data' by the role rule but is actually IV methods and results. Inference reporting conventions (stars or SEs) are not visible in the excerpts. Robustness checks (alternative IVs, exit zeros, controls) appear within the results and extensions sections, and an internet appendix exists."
257 + },
258 + "deterministic_checks": {
259 + "roadmap_regex": false,
260 + "contribution_regex_in_intro": true,
261 + "magnitude_in_abstract": false,
262 + "abstract_n_sentences": 6,
263 + "intro_n_paragraphs": 12,
264 + "intro_n_words": 1457,
265 + "causal_words_in_abstract": 0,
266 + "hedge_words_in_abstract": 0,
267 + "stars_mentioned": false,
268 + "clustered_se_mentioned": false,
269 + "internet_appendix_mentioned": false,
270 + "section_roles_rule": {
271 + "0": "introduction",
272 + "1": "other",
273 + "2": "data",
274 + "3": "results",
275 + "4": "data",
276 + "5": "robustness",
277 + "6": "conclusion"
278 + }
279 + },
280 + "schema_valid": true,
281 + "code_version": "feb6c9e-dirty"
282 +}
\ No newline at end of file
added data/annotations/documents/jof_12493.json +257 −0
@@ -0,0 +1,257 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12493",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:17+00:00",
9 + "source_sha256": "d96d0b1a12e49efd8bc1275856468407471d38d698f976c0e38de74a1155f54c",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20792,
12 + "usage": {
13 + "input": 6884,
14 + "output": 2035,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06903,
19 + "request_id": "req_011CggpajDPXTVamw2kLieaY",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "contract theory / costly enforcement",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; analytical costly-enforcement model with correlated borrower payoffs and no lender commitment, plus numerical examples",
28 + "key_assumptions": [
29 + "lenders cannot commit to enforcement ex ante",
30 + "borrower payoffs are correlated through an unobserved aggregate state",
31 + "enforcement is costly",
32 + "risk-neutral agents"
33 + ],
34 + "threats_addressed": [],
35 + "causal_language_strength": 1,
36 + "uses_causal_claim": false
37 + },
38 + "abstract_moves": {
39 + "question": {
40 + "present": false,
41 + "sentence_index": -1
42 + },
43 + "gap": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "method": {
48 + "present": true,
49 + "sentence_index": 1
50 + },
51 + "results": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "magnitude": {
56 + "present": false,
57 + "sentence_index": -1
58 + },
59 + "contribution": {
60 + "present": true,
61 + "sentence_index": 0
62 + }
63 + },
64 + "introduction": {
65 + "moves": [
66 + {
67 + "paragraph_index": 0,
68 + "move": "motivation"
69 + },
70 + {
71 + "paragraph_index": 0,
72 + "move": "tension_or_puzzle"
73 + },
74 + {
75 + "paragraph_index": 1,
76 + "move": "contribution_statement"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "mechanism_or_theory_sketch"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "motivation"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "motivation"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "related_literature_positioning"
97 + },
98 + {
99 + "paragraph_index": 5,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 6,
104 + "move": "preview_of_results"
105 + },
106 + {
107 + "paragraph_index": 6,
108 + "move": "related_literature_positioning"
109 + },
110 + {
111 + "paragraph_index": 7,
112 + "move": "related_literature_positioning"
113 + },
114 + {
115 + "paragraph_index": 8,
116 + "move": "related_literature_positioning"
117 + },
118 + {
119 + "paragraph_index": 8,
120 + "move": "roadmap"
121 + }
122 + ],
123 + "contribution_paragraph_index": 1,
124 + "contribution_types": [
125 + "new_theory_or_model"
126 + ],
127 + "roadmap_present": true,
128 + "literature_positioning_location": "inside_introduction"
129 + },
130 + "section_roles": [
131 + {
132 + "idx": 0,
133 + "role": "introduction"
134 + },
135 + {
136 + "idx": 1,
137 + "role": "theory"
138 + },
139 + {
140 + "idx": 2,
141 + "role": "theory"
142 + },
143 + {
144 + "idx": 3,
145 + "role": "other"
146 + },
147 + {
148 + "idx": 4,
149 + "role": "theory"
150 + },
151 + {
152 + "idx": 5,
153 + "role": "theory"
154 + },
155 + {
156 + "idx": 6,
157 + "role": "conclusion"
158 + }
159 + ],
160 + "results_presentation": {
161 + "economic_significance_discussed": false,
162 + "statistical_significance_language": "unclear",
163 + "magnitude_statements_in_intro": false
164 + },
165 + "robustness": {
166 + "location": "none_seen",
167 + "types": []
168 + },
169 + "tables": {
170 + "convention_notes": "Tables specify model primitives, giving payoff probability distributions by state, and report algebraic value-per-loan expressions; no estimates are shown.",
171 + "units_stated_in_notes": false,
172 + "standard_errors_description": "none"
173 + },
174 + "figures": {
175 + "uncertainty_shown": "no"
176 + },
177 + "conclusion": {
178 + "bounded_claims": true,
179 + "policy_implications": false,
180 + "open_questions": false,
181 + "restates_contribution": true
182 + },
183 + "hedging": {
184 + "level": "moderate",
185 + "typical_devices": [
186 + "'proposes a mechanism that would explain'",
187 + "modal verbs (may, can)",
188 + "'consistent with' framing for empirical regularities",
189 + "acknowledging model simplifications"
190 + ]
191 + },
192 + "claim_evidence": [
193 + {
194 + "claim": "When lenders cannot commit to enforce and borrower payoffs are correlated, intermediation dominates direct lending.",
195 + "evidence_type": "equation_or_proof",
196 + "evidence_ref": "Section III; proofs in Internet Appendix"
197 + },
198 + {
199 + "claim": "A large correlated portfolio provides ex post incentives to enforce because learning one loan's state informs on the others.",
200 + "evidence_type": "equation_or_proof",
201 + "evidence_ref": "Section II.A"
202 + },
203 + {
204 + "claim": "Higher correlation raises value per loan, and with increasing enforcement costs the optimal portfolio size is finite and varies with correlation.",
205 + "evidence_type": "figure",
206 + "evidence_ref": "Figures 1-2"
207 + },
208 + {
209 + "claim": "Returns to scale are strongest for small portfolios and diminish quickly, consistent with mixed empirical evidence on bank scale economies.",
210 + "evidence_type": "prior_literature",
211 + "evidence_ref": "Hughes and Mester (2013); Laeven, Ratnovski, and Tong (2014)"
212 + },
213 + {
214 + "claim": "Intermediaries issue risk-free deposits, earn positive profits, and hold systemic default risk.",
215 + "evidence_type": "equation_or_proof",
216 + "evidence_ref": "Section III.A"
217 + }
218 + ],
219 + "structural_summary": "The introduction opens with stylized facts about banks holding correlated, hard-to-value risks while still issuing safe deposits. It then states the key contribution together with the commitment mechanism and gives an extended informal motivation for the model's assumptions. Next it links the model to empirical regularities on returns to scale and on hard versus soft information, positions the paper against the delegated-monitoring, commitment-device, and safe-asset literatures, and closes with a roadmap. The body builds progressively: a one-borrower model with and without commitment; then many borrowers with correlated default plus numerical examples; then many lenders, where the main intermediation result is proved and empirical implications are drawn. The conclusion restates the mechanism and ends with an anecdotal banking illustration.",
220 + "writing_profile_keywords": [
221 + "theory",
222 + "costly enforcement",
223 + "commitment device",
224 + "correlated risk",
225 + "delegated monitoring",
226 + "numerical examples",
227 + "internet appendix proofs",
228 + "stylized facts motivation"
229 + ],
230 + "confidence": 0.85,
231 + "notes": "Sections idx 3 and 4 are parsing artifacts: 'A' is empty, and 'B' is a continuation of Section II containing the many-borrower equilibrium. Proofs are in the Internet Appendix. There is no empirical analysis; empirical regularities are cited only as consistent with the model."
232 + },
233 + "deterministic_checks": {
234 + "roadmap_regex": true,
235 + "contribution_regex_in_intro": true,
236 + "magnitude_in_abstract": false,
237 + "abstract_n_sentences": 5,
238 + "intro_n_paragraphs": 9,
239 + "intro_n_words": 1642,
240 + "causal_words_in_abstract": 0,
241 + "hedge_words_in_abstract": 0,
242 + "stars_mentioned": false,
243 + "clustered_se_mentioned": false,
244 + "internet_appendix_mentioned": true,
245 + "section_roles_rule": {
246 + "0": "introduction",
247 + "1": "theory",
248 + "2": "other",
249 + "3": "other",
250 + "4": "other",
251 + "5": "other",
252 + "6": "conclusion"
253 + }
254 + },
255 + "schema_valid": true,
256 + "code_version": "feb6c9e-dirty"
257 +}
\ No newline at end of file
added data/annotations/documents/jof_12494.json +259 −0
@@ -0,0 +1,259 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('figures_note_placeholder_removed' was unexpected) @ figures"
4 + ],
5 + "doc_id": "jof_12494",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:56:59+00:00",
11 + "source_sha256": "9d80b1fa06977d7cdb9c8a7496c73e9206a0ab277752c61c9c11d10ed1b30b47",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 17285,
14 + "usage": {
15 + "input": 6067,
16 + "output": 2216,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.069382,
21 + "request_id": "req_011CggpZDf7Ywsc1Kb3AhxNU",
22 + "annotation": {
23 + "genre": "mixed_theory_empirical",
24 + "subfield": "financial_intermediation_banking",
25 + "subfield_secondary": "household_finance",
26 + "primary_method": "instrumental_variables",
27 + "identification": {
28 + "estimand": "Effect of a lender's retained share of outstanding mortgages in a local area on the probability of foreclosing on 90+ day delinquent loans, and of zip-code concentration of outstanding mortgages on foreclosure rates and house price changes",
29 + "strategy": "Loan-level linear probability models with lender and location fixed effects, estimated conditional on serious delinquency; 2SLS using mergers of large nonfailing banks as an instrument for retained share and concentration; zip-code cross-sectional regressions with county fixed effects; heterogeneity across states with different foreclosure procedures",
30 + "key_assumptions": [
31 + "Mergers of large nonfailing banks are unrelated to local market characteristics (exclusion/exogeneity)",
32 + "Conditioning on 90+ day delinquency removes lenders' prior informational advantage",
33 + "Zip codes are the relevant area over which foreclosure externalities operate"
34 + ],
35 + "threats_addressed": [
36 + "Lender informational advantage in census tracts where it retains more mortgages",
37 + "Unobservable zip code or lender characteristics",
38 + "Alternative mechanisms such as lender organizational capabilities to foreclose or renegotiate"
39 + ],
40 + "causal_language_strength": 3,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": true,
46 + "sentence_index": 0
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 1
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 2
59 + },
60 + "magnitude": {
61 + "present": false,
62 + "sentence_index": -1
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 0
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "literature_gap"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "contribution_statement"
82 + },
83 + {
84 + "paragraph_index": 0,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "identification_defense"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "robustness_mention"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "mechanism_or_theory_sketch"
98 + },
99 + {
100 + "paragraph_index": 2,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 3,
105 + "move": "preview_of_approach"
106 + },
107 + {
108 + "paragraph_index": 3,
109 + "move": "preview_of_results"
110 + },
111 + {
112 + "paragraph_index": 3,
113 + "move": "identification_defense"
114 + }
115 + ],
116 + "contribution_paragraph_index": 0,
117 + "contribution_types": [
118 + "new_evidence",
119 + "new_theory_or_model"
120 + ],
121 + "roadmap_present": false,
122 + "literature_positioning_location": "inside_introduction"
123 + },
124 + "section_roles": [
125 + {
126 + "idx": 0,
127 + "role": "introduction"
128 + },
129 + {
130 + "idx": 1,
131 + "role": "theory"
132 + },
133 + {
134 + "idx": 2,
135 + "role": "data"
136 + },
137 + {
138 + "idx": 3,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 4,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 5,
147 + "role": "conclusion"
148 + }
149 + ],
150 + "results_presentation": {
151 + "economic_significance_discussed": true,
152 + "statistical_significance_language": "unclear",
153 + "magnitude_statements_in_intro": true
154 + },
155 + "robustness": {
156 + "location": "within_results",
157 + "types": [
158 + "alternative_specification",
159 + "subsample",
160 + "other"
161 + ]
162 + },
163 + "tables": {
164 + "convention_notes": "Each table note names the estimator (OLS or 2SLS), the dependent variable and its window (2007–2010), and the variable of interest measured over 2004–2006. Notes refer back to Table I for variable definitions and state which fixed effects are included (e.g., county fixed effects). First-stage results are reported in separate columns.",
165 + "units_stated_in_notes": true,
166 + "standard_errors_description": "unclear"
167 + },
168 + "figures": {
169 + "uncertainty_shown": "no",
170 + "figures_note_placeholder_removed": null
171 + },
172 + "conclusion": {
173 + "bounded_claims": false,
174 + "policy_implications": true,
175 + "open_questions": true,
176 + "restates_contribution": true
177 + },
178 + "hedging": {
179 + "level": "moderate",
180 + "typical_devices": [
181 + "'evidence consistent with' phrasing",
182 + "'arguably' qualifiers",
183 + "framing hypotheses as conjectures",
184 + "'may' and 'suggest' in implications"
185 + ]
186 + },
187 + "claim_evidence": [
188 + {
189 + "claim": "Lenders retaining a larger share of local outstanding mortgages are less likely to foreclose on seriously delinquent loans",
190 + "evidence_type": "table",
191 + "evidence_ref": "Table II"
192 + },
193 + {
194 + "claim": "The foreclosure result holds and is quantitatively stronger when retained share is instrumented with bank mergers",
195 + "evidence_type": "table",
196 + "evidence_ref": "Table III"
197 + },
198 + {
199 + "claim": "The retained-share effect is weaker where other lenders retain more, and differs in desirable versus distressed areas",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table IV"
202 + },
203 + {
204 + "claim": "Lenders with higher retained shares are more likely to modify delinquent mortgages",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table V"
207 + },
208 + {
209 + "claim": "Zip codes with higher Top4 concentration have lower foreclosure rates",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table VIII"
212 + },
213 + {
214 + "claim": "A one-SD increase in concentration is associated with about 3% smaller house price declines",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table IX"
217 + },
218 + {
219 + "claim": "Lenders' foreclosure decisions are strategic substitutes and respond to own retained share",
220 + "evidence_type": "equation_or_proof",
221 + "evidence_ref": "Section I.B"
222 + }
223 + ],
224 + "structural_summary": "The paper opens with the fire-sale literature and notes that lenders' ex post incentives have received little attention. It states its conjecture and gives reasons why the housing crisis is a good testing ground. A simple model then derives testable implications about lender concentration, foreclosures and prices. Loan-level LPM and IV tests follow, along with cross-sectional heterogeneity and mechanism tests (modifications, alternative channels). The paper then moves to zip-code-level aggregate tests of foreclosure rates and house prices using the merger instrument. It closes with policy implications for consolidation and bad banks, and calls for future research on other collateralized markets.",
225 + "writing_profile_keywords": [
226 + "fire-sale externalities",
227 + "model-motivated empirics",
228 + "instrumental variables with bank mergers",
229 + "loan-level and aggregate tests",
230 + "policy implications",
231 + "cautious causal language"
232 + ],
233 + "confidence": 0.8,
234 + "notes": "Standard-error type and significance-reporting conventions are not visible in the excerpts. The introduction has no visible roadmap paragraph. The literature is positioned within the introduction, and no separate literature section is seen."
235 + },
236 + "deterministic_checks": {
237 + "roadmap_regex": false,
238 + "contribution_regex_in_intro": false,
239 + "magnitude_in_abstract": false,
240 + "abstract_n_sentences": 6,
241 + "intro_n_paragraphs": 4,
242 + "intro_n_words": 761,
243 + "causal_words_in_abstract": 0,
244 + "hedge_words_in_abstract": 1,
245 + "stars_mentioned": false,
246 + "clustered_se_mentioned": false,
247 + "internet_appendix_mentioned": false,
248 + "section_roles_rule": {
249 + "0": "introduction",
250 + "1": "theory",
251 + "2": "data",
252 + "3": "other",
253 + "4": "other",
254 + "5": "conclusion"
255 + }
256 + },
257 + "schema_valid": false,
258 + "code_version": "feb6c9e-dirty"
259 +}
\ No newline at end of file
added data/annotations/documents/jof_12495.json +269 −0
@@ -0,0 +1,269 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12495",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:40+00:00",
9 + "source_sha256": "b754c860465f5a261346c3d9eb0d92277221061c4156ff2c08f5af8068255ffb",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22728,
12 + "usage": {
13 + "input": 7343,
14 + "output": 2140,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.072966,
19 + "request_id": "req_011CggpcETKLpNq2RhMDyQ4L",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "entrepreneurial_finance_private_equity",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Elasticity of net (and gross) employment creation in the nontradable sector, by firm age category, to local income growth at the commuting-zone level",
27 + "strategy": "Bartik-type shift-share instrument: national manufacturing changes interacted with each commuting zone's manufacturing composition instrument local income growth, which shifts local nontradable investment opportunities",
28 + "key_assumptions": [
29 + "Local manufacturing income changes driven by national shocks are unrelated to the region's own performance (exclusion)",
30 + "The nontradable sector responds only to local demand, so the effect operates through local income"
31 + ],
32 + "threats_addressed": [
33 + "Organizational arrangements specific to the nontradable sector such as franchising (construction sector check)",
34 + "Housing collateral and real estate value channels",
35 + "Local banking and deposit channels",
36 + "Short-lived jobs reflecting entrant overreaction",
37 + "Alternative shock measure (import penetration)",
38 + "Alternative geographic units, state-by-year fixed effects, alternative measurement periods",
39 + "Firm size confounding firm age"
40 + ],
41 + "causal_language_strength": 4,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 1
48 + },
49 + "gap": {
50 + "present": true,
51 + "sentence_index": 0
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": false,
67 + "sentence_index": -1
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "literature_gap"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "motivation"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "preview_of_approach"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "mechanism_or_theory_sketch"
103 + },
104 + {
105 + "paragraph_index": 5,
106 + "move": "robustness_mention"
107 + },
108 + {
109 + "paragraph_index": 5,
110 + "move": "related_literature_positioning"
111 + },
112 + {
113 + "paragraph_index": 6,
114 + "move": "roadmap"
115 + }
116 + ],
117 + "contribution_paragraph_index": 1,
118 + "contribution_types": [
119 + "new_evidence",
120 + "new_identification"
121 + ],
122 + "roadmap_present": true,
123 + "literature_positioning_location": "inside_introduction"
124 + },
125 + "section_roles": [
126 + {
127 + "idx": 0,
128 + "role": "introduction"
129 + },
130 + {
131 + "idx": 1,
132 + "role": "data"
133 + },
134 + {
135 + "idx": 2,
136 + "role": "results"
137 + },
138 + {
139 + "idx": 3,
140 + "role": "results"
141 + },
142 + {
143 + "idx": 4,
144 + "role": "results"
145 + },
146 + {
147 + "idx": 5,
148 + "role": "conclusion"
149 + }
150 + ],
151 + "results_presentation": {
152 + "economic_significance_discussed": true,
153 + "statistical_significance_language": "unclear",
154 + "magnitude_statements_in_intro": false
155 + },
156 + "robustness": {
157 + "location": "mixed",
158 + "types": [
159 + "alternative_measure",
160 + "alternative_specification",
161 + "subsample",
162 + "alternative_sample_period",
163 + "other"
164 + ]
165 + },
166 + "tables": {
167 + "convention_notes": "Notes give data source (QWI/LEHD, BDS), sector NAICS codes, unit of observation (CZ-year-firm age), dependent variable definition and horizon, and say controls are the same as earlier tables. Regressions are reported separately by firm age category, and descriptive tables come first.",
168 + "units_stated_in_notes": true,
169 + "standard_errors_description": "unclear"
170 + },
171 + "figures": {
172 + "uncertainty_shown": "no"
173 + },
174 + "conclusion": {
175 + "bounded_claims": true,
176 + "policy_implications": false,
177 + "open_questions": true,
178 + "restates_contribution": true
179 + },
180 + "hedging": {
181 + "level": "moderate",
182 + "typical_devices": [
183 + "suggesting",
184 + "consistent with",
185 + "beyond the scope of any single paper",
186 + "one specific channel framing",
187 + "presumably"
188 + ]
189 + },
190 + "claim_evidence": [
191 + {
192 + "claim": "New firms account for most net employment creation in response to local income shocks in the nontradable sector",
193 + "evidence_type": "table",
194 + "evidence_ref": "Table III, Table IV"
195 + },
196 + {
197 + "claim": "Existing young and old firms do not differ significantly in their elasticity to income shocks",
198 + "evidence_type": "table",
199 + "evidence_ref": "Table III"
200 + },
201 + {
202 + "claim": "Positive shocks raise both gross job creation and destruction among existing firms, increasing churn",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table V"
205 + },
206 + {
207 + "claim": "Results hold in the construction sector, where franchising is unimportant",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table VI"
210 + },
211 + {
212 + "claim": "Results are not driven by house price appreciation or collateral channels",
213 + "evidence_type": "table",
214 + "evidence_ref": "Table VIII"
215 + },
216 + {
217 + "claim": "Startup jobs created by shocks are not especially short-lived",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table IX"
220 + },
221 + {
222 + "claim": "Local bank access does not explain startup responsiveness",
223 + "evidence_type": "table",
224 + "evidence_ref": "Table X, Table XI"
225 + },
226 + {
227 + "claim": "Net job creation among older firms is concentrated in large firms, so the channel operates through age rather than size",
228 + "evidence_type": "table",
229 + "evidence_ref": "Table XII"
230 + }
231 + ],
232 + "structural_summary": "The introduction motivates the paper with the role of startups in job creation and the unclear mechanism behind it, then argues for comparing firms by age rather than size. It previews a Bartik-instrumented local demand design and states the main findings: startups drive net creation, existing firms of different ages do not differ, gross flows reveal churn, and age matters rather than size. Robustness and the literature follow, then a roadmap. The data and methodology section precedes a long results section with many subsections on short- and medium-term effects, gross flows, the construction sector, asymmetry, housing collateral and job durability. Separate sections then cover access to capital and size versus age. The conclusion restates the thought experiment, links the findings to the spawning literature and leaves the deeper reasons open.",
233 + "writing_profile_keywords": [
234 + "Bartik instrument",
235 + "firm age",
236 + "job creation",
237 + "nontradable sector",
238 + "commuting zones",
239 + "churn",
240 + "mechanism exploration",
241 + "q-theory framing"
242 + ],
243 + "confidence": 0.8,
244 + "notes": "The abstract text includes a stray first sentence of the introduction. The excerpts do not show standard error treatment or significance notation."
245 + },
246 + "deterministic_checks": {
247 + "roadmap_regex": true,
248 + "contribution_regex_in_intro": true,
249 + "magnitude_in_abstract": false,
250 + "abstract_n_sentences": 5,
251 + "intro_n_paragraphs": 7,
252 + "intro_n_words": 1049,
253 + "causal_words_in_abstract": 0,
254 + "hedge_words_in_abstract": 0,
255 + "stars_mentioned": false,
256 + "clustered_se_mentioned": false,
257 + "internet_appendix_mentioned": true,
258 + "section_roles_rule": {
259 + "0": "introduction",
260 + "1": "data",
261 + "2": "other",
262 + "3": "other",
263 + "4": "other",
264 + "5": "conclusion"
265 + }
266 + },
267 + "schema_valid": true,
268 + "code_version": "feb6c9e-dirty"
269 +}
\ No newline at end of file
added data/annotations/documents/jof_12496.json +258 −0
@@ -0,0 +1,258 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12496",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:11+00:00",
9 + "source_sha256": "650ab98d134ddd0d010f84f0fab453ecc4374ecff97fd68b9936955804262275",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22294,
12 + "usage": {
13 + "input": 8070,
14 + "output": 2001,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.073094,
19 + "request_id": "req_011CggpaGDvrU3R5jfFgrJGt",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "fixed_income_mbs_liquidity",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Effect of TBA eligibility and of parallel TBA trading volume on SP trading costs; dealer hedge ratios of SP inventory with TBA trades",
27 + "strategy": "Trading-cost regressions on consecutive trades; regression discontinuity at the 1.05 LTV eligibility threshold; propensity score matching on the estimated probability of eligibility; predictable dollar-roll volume before settlement used as exogenous variation in TBA volume; dealer-level inventory regressions",
28 + "key_assumptions": [
29 + "Trading costs vary smoothly in LTV except for the eligibility threshold",
30 + "Observable pool characteristics capture selection into eligibility",
31 + "Predicted dollar roll volume is exogenous to SP liquidity"
32 + ],
33 + "threats_addressed": [
34 + "Characteristics associated with TBA eligibility, rather than eligibility itself, driving liquidity",
35 + "Endogeneity of TBA trading volume"
36 + ],
37 + "causal_language_strength": 3,
38 + "uses_causal_claim": true
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": false,
43 + "sentence_index": -1
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 2
56 + },
57 + "magnitude": {
58 + "present": false,
59 + "sentence_index": -1
60 + },
61 + "contribution": {
62 + "present": true,
63 + "sentence_index": 3
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "tension_or_puzzle"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "mechanism_or_theory_sketch"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "related_literature_positioning"
83 + },
84 + {
85 + "paragraph_index": 2,
86 + "move": "contribution_statement"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "preview_of_approach"
91 + },
92 + {
93 + "paragraph_index": 2,
94 + "move": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "policy_or_broader_implication"
103 + },
104 + {
105 + "paragraph_index": 5,
106 + "move": "roadmap"
107 + }
108 + ],
109 + "contribution_paragraph_index": 2,
110 + "contribution_types": [
111 + "new_evidence"
112 + ],
113 + "roadmap_present": true,
114 + "literature_positioning_location": "inside_introduction"
115 + },
116 + "section_roles": [
117 + {
118 + "idx": 0,
119 + "role": "introduction"
120 + },
121 + {
122 + "idx": 1,
123 + "role": "other"
124 + },
125 + {
126 + "idx": 2,
127 + "role": "data"
128 + },
129 + {
130 + "idx": 3,
131 + "role": "results"
132 + },
133 + {
134 + "idx": 4,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 5,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 6,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 7,
147 + "role": "conclusion"
148 + }
149 + ],
150 + "results_presentation": {
151 + "economic_significance_discussed": true,
152 + "statistical_significance_language": "unclear",
153 + "magnitude_statements_in_intro": false
154 + },
155 + "robustness": {
156 + "location": "none_seen",
157 + "types": []
158 + },
159 + "tables": {
160 + "convention_notes": "Notes specify the regression equation, sample construction and variable definitions; some tables report coefficients with z-statistics and marginal effects.",
161 + "units_stated_in_notes": true,
162 + "standard_errors_description": "unclear"
163 + },
164 + "figures": {
165 + "uncertainty_shown": "unclear"
166 + },
167 + "conclusion": {
168 + "bounded_claims": true,
169 + "policy_implications": true,
170 + "open_questions": false,
171 + "restates_contribution": true
172 + },
173 + "hedging": {
174 + "level": "moderate",
175 + "typical_devices": [
176 + "suggest",
177 + "appears to",
178 + "may",
179 + "consistent with",
180 + "one (but not necessarily the only) way"
181 + ]
182 + },
183 + "claim_evidence": [
184 + {
185 + "claim": "TBA-eligible SPs have significantly lower trading costs than ineligible SPs",
186 + "evidence_type": "table",
187 + "evidence_ref": "Table II"
188 + },
189 + {
190 + "claim": "Trading costs jump when the LTV eligibility threshold is crossed, supporting an effect of eligibility itself",
191 + "evidence_type": "table",
192 + "evidence_ref": "Table III; Figure 3"
193 + },
194 + {
195 + "claim": "Eligibility lowers costs after matching on estimated eligibility probability",
196 + "evidence_type": "table",
197 + "evidence_ref": "Tables IV-V"
198 + },
199 + {
200 + "claim": "SP trading costs fall on days of predictably high TBA dollar roll volume",
201 + "evidence_type": "table",
202 + "evidence_ref": "Table VI; Figure 4"
203 + },
204 + {
205 + "claim": "Dealers hedge SP inventory changes with offsetting TBA trades, less so for ineligible SPs",
206 + "evidence_type": "table",
207 + "evidence_ref": "Table VII"
208 + },
209 + {
210 + "claim": "TBA trades hedge price changes of eligible SPs better than those of ineligible SPs",
211 + "evidence_type": "table",
212 + "evidence_ref": "Table VIII"
213 + },
214 + {
215 + "claim": "Prearranged trades are more common for hard-to-hedge SPs",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table IX"
218 + }
219 + ],
220 + "structural_summary": "The paper opens with the puzzle of high liquidity in a market of unique securities and proposes three channels through which TBA trading creates liquidity. It positions itself against prior trading-cost studies and previews the hedging result and a policy implication for bond markets. After institutional background and data sections, the results proceed in steps: trading-cost benchmarks, then the causal effect of eligibility using RD and matching, then spillovers using predictable dollar-roll volume, then evidence on dealer hedging and prearranged trades. The conclusion restates findings and proposes forward markets for municipal and corporate bonds.",
221 + "writing_profile_keywords": [
222 + "institutional detail",
223 + "OTC liquidity",
224 + "TRACE data",
225 + "regression discontinuity",
226 + "propensity score matching",
227 + "dealer hedging",
228 + "policy extension"
229 + ],
230 + "confidence": 0.8,
231 + "notes": "Robustness checks are not visible in the excerpts. Standard-error treatment is not described in the excerpts. The roadmap's description of Sections IV and V differs slightly from the section headings."
232 + },
233 + "deterministic_checks": {
234 + "roadmap_regex": true,
235 + "contribution_regex_in_intro": true,
236 + "magnitude_in_abstract": false,
237 + "abstract_n_sentences": 5,
238 + "intro_n_paragraphs": 6,
239 + "intro_n_words": 800,
240 + "causal_words_in_abstract": 0,
241 + "hedge_words_in_abstract": 1,
242 + "stars_mentioned": false,
243 + "clustered_se_mentioned": false,
244 + "internet_appendix_mentioned": false,
245 + "section_roles_rule": {
246 + "0": "introduction",
247 + "1": "other",
248 + "2": "data",
249 + "3": "other",
250 + "4": "other",
251 + "5": "other",
252 + "6": "other",
253 + "7": "conclusion"
254 + }
255 + },
256 + "schema_valid": true,
257 + "code_version": "feb6c9e-dirty"
258 +}
\ No newline at end of file
added data/annotations/documents/jof_12497.json +264 −0
@@ -0,0 +1,264 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12497",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:27+00:00",
9 + "source_sha256": "636dd198ce2f7c364fcf5cf643e8aabb6643bd45467ac431bb56bbf94f591e38",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20925,
12 + "usage": {
13 + "input": 7084,
14 + "output": 2132,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07177,
19 + "request_id": "req_011Cggpfr2oUQttk2HfTG2E9",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "international_finance",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of capital account openness on within-sector dispersion in the marginal revenue product of capital, differentially by sector external financial dependence, and implied aggregate TFP gains",
27 + "strategy": "Cross-country variation in the timing of liberalization (Chinn-Ito index, lagged), interacted with Rajan-Zingales sectoral external financial dependence, with country-year, country-sector and sector-year fixed effects. Aggregate country-year panel with country and year fixed effects. A misallocation model maps estimates into TFP.",
28 + "key_assumptions": [
29 + "Liberalization timing is driven largely by EU accession rather than sector-specific shocks",
30 + "Rajan-Zingales US-based financial dependence captures the technological need for external finance in these countries",
31 + "Dispersion in MRPK reflects misallocation from financial frictions"
32 + ],
33 + "threats_addressed": [
34 + "Concurrent institutional changes triggered by liberalization (alternative channels)",
35 + "Measurement of openness (alternative de jure and de facto measures)",
36 + "Political connections and corruption",
37 + "Pre-trends and dynamics (event-study dynamic effects)"
38 + ],
39 + "causal_language_strength": 4,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": true,
61 + "sentence_index": 3
62 + },
63 + "contribution": {
64 + "present": false,
65 + "sentence_index": -1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "preview_of_approach"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "identification_defense"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "robustness_mention"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "contribution_statement"
101 + },
102 + {
103 + "paragraph_index": 3,
104 + "move": "related_literature_positioning"
105 + },
106 + {
107 + "paragraph_index": 4,
108 + "move": "related_literature_positioning"
109 + }
110 + ],
111 + "contribution_paragraph_index": 3,
112 + "contribution_types": [
113 + "new_evidence"
114 + ],
115 + "roadmap_present": false,
116 + "literature_positioning_location": "inside_introduction"
117 + },
118 + "section_roles": [
119 + {
120 + "idx": 0,
121 + "role": "introduction"
122 + },
123 + {
124 + "idx": 1,
125 + "role": "results"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "theory"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "robustness"
138 + },
139 + {
140 + "idx": 5,
141 + "role": "robustness"
142 + },
143 + {
144 + "idx": 6,
145 + "role": "conclusion"
146 + }
147 + ],
148 + "results_presentation": {
149 + "economic_significance_discussed": true,
150 + "statistical_significance_language": "unclear",
151 + "magnitude_statements_in_intro": true
152 + },
153 + "robustness": {
154 + "location": "mixed",
155 + "types": [
156 + "alternative_measure",
157 + "alternative_specification",
158 + "subsample",
159 + "other"
160 + ]
161 + },
162 + "tables": {
163 + "convention_notes": "Each table note opens by writing out the full regression equation, with its fixed-effect structure and variable definitions. The aggregation table explains how its computation follows from a model equation. Excerpts are truncated, so standard-error and significance conventions are not visible.",
164 + "units_stated_in_notes": false,
165 + "standard_errors_description": "unclear"
166 + },
167 + "figures": {
168 + "uncertainty_shown": "unclear"
169 + },
170 + "conclusion": {
171 + "bounded_claims": false,
172 + "policy_implications": false,
173 + "open_questions": false,
174 + "restates_contribution": true
175 + },
176 + "hedging": {
177 + "level": "moderate",
178 + "typical_devices": [
179 + "'is associated with' for aggregate results",
180 + "'evidence consistent with the argument'",
181 + "'suggestive evidence'",
182 + "'we argue'"
183 + ]
184 + },
185 + "claim_evidence": [
186 + {
187 + "claim": "Liberalization raises capital inflows and private credit relative to GDP, narrows interest spreads, and is associated with higher aggregate TFP",
188 + "evidence_type": "table",
189 + "evidence_ref": "Table I"
190 + },
191 + {
192 + "claim": "Liberalization reduces MRPK dispersion more in sectors more dependent on external finance",
193 + "evidence_type": "table",
194 + "evidence_ref": "Table III"
195 + },
196 + {
197 + "claim": "Results survive controls for alternative institutional channels",
198 + "evidence_type": "table",
199 + "evidence_ref": "Table IV"
200 + },
201 + {
202 + "claim": "Young firms in financially dependent sectors expand most after liberalization",
203 + "evidence_type": "table",
204 + "evidence_ref": "Tables V-VI"
205 + },
206 + {
207 + "claim": "Decline in dispersion stems from the age-explained component of MRPK",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table VII"
210 + },
211 + {
212 + "claim": "Results are robust to alternative openness measures",
213 + "evidence_type": "table",
214 + "evidence_ref": "Table VIII"
215 + },
216 + {
217 + "claim": "Improved capital allocation raises aggregate TFP by 10-16%",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table IX; equation (15)"
220 + },
221 + {
222 + "claim": "Dynamic effects of liberalization on MRPK dispersion",
223 + "evidence_type": "figure",
224 + "evidence_ref": "Figure 3"
225 + }
226 + ],
227 + "structural_summary": "The paper motivates the question with cross-country growth evidence. It first presents aggregate panel evidence linking liberalization to financial development and TFP. It then builds a heterogeneous-firm misallocation model with bank-dependent firms. Firm-level tests follow, using a difference-in-differences design that interacts openness with sectoral financial dependence. Mechanism tests address alternative channels and firm age. Robustness checks cover corruption and alternative openness measures. The model then aggregates the reduced-form estimates into TFP gains, and the conclusion restates the contributions.",
228 + "writing_profile_keywords": [
229 + "cross-country firm-level panel",
230 + "misallocation",
231 + "Rajan-Zingales interaction",
232 + "model-guided empirics",
233 + "aggregation to TFP",
234 + "mechanism tests",
235 + "first-to claim"
236 + ],
237 + "confidence": 0.8,
238 + "notes": "The model section appears after the aggregate evidence rather than before it. Standard-error conventions are not visible in the excerpts. The introduction is truncated, and no roadmap is visible in it."
239 + },
240 + "deterministic_checks": {
241 + "roadmap_regex": false,
242 + "contribution_regex_in_intro": true,
243 + "magnitude_in_abstract": true,
244 + "abstract_n_sentences": 4,
245 + "intro_n_paragraphs": 5,
246 + "intro_n_words": 831,
247 + "causal_words_in_abstract": 0,
248 + "hedge_words_in_abstract": 0,
249 + "stars_mentioned": false,
250 + "clustered_se_mentioned": false,
251 + "internet_appendix_mentioned": false,
252 + "section_roles_rule": {
253 + "0": "introduction",
254 + "1": "results",
255 + "2": "theory",
256 + "3": "results",
257 + "4": "robustness",
258 + "5": "robustness",
259 + "6": "conclusion"
260 + }
261 + },
262 + "schema_valid": true,
263 + "code_version": "feb6c9e-dirty"
264 +}
\ No newline at end of file
added data/annotations/documents/jof_12498.json +228 −0
@@ -0,0 +1,228 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12498",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:33+00:00",
9 + "source_sha256": "aff1d3fb5dcc6454808b3f4b46c378903ad890d944e003ecbd3d10c320f2f2b6",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 16996,
12 + "usage": {
13 + "input": 6018,
14 + "output": 1903,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.062926,
19 + "request_id": "req_011CggpbvDTrBjcqYfxnxnAv",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "high-frequency trading / market intermediation",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "How second-by-second inventory changes of intraday intermediaries (HFTs, Market Makers) comove with contemporaneous and lagged price changes, and whether that relationship shifts during the Flash Crash.",
27 + "strategy": "Classify accounts into trader categories using audit-trail data. Regress inventory changes on contemporaneous and 20 lagged price changes, with dummy interactions for the crash down and up periods, comparing May 6 with May 3–5. Complemented by an event-type analysis of trading-volume shares around price-level changes.",
28 + "key_assumptions": [
29 + "The account classification correctly identifies intermediaries and HFTs.",
30 + "May 3–5 serve as a valid baseline for normal-day behavior."
31 + ],
32 + "threats_addressed": [
33 + "Change in behavior during the crash is tested with interaction dummies."
34 + ],
35 + "causal_language_strength": 2,
36 + "uses_causal_claim": false
37 + },
38 + "abstract_moves": {
39 + "question": {
40 + "present": true,
41 + "sentence_index": 0
42 + },
43 + "gap": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "method": {
48 + "present": true,
49 + "sentence_index": 2
50 + },
51 + "results": {
52 + "present": true,
53 + "sentence_index": 2
54 + },
55 + "magnitude": {
56 + "present": false,
57 + "sentence_index": -1
58 + },
59 + "contribution": {
60 + "present": false,
61 + "sentence_index": -1
62 + }
63 + },
64 + "introduction": {
65 + "moves": [
66 + {
67 + "paragraph_index": 0,
68 + "move": "motivation"
69 + },
70 + {
71 + "paragraph_index": 41,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 42,
76 + "move": "mechanism_or_theory_sketch"
77 + },
78 + {
79 + "paragraph_index": 43,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 44,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 44,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 45,
92 + "move": "related_literature_positioning"
93 + },
94 + {
95 + "paragraph_index": 45,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 45,
100 + "move": "contribution_statement"
101 + },
102 + {
103 + "paragraph_index": 46,
104 + "move": "roadmap"
105 + }
106 + ],
107 + "contribution_paragraph_index": 45,
108 + "contribution_types": [
109 + "new_evidence"
110 + ],
111 + "roadmap_present": true,
112 + "literature_positioning_location": "inside_introduction"
113 + },
114 + "section_roles": [
115 + {
116 + "idx": 0,
117 + "role": "introduction"
118 + },
119 + {
120 + "idx": 1,
121 + "role": "data"
122 + },
123 + {
124 + "idx": 2,
125 + "role": "results"
126 + },
127 + {
128 + "idx": 3,
129 + "role": "conclusion"
130 + }
131 + ],
132 + "results_presentation": {
133 + "economic_significance_discussed": true,
134 + "statistical_significance_language": "unclear",
135 + "magnitude_statements_in_intro": true
136 + },
137 + "robustness": {
138 + "location": "none_seen",
139 + "types": []
140 + },
141 + "tables": {
142 + "convention_notes": "Table notes state the regression equation explicitly, give units (contracts, ticks), define the sample windows (May 3–5 versus May 6, 8:30–15:15 CT), and report descriptive panels by trader category.",
143 + "units_stated_in_notes": true,
144 + "standard_errors_description": "unclear"
145 + },
146 + "figures": {
147 + "uncertainty_shown": "unclear"
148 + },
149 + "conclusion": {
150 + "bounded_claims": true,
151 + "policy_implications": true,
152 + "open_questions": false,
153 + "restates_contribution": true
154 + },
155 + "hedging": {
156 + "level": "moderate",
157 + "typical_devices": [
158 + "'consistent with' theory phrasing",
159 + "'our results suggest'",
160 + "'theory suggests'"
161 + ]
162 + },
163 + "claim_evidence": [
164 + {
165 + "claim": "Combined net inventories of intraday intermediaries stayed far below the size of the large sell program, consistent with limited risk-bearing capacity.",
166 + "evidence_type": "descriptive_statistic",
167 + "evidence_ref": "Table II; Figure 3"
168 + },
169 + {
170 + "claim": "The relationship between HFT inventory changes and price changes did not change during the Flash Crash, whereas Market Makers' relationship did.",
171 + "evidence_type": "table",
172 + "evidence_ref": "Table IV"
173 + },
174 + {
175 + "claim": "HFT inventory changes comove positively with contemporaneous price changes; Market Maker inventory changes comove negatively.",
176 + "evidence_type": "table",
177 + "evidence_ref": "Table III"
178 + },
179 + {
180 + "claim": "Prices rise after seconds in which HFTs are net buyers and remain higher for about 20 seconds.",
181 + "evidence_type": "figure",
182 + "evidence_ref": "Figure 4"
183 + },
184 + {
185 + "claim": "HFTs lift a disproportionate share of the final best-ask depth before price increases, consistent with stale-quote sniping.",
186 + "evidence_type": "table",
187 + "evidence_ref": "Table V"
188 + },
189 + {
190 + "claim": "Flash Crash price and volume dynamics are documented as the motivating event.",
191 + "evidence_type": "figure",
192 + "evidence_ref": "Figure 1"
193 + }
194 + ],
195 + "structural_summary": "Opens with a long quoted regulatory description of the event, then sketches inventory-theory mechanisms (limited risk-bearing capacity, latency arbitrage) and previews findings for each. The data section presents the market and trader classification. A single combined methods-and-results section covers trader categories, inventory–price regressions with crash interactions, and an HFT trading-pattern analysis. A short conclusion restates the findings and extends them to later flash events.",
196 + "writing_profile_keywords": [
197 + "event-driven",
198 + "theory-consistency framing",
199 + "audit-trail data",
200 + "high-frequency regressions",
201 + "regulatory report quotation",
202 + "short conclusion"
203 + ],
204 + "confidence": 0.72,
205 + "notes": "The introduction paragraph list is fragmented by the block quote: indices 1–40 are quote lines and are not annotated as separate moves. The abstract appears truncated and duplicates the opening of the introduction. Neither standard-error type nor robustness checks are visible in the excerpts."
206 + },
207 + "deterministic_checks": {
208 + "roadmap_regex": true,
209 + "contribution_regex_in_intro": true,
210 + "magnitude_in_abstract": false,
211 + "abstract_n_sentences": 6,
212 + "intro_n_paragraphs": 46,
213 + "intro_n_words": 1247,
214 + "causal_words_in_abstract": 0,
215 + "hedge_words_in_abstract": 3,
216 + "stars_mentioned": false,
217 + "clustered_se_mentioned": false,
218 + "internet_appendix_mentioned": false,
219 + "section_roles_rule": {
220 + "0": "introduction",
221 + "1": "literature",
222 + "2": "empirical_design",
223 + "3": "conclusion"
224 + }
225 + },
226 + "schema_valid": true,
227 + "code_version": "feb6c9e-dirty"
228 +}
\ No newline at end of file
added data/annotations/documents/jof_12499.json +300 −0
@@ -0,0 +1,300 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('convention_notes_placeholder_removed' was unexpected) @ figures"
4 + ],
5 + "doc_id": "jof_12499",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:57:21+00:00",
11 + "source_sha256": "4b2f62ba52a66448b39539a0d6864e9f14617c2e3e779fcd7b7fc526baca0750",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 24119,
14 + "usage": {
15 + "input": 8219,
16 + "output": 2639,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.08645,
21 + "request_id": "req_011CggpaZC8nFHh4jFxnj4se",
22 + "annotation": {
23 + "genre": "mixed_theory_empirical",
24 + "subfield": "international_finance",
25 + "subfield_secondary": "asset_pricing; macro_finance (monetary policy announcements)",
26 + "primary_method": "natural_experiment_event_study",
27 + "identification": {
28 + "estimand": "Difference in average currency excess returns (vs. USD) on scheduled FOMC announcement days relative to nonannouncement days, and how this difference varies with interest rate differentials, monetary policy uncertainty, and the sign of the policy shock",
29 + "strategy": "High-frequency event-study comparison of announcement and nonannouncement days. Time-series regressions use an announcement dummy interacted with uncertainty proxies and a monetary policy indicator, with returns split into pre- and postannouncement windows. Results are interpreted through a constrained-financier model.",
30 + "key_assumptions": [
31 + "Scheduled FOMC dates are predetermined, so announcement-day timing is exogenous to currency returns",
32 + "The uncertainty proxies capture market uncertainty about monetary policy",
33 + "The high-frequency monetary policy indicator measures the policy surprise"
34 + ],
35 + "threats_addressed": [
36 + "Unscheduled meetings excluded",
37 + "Outliers handled with a truncated sample",
38 + "Inference checked with bootstrapped standard errors and random sampling",
39 + "Transaction costs",
40 + "Other macroeconomic announcements (payroll, CPI, PPI)",
41 + "Other central bank announcements (Bank of Japan)",
42 + "Alternative measures of uncertainty"
43 + ],
44 + "causal_language_strength": 2,
45 + "uses_causal_claim": false
46 + },
47 + "abstract_moves": {
48 + "question": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "gap": {
53 + "present": false,
54 + "sentence_index": -1
55 + },
56 + "method": {
57 + "present": true,
58 + "sentence_index": 2
59 + },
60 + "results": {
61 + "present": true,
62 + "sentence_index": 0
63 + },
64 + "magnitude": {
65 + "present": false,
66 + "sentence_index": -1
67 + },
68 + "contribution": {
69 + "present": true,
70 + "sentence_index": 2
71 + }
72 + },
73 + "introduction": {
74 + "moves": [
75 + {
76 + "paragraph_index": 0,
77 + "move": "motivation"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "preview_of_results"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 2,
93 + "move": "preview_of_results"
94 + },
95 + {
96 + "paragraph_index": 3,
97 + "move": "mechanism_or_theory_sketch"
98 + },
99 + {
100 + "paragraph_index": 3,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 4,
105 + "move": "preview_of_approach"
106 + },
107 + {
108 + "paragraph_index": 4,
109 + "move": "preview_of_results"
110 + },
111 + {
112 + "paragraph_index": 5,
113 + "move": "preview_of_results"
114 + },
115 + {
116 + "paragraph_index": 5,
117 + "move": "robustness_mention"
118 + },
119 + {
120 + "paragraph_index": 6,
121 + "move": "contribution_statement"
122 + },
123 + {
124 + "paragraph_index": 6,
125 + "move": "related_literature_positioning"
126 + },
127 + {
128 + "paragraph_index": 7,
129 + "move": "related_literature_positioning"
130 + },
131 + {
132 + "paragraph_index": 8,
133 + "move": "related_literature_positioning"
134 + },
135 + {
136 + "paragraph_index": 8,
137 + "move": "roadmap"
138 + }
139 + ],
140 + "contribution_paragraph_index": 6,
141 + "contribution_types": [
142 + "new_evidence",
143 + "new_theory_or_model"
144 + ],
145 + "roadmap_present": true,
146 + "literature_positioning_location": "inside_introduction"
147 + },
148 + "section_roles": [
149 + {
150 + "idx": 0,
151 + "role": "introduction"
152 + },
153 + {
154 + "idx": 1,
155 + "role": "theory"
156 + },
157 + {
158 + "idx": 2,
159 + "role": "data"
160 + },
161 + {
162 + "idx": 3,
163 + "role": "results"
164 + },
165 + {
166 + "idx": 4,
167 + "role": "other"
168 + },
169 + {
170 + "idx": 5,
171 + "role": "other"
172 + },
173 + {
174 + "idx": 6,
175 + "role": "results"
176 + },
177 + {
178 + "idx": 7,
179 + "role": "conclusion"
180 + },
181 + {
182 + "idx": 8,
183 + "role": "appendix"
184 + }
185 + ],
186 + "results_presentation": {
187 + "economic_significance_discussed": true,
188 + "statistical_significance_language": "stars_and_tstats",
189 + "magnitude_statements_in_intro": true
190 + },
191 + "robustness": {
192 + "location": "mixed",
193 + "types": [
194 + "alternative_measure",
195 + "falsification",
196 + "other",
197 + "alternative_specification"
198 + ]
199 + },
200 + "tables": {
201 + "convention_notes": "Table notes define panels (full sample, nonannouncement days, announcement days), the return windows in ET clock times, portfolio labels (pf1–pf3, dol), and regressors such as the announcement dummy and the standardized uncertainty measure (TIV). Summary statistics are reported in daily basis points alongside Sharpe ratios. Regression tables are time-series regressions of portfolio excess returns.",
202 + "units_stated_in_notes": true,
203 + "standard_errors_description": "unclear"
204 + },
205 + "figures": {
206 + "uncertainty_shown": "mixed",
207 + "convention_notes_placeholder_removed": null
208 + },
209 + "conclusion": {
210 + "bounded_claims": true,
211 + "policy_implications": false,
212 + "open_questions": false,
213 + "restates_contribution": true
214 + },
215 + "hedging": {
216 + "level": "moderate",
217 + "typical_devices": [
218 + "'we interpret' framing of results through the model",
219 + "'consistent with / in line with our model'",
220 + "'associated with' wording for the uncertainty link",
221 + "'potentially' in the motivation"
222 + ]
223 + },
224 + "claim_evidence": [
225 + {
226 + "claim": "Short-USD strategy earns larger excess returns on FOMC announcement days than on other days",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table I, Table II, Figure 1"
229 + },
230 + {
231 + "claim": "Announcement-day returns are larger for high interest rate currency portfolios (14.47 vs 1.73 bps)",
232 + "evidence_type": "table",
233 + "evidence_ref": "Table II, Table III, Figure 3"
234 + },
235 + {
236 + "claim": "Higher monetary policy uncertainty is associated with higher preannouncement returns",
237 + "evidence_type": "table",
238 + "evidence_ref": "Table V, Table IX"
239 + },
240 + {
241 + "claim": "Postannouncement returns are higher after monetary easing than after tightening",
242 + "evidence_type": "table",
243 + "evidence_ref": "Table VI, Figure 4"
244 + },
245 + {
246 + "claim": "Conditioning the strategy on the policy stance and interest rate differential raises announcement-day returns from 10.77 to 20.54 bps",
247 + "evidence_type": "table",
248 + "evidence_ref": "Table VII"
249 + },
250 + {
251 + "claim": "Volatility does not rise proportionally on announcement days, so Sharpe ratios are higher",
252 + "evidence_type": "figure",
253 + "evidence_ref": "Figure 2, Table I"
254 + },
255 + {
256 + "claim": "In the model, monetary policy uncertainty raises currency risk premia, more so for high interest rate currencies",
257 + "evidence_type": "equation_or_proof",
258 + "evidence_ref": "Section I, Appendix A"
259 + }
260 + ],
261 + "structural_summary": "The introduction motivates the paper with the importance of FOMC announcements and the size and leverage of currency markets. It then previews a sequence of results with magnitudes: announcement-day excess returns, the interest-differential cross-section, the uncertainty link, the pre/post decomposition, and improved trading strategies. A contribution paragraph contrasts the paper with equity-market announcement work, followed by literature positioning and a roadmap. A Gabaix–Maggiori-style constrained-financier model comes first and generates predictions. Data and empirical sections follow: summary statistics, time-series regressions with announcement dummies interacted with uncertainty and policy-shock measures, trading strategies, other central banks, and an in-text robustness subsection. A brief conclusion restates findings through the model, with proofs in the appendix.",
262 + "writing_profile_keywords": [
263 + "announcement-day returns",
264 + "high-frequency data",
265 + "model-guided empirics",
266 + "risk premium interpretation",
267 + "currency portfolios",
268 + "trading strategy",
269 + "robustness battery"
270 + ],
271 + "confidence": 0.85,
272 + "notes": "Sections idx 4–6 are parsing artifacts (currency labels taken as headings); idx 6 actually holds the time-series analysis, trading strategy, other-central-bank, and robustness subsections. The Internet Appendix holds further robustness checks. The standard error type is not visible in the excerpts. In the figures field, the 'convention_notes_placeholder_removed' key is not in the schema and should be dropped."
273 + },
274 + "deterministic_checks": {
275 + "roadmap_regex": true,
276 + "contribution_regex_in_intro": true,
277 + "magnitude_in_abstract": false,
278 + "abstract_n_sentences": 4,
279 + "intro_n_paragraphs": 9,
280 + "intro_n_words": 1341,
281 + "causal_words_in_abstract": 0,
282 + "hedge_words_in_abstract": 0,
283 + "stars_mentioned": false,
284 + "clustered_se_mentioned": false,
285 + "internet_appendix_mentioned": true,
286 + "section_roles_rule": {
287 + "0": "introduction",
288 + "1": "theory",
289 + "2": "data",
290 + "3": "other",
291 + "4": "other",
292 + "5": "other",
293 + "6": "other",
294 + "7": "conclusion",
295 + "8": "appendix"
296 + }
297 + },
298 + "schema_valid": false,
299 + "code_version": "feb6c9e-dirty"
300 +}
\ No newline at end of file
added data/annotations/documents/jof_12500.json +302 −0
@@ -0,0 +1,302 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12500",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:13+00:00",
9 + "source_sha256": "b755ad9c4d88c62feb7694a9049b791bb43e63b3ee5a647e1ca09bf757181cbf",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20585,
12 + "usage": {
13 + "input": 6950,
14 + "output": 2256,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.073714,
19 + "request_id": "req_011CggpaJcYT4sHsYt5K6xW6",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Causal effect of property value (mortgage debt), holding home equity fixed, and of home equity, holding property value fixed, on the stock share of liquid wealth and on stock participation",
27 + "strategy": "Three IV designs: (1) state house price indices at the current year and at the year of purchase; (2) national house prices interacted with state housing supply elasticity (Saiz 2010); (3) panel changes around home purchase, instrumented with the state price index at purchase and with household fixed effects",
28 + "key_assumptions": [
29 + "State house price variation is uncorrelated with other determinants of portfolio choice once fixed effects are included",
30 + "National price shocks interacted with supply elasticity affect portfolios only through housing",
31 + "Time-varying unobservables do not drive portfolio changes around home purchase"
32 + ],
33 + "threats_addressed": [
34 + "omitted variable bias from local labor market conditions",
35 + "selection bias from endogenous timing of home purchase",
36 + "unobserved labor income correlated with property value"
37 + ],
38 + "causal_language_strength": 4,
39 + "uses_causal_claim": true
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 0
45 + },
46 + "gap": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 1
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 2
57 + },
58 + "magnitude": {
59 + "present": true,
60 + "sentence_index": 3
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 0
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "contribution_statement"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "mechanism_or_theory_sketch"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "identification_defense"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "preview_of_approach"
96 + },
97 + {
98 + "paragraph_index": 3,
99 + "move": "preview_of_results"
100 + },
101 + {
102 + "paragraph_index": 4,
103 + "move": "preview_of_approach"
104 + },
105 + {
106 + "paragraph_index": 4,
107 + "move": "identification_defense"
108 + },
109 + {
110 + "paragraph_index": 4,
111 + "move": "preview_of_results"
112 + },
113 + {
114 + "paragraph_index": 5,
115 + "move": "policy_or_broader_implication"
116 + },
117 + {
118 + "paragraph_index": 6,
119 + "move": "roadmap"
120 + }
121 + ],
122 + "contribution_paragraph_index": 1,
123 + "contribution_types": [
124 + "new_evidence",
125 + "new_identification",
126 + "new_theory_or_model",
127 + "resolution_of_puzzle"
128 + ],
129 + "roadmap_present": true,
130 + "literature_positioning_location": "inside_introduction"
131 + },
132 + "section_roles": [
133 + {
134 + "idx": 0,
135 + "role": "introduction"
136 + },
137 + {
138 + "idx": 1,
139 + "role": "theory"
140 + },
141 + {
142 + "idx": 2,
143 + "role": "data"
144 + },
145 + {
146 + "idx": 3,
147 + "role": "data"
148 + },
149 + {
150 + "idx": 4,
151 + "role": "empirical_design"
152 + },
153 + {
154 + "idx": 5,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 6,
159 + "role": "conclusion"
160 + },
161 + {
162 + "idx": 7,
163 + "role": "appendix"
164 + },
165 + {
166 + "idx": 8,
167 + "role": "appendix"
168 + },
169 + {
170 + "idx": 9,
171 + "role": "appendix"
172 + },
173 + {
174 + "idx": 10,
175 + "role": "appendix"
176 + },
177 + {
178 + "idx": 11,
179 + "role": "appendix"
180 + },
181 + {
182 + "idx": 12,
183 + "role": "appendix"
184 + },
185 + {
186 + "idx": 13,
187 + "role": "appendix"
188 + }
189 + ],
190 + "results_presentation": {
191 + "economic_significance_discussed": true,
192 + "statistical_significance_language": "unclear",
193 + "magnitude_statements_in_intro": true
194 + },
195 + "robustness": {
196 + "location": "mixed",
197 + "types": [
198 + "alternative_specification",
199 + "subsample",
200 + "sensitivity_to_parameters",
201 + "other"
202 + ]
203 + },
204 + "tables": {
205 + "convention_notes": "Table notes list the fixed effects included (age, current year, purchase year, state) and controls such as a 10-piece liquid wealth spline. IV tables report first-stage OLS estimates alongside the second stage. The model table reports simulated optimal stock shares across parameter grids.",
206 + "units_stated_in_notes": false,
207 + "standard_errors_description": "unclear"
208 + },
209 + "figures": {
210 + "uncertainty_shown": "no"
211 + },
212 + "conclusion": {
213 + "bounded_claims": true,
214 + "policy_implications": false,
215 + "open_questions": false,
216 + "restates_contribution": true
217 + },
218 + "hedging": {
219 + "level": "moderate",
220 + "typical_devices": [
221 + "acknowledges each design is not definitive alone",
222 + "stability across designs used as a robustness argument",
223 + "'may have' for broader implications",
224 + "'suggests' for interpretation"
225 + ]
226 + },
227 + "claim_evidence": [
228 + {
229 + "claim": "Holding home equity fixed, higher property value lowers the stock share through illiquidity, risk exposure and lower lifetime wealth",
230 + "evidence_type": "equation_or_proof",
231 + "evidence_ref": "Section I; Appendix A"
232 + },
233 + {
234 + "claim": "A $10,000 rise in property value cuts the stock share by about 4%, while a $10,000 rise in home equity raises it by about 4.8%",
235 + "evidence_type": "table",
236 + "evidence_ref": "Table V"
237 + },
238 + {
239 + "claim": "Elasticities of the stock share are -0.2 with respect to mortgage debt and 0.3 with respect to home equity",
240 + "evidence_type": "table",
241 + "evidence_ref": "Table V"
242 + },
243 + {
244 + "claim": "Buying a house that costs $10,000 more lowers the stock share by about 5.2% in the year after purchase",
245 + "evidence_type": "table",
246 + "evidence_ref": "Table VI"
247 + },
248 + {
249 + "claim": "State house price variation yields similar estimates in the first design",
250 + "evidence_type": "table",
251 + "evidence_ref": "Table IV"
252 + },
253 + {
254 + "claim": "Model calibrations imply sizable stock share responses that vary with mobility and the stock-housing return correlation",
255 + "evidence_type": "table",
256 + "evidence_ref": "Table I"
257 + }
258 + ],
259 + "structural_summary": "The introduction opens with a puzzle: theory predicts that housing crowds out risky assets, yet prior evidence finds no systematic link. The paper resolves this by separating property value from home equity in a stylized two-period model, then tests the model with three successively refined IV designs using SIPP data. The intro previews magnitudes and walks through the identification threats each design addresses. A short conclusion restates the elasticities and argues that stability across designs makes confounding unlikely. Appendices cover proofs, model extensions and numerical methods.",
260 + "writing_profile_keywords": [
261 + "puzzle resolution",
262 + "theory-guided empirics",
263 + "multiple research designs",
264 + "instrumental variables",
265 + "elasticity reporting",
266 + "household portfolios"
267 + ],
268 + "confidence": 0.8,
269 + "notes": "Section 5 is labeled as summary statistics but contains the main results subsections. The excerpts do not show how standard errors or significance are reported. The intro briefly mentions implications for welfare and policy, but the conclusion excerpt does not."
270 + },
271 + "deterministic_checks": {
272 + "roadmap_regex": true,
273 + "contribution_regex_in_intro": true,
274 + "magnitude_in_abstract": true,
275 + "abstract_n_sentences": 4,
276 + "intro_n_paragraphs": 7,
277 + "intro_n_words": 1045,
278 + "causal_words_in_abstract": 1,
279 + "hedge_words_in_abstract": 0,
280 + "stars_mentioned": false,
281 + "clustered_se_mentioned": false,
282 + "internet_appendix_mentioned": false,
283 + "section_roles_rule": {
284 + "0": "introduction",
285 + "1": "theory",
286 + "2": "data",
287 + "3": "data",
288 + "4": "other",
289 + "5": "data",
290 + "6": "conclusion",
291 + "7": "appendix",
292 + "8": "appendix",
293 + "9": "appendix",
294 + "10": "appendix",
295 + "11": "appendix",
296 + "12": "appendix",
297 + "13": "appendix"
298 + }
299 + },
300 + "schema_valid": true,
301 + "code_version": "feb6c9e-dirty"
302 +}
\ No newline at end of file
added data/annotations/documents/jof_12501.json +300 −0
@@ -0,0 +1,300 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12501",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:00+00:00",
9 + "source_sha256": "a5bcb20feb92041e3cb674521d8e69fca008f3eab01e4f7c23ad1819a7b6e7fc",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23766,
12 + "usage": {
13 + "input": 8419,
14 + "output": 2343,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.08133,
19 + "request_id": "req_011CggpdcTRn5zCxq5bG1H3D",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "international_finance",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Multiperiod exposures of interest-rate-sorted currency baskets to identified macroeconomic shocks, and the horizon-dependent prices of those exposures (shock exposure and shock price elasticities).",
27 + "strategy": "A structural VAR with stochastic variance for U.S. consumption growth, inflation, the nominal rate and basket exchange-rate growth, with cross-equation restrictions from recursive preferences. Structural shocks are identified through zero restrictions on the impact matrix. The model appears to be estimated with Bayesian methods, given that credible intervals are reported.",
28 + "key_assumptions": [
29 + "A representative U.S. agent with recursive preferences prices currency returns.",
30 + "The zero restrictions on the impact matrix correctly identify the structural shocks.",
31 + "Macro and exchange-rate dynamics follow a VAR with a single common stochastic variance factor.",
32 + "The foreign pricing kernel need not be modeled, so market completeness is not assumed."
33 + ],
34 + "threats_addressed": [
35 + "Comparison of unrestricted and restricted exposure elasticities.",
36 + "Statistical insignificance of risk premia at long horizons is acknowledged."
37 + ],
38 + "causal_language_strength": 2,
39 + "uses_causal_claim": false
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 0
45 + },
46 + "gap": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 0
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 1
57 + },
58 + "magnitude": {
59 + "present": false,
60 + "sentence_index": -1
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 4
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "preview_of_results"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "preview_of_results"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "preview_of_approach"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 4,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 4,
99 + "move": "related_literature_positioning"
100 + },
101 + {
102 + "paragraph_index": 5,
103 + "move": "preview_of_results"
104 + },
105 + {
106 + "paragraph_index": 5,
107 + "move": "policy_or_broader_implication"
108 + },
109 + {
110 + "paragraph_index": 6,
111 + "move": "preview_of_results"
112 + },
113 + {
114 + "paragraph_index": 6,
115 + "move": "related_literature_positioning"
116 + },
117 + {
118 + "paragraph_index": 7,
119 + "move": "literature_gap"
120 + },
121 + {
122 + "paragraph_index": 8,
123 + "move": "related_literature_positioning"
124 + },
125 + {
126 + "paragraph_index": 9,
127 + "move": "contribution_statement"
128 + },
129 + {
130 + "paragraph_index": 10,
131 + "move": "related_literature_positioning"
132 + },
133 + {
134 + "paragraph_index": 10,
135 + "move": "roadmap"
136 + }
137 + ],
138 + "contribution_paragraph_index": 9,
139 + "contribution_types": [
140 + "new_evidence",
141 + "new_theory_or_model"
142 + ],
143 + "roadmap_present": true,
144 + "literature_positioning_location": "inside_introduction"
145 + },
146 + "section_roles": [
147 + {
148 + "idx": 0,
149 + "role": "introduction"
150 + },
151 + {
152 + "idx": 1,
153 + "role": "results"
154 + },
155 + {
156 + "idx": 2,
157 + "role": "theory"
158 + },
159 + {
160 + "idx": 3,
161 + "role": "results"
162 + },
163 + {
164 + "idx": 4,
165 + "role": "conclusion"
166 + },
167 + {
168 + "idx": 5,
169 + "role": "appendix"
170 + },
171 + {
172 + "idx": 6,
173 + "role": "appendix"
174 + },
175 + {
176 + "idx": 7,
177 + "role": "appendix"
178 + },
179 + {
180 + "idx": 8,
181 + "role": "appendix"
182 + },
183 + {
184 + "idx": 9,
185 + "role": "appendix"
186 + },
187 + {
188 + "idx": 10,
189 + "role": "appendix"
190 + }
191 + ],
192 + "results_presentation": {
193 + "economic_significance_discussed": true,
194 + "statistical_significance_language": "confidence_intervals",
195 + "magnitude_statements_in_intro": true
196 + },
197 + "robustness": {
198 + "location": "appendix",
199 + "types": [
200 + "alternative_specification"
201 + ]
202 + },
203 + "tables": {
204 + "convention_notes": "Notes state model equations and identifying restrictions, data frequency and sample period. Uncertainty is reported as 95% credible intervals in parentheses rather than stars.",
205 + "units_stated_in_notes": true,
206 + "standard_errors_description": "none"
207 + },
208 + "figures": {
209 + "uncertainty_shown": "yes"
210 + },
211 + "conclusion": {
212 + "bounded_claims": true,
213 + "policy_implications": false,
214 + "open_questions": false,
215 + "restates_contribution": true
216 + },
217 + "hedging": {
218 + "level": "moderate",
219 + "typical_devices": [
220 + "'suggests' and 'qualitatively similar'",
221 + "explicitly noting statistical insignificance at long horizons",
222 + "deferring the economic origins of the shock to future research",
223 + "describing the study as complementary to existing work"
224 + ]
225 + },
226 + "claim_evidence": [
227 + {
228 + "claim": "The term structure of average carry log returns slopes slightly downward, from about 1.13% to 0.80% per quarter.",
229 + "evidence_type": "figure",
230 + "evidence_ref": "Figure 1"
231 + },
232 + {
233 + "claim": "Higher yield currencies have larger multiperiod exposures to interest rate and inflation shocks.",
234 + "evidence_type": "figure",
235 + "evidence_ref": "Figures 2-3; Table III"
236 + },
237 + {
238 + "claim": "Prices of the interest rate, inflation and consumption growth risks are positive, with the interest rate risk priced highest.",
239 + "evidence_type": "table",
240 + "evidence_ref": "Table VI; Figure 4"
241 + },
242 + {
243 + "claim": "The interest rate shock affects consumption with a lag and behaves like Bansal-Yaron long-run risk.",
244 + "evidence_type": "figure",
245 + "evidence_ref": "Figure 5"
246 + },
247 + {
248 + "claim": "Three-month currency risk premia decompose into contributions from the separate macro shocks.",
249 + "evidence_type": "table",
250 + "evidence_ref": "Table IV"
251 + },
252 + {
253 + "claim": "The variance shock is the sole driver of time variation in risk premia.",
254 + "evidence_type": "figure",
255 + "evidence_ref": "Figure D1"
256 + }
257 + ],
258 + "structural_summary": "Opens with a new stylized fact (a downward-sloping carry return term structure) and asks which macro risks it compensates. The introduction previews findings shock by shock, then positions the paper against term-structure and long-run risk work and lists facts future models must match. Section I sorts currencies into baskets and estimates a reduced-form VAR with exposure elasticities. Section II builds a recursive-preference pricing kernel and describes estimation. Section III reports exposure and price elasticities and links the interest rate shock to long-run risk. A short conclusion restates the findings, and appendices cover data, model solution and parameter estimates.",
259 + "writing_profile_keywords": [
260 + "first-person singular",
261 + "shock elasticities",
262 + "term structure",
263 + "structural VAR",
264 + "recursive preferences",
265 + "credible intervals",
266 + "stylized-fact motivation",
267 + "complementarity framing"
268 + ],
269 + "confidence": 0.8,
270 + "notes": "Section I mixes reduced-form evidence with model setup and is labeled results. The 'credible intervals' suggest Bayesian estimation. Robustness is limited to the unrestricted-versus-restricted comparison in Figure C1."
271 + },
272 + "deterministic_checks": {
273 + "roadmap_regex": true,
274 + "contribution_regex_in_intro": true,
275 + "magnitude_in_abstract": false,
276 + "abstract_n_sentences": 6,
277 + "intro_n_paragraphs": 11,
278 + "intro_n_words": 1501,
279 + "causal_words_in_abstract": 0,
280 + "hedge_words_in_abstract": 0,
281 + "stars_mentioned": false,
282 + "clustered_se_mentioned": false,
283 + "internet_appendix_mentioned": false,
284 + "section_roles_rule": {
285 + "0": "introduction",
286 + "1": "other",
287 + "2": "theory",
288 + "3": "results",
289 + "4": "conclusion",
290 + "5": "appendix",
291 + "6": "appendix",
292 + "7": "appendix",
293 + "8": "appendix",
294 + "9": "appendix",
295 + "10": "appendix"
296 + }
297 + },
298 + "schema_valid": true,
299 + "code_version": "feb6c9e-dirty"
300 +}
\ No newline at end of file
added data/annotations/documents/jof_12504.json +51 −0
@@ -0,0 +1,51 @@
1 +{
2 + "validation_errors": [
3 + "no json"
4 + ],
5 + "doc_id": "jof_12504",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:58:59+00:00",
11 + "source_sha256": "52d5b46ca492ea351913566bc1147928b95420efc304b364cee587e47e003f1e",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 26051,
14 + "usage": {
15 + "input": 8938,
16 + "output": 2405,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.084646,
21 + "request_id": "req_011Cggpi1Y7G1FaQpkW64PC5",
22 + "annotation": null,
23 + "deterministic_checks": {
24 + "roadmap_regex": true,
25 + "contribution_regex_in_intro": true,
26 + "magnitude_in_abstract": true,
27 + "abstract_n_sentences": 5,
28 + "intro_n_paragraphs": 11,
29 + "intro_n_words": 1440,
30 + "causal_words_in_abstract": 0,
31 + "hedge_words_in_abstract": 0,
32 + "stars_mentioned": false,
33 + "clustered_se_mentioned": false,
34 + "internet_appendix_mentioned": true,
35 + "section_roles_rule": {
36 + "0": "introduction",
37 + "1": "results",
38 + "2": "other",
39 + "3": "theory",
40 + "4": "results",
41 + "5": "theory",
42 + "6": "conclusion",
43 + "7": "appendix",
44 + "8": "appendix",
45 + "9": "appendix"
46 + }
47 + },
48 + "schema_valid": false,
49 + "code_version": "feb6c9e-dirty",
50 + "raw_text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance (labor search frictions, production-based asset pricing)\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Cross-sectional return premium associated with firm stock-return loadings on changes in aggregate labor market tightness, interpreted as compensation for exposure to matching efficiency shocks\",\"strategy\":\"Rolling firm-level regressions estimate loadings on log changes in labor market tightness, controlling for the market. These feed portfolio sorts and Fama-MacBeth regressions with factor-model alphas. A calibrated partial-equilibrium search model is then used to interpret the premium.\",\"key_assumptions\":[\"Loadings on labor market tightness proxy for exposure to priced matching efficiency shocks\",\"Exogenous pricing kernel with both productivity and matching efficiency shocks priced\",\"Positive matching efficiency shocks raise discount rates\",\"Approximate aggregation via equilibrium forecasting rules\"],\"threats_addressed\":[\"Explanation by standard factor models (four-factor alphas)\",\"Overlap with known characteristics such as size, value, momentum, hiring rate, investment and asset growth\",\"Alternative portfolio formation timing\",\"Alternative calibrations shutting down priced shocks\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":false,\"sentence_index\":-1},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":1},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_th"
51 +}
\ No newline at end of file
added data/annotations/documents/jof_12505.json +260 −0
@@ -0,0 +1,260 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12505",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:26+00:00",
9 + "source_sha256": "b466f4f47753f3f75e0e44fd53a808d0a0fa49dfa3cbd1d724464cdc6497d573",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 16030,
12 + "usage": {
13 + "input": 5540,
14 + "output": 2121,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.065374,
19 + "request_id": "req_011CggpfhfGLy8SvPtJsyyAY",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Difference in crisis-period stock returns and operating performance between firms with high and low pre-crisis CSR (social capital), interpreted as the payoff to trust during a trust shock",
27 + "strategy": "Cross-sectional regressions of crisis-period raw and abnormal returns on pre-crisis (2006) CSR with industry dummies and controls. Complemented by panel firm fixed effects models interacting CSR with crisis and post-crisis dummies, and a continuous-treatment difference-in-differences design for operating outcomes.",
28 + "key_assumptions": [
29 + "The 2008-2009 crisis was an unexpected shock to the importance of trust",
30 + "Pre-crisis CSR is not proxying for omitted firm characteristics that independently drive crisis returns",
31 + "CSR ratings capture firm-level social capital"
32 + ],
33 + "threats_addressed": [
34 + "confounding by the credit supply shock (tested July 2007-July 2008)",
35 + "CSR firms outperforming generally rather than only in crises (tested via pre/post comparison with firm fixed effects)",
36 + "time-invariant unobservables (firm fixed effects)",
37 + "alternative trust shock as external validity check (Enron/Worldcom)",
38 + "measurement timing of CSR (2005, 2007, 2008 measures)",
39 + "industry differences (industry dummies)",
40 + "governance and transparency controls"
41 + ],
42 + "causal_language_strength": 3,
43 + "uses_causal_claim": true
44 + },
45 + "abstract_moves": {
46 + "question": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "gap": {
51 + "present": false,
52 + "sentence_index": -1
53 + },
54 + "method": {
55 + "present": false,
56 + "sentence_index": -1
57 + },
58 + "results": {
59 + "present": true,
60 + "sentence_index": 0
61 + },
62 + "magnitude": {
63 + "present": true,
64 + "sentence_index": 0
65 + },
66 + "contribution": {
67 + "present": true,
68 + "sentence_index": 2
69 + }
70 + },
71 + "introduction": {
72 + "moves": [
73 + {
74 + "paragraph_index": 0,
75 + "move": "motivation"
76 + },
77 + {
78 + "paragraph_index": 13,
79 + "move": "motivation"
80 + },
81 + {
82 + "paragraph_index": 13,
83 + "move": "literature_gap"
84 + },
85 + {
86 + "paragraph_index": 13,
87 + "move": "identification_defense"
88 + },
89 + {
90 + "paragraph_index": 14,
91 + "move": "preview_of_approach"
92 + },
93 + {
94 + "paragraph_index": 14,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 15,
99 + "move": "identification_defense"
100 + },
101 + {
102 + "paragraph_index": 15,
103 + "move": "robustness_mention"
104 + },
105 + {
106 + "paragraph_index": 16,
107 + "move": "identification_defense"
108 + },
109 + {
110 + "paragraph_index": 16,
111 + "move": "preview_of_results"
112 + },
113 + {
114 + "paragraph_index": 16,
115 + "move": "mechanism_or_theory_sketch"
116 + },
117 + {
118 + "paragraph_index": 16,
119 + "move": "contribution_statement"
120 + }
121 + ],
122 + "contribution_paragraph_index": 16,
123 + "contribution_types": [
124 + "new_evidence"
125 + ],
126 + "roadmap_present": false,
127 + "literature_positioning_location": "both"
128 + },
129 + "section_roles": [
130 + {
131 + "idx": 0,
132 + "role": "introduction"
133 + },
134 + {
135 + "idx": 1,
136 + "role": "literature"
137 + },
138 + {
139 + "idx": 2,
140 + "role": "data"
141 + },
142 + {
143 + "idx": 3,
144 + "role": "results"
145 + },
146 + {
147 + "idx": 4,
148 + "role": "results"
149 + },
150 + {
151 + "idx": 5,
152 + "role": "conclusion"
153 + }
154 + ],
155 + "results_presentation": {
156 + "economic_significance_discussed": true,
157 + "statistical_significance_language": "unclear",
158 + "magnitude_statements_in_intro": true
159 + },
160 + "robustness": {
161 + "location": "within_results",
162 + "types": [
163 + "alternative_sample_period",
164 + "falsification",
165 + "alternative_measure",
166 + "subsample",
167 + "alternative_specification"
168 + ]
169 + },
170 + "tables": {
171 + "convention_notes": "Table notes define the sample, the CSR construction (strengths minus concerns), return windows, and the abnormal return model, and they write out the panel regression equations. Dependent variables are given as both raw and market-model-adjusted returns.",
172 + "units_stated_in_notes": false,
173 + "standard_errors_description": "unclear"
174 + },
175 + "figures": {
176 + "uncertainty_shown": "no_figures"
177 + },
178 + "conclusion": {
179 + "bounded_claims": true,
180 + "policy_implications": false,
181 + "open_questions": false,
182 + "restates_contribution": true
183 + },
184 + "hedging": {
185 + "level": "moderate",
186 + "typical_devices": [
187 + "suggests",
188 + "consistent with",
189 + "indicate",
190 + "to alleviate concerns"
191 + ]
192 + },
193 + "claim_evidence": [
194 + {
195 + "claim": "High-CSR firms earned 4-7 percentage points higher returns during the 2008-09 crisis",
196 + "evidence_type": "table",
197 + "evidence_ref": "Table II"
198 + },
199 + {
200 + "claim": "High-CSR firms also outperformed during the Enron/Worldcom trust shock",
201 + "evidence_type": "table",
202 + "evidence_ref": "Table III"
203 + },
204 + {
205 + "claim": "The CSR return effect appears only in the crisis period, not in pre- or post-crisis periods, under firm fixed effects",
206 + "evidence_type": "table",
207 + "evidence_ref": "Table IV"
208 + },
209 + {
210 + "claim": "Results are robust to measuring CSR in other years",
211 + "evidence_type": "table",
212 + "evidence_ref": "Table V"
213 + },
214 + {
215 + "claim": "High-CSR firms had higher profitability, sales growth, sales per employee, and debt raising during the crisis",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table VI"
218 + },
219 + {
220 + "claim": "The relation between CSR and returns is stronger in high-trust regions",
221 + "evidence_type": "none_stated",
222 + "evidence_ref": "Section III.E"
223 + }
224 + ],
225 + "structural_summary": "The paper opens with epigraphs on the crisis as a collapse of trust, then gives macro-level literature on social capital and identifies a gap at the firm level. The introduction previews the main return result with magnitudes and three tests that rule out alternatives, followed by panel and mechanism results. A separate conceptual section links trust, social capital, and CSR. After the data section, a long results section presents baseline regressions, alternative shock episodes, the credit-shock placebo, regional trust heterogeneity, CSR components, and robustness. A difference-in-differences section on operating outcomes and capital raising follows, and a brief conclusion restates the findings.",
226 + "writing_profile_keywords": [
227 + "epigraph opening",
228 + "trust shock framing",
229 + "alternative-explanation tests",
230 + "mechanism via real outcomes",
231 + "CSR measurement",
232 + "crisis-period returns"
233 + ],
234 + "confidence": 0.8,
235 + "notes": "The introduction paragraphs at indices 0-12 are epigraph fragments, so the substantive introduction text begins at index 13. Significance reporting conventions are not visible in the excerpts."
236 + },
237 + "deterministic_checks": {
238 + "roadmap_regex": false,
239 + "contribution_regex_in_intro": false,
240 + "magnitude_in_abstract": false,
241 + "abstract_n_sentences": 3,
242 + "intro_n_paragraphs": 17,
243 + "intro_n_words": 947,
244 + "causal_words_in_abstract": 0,
245 + "hedge_words_in_abstract": 0,
246 + "stars_mentioned": false,
247 + "clustered_se_mentioned": false,
248 + "internet_appendix_mentioned": false,
249 + "section_roles_rule": {
250 + "0": "introduction",
251 + "1": "other",
252 + "2": "data",
253 + "3": "other",
254 + "4": "other",
255 + "5": "conclusion"
256 + }
257 + },
258 + "schema_valid": true,
259 + "code_version": "feb6c9e-dirty"
260 +}
\ No newline at end of file
added data/annotations/documents/jof_12506.json +284 −0
@@ -0,0 +1,284 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12506",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:10+00:00",
9 + "source_sha256": "9e7140acf9652fa8c482473e5172aa1150ccebc6801e6de6628b916c6fdcac99",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24607,
12 + "usage": {
13 + "input": 8026,
14 + "output": 2170,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.076298,
19 + "request_id": "req_011CggpeZe3Vsp9duoyQgftp",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "public/municipal finance; fixed income valuation",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "NPV of advance refundings measured with option-adjusted callable bond valuation; associations between aggregate refunding activity or NPV losses and fiscal and macroeconomic conditions; cross-sectional probability of refunding",
27 + "strategy": "Each transaction is valued using municipal and SLGS term structures and Black-model volatilities. Aggregate time-series regressions use lagged explanatory variables, and a cross-sectional probit relates refunding probability to potential NPV and cash flow savings.",
28 + "key_assumptions": [
29 + "Valuation model and term structures correctly measure the continuation value of the callable bond",
30 + "Lagged fiscal variables proxy for financial constraints"
31 + ],
32 + "threats_addressed": [
33 + "Covenant-elimination motive",
34 + "Informational frictions about the implicit option value (option ignored entirely)",
35 + "Transaction costs omitted, so losses are likely understated"
36 + ],
37 + "causal_language_strength": 3,
38 + "uses_causal_claim": true
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": true,
43 + "sentence_index": 3
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 2
56 + },
57 + "magnitude": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "contribution": {
62 + "present": false,
63 + "sentence_index": -1
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "mechanism_or_theory_sketch"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "preview_of_results"
79 + },
80 + {
81 + "paragraph_index": 2,
82 + "move": "identification_defense"
83 + },
84 + {
85 + "paragraph_index": 2,
86 + "move": "robustness_mention"
87 + },
88 + {
89 + "paragraph_index": 3,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 4,
94 + "move": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 5,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 6,
102 + "move": "preview_of_results"
103 + },
104 + {
105 + "paragraph_index": 6,
106 + "move": "policy_or_broader_implication"
107 + },
108 + {
109 + "paragraph_index": 7,
110 + "move": "roadmap"
111 + }
112 + ],
113 + "contribution_paragraph_index": -1,
114 + "contribution_types": [
115 + "new_evidence"
116 + ],
117 + "roadmap_present": true,
118 + "literature_positioning_location": "inside_introduction"
119 + },
120 + "section_roles": [
121 + {
122 + "idx": 0,
123 + "role": "introduction"
124 + },
125 + {
126 + "idx": 1,
127 + "role": "other"
128 + },
129 + {
130 + "idx": 2,
131 + "role": "theory"
132 + },
133 + {
134 + "idx": 3,
135 + "role": "other"
136 + },
137 + {
138 + "idx": 4,
139 + "role": "data"
140 + },
141 + {
142 + "idx": 5,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 6,
147 + "role": "robustness"
148 + },
149 + {
150 + "idx": 7,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 8,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 9,
159 + "role": "conclusion"
160 + },
161 + {
162 + "idx": 10,
163 + "role": "appendix"
164 + }
165 + ],
166 + "results_presentation": {
167 + "economic_significance_discussed": true,
168 + "statistical_significance_language": "unclear",
169 + "magnitude_statements_in_intro": true
170 + },
171 + "robustness": {
172 + "location": "main_text_section",
173 + "types": [
174 + "sensitivity_to_parameters",
175 + "alternative_measure"
176 + ]
177 + },
178 + "tables": {
179 + "convention_notes": "Notes define variables and units (dollars, percentages, per $100 par, per CUSIP). Regression tables report lagged explanatory variables and include a probit model. A numerical example table illustrates the valuation logic. Standard error treatment is not visible.",
180 + "units_stated_in_notes": true,
181 + "standard_errors_description": "unclear"
182 + },
183 + "figures": {
184 + "uncertainty_shown": "no"
185 + },
186 + "conclusion": {
187 + "bounded_claims": true,
188 + "policy_implications": true,
189 + "open_questions": false,
190 + "restates_contribution": true
191 + },
192 + "hedging": {
193 + "level": "moderate",
194 + "typical_devices": [
195 + "'suggest' and 'consistent with the view'",
196 + "'may' for mechanisms",
197 + "explicit note that losses are likely underestimated",
198 + "'it is tempting to conclude ... however'"
199 + ]
200 + },
201 + "claim_evidence": [
202 + {
203 + "claim": "About 85% of advance refundings occur at an NPV loss, with aggregate losses exceeding $15 billion",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table III; Figures 2-3"
206 + },
207 + {
208 + "claim": "More than 96% of refundings give short-term cash flow savings, averaging 141 bp of par",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table III; Figure 4"
211 + },
212 + {
213 + "claim": "Even if the implicit option is ignored, over half of refundings occur at a loss",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table IV"
216 + },
217 + {
218 + "claim": "Refunding activity responds to potential cash flow savings but not to potential NPV",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table V"
221 + },
222 + {
223 + "claim": "Refunding activity rises with tax revenue shortfalls and budget deficits",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table VI"
226 + },
227 + {
228 + "claim": "NPV losses grow as tax collections fall and unfunded pension liabilities rise",
229 + "evidence_type": "table",
230 + "evidence_ref": "Table VII"
231 + },
232 + {
233 + "claim": "Potential cash flow savings drive the cross-sectional probability of refunding",
234 + "evidence_type": "table",
235 + "evidence_ref": "Table VIII"
236 + },
237 + {
238 + "claim": "Average refunding NPV is negative in all 50 states, with heterogeneity across states",
239 + "evidence_type": "table",
240 + "evidence_ref": "Table IX"
241 + }
242 + ],
243 + "structural_summary": "The paper opens with the rapid growth of advance refundings and frames each one as an early synthetic call akin to American option exercise. It sets out an optimal-refunding valuation framework and gives a case study of political context. It then values over 206,000 transactions and documents widespread NPV losses alongside short-term cash flow savings. Alternative frictions (covenants, informational problems) are ruled out. Aggregate time-series and cross-sectional probit regressions link refunding activity to fiscal constraints. The conclusion restates the findings and discusses which other parties benefit from suboptimal refundings.",
244 + "writing_profile_keywords": [
245 + "municipal bonds",
246 + "option valuation",
247 + "NPV losses",
248 + "financial constraints",
249 + "time-series regressions",
250 + "case study",
251 + "policy relevance"
252 + ],
253 + "confidence": 0.8,
254 + "notes": "No explicit contribution paragraph is visible. Standard error and significance conventions are not visible in the excerpts. Section VI is treated as the robustness section because it rules out alternative explanations."
255 + },
256 + "deterministic_checks": {
257 + "roadmap_regex": true,
258 + "contribution_regex_in_intro": false,
259 + "magnitude_in_abstract": true,
260 + "abstract_n_sentences": 6,
261 + "intro_n_paragraphs": 8,
262 + "intro_n_words": 956,
263 + "causal_words_in_abstract": 0,
264 + "hedge_words_in_abstract": 1,
265 + "stars_mentioned": false,
266 + "clustered_se_mentioned": false,
267 + "internet_appendix_mentioned": false,
268 + "section_roles_rule": {
269 + "0": "introduction",
270 + "1": "other",
271 + "2": "other",
272 + "3": "other",
273 + "4": "data",
274 + "5": "other",
275 + "6": "other",
276 + "7": "other",
277 + "8": "other",
278 + "9": "conclusion",
279 + "10": "appendix"
280 + }
281 + },
282 + "schema_valid": true,
283 + "code_version": "feb6c9e-dirty"
284 +}
\ No newline at end of file
added data/annotations/documents/jof_12507.json +253 −0
@@ -0,0 +1,253 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12507",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:03+00:00",
9 + "source_sha256": "7db274be305c6bdd8e90034fa01020a60724b9474c7cb1fbaa87a78f32ee899c",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 18912,
12 + "usage": {
13 + "input": 6438,
14 + "output": 2012,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.066786,
19 + "request_id": "req_011Cggpe79LKkBbdTs8VTKky",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "regression_discontinuity",
25 + "identification": {
26 + "estimand": "Local effect of a stock's margin trading eligibility (Group 1 status) on its liquidity (effective spread, price impact), and how this effect varies with market conditions",
27 + "strategy": "Regression discontinuity around the impact-cost eligibility cutoff of 1% for margin trading in India. The sample is local to the cutoff with CCT bandwidths, and the regressions include lagged controls.",
28 + "key_assumptions": [
29 + "Stocks near the impact-cost cutoff are comparable apart from eligibility",
30 + "No manipulation or sorting of stocks around the cutoff",
31 + "Other characteristics are continuous at the threshold"
32 + ],
33 + "threats_addressed": [
34 + "Common shocks driving both deleveraging and illiquidity",
35 + "Effects of Group 1 status other than margin trading",
36 + "Cannibalization of liquidity from ineligible stocks",
37 + "Mechanical variation in measured impact cost (addressed with placebo cutoffs)",
38 + "Bandwidth and functional-form sensitivity"
39 + ],
40 + "causal_language_strength": 4,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": true,
46 + "sentence_index": 0
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 2
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 3
59 + },
60 + "magnitude": {
61 + "present": false,
62 + "sentence_index": -1
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 1
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "tension_or_puzzle"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "related_literature_positioning"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "identification_defense"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "mechanism_or_theory_sketch"
98 + },
99 + {
100 + "paragraph_index": 2,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 3,
105 + "move": "related_literature_positioning"
106 + },
107 + {
108 + "paragraph_index": 3,
109 + "move": "literature_gap"
110 + },
111 + {
112 + "paragraph_index": 4,
113 + "move": "roadmap"
114 + }
115 + ],
116 + "contribution_paragraph_index": 0,
117 + "contribution_types": [
118 + "new_evidence",
119 + "new_identification"
120 + ],
121 + "roadmap_present": true,
122 + "literature_positioning_location": "inside_introduction"
123 + },
124 + "section_roles": [
125 + {
126 + "idx": 0,
127 + "role": "introduction"
128 + },
129 + {
130 + "idx": 1,
131 + "role": "other"
132 + },
133 + {
134 + "idx": 2,
135 + "role": "data"
136 + },
137 + {
138 + "idx": 3,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 4,
143 + "role": "conclusion"
144 + }
145 + ],
146 + "results_presentation": {
147 + "economic_significance_discussed": true,
148 + "statistical_significance_language": "unclear",
149 + "magnitude_statements_in_intro": false
150 + },
151 + "robustness": {
152 + "location": "within_results",
153 + "types": [
154 + "alternative_specification",
155 + "alternative_measure",
156 + "placebo",
157 + "sensitivity_to_parameters"
158 + ]
159 + },
160 + "tables": {
161 + "convention_notes": "Table notes define dependent variables (Espread, Pimpact), the timing of eligibility, the local sample defined by CCT bandwidths, and how specifications differ across columns. Robustness tables refer back to specific columns of the main table.",
162 + "units_stated_in_notes": true,
163 + "standard_errors_description": "unclear"
164 + },
165 + "figures": {
166 + "uncertainty_shown": "unclear"
167 + },
168 + "conclusion": {
169 + "bounded_claims": true,
170 + "policy_implications": true,
171 + "open_questions": false,
172 + "restates_contribution": true
173 + },
174 + "hedging": {
175 + "level": "moderate",
176 + "typical_devices": [
177 + "evidence consistent with a causal effect",
178 + "potential effects",
179 + "suggest"
180 + ]
181 + },
182 + "claim_evidence": [
183 + {
184 + "claim": "Margin eligibility lowers effective spreads by about 2.4 basis points",
185 + "evidence_type": "table",
186 + "evidence_ref": "Table II"
187 + },
188 + {
189 + "claim": "The liquidity benefit of margin eligibility reverses during severe market downturns",
190 + "evidence_type": "table",
191 + "evidence_ref": "Table VIII"
192 + },
193 + {
194 + "claim": "Margin traders follow contrarian strategies, with position changes negatively related to returns",
195 + "evidence_type": "table",
196 + "evidence_ref": "Table IX"
197 + },
198 + {
199 + "claim": "Contrarian margin trading reduces returns to reversal strategies in eligible stocks",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table XI"
202 + },
203 + {
204 + "claim": "Liquidity improvements are larger when margin trading is more intense",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table VII"
207 + },
208 + {
209 + "claim": "Results are robust to alternative bandwidths, polynomials and placebo cutoffs",
210 + "evidence_type": "table",
211 + "evidence_ref": "Tables IV-VI"
212 + },
213 + {
214 + "claim": "Liquidity measures are visibly discontinuous at the eligibility cutoff",
215 + "evidence_type": "figure",
216 + "evidence_ref": "Figure 2"
217 + }
218 + ],
219 + "structural_summary": "The paper opens with a question about leverage and liquidity, motivated by theory and by the identification difficulty that common shocks drive both. It previews an RDD built on India's margin eligibility rules, followed by a crisis-reversal result and mechanism evidence on contrarian margin traders. An institutional section precedes the data and RDD design, including bandwidth selection. A long results section covers the baseline, robustness, placebos, alternative interpretations, downturn interactions and mechanisms. The conclusion summarizes three findings and draws policy relevance.",
220 + "writing_profile_keywords": [
221 + "regression discontinuity",
222 + "margin trading",
223 + "liquidity spirals",
224 + "institutional laboratory",
225 + "causal framing",
226 + "India"
227 + ],
228 + "confidence": 0.85,
229 + "notes": "Section I (institutional setting) is labeled 'theory' by the rule but is coded here as 'other'. Standard error treatment is not visible in the excerpts. Figures 2–4 are RDD plots, but the captions do not say whether confidence bands are shown."
230 + },
231 + "deterministic_checks": {
232 + "roadmap_regex": true,
233 + "contribution_regex_in_intro": false,
234 + "magnitude_in_abstract": false,
235 + "abstract_n_sentences": 5,
236 + "intro_n_paragraphs": 5,
237 + "intro_n_words": 831,
238 + "causal_words_in_abstract": 2,
239 + "hedge_words_in_abstract": 1,
240 + "stars_mentioned": false,
241 + "clustered_se_mentioned": false,
242 + "internet_appendix_mentioned": false,
243 + "section_roles_rule": {
244 + "0": "introduction",
245 + "1": "theory",
246 + "2": "data",
247 + "3": "results",
248 + "4": "conclusion"
249 + }
250 + },
251 + "schema_valid": true,
252 + "code_version": "feb6c9e-dirty"
253 +}
\ No newline at end of file
added data/annotations/documents/jof_12508.json +288 −0
@@ -0,0 +1,288 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12508",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:57+00:00",
9 + "source_sha256": "3b28837fe16b036e94dc6b267b5a8a91f5733c530dda29fec36947f7c6aea504",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 25391,
12 + "usage": {
13 + "input": 9197,
14 + "output": 2236,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.082302,
19 + "request_id": "req_011Cggphu79gT5xDcCNv15Fo",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "calibration_simulation",
25 + "identification": {
26 + "estimand": "Predictive association between the labor-share and term-structure moments: variance ratios of wages and dividends, and the risk, premium and slope of dividend strips.",
27 + "strategy": "Time-series predictive regressions of variance ratios and dividend strip returns on the labor-share, with controls and Newey-West t-statistics, combined with a calibrated general equilibrium model whose predictions are compared with the data.",
28 + "key_assumptions": [
29 + "Income insurance concerns transitory shocks only.",
30 + "Workers and shareholders share permanent shocks, with cointegrated levels.",
31 + "Limited market participation: workers consume their wages."
32 + ],
33 + "threats_addressed": [
34 + "Financial leverage as an alternative channel for downward-sloping dividend risk (EBIT term structure shown to be similar).",
35 + "Other sources of transitory dividend risk not attributable to income insurance.",
36 + "Belief-formation channel as an alternative explanation."
37 + ],
38 + "causal_language_strength": 3,
39 + "uses_causal_claim": true
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": false,
44 + "sentence_index": -1
45 + },
46 + "gap": {
47 + "present": true,
48 + "sentence_index": 4
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 4
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 0
57 + },
58 + "magnitude": {
59 + "present": false,
60 + "sentence_index": -1
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 1
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "literature_gap"
80 + },
81 + {
82 + "paragraph_index": 0,
83 + "move": "contribution_statement"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 1,
91 + "move": "related_literature_positioning"
92 + },
93 + {
94 + "paragraph_index": 2,
95 + "move": "mechanism_or_theory_sketch"
96 + },
97 + {
98 + "paragraph_index": 2,
99 + "move": "preview_of_results"
100 + },
101 + {
102 + "paragraph_index": 3,
103 + "move": "preview_of_results"
104 + },
105 + {
106 + "paragraph_index": 4,
107 + "move": "preview_of_results"
108 + },
109 + {
110 + "paragraph_index": 4,
111 + "move": "robustness_mention"
112 + },
113 + {
114 + "paragraph_index": 5,
115 + "move": "preview_of_approach"
116 + },
117 + {
118 + "paragraph_index": 5,
119 + "move": "mechanism_or_theory_sketch"
120 + },
121 + {
122 + "paragraph_index": 6,
123 + "move": "related_literature_positioning"
124 + },
125 + {
126 + "paragraph_index": 7,
127 + "move": "other"
128 + },
129 + {
130 + "paragraph_index": 8,
131 + "move": "roadmap"
132 + }
133 + ],
134 + "contribution_paragraph_index": 0,
135 + "contribution_types": [
136 + "new_evidence",
137 + "new_theory_or_model",
138 + "resolution_of_puzzle"
139 + ],
140 + "roadmap_present": true,
141 + "literature_positioning_location": "inside_introduction"
142 + },
143 + "section_roles": [
144 + {
145 + "idx": 0,
146 + "role": "introduction"
147 + },
148 + {
149 + "idx": 1,
150 + "role": "results"
151 + },
152 + {
153 + "idx": 2,
154 + "role": "theory"
155 + },
156 + {
157 + "idx": 3,
158 + "role": "results"
159 + },
160 + {
161 + "idx": 4,
162 + "role": "results"
163 + },
164 + {
165 + "idx": 5,
166 + "role": "other"
167 + },
168 + {
169 + "idx": 6,
170 + "role": "conclusion"
171 + },
172 + {
173 + "idx": 7,
174 + "role": "appendix"
175 + }
176 + ],
177 + "results_presentation": {
178 + "economic_significance_discussed": true,
179 + "statistical_significance_language": "mixed",
180 + "magnitude_statements_in_intro": true
181 + },
182 + "robustness": {
183 + "location": "within_results",
184 + "types": [
185 + "alternative_specification",
186 + "alternative_measure",
187 + "other"
188 + ]
189 + },
190 + "tables": {
191 + "convention_notes": "Notes state the regression equations, report coefficients with Newey-West (and Hansen-Hodrick) t-statistics in parentheses and adjusted R2; calibration tables set data moments against model moments.",
192 + "units_stated_in_notes": false,
193 + "standard_errors_description": "newey_west"
194 + },
195 + "figures": {
196 + "uncertainty_shown": "mixed"
197 + },
198 + "conclusion": {
199 + "bounded_claims": false,
200 + "policy_implications": false,
201 + "open_questions": true,
202 + "restates_contribution": true
203 + },
204 + "hedging": {
205 + "level": "moderate",
206 + "typical_devices": [
207 + "'supports the idea'",
208 + "'consistent with'",
209 + "'can have only a limited effect'",
210 + "explicit acknowledgment of model limitations"
211 + ]
212 + },
213 + "claim_evidence": [
214 + {
215 + "claim": "Variance-ratio term structures of output, wages and dividends are flat, increasing and decreasing respectively.",
216 + "evidence_type": "figure",
217 + "evidence_ref": "Figure 1"
218 + },
219 + {
220 + "claim": "Labor-share predicts the wage-dividend variance-ratio gap and the dividend variance ratios.",
221 + "evidence_type": "table",
222 + "evidence_ref": "Table I; Figure 2"
223 + },
224 + {
225 + "claim": "Labor-share forecasts dividend strip volatility, premium and slope, with high correlations.",
226 + "evidence_type": "table",
227 + "evidence_ref": "Tables II-III; Figure 3"
228 + },
229 + {
230 + "claim": "Labor-share positively forecasts dividend and consumption growth.",
231 + "evidence_type": "table",
232 + "evidence_ref": "Table IV"
233 + },
234 + {
235 + "claim": "The transitory dividend component not tied to income insurance does not explain the slope of dividend risk.",
236 + "evidence_type": "table",
237 + "evidence_ref": "Table V"
238 + },
239 + {
240 + "claim": "The calibrated model matches cash-flow moments and standard asset pricing moments.",
241 + "evidence_type": "table",
242 + "evidence_ref": "Tables VI, VII, IX"
243 + },
244 + {
245 + "claim": "The model generates downward-sloping dividend strip volatility and premia.",
246 + "evidence_type": "figure",
247 + "evidence_ref": "Figure 6"
248 + }
249 + ],
250 + "structural_summary": "The introduction frames a puzzle: leading models imply long-run priced risk, while the data show downward-sloping equity term structures. It poses questions, previews four empirical findings, sketches the income-insurance model, contrasts the paper with competing mechanisms, admits a limitation on interest rates, and gives a roadmap. Section I presents the regression evidence on variance ratios and dividend strips. Section II builds the general equilibrium model. Section III calibrates it and compares asset pricing results with the data. Section IV extends the model to time-varying uncertainty. The conclusion restates the contribution and points to possible extensions.",
251 + "writing_profile_keywords": [
252 + "term structure of equity",
253 + "variance ratios",
254 + "income insurance",
255 + "labor-share",
256 + "dividend strips",
257 + "general equilibrium calibration",
258 + "recursive preferences"
259 + ],
260 + "confidence": 0.8,
261 + "notes": "The empirical part precedes the model. Figure 1 shows confidence bands, but the model figures appear not to show uncertainty. Robustness is located within Section I.D. The magnitudes stated in the introduction are correlations of 86%, 57% and 52%."
262 + },
263 + "deterministic_checks": {
264 + "roadmap_regex": true,
265 + "contribution_regex_in_intro": false,
266 + "magnitude_in_abstract": false,
267 + "abstract_n_sentences": 5,
268 + "intro_n_paragraphs": 9,
269 + "intro_n_words": 1270,
270 + "causal_words_in_abstract": 0,
271 + "hedge_words_in_abstract": 0,
272 + "stars_mentioned": false,
273 + "clustered_se_mentioned": false,
274 + "internet_appendix_mentioned": false,
275 + "section_roles_rule": {
276 + "0": "introduction",
277 + "1": "other",
278 + "2": "theory",
279 + "3": "results",
280 + "4": "other",
281 + "5": "other",
282 + "6": "conclusion",
283 + "7": "appendix"
284 + }
285 + },
286 + "schema_valid": true,
287 + "code_version": "feb6c9e-dirty"
288 +}
\ No newline at end of file
added data/annotations/documents/jof_12509.json +285 −0
@@ -0,0 +1,285 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12509",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:53+00:00",
9 + "source_sha256": "d97f9e3543f798e3a98fead0220fb2961999c374299fcf55a8f22dfbc3696440",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21239,
12 + "usage": {
13 + "input": 7288,
14 + "output": 2326,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.076466,
19 + "request_id": "req_011CggpdBTWWQeqXj48BGve5",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "mutual fund performance and skill",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Within-fund time-series slope of subsequent benchmark-adjusted fund return on lagged fund turnover, interpreted as evidence of skill in timing profit opportunities",
27 + "strategy": "Panel regressions with fund fixed effects isolate the time-series turnover-performance relation, contrasted with cross-sectional slopes; model-derived predictions are tested by splitting funds into categories; turnover of similar funds is used as an additional predictor",
28 + "key_assumptions": [
29 + "Profit opportunities vary over time and jointly determine turnover and performance",
30 + "Funds choose turnover optimally given trading costs (relaxed via the suboptimal-trading extension)",
31 + "More-skilled funds charge higher fees",
32 + "Measured turnover is a noisy proxy for true discretionary turnover"
33 + ],
34 + "threats_addressed": [
35 + "Cross-sectional confounds between fund types, handled by fund fixed effects and the time-series design",
36 + "Alternative ways of measuring turnover",
37 + "Alternative benchmark models",
38 + "Out-of-sample validity",
39 + "Measurement error in turnover"
40 + ],
41 + "causal_language_strength": 2,
42 + "uses_causal_claim": false
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 0
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": false,
67 + "sentence_index": -1
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "mechanism_or_theory_sketch"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "preview_of_results"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "mechanism_or_theory_sketch"
99 + },
100 + {
101 + "paragraph_index": 5,
102 + "move": "mechanism_or_theory_sketch"
103 + },
104 + {
105 + "paragraph_index": 6,
106 + "move": "related_literature_positioning"
107 + },
108 + {
109 + "paragraph_index": 6,
110 + "move": "contribution_statement"
111 + },
112 + {
113 + "paragraph_index": 7,
114 + "move": "related_literature_positioning"
115 + },
116 + {
117 + "paragraph_index": 8,
118 + "move": "related_literature_positioning"
119 + },
120 + {
121 + "paragraph_index": 9,
122 + "move": "roadmap"
123 + }
124 + ],
125 + "contribution_paragraph_index": 6,
126 + "contribution_types": [
127 + "new_evidence",
128 + "new_theory_or_model"
129 + ],
130 + "roadmap_present": true,
131 + "literature_positioning_location": "inside_introduction"
132 + },
133 + "section_roles": [
134 + {
135 + "idx": 0,
136 + "role": "introduction"
137 + },
138 + {
139 + "idx": 1,
140 + "role": "theory"
141 + },
142 + {
143 + "idx": 2,
144 + "role": "results"
145 + },
146 + {
147 + "idx": 3,
148 + "role": "results"
149 + },
150 + {
151 + "idx": 4,
152 + "role": "results"
153 + },
154 + {
155 + "idx": 5,
156 + "role": "conclusion"
157 + },
158 + {
159 + "idx": 6,
160 + "role": "appendix"
161 + },
162 + {
163 + "idx": 7,
164 + "role": "appendix"
165 + },
166 + {
167 + "idx": 8,
168 + "role": "appendix"
169 + }
170 + ],
171 + "results_presentation": {
172 + "economic_significance_discussed": true,
173 + "statistical_significance_language": "unclear",
174 + "magnitude_statements_in_intro": true
175 + },
176 + "robustness": {
177 + "location": "main_text_section",
178 + "types": [
179 + "alternative_measure",
180 + "alternative_specification",
181 + "alternative_sample_period"
182 + ]
183 + },
184 + "tables": {
185 + "convention_notes": "Table notes define the dependent variable precisely (net return plus expense ratio minus the Morningstar benchmark), describe the regressors and regression variants, and report correlation and summary-statistics tables by fund category. How standard errors are computed is not visible in the excerpts.",
186 + "units_stated_in_notes": false,
187 + "standard_errors_description": "unclear"
188 + },
189 + "figures": {
190 + "uncertainty_shown": "no"
191 + },
192 + "conclusion": {
193 + "bounded_claims": true,
194 + "policy_implications": false,
195 + "open_questions": false,
196 + "restates_contribution": true
197 + },
198 + "hedging": {
199 + "level": "moderate",
200 + "typical_devices": [
201 + "consistent with",
202 + "suggesting",
203 + "appears to be related",
204 + "under the plausible assumption",
205 + "seems unique"
206 + ]
207 + },
208 + "claim_evidence": [
209 + {
210 + "claim": "A fund's turnover positively predicts its subsequent benchmark-adjusted return in the time series",
211 + "evidence_type": "table",
212 + "evidence_ref": "Table I"
213 + },
214 + {
215 + "claim": "The time-series turnover-performance slope exceeds the cross-sectional slope, as the model predicts",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table I"
218 + },
219 + {
220 + "claim": "The relation is stronger for small-stock funds, small funds, and high-fee funds",
221 + "evidence_type": "table",
222 + "evidence_ref": "Table II"
223 + },
224 + {
225 + "claim": "Fund turnover comoves strongly across funds, especially among similar funds",
226 + "evidence_type": "table",
227 + "evidence_ref": "Table IV, Table V, Figure 1"
228 + },
229 + {
230 + "claim": "Average turnover rises with investor sentiment and return dispersion and falls with market liquidity",
231 + "evidence_type": "table",
232 + "evidence_ref": "Table VI, Figure 2"
233 + },
234 + {
235 + "claim": "A one-SD rise in similar funds' average turnover is associated with about 0.43% per year higher performance",
236 + "evidence_type": "table",
237 + "evidence_ref": "Table VII"
238 + },
239 + {
240 + "claim": "The model implies a positive time-series turnover-performance relation",
241 + "evidence_type": "equation_or_proof",
242 + "evidence_ref": "Section I, equation (11)"
243 + }
244 + ],
245 + "structural_summary": "The paper opens with motivation about active fund skill and immediately sketches the model's intuition. The introduction then previews results (main relation, heterogeneity, commonality, predictive power of peer turnover), offers two model-based interpretations (measurement error and suboptimal trading), and positions the work against three literatures before a roadmap. Section I presents an analytical model of optimal turnover. Section II estimates time-series versus cross-sectional slopes and includes robustness subsections. Section III tests heterogeneity across fund categories. Section IV studies common turnover, its links to mispricing proxies, and its predictive power. The conclusion restates the findings, and the appendices hold the model extension and the econometric and proof derivations.",
246 + "writing_profile_keywords": [
247 + "model-then-test",
248 + "prediction-confirmation framing",
249 + "time-series vs cross-section",
250 + "fund fixed effects",
251 + "mutual fund skill",
252 + "commonality",
253 + "mispricing proxies",
254 + "appendix derivations"
255 + ],
256 + "confidence": 0.85,
257 + "notes": "Results are framed as confirming model predictions rather than as causal estimates. Robustness appears as subsections within Section II (turnover measurement, benchmark models, out-of-sample evidence). The out-of-sample check is mapped to the alternative_sample_period type, which is an approximation. Significance conventions are not visible in the excerpts."
258 + },
259 + "deterministic_checks": {
260 + "roadmap_regex": true,
261 + "contribution_regex_in_intro": false,
262 + "magnitude_in_abstract": false,
263 + "abstract_n_sentences": 8,
264 + "intro_n_paragraphs": 10,
265 + "intro_n_words": 1171,
266 + "causal_words_in_abstract": 0,
267 + "hedge_words_in_abstract": 1,
268 + "stars_mentioned": false,
269 + "clustered_se_mentioned": false,
270 + "internet_appendix_mentioned": false,
271 + "section_roles_rule": {
272 + "0": "introduction",
273 + "1": "theory",
274 + "2": "other",
275 + "3": "other",
276 + "4": "other",
277 + "5": "conclusion",
278 + "6": "appendix",
279 + "7": "appendix",
280 + "8": "appendix"
281 + }
282 + },
283 + "schema_valid": true,
284 + "code_version": "feb6c9e-dirty"
285 +}
\ No newline at end of file
added data/annotations/documents/jof_12510.json +246 −0
@@ -0,0 +1,246 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12510",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:51+00:00",
9 + "source_sha256": "822f19bea05c40295dc03cc38fb2c2725b408a745a7d1c8a5790aab16b438cb9",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19485,
12 + "usage": {
13 + "input": 6779,
14 + "output": 1998,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06787,
19 + "request_id": "req_011Cggpn756kc6isVapqRaMf",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "asset_pricing (transaction costs, investor clienteles); policy/taxation",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Average treatment effect of the 2012 French FTT on market quality measures of large French stocks, and on trading by trader type and on fund holdings and turnover",
27 + "strategy": "Difference-in-differences comparing French stocks above the 1bn EUR threshold with non-French control stocks, using stock and time fixed effects. Separate splits for SLP and non-SLP stocks. Exploits the exemption of intraday (HFT) trading to isolate indirect liquidity effects. Fund-level treatment effects are regressed on fund characteristics.",
28 + "key_assumptions": [
29 + "parallel trends between French and non-French control stocks",
30 + "control stocks not affected by the French tax",
31 + "HFTs affected only indirectly because intraday trading is exempt"
32 + ],
33 + "threats_addressed": [
34 + "mechanical decrease in HFT activity (Table VIII)",
35 + "heterogeneity across liquidity-provision regimes (SLP vs non-SLP)"
36 + ],
37 + "causal_language_strength": 4,
38 + "uses_causal_claim": true
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": true,
43 + "sentence_index": 0
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": true,
51 + "sentence_index": 0
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "magnitude": {
58 + "present": false,
59 + "sentence_index": -1
60 + },
61 + "contribution": {
62 + "present": true,
63 + "sentence_index": 4
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "tension_or_puzzle"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "literature_gap"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "preview_of_approach"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "preview_of_results"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "mechanism_or_theory_sketch"
91 + },
92 + {
93 + "paragraph_index": 2,
94 + "move": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "contribution_statement"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "policy_or_broader_implication"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "related_literature_positioning"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "roadmap"
111 + }
112 + ],
113 + "contribution_paragraph_index": 2,
114 + "contribution_types": [
115 + "new_evidence",
116 + "new_data_or_measure"
117 + ],
118 + "roadmap_present": true,
119 + "literature_positioning_location": "inside_introduction"
120 + },
121 + "section_roles": [
122 + {
123 + "idx": 0,
124 + "role": "introduction"
125 + },
126 + {
127 + "idx": 1,
128 + "role": "theory"
129 + },
130 + {
131 + "idx": 2,
132 + "role": "results"
133 + },
134 + {
135 + "idx": 3,
136 + "role": "results"
137 + },
138 + {
139 + "idx": 4,
140 + "role": "results"
141 + },
142 + {
143 + "idx": 5,
144 + "role": "conclusion"
145 + }
146 + ],
147 + "results_presentation": {
148 + "economic_significance_discussed": true,
149 + "statistical_significance_language": "unclear",
150 + "magnitude_statements_in_intro": false
151 + },
152 + "robustness": {
153 + "location": "mixed",
154 + "types": [
155 + "falsification",
156 + "subsample"
157 + ]
158 + },
159 + "tables": {
160 + "convention_notes": "Table notes restate the estimating regression equation, define the treated and control groups, and state the sample period. Summary tables report standard deviations in parentheses. Some units are given, such as bps and billions of USD. Several table titles use the word 'Causal Impact'.",
161 + "units_stated_in_notes": true,
162 + "standard_errors_description": "unclear"
163 + },
164 + "figures": {
165 + "uncertainty_shown": "unclear"
166 + },
167 + "conclusion": {
168 + "bounded_claims": true,
169 + "policy_implications": true,
170 + "open_questions": true,
171 + "restates_contribution": true
172 + },
173 + "hedging": {
174 + "level": "moderate",
175 + "typical_devices": [
176 + "'in line with' / 'consistent with' theory",
177 + "'we find no support/evidence'",
178 + "'suggest'",
179 + "'it is unclear whether'",
180 + "'to our knowledge'"
181 + ]
182 + },
183 + "claim_evidence": [
184 + {
185 + "claim": "No evidence that the FTT improves market quality through a change in volume composition; market quality declines via a liquidity effect.",
186 + "evidence_type": "table",
187 + "evidence_ref": "Tables III, IV; Figure 1"
188 + },
189 + {
190 + "claim": "Negative effects are muted for the most liquid stocks and more severe for less liquid (non-SLP) stocks.",
191 + "evidence_type": "table",
192 + "evidence_ref": "Table IV"
193 + },
194 + {
195 + "claim": "Exempt HFTs are indirectly affected through the liquidity channel.",
196 + "evidence_type": "table",
197 + "evidence_ref": "Tables VII, VIII, IX"
198 + },
199 + {
200 + "claim": "High-turnover funds reduced holdings of French stocks while low-turnover funds increased them, and funds reduced turnover.",
201 + "evidence_type": "table",
202 + "evidence_ref": "Table XI"
203 + },
204 + {
205 + "claim": "FTT revenue fell short of expectations.",
206 + "evidence_type": "descriptive_statistic",
207 + "evidence_ref": "official revenue figures, footnote 3"
208 + }
209 + ],
210 + "structural_summary": "The introduction frames competing composition vs liquidity theories of FTTs, notes mixed aggregate evidence, and motivates granular data. Section I describes the policy experiment, develops competing hypotheses, and sets out the DiD design. Section II estimates aggregate market quality effects, with heterogeneity across SLP and non-SLP stocks. Section III examines trader types, especially exempt HFTs, as direct evidence on the liquidity channel. Section IV tests the extensive vs intensive margins of adjustment using fund holdings. The conclusion draws out policy implications and future research.",
211 + "writing_profile_keywords": [
212 + "policy experiment",
213 + "competing hypotheses",
214 + "difference-in-differences",
215 + "theory-guided testing",
216 + "granular data",
217 + "heterogeneity",
218 + "policy debate"
219 + ],
220 + "confidence": 0.8,
221 + "notes": "Standard error type and significance notation are not visible in the excerpts. Robustness is inferred from Table VIII and the internet appendix mention. The paper uses explicit causal phrasing in its table titles."
222 + },
223 + "deterministic_checks": {
224 + "roadmap_regex": true,
225 + "contribution_regex_in_intro": false,
226 + "magnitude_in_abstract": false,
227 + "abstract_n_sentences": 5,
228 + "intro_n_paragraphs": 5,
229 + "intro_n_words": 849,
230 + "causal_words_in_abstract": 0,
231 + "hedge_words_in_abstract": 1,
232 + "stars_mentioned": false,
233 + "clustered_se_mentioned": false,
234 + "internet_appendix_mentioned": false,
235 + "section_roles_rule": {
236 + "0": "introduction",
237 + "1": "theory",
238 + "2": "other",
239 + "3": "other",
240 + "4": "other",
241 + "5": "conclusion"
242 + }
243 + },
244 + "schema_valid": true,
245 + "code_version": "feb6c9e-dirty"
246 +}
\ No newline at end of file
added data/annotations/documents/jof_12511.json +256 −0
@@ -0,0 +1,256 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12511",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:18+00:00",
9 + "source_sha256": "f0eaf0c72163d1961b267e344f27c1293d0d94850eb661c980d2051bf2bf5674",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20398,
12 + "usage": {
13 + "input": 7027,
14 + "output": 2049,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.069882,
19 + "request_id": "req_011Cggpf7iLfvmEhFxGrLnum",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "fixed income / public finance (municipal bonds)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Share of tax-adjusted municipal bond spreads attributable to default risk versus liquidity, and the implied default risk premium.",
27 + "strategy": "Triangulates three decompositions: (1) transaction-based regressions of spreads on liquidity measures to back out a liquidity component; (2) CDS premia used to measure the default component; (3) a quasi-natural experiment on within-bond spread changes around pre-refunding events, with bond and time controls.",
28 + "key_assumptions": [
29 + "Pre-refunding removes default risk without an offsetting liquidity effect (lower bound interpretation)",
30 + "Tax adjustment using top statutory rates correctly converts tax-exempt yields to taxable equivalents",
31 + "CDS premia measure pure default compensation",
32 + "Swap rate is an appropriate riskless benchmark"
33 + ],
34 + "threats_addressed": [
35 + "omitted issuer/bond-specific factors in cross-pool comparisons (within-bond design)",
36 + "sensitivity to the tax adjustment method",
37 + "benchmark and sample selection"
38 + ],
39 + "causal_language_strength": 2,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 2
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": true,
61 + "sentence_index": 1
62 + },
63 + "contribution": {
64 + "present": false,
65 + "sentence_index": -1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "related_literature_positioning"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "literature_gap"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "related_literature_positioning"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "related_literature_positioning"
101 + },
102 + {
103 + "paragraph_index": 3,
104 + "move": "identification_defense"
105 + },
106 + {
107 + "paragraph_index": 4,
108 + "move": "related_literature_positioning"
109 + },
110 + {
111 + "paragraph_index": 5,
112 + "move": "contribution_statement"
113 + },
114 + {
115 + "paragraph_index": 5,
116 + "move": "roadmap"
117 + }
118 + ],
119 + "contribution_paragraph_index": 5,
120 + "contribution_types": [
121 + "new_evidence",
122 + "new_identification"
123 + ],
124 + "roadmap_present": true,
125 + "literature_positioning_location": "inside_introduction"
126 + },
127 + "section_roles": [
128 + {
129 + "idx": 0,
130 + "role": "introduction"
131 + },
132 + {
133 + "idx": 1,
134 + "role": "data"
135 + },
136 + {
137 + "idx": 2,
138 + "role": "empirical_design"
139 + },
140 + {
141 + "idx": 3,
142 + "role": "results"
143 + },
144 + {
145 + "idx": 4,
146 + "role": "robustness"
147 + },
148 + {
149 + "idx": 5,
150 + "role": "conclusion"
151 + },
152 + {
153 + "idx": 6,
154 + "role": "appendix"
155 + }
156 + ],
157 + "results_presentation": {
158 + "economic_significance_discussed": true,
159 + "statistical_significance_language": "unclear",
160 + "magnitude_statements_in_intro": true
161 + },
162 + "robustness": {
163 + "location": "main_text_section",
164 + "types": [
165 + "alternative_measure",
166 + "subsample",
167 + "alternative_specification"
168 + ]
169 + },
170 + "tables": {
171 + "convention_notes": "Notes define dependent variables (tax-adjusted spread over interpolated swap rate), explain the observation level (bond-month, issuer-month), and describe sample filters. Tables move from sample construction and summary statistics to pricing regressions, cross-method comparisons, and implied premia.",
172 + "units_stated_in_notes": false,
173 + "standard_errors_description": "unclear"
174 + },
175 + "figures": {
176 + "uncertainty_shown": "no"
177 + },
178 + "conclusion": {
179 + "bounded_claims": true,
180 + "policy_implications": true,
181 + "open_questions": true,
182 + "restates_contribution": true
183 + },
184 + "hedging": {
185 + "level": "moderate",
186 + "typical_devices": [
187 + "'likely due to' for mechanism attribution",
188 + "lower-bound framing ('at least', 'assuming no offsetting effect')",
189 + "ranges rather than point estimates",
190 + "'could' for candidate channels"
191 + ]
192 + },
193 + "claim_evidence": [
194 + {
195 + "claim": "Default risk explains roughly three-quarters or more of the average tax-adjusted muni spread.",
196 + "evidence_type": "table",
197 + "evidence_ref": "Tables IV, VII, VIII"
198 + },
199 + {
200 + "claim": "Pre-refunding lowers within-bond spreads by at least about 41%.",
201 + "evidence_type": "table",
202 + "evidence_ref": "Table IX"
203 + },
204 + {
205 + "claim": "Implied default risk premium is an order of magnitude larger than in corporate bonds.",
206 + "evidence_type": "table",
207 + "evidence_ref": "Table X; Berndt et al. (2005)"
208 + },
209 + {
210 + "claim": "Distressed issuers such as Detroit, Chicago and Illinois carry elevated default spreads, consistent with pension underfunding.",
211 + "evidence_type": "table",
212 + "evidence_ref": "Table V"
213 + },
214 + {
215 + "claim": "Municipal defaults are historically extremely rare.",
216 + "evidence_type": "descriptive_statistic",
217 + "evidence_ref": "Moody's (2015) default rates"
218 + }
219 + ],
220 + "structural_summary": "The paper opens by contrasting recent municipal distress with very low historical default rates, then notes that the literature emphasizes taxes and illiquidity. It poses the question of how much of the spread is default versus liquidity and previews the headline share and the implied premium. It positions itself against a contemporaneous paper that reaches opposing conclusions, defending its own within-bond design. A data section is followed by a methods section covering the tax adjustment and three decompositions, then results, robustness, and a conclusion. The conclusion speculates on tax-related channels and draws policy implications.",
221 + "writing_profile_keywords": [
222 + "triangulation of methods",
223 + "decomposition",
224 + "contrast with contemporaneous work",
225 + "quasi-natural experiment",
226 + "policy implications",
227 + "single-author first person"
228 + ],
229 + "confidence": 0.82,
230 + "notes": "Standard-error conventions are not visible in the table note excerpts. Causal language is limited to the pre-refunding experiment; the main claims are decomposition shares."
231 + },
232 + "deterministic_checks": {
233 + "roadmap_regex": true,
234 + "contribution_regex_in_intro": true,
235 + "magnitude_in_abstract": true,
236 + "abstract_n_sentences": 4,
237 + "intro_n_paragraphs": 6,
238 + "intro_n_words": 909,
239 + "causal_words_in_abstract": 0,
240 + "hedge_words_in_abstract": 0,
241 + "stars_mentioned": false,
242 + "clustered_se_mentioned": false,
243 + "internet_appendix_mentioned": false,
244 + "section_roles_rule": {
245 + "0": "introduction",
246 + "1": "data",
247 + "2": "other",
248 + "3": "results",
249 + "4": "robustness",
250 + "5": "conclusion",
251 + "6": "appendix"
252 + }
253 + },
254 + "schema_valid": true,
255 + "code_version": "feb6c9e-dirty"
256 +}
\ No newline at end of file
added data/annotations/documents/jof_12512.json +294 −0
@@ -0,0 +1,294 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12512",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:23+00:00",
9 + "source_sha256": "7dc1ad489e2990da640cd11f07559afbd55ad60da879aa37af398533fb333700",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 29142,
12 + "usage": {
13 + "input": 9752,
14 + "output": 2470,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.089202,
19 + "request_id": "req_011CggpfM1YjYEn5V5xt9y6b",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "corporate_finance (M&A / auctions)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "How potential acquirers' characteristics (distance, asset overlap, concentration gains) and capitalization relate to the probability of acquiring a failed bank, and how local acquirer capitalization relates to FDIC resolution costs.",
27 + "strategy": "Fixed-effects logit models on failed bank × potential acquirer pairs, with failed bank and acquirer-quarter fixed effects; interactions with acquirer capitalization; OLS regressions of resolution cost on local acquirer capitalization; samples restricted to eligible and actual bidders to rule out an FDIC-eligibility explanation.",
28 + "key_assumptions": [
29 + "Fixed effects absorb inherent failed bank and acquirer characteristics.",
30 + "Acquisition likelihood reveals relative willingness to pay, per the auction model.",
31 + "Capitalization of local potential acquirers is not driven by omitted regional shocks that also drive costs."
32 + ],
33 + "threats_addressed": [
34 + "FDIC eligibility criteria as an alternative explanation",
35 + "regional shocks",
36 + "bid characteristics and measurement of resolution costs",
37 + "selection from focusing only on participating bidders",
38 + "soft information vs. other interpretations of distance"
39 + ],
40 + "causal_language_strength": 3,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": false,
46 + "sentence_index": -1
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 3
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 1
59 + },
60 + "magnitude": {
61 + "present": true,
62 + "sentence_index": 0
63 + },
64 + "contribution": {
65 + "present": false,
66 + "sentence_index": -1
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "literature_gap"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 2,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 3,
93 + "move": "preview_of_results"
94 + },
95 + {
96 + "paragraph_index": 3,
97 + "move": "identification_defense"
98 + },
99 + {
100 + "paragraph_index": 4,
101 + "move": "identification_defense"
102 + },
103 + {
104 + "paragraph_index": 5,
105 + "move": "identification_defense"
106 + },
107 + {
108 + "paragraph_index": 5,
109 + "move": "mechanism_or_theory_sketch"
110 + },
111 + {
112 + "paragraph_index": 5,
113 + "move": "preview_of_results"
114 + },
115 + {
116 + "paragraph_index": 6,
117 + "move": "related_literature_positioning"
118 + },
119 + {
120 + "paragraph_index": 6,
121 + "move": "contribution_statement"
122 + },
123 + {
124 + "paragraph_index": 7,
125 + "move": "related_literature_positioning"
126 + }
127 + ],
128 + "contribution_paragraph_index": 6,
129 + "contribution_types": [
130 + "new_evidence",
131 + "new_theory_or_model"
132 + ],
133 + "roadmap_present": false,
134 + "literature_positioning_location": "inside_introduction"
135 + },
136 + "section_roles": [
137 + {
138 + "idx": 0,
139 + "role": "introduction"
140 + },
141 + {
142 + "idx": 1,
143 + "role": "other"
144 + },
145 + {
146 + "idx": 2,
147 + "role": "data"
148 + },
149 + {
150 + "idx": 3,
151 + "role": "theory"
152 + },
153 + {
154 + "idx": 4,
155 + "role": "data"
156 + },
157 + {
158 + "idx": 5,
159 + "role": "results"
160 + },
161 + {
162 + "idx": 6,
163 + "role": "results"
164 + },
165 + {
166 + "idx": 7,
167 + "role": "conclusion"
168 + },
169 + {
170 + "idx": 8,
171 + "role": "appendix"
172 + }
173 + ],
174 + "results_presentation": {
175 + "economic_significance_discussed": true,
176 + "statistical_significance_language": "unclear",
177 + "magnitude_statements_in_intro": true
178 + },
179 + "robustness": {
180 + "location": "mixed",
181 + "types": [
182 + "subsample",
183 + "alternative_measure",
184 + "alternative_specification",
185 + "other"
186 + ]
187 + },
188 + "tables": {
189 + "convention_notes": "Each table note names the estimator (logit, FE logit, OLS) and defines the dependent variable and key regressors, including units (e.g., distance in 100-mile increments, cost as % of assets, constant 2009 prices). The excerpts do not show standard error details.",
190 + "units_stated_in_notes": true,
191 + "standard_errors_description": "unclear"
192 + },
193 + "figures": {
194 + "uncertainty_shown": "unclear"
195 + },
196 + "conclusion": {
197 + "bounded_claims": true,
198 + "policy_implications": true,
199 + "open_questions": false,
200 + "restates_contribution": true
201 + },
202 + "hedging": {
203 + "level": "moderate",
204 + "typical_devices": [
205 + "'suggests'",
206 + "'consistent with'",
207 + "'unlikely to be driven by'",
208 + "'partially explains'",
209 + "'may lead to'",
210 + "framework 'rationalizes' rather than proves"
211 + ]
212 + },
213 + "claim_evidence": [
214 + {
215 + "claim": "Average FDIC loss from selling a failed bank is about 28% of assets.",
216 + "evidence_type": "descriptive_statistic",
217 + "evidence_ref": "Table I; Figure 2"
218 + },
219 + {
220 + "claim": "Failed bank sales are predominantly local; acquisition likelihood falls with distance.",
221 + "evidence_type": "table",
222 + "evidence_ref": "Table III; Figure 3"
223 + },
224 + {
225 + "claim": "Banks with similar loan portfolios are more likely to acquire failed banks.",
226 + "evidence_type": "table",
227 + "evidence_ref": "Table VI; Figure 4"
228 + },
229 + {
230 + "claim": "Market concentration gains have an economically small but statistically significant effect on acquisition.",
231 + "evidence_type": "table",
232 + "evidence_ref": "Table VII; Figure 5"
233 + },
234 + {
235 + "claim": "Best-capitalized local buyers acquire over 30% of failed banks versus 10% for the worst capitalized.",
236 + "evidence_type": "figure",
237 + "evidence_ref": "Figure 6; Table IX"
238 + },
239 + {
240 + "claim": "Results persist when the sample is restricted to FDIC-eligible bidders, ruling out the eligibility explanation.",
241 + "evidence_type": "table",
242 + "evidence_ref": "Table X"
243 + },
244 + {
245 + "claim": "Poor capitalization of local potential acquirers is associated with higher FDIC resolution costs.",
246 + "evidence_type": "table",
247 + "evidence_ref": "Table XI; Figure 7"
248 + },
249 + {
250 + "claim": "Higher-valuation bidders bid weakly more; budget constraints create a wedge between willingness and ability to pay.",
251 + "evidence_type": "equation_or_proof",
252 + "evidence_ref": "Section III; Appendix"
253 + }
254 + ],
255 + "structural_summary": "The paper opens with the scale of FDIC losses in the Great Recession and notes how little is known about failed bank allocation since the S&L crisis. It then introduces an auction model with budget constraints. Next it documents three willingness-to-pay determinants (distance, asset overlap, concentration gains) and shows that capitalization limits ability to pay, which distorts allocation toward lower-valuation acquirers. An FDIC-eligibility alternative is ruled out using bidder data, and the misallocation is linked to resolution costs, with robustness checks for regional shocks and cost measurement. The paper closes with links to fire-sale and banking theories and with policy implications for capital regulation and resolution.",
256 + "writing_profile_keywords": [
257 + "policy-motivated",
258 + "auction framework",
259 + "fixed-effects logit",
260 + "pair-level data",
261 + "alternative hypothesis testing",
262 + "misallocation",
263 + "bank resolution"
264 + ],
265 + "confidence": 0.8,
266 + "notes": "Institutional background (idx 1) is labeled 'literature' by the role rule but is descriptive institutional content, so it is coded as 'other'. Sections V and VI are both results; Section VI contains robustness subsections. No equations are counted, despite a stylized model and an appendix proof."
267 + },
268 + "deterministic_checks": {
269 + "roadmap_regex": false,
270 + "contribution_regex_in_intro": true,
271 + "magnitude_in_abstract": true,
272 + "abstract_n_sentences": 5,
273 + "intro_n_paragraphs": 8,
274 + "intro_n_words": 1343,
275 + "causal_words_in_abstract": 0,
276 + "hedge_words_in_abstract": 0,
277 + "stars_mentioned": false,
278 + "clustered_se_mentioned": false,
279 + "internet_appendix_mentioned": false,
280 + "section_roles_rule": {
281 + "0": "introduction",
282 + "1": "literature",
283 + "2": "data",
284 + "3": "theory",
285 + "4": "data",
286 + "5": "other",
287 + "6": "other",
288 + "7": "conclusion",
289 + "8": "appendix"
290 + }
291 + },
292 + "schema_valid": true,
293 + "code_version": "feb6c9e-dirty"
294 +}
\ No newline at end of file
added data/annotations/documents/jof_12513.json +267 −0
@@ -0,0 +1,267 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12513",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:08+00:00",
9 + "source_sha256": "ab6176d011452d4b3fafa0f5ee08edc6b15134bdaabbb5faee6237c71b5af5c1",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 17284,
12 + "usage": {
13 + "input": 5877,
14 + "output": 2016,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.064622,
19 + "request_id": "req_011CggpeTfcrxHG5drcG6sTs",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Time-series alpha of volatility-scaled factor returns relative to the unmanaged factors, plus Sharpe ratio and utility gains.",
27 + "strategy": "Scale monthly factor returns by the inverse of the prior month's realized variance and regress the managed returns on the original factors. Compare the resulting alphas with simulations from structural macrofinance models.",
28 + "key_assumptions": [
29 + "Realized variance is forecastable at short horizons",
30 + "Lagged variance only weakly forecasts returns",
31 + "Strategy is implementable in real time"
32 + ],
33 + "threats_addressed": [
34 + "transaction costs",
35 + "leverage constraints",
36 + "overlap with cross-sectional low-risk anomalies and risk parity",
37 + "common volatility component",
38 + "rebalancing horizon",
39 + "business cycle risk"
40 + ],
41 + "causal_language_strength": 2,
42 + "uses_causal_claim": false
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 0
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 4
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "preview_of_approach"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "preview_of_results"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "motivation"
83 + },
84 + {
85 + "paragraph_index": 2,
86 + "move": "preview_of_results"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "mechanism_or_theory_sketch"
91 + },
92 + {
93 + "paragraph_index": 4,
94 + "move": "mechanism_or_theory_sketch"
95 + },
96 + {
97 + "paragraph_index": 5,
98 + "move": "mechanism_or_theory_sketch"
99 + },
100 + {
101 + "paragraph_index": 6,
102 + "move": "preview_of_results"
103 + },
104 + {
105 + "paragraph_index": 7,
106 + "move": "tension_or_puzzle"
107 + },
108 + {
109 + "paragraph_index": 7,
110 + "move": "related_literature_positioning"
111 + },
112 + {
113 + "paragraph_index": 8,
114 + "move": "related_literature_positioning"
115 + },
116 + {
117 + "paragraph_index": 8,
118 + "move": "contribution_statement"
119 + },
120 + {
121 + "paragraph_index": 9,
122 + "move": "roadmap"
123 + }
124 + ],
125 + "contribution_paragraph_index": 8,
126 + "contribution_types": [
127 + "new_evidence",
128 + "new_theory_or_model"
129 + ],
130 + "roadmap_present": true,
131 + "literature_positioning_location": "inside_introduction"
132 + },
133 + "section_roles": [
134 + {
135 + "idx": 0,
136 + "role": "introduction"
137 + },
138 + {
139 + "idx": 1,
140 + "role": "results"
141 + },
142 + {
143 + "idx": 2,
144 + "role": "robustness"
145 + },
146 + {
147 + "idx": 3,
148 + "role": "theory"
149 + },
150 + {
151 + "idx": 4,
152 + "role": "theory"
153 + },
154 + {
155 + "idx": 5,
156 + "role": "conclusion"
157 + }
158 + ],
159 + "results_presentation": {
160 + "economic_significance_discussed": true,
161 + "statistical_significance_language": "unclear",
162 + "magnitude_statements_in_intro": true
163 + },
164 + "robustness": {
165 + "location": "mixed",
166 + "types": [
167 + "alternative_specification",
168 + "alternative_measure",
169 + "other"
170 + ]
171 + },
172 + "tables": {
173 + "convention_notes": "Table notes state the regression equation explicitly (managed factor on original factor, with optional controls or recession interactions) and describe how each managed factor is constructed. The visible excerpts do not describe standard errors.",
174 + "units_stated_in_notes": false,
175 + "standard_errors_description": "unclear"
176 + },
177 + "figures": {
178 + "uncertainty_shown": "mixed"
179 + },
180 + "conclusion": {
181 + "bounded_claims": false,
182 + "policy_implications": false,
183 + "open_questions": false,
184 + "restates_contribution": true
185 + },
186 + "hedging": {
187 + "level": "low",
188 + "typical_devices": [
189 + "'suggesting'",
190 + "'challenge to' rather than rejection",
191 + "'weakly related'"
192 + ]
193 + },
194 + "claim_evidence": [
195 + {
196 + "claim": "Volatility-managed factors earn large positive alphas against the original factors.",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table I"
199 + },
200 + {
201 + "claim": "Lagged volatility predicts future volatility but barely predicts average returns, so the mean-variance trade-off worsens when volatility is high.",
202 + "evidence_type": "figure",
203 + "evidence_ref": "Figure 1"
204 + },
205 + {
206 + "claim": "The managed strategy takes relatively less risk in recessions.",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table III"
209 + },
210 + {
211 + "claim": "Alphas survive transaction costs and leverage limits.",
212 + "evidence_type": "table",
213 + "evidence_ref": "Tables IV, V"
214 + },
215 + {
216 + "claim": "Structural macrofinance models rarely generate alphas as large as the historical ones.",
217 + "evidence_type": "figure",
218 + "evidence_ref": "Figure 7"
219 + },
220 + {
221 + "claim": "Alphas are proportional to the covariance between variance and the factor price of risk.",
222 + "evidence_type": "equation_or_proof",
223 + "evidence_ref": "Section III"
224 + },
225 + {
226 + "claim": "Utility gains of about 65% for mean-variance investors.",
227 + "evidence_type": "figure",
228 + "evidence_ref": "Figure 4"
229 + }
230 + ],
231 + "structural_summary": "The paper opens with a simple strategy and its headline result: alphas across many factors. It motivates the strategy with mean-variance logic and gives intuition from volatility sorts. It then previews a theory link from alphas to price-of-risk dynamics and the stochastic discount factor, and frames the findings as a challenge to structural models. The body presents empirical results, followed by a long section of mechanism and robustness checks. Theory and general-equilibrium sections close the paper, ending with a brief conclusion.",
232 + "writing_profile_keywords": [
233 + "trading strategy",
234 + "factor alphas",
235 + "Sharpe ratio",
236 + "volatility timing",
237 + "macrofinance challenge",
238 + "simulation of models",
239 + "self-contained robustness subsections"
240 + ],
241 + "confidence": 0.8,
242 + "notes": "The abstract is truncated, and part of the introduction (sentence 5 onward) is merged into it. Introduction paragraphs are fragmented by line breaks in the extraction, which affects paragraph indices. Statistical reporting conventions are not visible in the excerpts."
243 + },
244 + "deterministic_checks": {
245 + "roadmap_regex": true,
246 + "contribution_regex_in_intro": false,
247 + "magnitude_in_abstract": false,
248 + "abstract_n_sentences": 6,
249 + "intro_n_paragraphs": 10,
250 + "intro_n_words": 781,
251 + "causal_words_in_abstract": 0,
252 + "hedge_words_in_abstract": 0,
253 + "stars_mentioned": false,
254 + "clustered_se_mentioned": false,
255 + "internet_appendix_mentioned": false,
256 + "section_roles_rule": {
257 + "0": "introduction",
258 + "1": "results",
259 + "2": "other",
260 + "3": "theory",
261 + "4": "theory",
262 + "5": "conclusion"
263 + }
264 + },
265 + "schema_valid": true,
266 + "code_version": "feb6c9e-dirty"
267 +}
\ No newline at end of file
added data/annotations/documents/jof_12514.json +256 −0
@@ -0,0 +1,256 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12514",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:52+00:00",
9 + "source_sha256": "7dd8bfbef11797e52089a25c38b62f4ce1788c8e2a4450f2c2af25eda6b22d01",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 18515,
12 + "usage": {
13 + "input": 6266,
14 + "output": 2147,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.068798,
19 + "request_id": "req_011Cggpd87721PchFpgzdAHV",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "The share of cross-sectional variation in client portfolio risky share and home bias that is attributable to advisor-specific effects, compared with the share explained by investor attributes.",
27 + "strategy": "Panel regressions with advisor fixed effects and year fixed effects. A two-way fixed effects design adds investor fixed effects and uses clients forced to switch advisors after the advisor's death, retirement or resignation, then tracks convergence toward the new advisor's allocation. Advisor fixed effects are then regressed on advisor attributes, including the advisor's own portfolio.",
28 + "key_assumptions": [
29 + "Forced switches are unrelated to changes in client preferences.",
30 + "Investor fixed effects absorb persistent unobserved client preferences.",
31 + "Investors are not matched to advisors on unobserved characteristics, or such matching is controlled for by the switcher design."
32 + ],
33 + "threats_addressed": [
34 + "Matching between investors and advisors on omitted characteristics",
35 + "Unobserved client preferences",
36 + "Self-directed investments initiated by the client",
37 + "Client-initiated switches that coincide with preference changes"
38 + ],
39 + "causal_language_strength": 3,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": true,
61 + "sentence_index": 4
62 + },
63 + "contribution": {
64 + "present": false,
65 + "sentence_index": -1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "literature_gap"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "tension_or_puzzle"
81 + },
82 + {
83 + "paragraph_index": 0,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "other"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "identification_defense"
101 + },
102 + {
103 + "paragraph_index": 3,
104 + "move": "preview_of_results"
105 + },
106 + {
107 + "paragraph_index": 4,
108 + "move": "contribution_statement"
109 + },
110 + {
111 + "paragraph_index": 5,
112 + "move": "roadmap"
113 + }
114 + ],
115 + "contribution_paragraph_index": 4,
116 + "contribution_types": [
117 + "new_evidence",
118 + "new_data_or_measure"
119 + ],
120 + "roadmap_present": true,
121 + "literature_positioning_location": "inside_introduction"
122 + },
123 + "section_roles": [
124 + {
125 + "idx": 0,
126 + "role": "introduction"
127 + },
128 + {
129 + "idx": 1,
130 + "role": "data"
131 + },
132 + {
133 + "idx": 2,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 3,
138 + "role": "results"
139 + },
140 + {
141 + "idx": 4,
142 + "role": "results"
143 + },
144 + {
145 + "idx": 5,
146 + "role": "conclusion"
147 + }
148 + ],
149 + "results_presentation": {
150 + "economic_significance_discussed": true,
151 + "statistical_significance_language": "unclear",
152 + "magnitude_statements_in_intro": true
153 + },
154 + "robustness": {
155 + "location": "within_results",
156 + "types": [
157 + "other"
158 + ]
159 + },
160 + "tables": {
161 + "convention_notes": "Notes define the dependent variables (risky share, home bias), the sample period and the fixed effects included. Tables are organized in panels by outcome (risky share vs. home bias). Descriptive statistics are reported as percentiles.",
162 + "units_stated_in_notes": true,
163 + "standard_errors_description": "unclear"
164 + },
165 + "figures": {
166 + "uncertainty_shown": "mixed"
167 + },
168 + "conclusion": {
169 + "bounded_claims": false,
170 + "policy_implications": false,
171 + "open_questions": false,
172 + "restates_contribution": true
173 + },
174 + "hedging": {
175 + "level": "moderate",
176 + "typical_devices": [
177 + "Offering alternative interpretations of the advisor fixed effects",
178 + "An explicit caveats-and-robustness subsection",
179 + "Framing the matching concern as a hypothesis that is then tested"
180 + ]
181 + },
182 + "claim_evidence": [
183 + {
184 + "claim": "Investor characteristics explain only about 12% of the variation in risky share.",
185 + "evidence_type": "table",
186 + "evidence_ref": "Table II"
187 + },
188 + {
189 + "claim": "Advisor fixed effects raise adjusted R2 from 12% to 30% for risky share, and add 24% for home bias.",
190 + "evidence_type": "table",
191 + "evidence_ref": "Table II"
192 + },
193 + {
194 + "claim": "Moving from the 25th to the 75th percentile of advisor effects shifts risky share by 20 pp and home bias by 32 pp.",
195 + "evidence_type": "figure",
196 + "evidence_ref": "Figure 3"
197 + },
198 + {
199 + "claim": "Advisor effects remain important after controlling for investor fixed effects among forced switchers.",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table III"
202 + },
203 + {
204 + "claim": "An advisor's own risky share is the strongest predictor of the advisor fixed effect.",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table IV; Figure 5"
207 + },
208 + {
209 + "claim": "Advised portfolios cost about 2.5% per year, roughly 1.5% more than life cycle funds.",
210 + "evidence_type": "figure",
211 + "evidence_ref": "Figure 6; Section IV"
212 + },
213 + {
214 + "claim": "Advisors' gross and net alphas are estimated against passive factor benchmarks.",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table V"
217 + }
218 + ],
219 + "structural_summary": "The introduction motivates the topic with how widely households use advisors, then sets the competing views of underperformance and \"interior decoration\" against each other. It previews variance-decomposition results with R2 magnitudes and raises the matching interpretation. It addresses that interpretation with a forced-switcher two-way fixed effects design, then states three contributions and gives a roadmap. The body proceeds through the data, customization regressions (with caveats, investor fixed effects, and advisor attributes), performance alphas, and a fee decomposition. The conclusion restates three findings with magnitudes.",
220 + "writing_profile_keywords": [
221 + "administrative data",
222 + "variance decomposition",
223 + "advisor fixed effects",
224 + "two-way fixed effects",
225 + "forced switchers",
226 + "adjusted R2 magnitudes",
227 + "fee decomposition",
228 + "one-size-fits-all"
229 + ],
230 + "confidence": 0.8,
231 + "notes": "The conclusion excerpt is truncated, so policy implications may appear later in it but are not visible. The standard error type is not visible in the table note excerpts. Some figures show 95% confidence intervals (Figures 1 and 4), while others show distributions or averages."
232 + },
233 + "deterministic_checks": {
234 + "roadmap_regex": true,
235 + "contribution_regex_in_intro": true,
236 + "magnitude_in_abstract": true,
237 + "abstract_n_sentences": 5,
238 + "intro_n_paragraphs": 6,
239 + "intro_n_words": 797,
240 + "causal_words_in_abstract": 0,
241 + "hedge_words_in_abstract": 0,
242 + "stars_mentioned": false,
243 + "clustered_se_mentioned": false,
244 + "internet_appendix_mentioned": false,
245 + "section_roles_rule": {
246 + "0": "introduction",
247 + "1": "other",
248 + "2": "other",
249 + "3": "other",
250 + "4": "other",
251 + "5": "conclusion"
252 + }
253 + },
254 + "schema_valid": true,
255 + "code_version": "feb6c9e-dirty"
256 +}
\ No newline at end of file
added data/annotations/documents/jof_12518.json +50 −0
@@ -0,0 +1,50 @@
1 +{
2 + "validation_errors": [
3 + "no json"
4 + ],
5 + "doc_id": "jof_12518",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:59:16+00:00",
11 + "source_sha256": "f1de67b38763a6406a2a5f8ea6d7973c5b57e425764e66fe97346dd1ef029529",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 20898,
14 + "usage": {
15 + "input": 7763,
16 + "output": 2158,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.075006,
21 + "request_id": "req_011CggpjJVqsfs4uEEvQbLBR",
22 + "annotation": null,
23 + "deterministic_checks": {
24 + "roadmap_regex": false,
25 + "contribution_regex_in_intro": true,
26 + "magnitude_in_abstract": false,
27 + "abstract_n_sentences": 3,
28 + "intro_n_paragraphs": 5,
29 + "intro_n_words": 757,
30 + "causal_words_in_abstract": 0,
31 + "hedge_words_in_abstract": 1,
32 + "stars_mentioned": false,
33 + "clustered_se_mentioned": false,
34 + "internet_appendix_mentioned": false,
35 + "section_roles_rule": {
36 + "0": "introduction",
37 + "1": "data",
38 + "2": "other",
39 + "3": "other",
40 + "4": "other",
41 + "5": "robustness",
42 + "6": "conclusion",
43 + "7": "appendix",
44 + "8": "appendix"
45 + }
46 + },
47 + "schema_valid": false,
48 + "code_version": "feb6c9e-dirty",
49 + "raw_text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"value premium and cash-flow duration\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Differences in future cash-flow (dividend/earnings) growth rates between low and high book-to-market portfolios\",\"strategy\":\"Descriptive portfolio sorts (buy-and-hold and rebalanced) tracking average real dividends and earnings, with adjustments for survivorship and look-back bias; Fama-MacBeth firm-level regressions; valuation-model and accounting decompositions\",\"key_assumptions\":[\"Averages across formation years are representative of expected growth\",\"Reinvesting delisting proceeds properly corrects survivorship bias\",\"Terminal growth perpetuity assumption beyond year 20\"],\"threats_addressed\":[\"survivorship bias\",\"look-back bias\",\"rebalancing effect\",\"small-stock driven results (value vs equal weighting)\",\"negative earnings\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":false,\"sentence_index\":-1},\"results\":{\"present\":true,\"sentence_index\":0},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"other\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"resolution_of_puzzle\"===\"x\"?\"other\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"resolution_of_puzzle\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"da"
50 +}
\ No newline at end of file
added data/annotations/documents/jof_12519.json +284 −0
@@ -0,0 +1,284 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('typical_devices_note' was unexpected) @ hedging"
4 + ],
5 + "doc_id": "jof_12519",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:58:40+00:00",
11 + "source_sha256": "773110aa8eba11104516ee3a48ca9762debd2b6f03715e7e5a39842de147c771",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 22115,
14 + "usage": {
15 + "input": 7283,
16 + "output": 1962,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.069166,
21 + "request_id": "req_011CggpgpN7qPop4HAoqdN13",
22 + "annotation": {
23 + "genre": "theoretical",
24 + "subfield": "financial_intermediation_banking",
25 + "subfield_secondary": "corporate_finance (security design, securitization regulation)",
26 + "primary_method": "theoretical_model_analytical",
27 + "identification": {
28 + "estimand": "not applicable (theoretical model)",
29 + "strategy": "not applicable; analytical characterization of first-best, second-best optimal mechanism and market equilibria",
30 + "key_assumptions": [
31 + "risk-neutral agents, three periods",
32 + "manager more impatient than investors (gains from trade)",
33 + "screening effort and asset quality are manager's private information",
34 + "lack of commitment in market equilibria"
35 + ],
36 + "threats_addressed": [],
37 + "causal_language_strength": 1,
38 + "uses_causal_claim": false
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": true,
43 + "sentence_index": 0
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": true,
51 + "sentence_index": 1
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 3
56 + },
57 + "magnitude": {
58 + "present": false,
59 + "sentence_index": -1
60 + },
61 + "contribution": {
62 + "present": false,
63 + "sentence_index": -1
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "tension_or_puzzle"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "preview_of_approach"
79 + },
80 + {
81 + "paragraph_index": 2,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 3,
86 + "move": "preview_of_results"
87 + },
88 + {
89 + "paragraph_index": 3,
90 + "move": "policy_or_broader_implication"
91 + },
92 + {
93 + "paragraph_index": 4,
94 + "move": "mechanism_or_theory_sketch"
95 + },
96 + {
97 + "paragraph_index": 5,
98 + "move": "policy_or_broader_implication"
99 + },
100 + {
101 + "paragraph_index": 6,
102 + "move": "policy_or_broader_implication"
103 + },
104 + {
105 + "paragraph_index": 7,
106 + "move": "contribution_statement"
107 + },
108 + {
109 + "paragraph_index": 8,
110 + "move": "related_literature_positioning"
111 + },
112 + {
113 + "paragraph_index": 8,
114 + "move": "literature_gap"
115 + }
116 + ],
117 + "contribution_paragraph_index": 8,
118 + "contribution_types": [
119 + "new_theory_or_model"
120 + ],
121 + "roadmap_present": false,
122 + "literature_positioning_location": "inside_introduction"
123 + },
124 + "section_roles": [
125 + {
126 + "idx": 0,
127 + "role": "introduction"
128 + },
129 + {
130 + "idx": 1,
131 + "role": "theory"
132 + },
133 + {
134 + "idx": 2,
135 + "role": "theory"
136 + },
137 + {
138 + "idx": 3,
139 + "role": "theory"
140 + },
141 + {
142 + "idx": 4,
143 + "role": "theory"
144 + },
145 + {
146 + "idx": 5,
147 + "role": "conclusion"
148 + },
149 + {
150 + "idx": 6,
151 + "role": "appendix"
152 + },
153 + {
154 + "idx": 7,
155 + "role": "appendix"
156 + },
157 + {
158 + "idx": 8,
159 + "role": "appendix"
160 + },
161 + {
162 + "idx": 9,
163 + "role": "appendix"
164 + },
165 + {
166 + "idx": 10,
167 + "role": "appendix"
168 + },
169 + {
170 + "idx": 11,
171 + "role": "appendix"
172 + }
173 + ],
174 + "results_presentation": {
175 + "economic_significance_discussed": false,
176 + "statistical_significance_language": "unclear",
177 + "magnitude_statements_in_intro": false
178 + },
179 + "robustness": {
180 + "location": "appendix",
181 + "types": [
182 + "other"
183 + ]
184 + },
185 + "tables": {
186 + "convention_notes": "No tables; results are stated as propositions and lemmas, with proofs in appendices and a numerical illustration in a figure.",
187 + "units_stated_in_notes": false,
188 + "standard_errors_description": "none"
189 + },
190 + "figures": {
191 + "uncertainty_shown": "no"
192 + },
193 + "conclusion": {
194 + "bounded_claims": true,
195 + "policy_implications": true,
196 + "open_questions": false,
197 + "restates_contribution": true
198 + },
199 + "hedging": {
200 + "level": "moderate",
201 + "typical_devices": [
202 + "'can feature' / 'may' modal verbs",
203 + "acknowledging the model is stylized",
204 + "'consistent with' / 'in line with' when linking to evidence",
205 + "scope limits on where policy implications apply"
206 + ],
207 + "typical_devices_note": null
208 + },
209 + "claim_evidence": [
210 + {
211 + "claim": "Costly cash flow retention is essential to implement positive screening under the optimal mechanism",
212 + "evidence_type": "equation_or_proof",
213 + "evidence_ref": "Section II; Appendix D"
214 + },
215 + {
216 + "claim": "Market allocations can have too much or too little screening relative to second-best",
217 + "evidence_type": "figure",
218 + "evidence_ref": "Section III; Figure 3"
219 + },
220 + {
221 + "claim": "Multiple equilibria arise when gains from trade are large",
222 + "evidence_type": "equation_or_proof",
223 + "evidence_ref": "Section III.A; Appendix E"
224 + },
225 + {
226 + "claim": "Optimal mechanism decentralized via differential retention rules and transfers across securities",
227 + "evidence_type": "equation_or_proof",
228 + "evidence_ref": "Section IV"
229 + },
230 + {
231 + "claim": "Retention correlates with better CMBS performance, consistent with model",
232 + "evidence_type": "prior_literature",
233 + "evidence_ref": "Ashcraft, Gooriah, and Kermani (2014)"
234 + },
235 + {
236 + "claim": "Securitization associated with declining lending standards and originator private information",
237 + "evidence_type": "prior_literature",
238 + "evidence_ref": "Keys et al. (2010), Mian and Sufi (2009), others"
239 + }
240 + ],
241 + "structural_summary": "Theory paper. The introduction motivates the question with securitization facts and empirical literature on two frictions, sketches the model, and previews second-best and equilibrium results. It links predictions to crisis-era patterns, discusses policy and commitment, and closes with literature positioning; there is no roadmap. The body runs model, then optimal mechanism, then market equilibria (including multiplicity), then efficiency and policy with a securitization application. The conclusion restates findings and regulatory implications. Appendices hold the microfoundation, proofs, and alternative information structures.",
242 + "writing_profile_keywords": [
243 + "mechanism design",
244 + "adverse selection",
245 + "moral hazard",
246 + "securitization",
247 + "retention",
248 + "policy decentralization",
249 + "multiple equilibria",
250 + "first-person singular"
251 + ],
252 + "confidence": 0.85,
253 + "notes": "The role_rule labels for Section II (robustness) and Sections III–IV (other) were overridden to theory. Appendix F's alternative information structures serve as robustness/understanding exercises."
254 + },
255 + "deterministic_checks": {
256 + "roadmap_regex": false,
257 + "contribution_regex_in_intro": true,
258 + "magnitude_in_abstract": false,
259 + "abstract_n_sentences": 7,
260 + "intro_n_paragraphs": 9,
261 + "intro_n_words": 1552,
262 + "causal_words_in_abstract": 0,
263 + "hedge_words_in_abstract": 0,
264 + "stars_mentioned": false,
265 + "clustered_se_mentioned": false,
266 + "internet_appendix_mentioned": false,
267 + "section_roles_rule": {
268 + "0": "introduction",
269 + "1": "theory",
270 + "2": "robustness",
271 + "3": "other",
272 + "4": "other",
273 + "5": "conclusion",
274 + "6": "appendix",
275 + "7": "appendix",
276 + "8": "appendix",
277 + "9": "appendix",
278 + "10": "appendix",
279 + "11": "appendix"
280 + }
281 + },
282 + "schema_valid": false,
283 + "code_version": "feb6c9e-dirty"
284 +}
\ No newline at end of file
added data/annotations/documents/jof_12520.json +281 −0
@@ -0,0 +1,281 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12520",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:38+00:00",
9 + "source_sha256": "a137c7ec184e2efa692ff24f81d61d59d6960f30e14f8873a2c5a0bf7649fc1a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 26847,
12 + "usage": {
13 + "input": 9141,
14 + "output": 2343,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.084218,
19 + "request_id": "req_011CggpgQ3daNZ4LAGW4zepz",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "asset management / securitization (CDO collateral managers)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Association of a manager's specialization (CDO Share) with deal and collateral performance; association of past deal performance with future deal assignments; effect of CDO involvement on mutual fund flows during the crisis",
27 + "strategy": "Cross-sectional regressions with deal type × vintage fixed effects comparing similar deals closed at the same time. Poisson regressions of future deal counts on past deal characteristics. Security-level comparisons of collateral underwritten by the CDO's arranger versus other collateral, conditional on asset class × rating cells. Fund-flow regressions with objective × time fixed effects. Alternative explanations are tested.",
28 + "key_assumptions": [
29 + "CDO Share proxies for franchise-value and reputational concerns",
30 + "Deals within the same type and vintage are comparable after controls",
31 + "Observationally similar securities are comparable apart from arranger underwriting"
32 + ],
33 + "threats_addressed": [
34 + "Differences in direct incentives (fees, equity stakes)",
35 + "Employee skill and experience",
36 + "Manager optimism about housing",
37 + "Deal characteristics under manager discretion",
38 + "Magnetar-sponsored deals excluded",
39 + "Selection in securities with price data"
40 + ],
41 + "causal_language_strength": 2,
42 + "uses_causal_claim": false
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": false,
55 + "sentence_index": -1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 1
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 5
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "related_literature_positioning"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "literature_gap"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "mechanism_or_theory_sketch"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "mechanism_or_theory_sketch"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "preview_of_approach"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "preview_of_approach"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "preview_of_results"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "identification_defense"
111 + },
112 + {
113 + "paragraph_index": 5,
114 + "move": "preview_of_results"
115 + },
116 + {
117 + "paragraph_index": 5,
118 + "move": "other"
119 + },
120 + {
121 + "paragraph_index": 6,
122 + "move": "contribution_statement"
123 + },
124 + {
125 + "paragraph_index": 6,
126 + "move": "related_literature_positioning"
127 + },
128 + {
129 + "paragraph_index": 7,
130 + "move": "related_literature_positioning"
131 + },
132 + {
133 + "paragraph_index": 7,
134 + "move": "roadmap"
135 + }
136 + ],
137 + "contribution_paragraph_index": 6,
138 + "contribution_types": [
139 + "new_evidence"
140 + ],
141 + "roadmap_present": true,
142 + "literature_positioning_location": "inside_introduction"
143 + },
144 + "section_roles": [
145 + {
146 + "idx": 0,
147 + "role": "introduction"
148 + },
149 + {
150 + "idx": 1,
151 + "role": "other"
152 + },
153 + {
154 + "idx": 2,
155 + "role": "data"
156 + },
157 + {
158 + "idx": 3,
159 + "role": "results"
160 + },
161 + {
162 + "idx": 4,
163 + "role": "conclusion"
164 + },
165 + {
166 + "idx": 5,
167 + "role": "appendix"
168 + },
169 + {
170 + "idx": 6,
171 + "role": "appendix"
172 + }
173 + ],
174 + "results_presentation": {
175 + "economic_significance_discussed": true,
176 + "statistical_significance_language": "unclear",
177 + "magnitude_statements_in_intro": true
178 + },
179 + "robustness": {
180 + "location": "mixed",
181 + "types": [
182 + "alternative_measure",
183 + "alternative_specification",
184 + "subsample",
185 + "other"
186 + ]
187 + },
188 + "tables": {
189 + "convention_notes": "Table notes give the full regression equation with fixed-effect structure (type × vintage, asset class × rating × time), define variables, and state the sample and data window (e.g., IRR from cash flows through December 2015). Note openings often state the table's takeaway ('This table shows that...'). Appendix tables cover deal type mapping, alternative performance measures, and price-data selection.",
190 + "units_stated_in_notes": true,
191 + "standard_errors_description": "unclear"
192 + },
193 + "figures": {
194 + "uncertainty_shown": "unclear"
195 + },
196 + "conclusion": {
197 + "bounded_claims": true,
198 + "policy_implications": true,
199 + "open_questions": false,
200 + "restates_contribution": true
201 + },
202 + "hedging": {
203 + "level": "moderate",
204 + "typical_devices": [
205 + "'consistent with' framing of results",
206 + "'suggests' / 'potentially understate'",
207 + "explicit acknowledgement that time-series race-to-bottom explanations cannot be ruled out",
208 + "restricting claims to cross-sectional differences"
209 + ]
210 + },
211 + "claim_evidence": [
212 + {
213 + "claim": "Managers whose past CDOs performed poorly received more subsequent deal assignments",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table III; Figure 1"
216 + },
217 + {
218 + "claim": "Moving from 0% to 100% CDO Share is associated with 6.6% lower deal IRR",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table IV; Figure 2"
221 + },
222 + {
223 + "claim": "Specialists' underperformance is not explained by direct incentives, skill, or optimism",
224 + "evidence_type": "table",
225 + "evidence_ref": "Tables V, VI"
226 + },
227 + {
228 + "claim": "Arranger-underwritten collateral held by specialized managers performed worse than similar securities",
229 + "evidence_type": "table",
230 + "evidence_ref": "Table VIII; Figure 4"
231 + },
232 + {
233 + "claim": "Institutional investors withdrew from mutual funds of families that managed ABS CDOs during the crisis",
234 + "evidence_type": "table",
235 + "evidence_ref": "Table XII"
236 + },
237 + {
238 + "claim": "Specialist underperformance is robust to alternative performance measures",
239 + "evidence_type": "table",
240 + "evidence_ref": "Table A.III"
241 + }
242 + ],
243 + "structural_summary": "The introduction motivates the paper with securitization's role in the crisis and notes that collateral managers have been neglected. It then states a popular 'front men' narrative and derives four testable predictions from it, previews tests and results (including one magnitude), and openly bounds the claims to cross-sectional evidence. It closes by positioning against the securitization-conflicts and demand literatures and giving a roadmap. The body has three sections: institutional background, data built from merged sources, and a long results section. The results section tests the predictions in order, with an explicit 'alternative explanations' subsection. The conclusion restates the contribution, flags limitations, and draws implications for asset-manager policy.",
244 + "writing_profile_keywords": [
245 + "prediction-driven narrative",
246 + "quid pro quo hypothesis",
247 + "anecdotal motivation",
248 + "alternative explanations ruled out",
249 + "fixed-effects regressions",
250 + "consistent-with framing",
251 + "financial crisis",
252 + "single-author first person"
253 + ],
254 + "confidence": 0.8,
255 + "notes": "Standard-error type and significance-reporting conventions are not visible in the excerpts. The institutional background section is classed as 'other' rather than literature. Hypotheses come from a verbal narrative, not a formal model."
256 + },
257 + "deterministic_checks": {
258 + "roadmap_regex": true,
259 + "contribution_regex_in_intro": true,
260 + "magnitude_in_abstract": false,
261 + "abstract_n_sentences": 6,
262 + "intro_n_paragraphs": 8,
263 + "intro_n_words": 1384,
264 + "causal_words_in_abstract": 0,
265 + "hedge_words_in_abstract": 2,
266 + "stars_mentioned": false,
267 + "clustered_se_mentioned": false,
268 + "internet_appendix_mentioned": true,
269 + "section_roles_rule": {
270 + "0": "introduction",
271 + "1": "literature",
272 + "2": "data",
273 + "3": "results",
274 + "4": "conclusion",
275 + "5": "appendix",
276 + "6": "appendix"
277 + }
278 + },
279 + "schema_valid": true,
280 + "code_version": "feb6c9e-dirty"
281 +}
\ No newline at end of file
added data/annotations/documents/jof_12521.json +278 −0
@@ -0,0 +1,278 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12521",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:01+00:00",
9 + "source_sha256": "8a6a65a255be5521a3b22316f22349bfe9177fb2f5870fd32db2753d81aecb10",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 30294,
12 + "usage": {
13 + "input": 9999,
14 + "output": 2165,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.08409,
19 + "request_id": "req_011CggpiNCK2no8ULgMj8aPY",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Separate associations of adoptees' equity participation, risky share and portfolio volatility with those of their biological parents (prebirth/nature) and adoptive parents (postbirth/nurture)",
27 + "strategy": "Adoption design using Swedish administrative and multigenerational registers that link adoptees to both biological and adoptive parents; probit and linear regressions estimated separately for adoptees and own-birth children",
28 + "key_assumptions": [
29 + "Adoption occurs soon after birth, so biological parents contribute little postbirth environment",
30 + "Adoptees are as good as randomly assigned to adoptive families, or nonrandom placement is controlled for",
31 + "Results from adoptees and their parents generalize to the wider population"
32 + ],
33 + "threats_addressed": [
34 + "nonrandom assignment of adoptees",
35 + "timing of adoption",
36 + "cohort and age composition",
37 + "timing of when equity holding is measured",
38 + "external validity / representativeness",
39 + "omitted parental characteristics (wealth, income, education, entrepreneurship)"
40 + ],
41 + "causal_language_strength": 3,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 2
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 3
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "tension_or_puzzle"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "policy_or_broader_implication"
83 + },
84 + {
85 + "paragraph_index": 0,
86 + "move": "preview_of_approach"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 1,
94 + "move": "contribution_statement"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "related_literature_positioning"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "related_literature_positioning"
103 + },
104 + {
105 + "paragraph_index": 3,
106 + "move": "literature_gap"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "related_literature_positioning"
111 + }
112 + ],
113 + "contribution_paragraph_index": 1,
114 + "contribution_types": [
115 + "new_evidence",
116 + "new_identification"
117 + ],
118 + "roadmap_present": false,
119 + "literature_positioning_location": "inside_introduction"
120 + },
121 + "section_roles": [
122 + {
123 + "idx": 0,
124 + "role": "introduction"
125 + },
126 + {
127 + "idx": 1,
128 + "role": "other"
129 + },
130 + {
131 + "idx": 2,
132 + "role": "empirical_design"
133 + },
134 + {
135 + "idx": 3,
136 + "role": "data"
137 + },
138 + {
139 + "idx": 4,
140 + "role": "results"
141 + },
142 + {
143 + "idx": 5,
144 + "role": "robustness"
145 + },
146 + {
147 + "idx": 6,
148 + "role": "robustness"
149 + },
150 + {
151 + "idx": 7,
152 + "role": "results"
153 + },
154 + {
155 + "idx": 8,
156 + "role": "results"
157 + },
158 + {
159 + "idx": 9,
160 + "role": "conclusion"
161 + },
162 + {
163 + "idx": 10,
164 + "role": "appendix"
165 + }
166 + ],
167 + "results_presentation": {
168 + "economic_significance_discussed": true,
169 + "statistical_significance_language": "unclear",
170 + "magnitude_statements_in_intro": false
171 + },
172 + "robustness": {
173 + "location": "mixed",
174 + "types": [
175 + "subsample",
176 + "alternative_specification",
177 + "alternative_sample_period",
178 + "other"
179 + ]
180 + },
181 + "tables": {
182 + "convention_notes": "Table notes identify the estimator (probit marginal effects or linear regression), the 1999–2006 estimation window, observation and adoptee counts, and the controls included. Own-birth and adopted children are estimated separately in different panels. Adjusted R2 decompositions are reported in percentages. Appendix tables cover sample construction, reweighting and subsample splits.",
183 + "units_stated_in_notes": true,
184 + "standard_errors_description": "unclear"
185 + },
186 + "figures": {
187 + "uncertainty_shown": "no_figures"
188 + },
189 + "conclusion": {
190 + "bounded_claims": true,
191 + "policy_implications": false,
192 + "open_questions": true,
193 + "restates_contribution": true
194 + },
195 + "hedging": {
196 + "level": "moderate",
197 + "typical_devices": [
198 + "suggesting",
199 + "may play an important role",
200 + "is associated with",
201 + "it is plausible that",
202 + "not likely working through"
203 + ]
204 + },
205 + "claim_evidence": [
206 + {
207 + "claim": "Adoptees' equity participation relates to the participation of both their biological and adoptive parents",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table II"
210 + },
211 + {
212 + "claim": "Adoptive-parent (nurture) effects are larger than biological effects for the risky share and portfolio volatility",
213 + "evidence_type": "table",
214 + "evidence_ref": "Tables III, IV"
215 + },
216 + {
217 + "claim": "Estimates are robust to addressing nonrandom placement of adoptees",
218 + "evidence_type": "table",
219 + "evidence_ref": "Tables V-VII"
220 + },
221 + {
222 + "claim": "Controlling for parental education, earnings, wealth and entrepreneurship barely changes the intergenerational coefficients",
223 + "evidence_type": "table",
224 + "evidence_ref": "Tables VIII, IX"
225 + },
226 + {
227 + "claim": "Parental risk-taking variables add materially to the adjusted R2 of cross-sectional portfolio regressions",
228 + "evidence_type": "table",
229 + "evidence_ref": "Tables X, XI, A.X"
230 + },
231 + {
232 + "claim": "Twin studies attribute little role to shared environment in financial risk-taking",
233 + "evidence_type": "prior_literature",
234 + "evidence_ref": "Cesarini et al. (2010); Barnea, Cronqvist, and Siegel (2010)"
235 + }
236 + ],
237 + "structural_summary": "Opens with the documented intergenerational correlation in risk-taking and frames a nature-versus-nurture question, motivated by welfare and inequality concerns. Introduces the Swedish adoption design, in which both biological and adoptive parents are observed. Positions the paper against twin-study and adoption literatures, stressing that it targets a different estimand. Moves through institutional background, empirical specification and data, then main results for three outcomes. Follows with extensive assumption tests and external-validity checks, a mechanisms section and an explanatory-power analysis. Closes by reconciling the findings with twin studies and acknowledging limits related to age.",
238 + "writing_profile_keywords": [
239 + "adoption design",
240 + "nature vs nurture",
241 + "administrative registers",
242 + "intergenerational transmission",
243 + "external validity",
244 + "variance decomposition",
245 + "household portfolio choice"
246 + ],
247 + "confidence": 0.82,
248 + "notes": "No figures are present. Standard error type is not visible in the table excerpts. The Institutional Background section is assigned the role 'other' rather than literature. The Mechanisms section functions as results/extensions."
249 + },
250 + "deterministic_checks": {
251 + "roadmap_regex": false,
252 + "contribution_regex_in_intro": false,
253 + "magnitude_in_abstract": false,
254 + "abstract_n_sentences": 6,
255 + "intro_n_paragraphs": 5,
256 + "intro_n_words": 746,
257 + "causal_words_in_abstract": 0,
258 + "hedge_words_in_abstract": 1,
259 + "stars_mentioned": false,
260 + "clustered_se_mentioned": false,
261 + "internet_appendix_mentioned": true,
262 + "section_roles_rule": {
263 + "0": "introduction",
264 + "1": "literature",
265 + "2": "empirical_design",
266 + "3": "data",
267 + "4": "results",
268 + "5": "theory",
269 + "6": "other",
270 + "7": "robustness",
271 + "8": "other",
272 + "9": "conclusion",
273 + "10": "appendix"
274 + }
275 + },
276 + "schema_valid": true,
277 + "code_version": "feb6c9e-dirty"
278 +}
\ No newline at end of file
added data/annotations/documents/jof_12522.json +269 −0
@@ -0,0 +1,269 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12522",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:17+00:00",
9 + "source_sha256": "583feee465b88ccc5bebcb9ca367adf20c06c6c6c8ea43662c0bb8017cf71117",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21627,
12 + "usage": {
13 + "input": 7041,
14 + "output": 2274,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.074438,
19 + "request_id": "req_011CggpjPkMd3HEbAgrGHxnf",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "regression_discontinuity",
25 + "identification": {
26 + "estimand": "Effect of an information-free credit rating downgrade on consumers' subsequent borrowing, borrower actions and market exit, plus the share of the post-default credit decline attributable to the credit-reporting downgrade itself.",
27 + "strategy": "Regression discontinuity around Peru's 'rule of twenty' rating-alignment regulation. The running variable is an exchange-rate-adjusted debt share, so currency movements push a bank's share across the 20% threshold. Complete-default outcomes are then decomposed by interacting default with the expected exogenous classification change.",
28 + "key_assumptions": [
29 + "Currency movements are exogenous to individual consumers",
30 + "No manipulation of the running variable around the 20% threshold",
31 + "Borrowers just above and just below the threshold are comparable",
32 + "Threshold crossing affects outcomes only through the rating alignment"
33 + ],
34 + "threats_addressed": [
35 + "Endogeneity of current-month balance changes (handled by using lagged, exchange-rate-adjusted balances)",
36 + "Covariate balance around the threshold",
37 + "Density manipulation (McCrary test)",
38 + "Placebo sample of borrowers with homogeneous loan ratings who should be unaffected"
39 + ],
40 + "causal_language_strength": 4,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": true,
46 + "sentence_index": 0
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 1
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 1
59 + },
60 + "magnitude": {
61 + "present": true,
62 + "sentence_index": 3
63 + },
64 + "contribution": {
65 + "present": false,
66 + "sentence_index": -1
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "tension_or_puzzle"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 2,
89 + "move": "identification_defense"
90 + },
91 + {
92 + "paragraph_index": 3,
93 + "move": "preview_of_results"
94 + },
95 + {
96 + "paragraph_index": 4,
97 + "move": "related_literature_positioning"
98 + },
99 + {
100 + "paragraph_index": 4,
101 + "move": "contribution_statement"
102 + },
103 + {
104 + "paragraph_index": 5,
105 + "move": "preview_of_results"
106 + },
107 + {
108 + "paragraph_index": 5,
109 + "move": "mechanism_or_theory_sketch"
110 + },
111 + {
112 + "paragraph_index": 6,
113 + "move": "related_literature_positioning"
114 + },
115 + {
116 + "paragraph_index": 6,
117 + "move": "policy_or_broader_implication"
118 + },
119 + {
120 + "paragraph_index": 7,
121 + "move": "roadmap"
122 + }
123 + ],
124 + "contribution_paragraph_index": 4,
125 + "contribution_types": [
126 + "new_evidence",
127 + "new_identification"
128 + ],
129 + "roadmap_present": true,
130 + "literature_positioning_location": "inside_introduction"
131 + },
132 + "section_roles": [
133 + {
134 + "idx": 0,
135 + "role": "introduction"
136 + },
137 + {
138 + "idx": 1,
139 + "role": "data"
140 + },
141 + {
142 + "idx": 2,
143 + "role": "empirical_design"
144 + },
145 + {
146 + "idx": 3,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 4,
151 + "role": "conclusion"
152 + }
153 + ],
154 + "results_presentation": {
155 + "economic_significance_discussed": true,
156 + "statistical_significance_language": "stars_and_tstats",
157 + "magnitude_statements_in_intro": true
158 + },
159 + "robustness": {
160 + "location": "within_results",
161 + "types": [
162 + "placebo",
163 + "alternative_specification",
164 + "falsification"
165 + ]
166 + },
167 + "tables": {
168 + "convention_notes": "Notes give the observation level (client-bank-month), the sample definition, and the equation estimated. Later tables point back to the baseline specification of Table IV. Stars mark significance levels. Outcomes are measured at 12-, 24- and 36-month horizons. Some variables are stated in soles.",
169 + "units_stated_in_notes": true,
170 + "standard_errors_description": "unclear"
171 + },
172 + "figures": {
173 + "uncertainty_shown": "unclear"
174 + },
175 + "conclusion": {
176 + "bounded_claims": true,
177 + "policy_implications": true,
178 + "open_questions": false,
179 + "restates_contribution": true
180 + },
181 + "hedging": {
182 + "level": "moderate",
183 + "typical_devices": [
184 + "suggest",
185 + "may",
186 + "appears to",
187 + "potentially",
188 + "in our sample"
189 + ]
190 + },
191 + "claim_evidence": [
192 + {
193 + "claim": "Complete default is associated with medium-term reductions in borrowing in a random sample of the full data set",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table III"
196 + },
197 + {
198 + "claim": "Exchange-rate-adjusted share predicts actual crossing of the 20% threshold",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table IV, Figure 1"
201 + },
202 + {
203 + "claim": "Borrowers just above and below the threshold look similar on observables, with no density manipulation",
204 + "evidence_type": "figure",
205 + "evidence_ref": "Table V, Figures 2-3"
206 + },
207 + {
208 + "claim": "Crossing the threshold produces a moderate-duration rating downgrade only for borrowers with heterogeneous ratings",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table VI, Figure 4"
211 + },
212 + {
213 + "claim": "Shocked borrowers see lower loan balances, less new financing and fewer new relationships for three years",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table VII, Figure 5"
216 + },
217 + {
218 + "claim": "Shocked borrowers pay down judicial-collection loans and zero out credit cards",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table VIII, Figure 6"
221 + },
222 + {
223 + "claim": "Shocked borrowers are more likely to exit the consumer loan market at two- and three-year horizons",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table IX, Figure 7"
226 + },
227 + {
228 + "claim": "The reporting downgrade accounts for 25-65% of the post-default credit decline",
229 + "evidence_type": "table",
230 + "evidence_ref": "Tables X-XI"
231 + }
232 + ],
233 + "structural_summary": "Opens from the macro debate on weak post-2008 recovery and household borrowing constraints, then asks why negative credit events have lasting effects. Establishes a descriptive default–credit association first. It then uses a regulatory threshold combined with currency-driven variation as an RD to isolate information-free downgrades. The design is validated through first stage, covariate balance, density and placebo checks. Effects are traced on financing, borrower actions and broader outcomes, and the paper ends with a decomposition of the post-default decline. Policy implications about macro-contingent credit-history forgiveness close both the introduction and the conclusion.",
234 + "writing_profile_keywords": [
235 + "regression discontinuity",
236 + "regulatory threshold",
237 + "credit reporting",
238 + "consumer credit",
239 + "Peru",
240 + "exchange rate variation",
241 + "decomposition",
242 + "policy implication"
243 + ],
244 + "confidence": 0.85,
245 + "notes": "Standard-error type is not visible in the excerpts. Robustness checks are embedded in results subsections rather than a separate section. An Internet Appendix exists, but its content is not visible."
246 + },
247 + "deterministic_checks": {
248 + "roadmap_regex": true,
249 + "contribution_regex_in_intro": false,
250 + "magnitude_in_abstract": true,
251 + "abstract_n_sentences": 6,
252 + "intro_n_paragraphs": 8,
253 + "intro_n_words": 1114,
254 + "causal_words_in_abstract": 1,
255 + "hedge_words_in_abstract": 1,
256 + "stars_mentioned": true,
257 + "clustered_se_mentioned": false,
258 + "internet_appendix_mentioned": false,
259 + "section_roles_rule": {
260 + "0": "introduction",
261 + "1": "data",
262 + "2": "empirical_design",
263 + "3": "results",
264 + "4": "conclusion"
265 + }
266 + },
267 + "schema_valid": true,
268 + "code_version": "feb6c9e-dirty"
269 +}
\ No newline at end of file
added data/annotations/documents/jof_12523.json +268 −0
@@ -0,0 +1,268 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12523",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:19+00:00",
9 + "source_sha256": "97af167555c9e789ed0e239a82058b8f998cc2c8d1d49e45dfd02b4347f8a71b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23886,
12 + "usage": {
13 + "input": 7965,
14 + "output": 2234,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.077334,
19 + "request_id": "req_011CggpjbeyRwnQVvhd3o8eJ",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "real estate / mortgage default",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Distribution across borrowers of walk-away (threshold) negative equity / CLTV levels at which ruthless default occurs",
27 + "strategy": "Two steps: a discrete-time logit hazard model of default on flexible CLTV dummies, exploiting ZIP-level variation in the timing and size of house price declines among zero-down-payment borrowers; then maximum likelihood estimation of a gamma distribution of default costs, with defaults probabilistically classified as liquidity-driven or ruthless.",
28 + "key_assumptions": [
29 + "Default arises only from unwillingness to pay (an equity response) or inability to pay (liquidity)",
30 + "The effect of equity on default can be separated from liquidity shocks once loan-age fixed effects, quarter fixed effects and county economic conditions are controlled",
31 + "ZIP-code HPI accurately measures individual home equity",
32 + "Borrowers walk away when the financial benefits exceed the costs"
33 + ],
34 + "threats_addressed": [
35 + "liquidity/income shocks confounding default timing",
36 + "endogenous down payment choice (all borrowers start with CLTV of 100)",
37 + "lender recourse",
38 + "measurement error in house values",
39 + "non-representativeness of the nonprime sample (external validity via LPS sample)"
40 + ],
41 + "causal_language_strength": 3,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 1
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 1
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 1
60 + },
61 + "magnitude": {
62 + "present": false,
63 + "sentence_index": -1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 4
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "preview_of_approach"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "literature_gap"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "contribution_statement"
91 + },
92 + {
93 + "paragraph_index": 2,
94 + "move": "related_literature_positioning"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "tension_or_puzzle"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "policy_or_broader_implication"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "preview_of_approach"
107 + },
108 + {
109 + "paragraph_index": 5,
110 + "move": "robustness_mention"
111 + },
112 + {
113 + "paragraph_index": 6,
114 + "move": "identification_defense"
115 + },
116 + {
117 + "paragraph_index": 7,
118 + "move": "mechanism_or_theory_sketch"
119 + },
120 + {
121 + "paragraph_index": 8,
122 + "move": "policy_or_broader_implication"
123 + },
124 + {
125 + "paragraph_index": 9,
126 + "move": "roadmap"
127 + }
128 + ],
129 + "contribution_paragraph_index": 1,
130 + "contribution_types": [
131 + "new_evidence",
132 + "new_method"
133 + ],
134 + "roadmap_present": true,
135 + "literature_positioning_location": "inside_introduction"
136 + },
137 + "section_roles": [
138 + {
139 + "idx": 0,
140 + "role": "introduction"
141 + },
142 + {
143 + "idx": 1,
144 + "role": "theory"
145 + },
146 + {
147 + "idx": 2,
148 + "role": "data"
149 + },
150 + {
151 + "idx": 3,
152 + "role": "results"
153 + },
154 + {
155 + "idx": 4,
156 + "role": "robustness"
157 + },
158 + {
159 + "idx": 5,
160 + "role": "conclusion"
161 + }
162 + ],
163 + "results_presentation": {
164 + "economic_significance_discussed": true,
165 + "statistical_significance_language": "unclear",
166 + "magnitude_statements_in_intro": true
167 + },
168 + "robustness": {
169 + "location": "mixed",
170 + "types": [
171 + "alternative_specification",
172 + "subsample",
173 + "other"
174 + ]
175 + },
176 + "tables": {
177 + "convention_notes": "Notes describe the estimating equation, the number of observations and the fixed effects that are included but not shown. The logit table reports coefficients, standard errors and odds ratios. Walk-away distributions are reported as CLTV percentiles (25th, median, 75th) with N, and the baseline row is repeated across the mechanism, robustness and broader-sample tables.",
178 + "units_stated_in_notes": true,
179 + "standard_errors_description": "unclear"
180 + },
181 + "figures": {
182 + "uncertainty_shown": "unclear"
183 + },
184 + "conclusion": {
185 + "bounded_claims": true,
186 + "policy_implications": false,
187 + "open_questions": true,
188 + "restates_contribution": true
189 + },
190 + "hedging": {
191 + "level": "moderate",
192 + "typical_devices": [
193 + "suggests/appears",
194 + "probably",
195 + "we argue",
196 + "may be",
197 + "it is difficult to pin down"
198 + ]
199 + },
200 + "claim_evidence": [
201 + {
202 + "claim": "Median borrower does not walk away until CLTV reaches about 174 (equity of -74%)",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table III; Figure 7"
205 + },
206 + {
207 + "claim": "CLTV alone does not drive default until it exceeds roughly 120",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table II; Figure 6"
210 + },
211 + {
212 + "claim": "Frictionless option model predicts certain default at about -20% equity",
213 + "evidence_type": "prior_literature",
214 + "evidence_ref": "Kau, Keenan, and Kim (1994)"
215 + },
216 + {
217 + "claim": "Lender recourse is not the main driver of high walk-away thresholds",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table IV"
220 + },
221 + {
222 + "claim": "High thresholds also hold in a broader sample including GSE and portfolio loans",
223 + "evidence_type": "table",
224 + "evidence_ref": "Table VI"
225 + },
226 + {
227 + "claim": "House prices fell sharply in the four studied states",
228 + "evidence_type": "figure",
229 + "evidence_ref": "Figures 1-2"
230 + }
231 + ],
232 + "structural_summary": "The introduction moves from the crisis facts to the ruthless-default debate, then previews the headline magnitude against model predictions. It defends a two-step method that separates liquidity-driven from ruthless defaults and closes with policy implications on moral hazard. A short conceptual section defines ruthless default, followed by the data section. The empirical section estimates a hazard model and then a maximum likelihood distribution of walk-away thresholds. A long discussion section weighs financial versus behavioral mechanisms, identification, external validity and robustness. The conclusion restates the findings and leaves the behavioral drivers to future research.",
233 + "writing_profile_keywords": [
234 + "walk-away threshold",
235 + "negative equity",
236 + "hazard model",
237 + "maximum likelihood",
238 + "behavioral explanation",
239 + "policy implication",
240 + "mechanism discussion"
241 + ],
242 + "confidence": 0.8,
243 + "notes": "Primary method could also be classed as reduced_form_regression, since step 1 is a logit hazard; it was labelled structural because step 2 estimates a parametric default-cost distribution by maximum likelihood. The Discussion section's role_rule says conclusion, but its content is mechanisms and robustness, so it was assigned robustness. The way statistical significance is reported is not visible in the excerpts."
244 + },
245 + "deterministic_checks": {
246 + "roadmap_regex": true,
247 + "contribution_regex_in_intro": false,
248 + "magnitude_in_abstract": false,
249 + "abstract_n_sentences": 6,
250 + "intro_n_paragraphs": 10,
251 + "intro_n_words": 1349,
252 + "causal_words_in_abstract": 0,
253 + "hedge_words_in_abstract": 2,
254 + "stars_mentioned": false,
255 + "clustered_se_mentioned": false,
256 + "internet_appendix_mentioned": false,
257 + "section_roles_rule": {
258 + "0": "introduction",
259 + "1": "other",
260 + "2": "data",
261 + "3": "other",
262 + "4": "conclusion",
263 + "5": "conclusion"
264 + }
265 + },
266 + "schema_valid": true,
267 + "code_version": "feb6c9e-dirty"
268 +}
\ No newline at end of file
added data/annotations/documents/jof_12524.json +257 −0
@@ -0,0 +1,257 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12524",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:45+00:00",
9 + "source_sha256": "184ca3049b751f1f413b6e6a9d87cd178f37fa2864866bd9ca6d081847fe657c",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 16632,
12 + "usage": {
13 + "input": 5930,
14 + "output": 2034,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.065194,
19 + "request_id": "req_011Cggph8QKmdoupfMwz4MyG",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "fixed income / credit risk",
24 + "primary_method": "other",
25 + "identification": {
26 + "estimand": "Shares of cross-sectional and time-series variance of credit spreads attributable to long-run expected excess returns versus long-run expected credit loss",
27 + "strategy": "Campbell-Shiller-style log-linearized present-value identity combined with a panel VAR of credit spreads, excess returns, default probability, and ratings; expectations are taken from VAR forecasts",
28 + "key_assumptions": [
29 + "Log-linear approximation of bond excess returns is accurate",
30 + "VAR adequately captures conditional expectations",
31 + "Recovery rate on coupons equals recovery rate on principal",
32 + "Price data are complete upon default"
33 + ],
34 + "threats_addressed": [
35 + "Missing default prices, filled from Moody's DRS",
36 + "Sensitivity to the priority order across overlapping databases (Appendix B)",
37 + "Infrequent default, handled with a large panel"
38 + ],
39 + "causal_language_strength": 2,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": false,
65 + "sentence_index": -1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "literature_gap"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "preview_of_approach"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "related_literature_positioning"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "related_literature_positioning"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "contribution_statement"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "roadmap"
105 + }
106 + ],
107 + "contribution_paragraph_index": 2,
108 + "contribution_types": [
109 + "new_evidence",
110 + "new_data_or_measure"
111 + ],
112 + "roadmap_present": true,
113 + "literature_positioning_location": "inside_introduction"
114 + },
115 + "section_roles": [
116 + {
117 + "idx": 0,
118 + "role": "introduction"
119 + },
120 + {
121 + "idx": 1,
122 + "role": "theory"
123 + },
124 + {
125 + "idx": 2,
126 + "role": "results"
127 + },
128 + {
129 + "idx": 3,
130 + "role": "results"
131 + },
132 + {
133 + "idx": 4,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 5,
138 + "role": "conclusion"
139 + },
140 + {
141 + "idx": 6,
142 + "role": "appendix"
143 + },
144 + {
145 + "idx": 7,
146 + "role": "appendix"
147 + },
148 + {
149 + "idx": 8,
150 + "role": "appendix"
151 + }
152 + ],
153 + "results_presentation": {
154 + "economic_significance_discussed": true,
155 + "statistical_significance_language": "unclear",
156 + "magnitude_statements_in_intro": true
157 + },
158 + "robustness": {
159 + "location": "appendix",
160 + "types": [
161 + "alternative_specification"
162 + ]
163 + },
164 + "tables": {
165 + "convention_notes": "Notes state the monthly sample period 1973-2011 and define the VAR-implied long-run coefficient formula. Panels separate long-run credit-loss and excess-return coefficients and report volatility ratios. Summary statistics are shown both raw and cross-sectionally demeaned, with variables in percent.",
166 + "units_stated_in_notes": true,
167 + "standard_errors_description": "unclear"
168 + },
169 + "figures": {
170 + "uncertainty_shown": "unclear"
171 + },
172 + "conclusion": {
173 + "bounded_claims": true,
174 + "policy_implications": false,
175 + "open_questions": true,
176 + "restates_contribution": true
177 + },
178 + "hedging": {
179 + "level": "low",
180 + "typical_devices": [
181 + "'nearly as much' / 'about half' approximations",
182 + "'can shed light' / 'can be useful' for implications",
183 + "future research deferrals"
184 + ]
185 + },
186 + "claim_evidence": [
187 + {
188 + "claim": "Long-run expected credit loss volatility is about two-thirds of credit spread volatility in the cross-section.",
189 + "evidence_type": "table",
190 + "evidence_ref": "Table II"
191 + },
192 + {
193 + "claim": "Expected returns account for about half of cross-sectional credit spread variation, nearly as much as expected credit loss.",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table II"
196 + },
197 + {
198 + "claim": "Risk premiums dominate time-series variation of the aggregate market portfolio's credit spread, as credit loss diversifies away.",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table IV"
201 + },
202 + {
203 + "claim": "The joint decomposition links bond credit spreads to stock book-to-market components.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table III"
206 + },
207 + {
208 + "claim": "Long-run expected credit loss and returns vary with credit spreads differently across ratings.",
209 + "evidence_type": "figure",
210 + "evidence_ref": "Figure 1"
211 + },
212 + {
213 + "claim": "Results are not driven by the priority order among overlapping databases.",
214 + "evidence_type": "table",
215 + "evidence_ref": "Appendix B comparison of monthly returns"
216 + }
217 + ],
218 + "structural_summary": "Opens by asking what drives credit spread variation and notes that structural models disagree on the components. Adapts the Campbell-Shiller present-value decomposition to bonds, with credit spread in place of the dividend-price ratio and credit loss in place of dividend growth. Builds a large merged bond panel and previews the volatility-ratio results, then positions the paper against structural, reduced-form, and price-ratio literatures. Section I derives the identity. Section II covers data, VAR estimation, and main results. Section III extends to a joint stock-bond decomposition, and Section IV turns to the aggregate portfolio, where risk premiums dominate. The conclusion restates findings and points to investment and illiquidity extensions. Appendices give the derivation and data details.",
219 + "writing_profile_keywords": [
220 + "variance decomposition",
221 + "present-value identity",
222 + "VAR",
223 + "model-free",
224 + "credit spreads",
225 + "cross-section vs aggregate",
226 + "concise"
227 + ],
228 + "confidence": 0.8,
229 + "notes": "The primary method is coded 'other' because the approach is a VAR-based present-value variance decomposition. Robustness is visible only through the database priority-order check in Appendix B. Significance reporting conventions are not visible in the excerpts."
230 + },
231 + "deterministic_checks": {
232 + "roadmap_regex": true,
233 + "contribution_regex_in_intro": true,
234 + "magnitude_in_abstract": false,
235 + "abstract_n_sentences": 5,
236 + "intro_n_paragraphs": 5,
237 + "intro_n_words": 646,
238 + "causal_words_in_abstract": 0,
239 + "hedge_words_in_abstract": 0,
240 + "stars_mentioned": false,
241 + "clustered_se_mentioned": false,
242 + "internet_appendix_mentioned": false,
243 + "section_roles_rule": {
244 + "0": "introduction",
245 + "1": "other",
246 + "2": "results",
247 + "3": "other",
248 + "4": "other",
249 + "5": "conclusion",
250 + "6": "appendix",
251 + "7": "appendix",
252 + "8": "appendix"
253 + }
254 + },
255 + "schema_valid": true,
256 + "code_version": "feb6c9e-dirty"
257 +}
\ No newline at end of file
added data/annotations/documents/jof_12525.json +259 −0
@@ -0,0 +1,259 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12525",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:41+00:00",
9 + "source_sha256": "764af416d84f7a330ab726d5ab8f4375cd64db234044768553342aa78671a0ff",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 17831,
12 + "usage": {
13 + "input": 6278,
14 + "output": 1937,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.064646,
19 + "request_id": "req_011Cggph274eHbEbFRmDdQqx",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical welfare comparison of benchmark vs. no-benchmark equilibria)",
27 + "strategy": "not applicable; search-model equilibrium comparison and mechanism design",
28 + "key_assumptions": [
29 + "search frictions with slow and fast traders",
30 + "dealer cost is common (homogeneous) or heterogeneous",
31 + "parameter conditions vary by result"
32 + ],
33 + "threats_addressed": [],
34 + "causal_language_strength": 1,
35 + "uses_causal_claim": false
36 + },
37 + "abstract_moves": {
38 + "question": {
39 + "present": true,
40 + "sentence_index": 0
41 + },
42 + "gap": {
43 + "present": false,
44 + "sentence_index": -1
45 + },
46 + "method": {
47 + "present": true,
48 + "sentence_index": 4
49 + },
50 + "results": {
51 + "present": true,
52 + "sentence_index": 1
53 + },
54 + "magnitude": {
55 + "present": false,
56 + "sentence_index": -1
57 + },
58 + "contribution": {
59 + "present": true,
60 + "sentence_index": 0
61 + }
62 + },
63 + "introduction": {
64 + "moves": [
65 + {
66 + "paragraph_index": 0,
67 + "move": "motivation"
68 + },
69 + {
70 + "paragraph_index": 0,
71 + "move": "preview_of_approach"
72 + },
73 + {
74 + "paragraph_index": 1,
75 + "move": "motivation"
76 + },
77 + {
78 + "paragraph_index": 2,
79 + "move": "preview_of_results"
80 + },
81 + {
82 + "paragraph_index": 3,
83 + "move": "mechanism_or_theory_sketch"
84 + },
85 + {
86 + "paragraph_index": 4,
87 + "move": "mechanism_or_theory_sketch"
88 + },
89 + {
90 + "paragraph_index": 5,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 6,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 7,
99 + "move": "preview_of_results"
100 + },
101 + {
102 + "paragraph_index": 7,
103 + "move": "policy_or_broader_implication"
104 + },
105 + {
106 + "paragraph_index": 8,
107 + "move": "related_literature_positioning"
108 + },
109 + {
110 + "paragraph_index": 9,
111 + "move": "roadmap"
112 + }
113 + ],
114 + "contribution_paragraph_index": 0,
115 + "contribution_types": [
116 + "new_theory_or_model"
117 + ],
118 + "roadmap_present": true,
119 + "literature_positioning_location": "inside_introduction"
120 + },
121 + "section_roles": [
122 + {
123 + "idx": 0,
124 + "role": "introduction"
125 + },
126 + {
127 + "idx": 1,
128 + "role": "other"
129 + },
130 + {
131 + "idx": 2,
132 + "role": "theory"
133 + },
134 + {
135 + "idx": 3,
136 + "role": "theory"
137 + },
138 + {
139 + "idx": 4,
140 + "role": "other"
141 + },
142 + {
143 + "idx": 5,
144 + "role": "conclusion"
145 + },
146 + {
147 + "idx": 6,
148 + "role": "appendix"
149 + },
150 + {
151 + "idx": 7,
152 + "role": "appendix"
153 + }
154 + ],
155 + "results_presentation": {
156 + "economic_significance_discussed": false,
157 + "statistical_significance_language": "unclear",
158 + "magnitude_statements_in_intro": false
159 + },
160 + "robustness": {
161 + "location": "none_seen",
162 + "types": []
163 + },
164 + "tables": {
165 + "convention_notes": "No tables. Results are stated as propositions and theorems, with proofs in the appendices. Figures illustrate numerical examples of surplus, entry probability, prices, and equilibrium regimes.",
166 + "units_stated_in_notes": false,
167 + "standard_errors_description": "none"
168 + },
169 + "figures": {
170 + "uncertainty_shown": "no"
171 + },
172 + "conclusion": {
173 + "bounded_claims": true,
174 + "policy_implications": true,
175 + "open_questions": false,
176 + "restates_contribution": true
177 + },
178 + "hedging": {
179 + "level": "moderate",
180 + "typical_devices": [
181 + "under natural parameter assumptions",
182 + "conditions under which",
183 + "may",
184 + "in some cases",
185 + "is likely to be",
186 + "suggests that there may be"
187 + ]
188 + },
189 + "claim_evidence": [
190 + {
191 + "claim": "Publishing a benchmark encourages efficient trader entry and mitigates dealer hold-up",
192 + "evidence_type": "equation_or_proof",
193 + "evidence_ref": "Section II, Appendix A"
194 + },
195 + {
196 + "claim": "Benchmarks improve matching efficiency and raise the market share of low-cost dealers",
197 + "evidence_type": "equation_or_proof",
198 + "evidence_ref": "Section III, Appendix B"
199 + },
200 + {
201 + "claim": "Dealers may introduce benchmarks when volume gains offset reduced margins",
202 + "evidence_type": "equation_or_proof",
203 + "evidence_ref": "Section II.E"
204 + },
205 + {
206 + "claim": "The optimal revelation mechanism coincides with announcing the benchmark under stated conditions",
207 + "evidence_type": "equation_or_proof",
208 + "evidence_ref": "Section II.F-G"
209 + },
210 + {
211 + "claim": "Benchmarks underlie trillions of dollars in loans and derivatives",
212 + "evidence_type": "prior_literature",
213 + "evidence_ref": "MPG (2014); BIS (2014)"
214 + },
215 + {
216 + "claim": "Benchmarks increase surplus relative to no benchmark in a numerical example",
217 + "evidence_type": "figure",
218 + "evidence_ref": "Figures 1-3"
219 + }
220 + ],
221 + "structural_summary": "Theory paper. The introduction motivates the question with the scale of benchmark-referenced OTC markets. It then previews three welfare channels (entry, matching, search costs), cases where a benchmark lowers welfare, an optimal-mechanism result, and dealers' incentives to introduce benchmarks. A brief literature comparison and a roadmap follow. An institutional section describes how benchmarks are used. The core search model with homogeneous dealer costs compares equilibria with and without a benchmark, derives the welfare comparison and dealer incentives, and characterizes the optimal mechanism. An extension adds heterogeneous dealer costs to study matching efficiency. A short discussion of manipulation precedes a conclusion with policy remarks. Proofs are in the appendices.",
222 + "writing_profile_keywords": [
223 + "search model",
224 + "OTC markets",
225 + "price transparency",
226 + "mechanism design",
227 + "welfare comparison",
228 + "proposition-proof structure",
229 + "policy relevance"
230 + ],
231 + "confidence": 0.88,
232 + "notes": "Purely theoretical; there are no tables or empirical estimates. Institutional magnitudes (trillions of dollars) are used only as motivation. One author chaired the MPG report."
233 + },
234 + "deterministic_checks": {
235 + "roadmap_regex": true,
236 + "contribution_regex_in_intro": false,
237 + "magnitude_in_abstract": false,
238 + "abstract_n_sentences": 5,
239 + "intro_n_paragraphs": 10,
240 + "intro_n_words": 978,
241 + "causal_words_in_abstract": 0,
242 + "hedge_words_in_abstract": 2,
243 + "stars_mentioned": false,
244 + "clustered_se_mentioned": false,
245 + "internet_appendix_mentioned": true,
246 + "section_roles_rule": {
247 + "0": "introduction",
248 + "1": "other",
249 + "2": "theory",
250 + "3": "other",
251 + "4": "other",
252 + "5": "conclusion",
253 + "6": "appendix",
254 + "7": "appendix"
255 + }
256 + },
257 + "schema_valid": true,
258 + "code_version": "feb6c9e-dirty"
259 +}
\ No newline at end of file
added data/annotations/documents/jof_12526.json +221 −0
@@ -0,0 +1,221 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12526",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:58:58+00:00",
9 + "source_sha256": "a9f703cc3137b18980c93979d154cd4b0e1a21b2a6f33fe4c2e11f728ca906ba",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 10936,
12 + "usage": {
13 + "input": 3773,
14 + "output": 1852,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.052926,
19 + "request_id": "req_011Cggpi94nVZTmVD7qfL9Fi",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "organizational economics / internal capital markets; contract and mechanism design",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; the paper does an analytical mechanism-design analysis of a principal-agent model with private information",
28 + "key_assumptions": [
29 + "risk-neutral economy",
30 + "worker (agent) is penniless, has limited liability and a zero reservation wage",
31 + "agent's effort is unobservable",
32 + "executive privately observes a binary productivity shock",
33 + "effort is more productive when the firm's prospects are better"
34 + ],
35 + "threats_addressed": [],
36 + "causal_language_strength": 1,
37 + "uses_causal_claim": false
38 + },
39 + "abstract_moves": {
40 + "question": {
41 + "present": true,
42 + "sentence_index": 2
43 + },
44 + "gap": {
45 + "present": false,
46 + "sentence_index": -1
47 + },
48 + "method": {
49 + "present": true,
50 + "sentence_index": 0
51 + },
52 + "results": {
53 + "present": true,
54 + "sentence_index": 2
55 + },
56 + "magnitude": {
57 + "present": false,
58 + "sentence_index": -1
59 + },
60 + "contribution": {
61 + "present": true,
62 + "sentence_index": 0
63 + }
64 + },
65 + "introduction": {
66 + "moves": [
67 + {
68 + "paragraph_index": 0,
69 + "move": "preview_of_approach"
70 + },
71 + {
72 + "paragraph_index": 0,
73 + "move": "mechanism_or_theory_sketch"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "preview_of_results"
78 + }
79 + ],
80 + "contribution_paragraph_index": 0,
81 + "contribution_types": [
82 + "new_theory_or_model"
83 + ],
84 + "roadmap_present": false,
85 + "literature_positioning_location": "inside_introduction"
86 + },
87 + "section_roles": [
88 + {
89 + "idx": 0,
90 + "role": "introduction"
91 + },
92 + {
93 + "idx": 1,
94 + "role": "theory"
95 + },
96 + {
97 + "idx": 2,
98 + "role": "theory"
99 + },
100 + {
101 + "idx": 3,
102 + "role": "theory"
103 + },
104 + {
105 + "idx": 4,
106 + "role": "other"
107 + },
108 + {
109 + "idx": 5,
110 + "role": "other"
111 + },
112 + {
113 + "idx": 6,
114 + "role": "conclusion"
115 + },
116 + {
117 + "idx": 7,
118 + "role": "appendix"
119 + }
120 + ],
121 + "results_presentation": {
122 + "economic_significance_discussed": false,
123 + "statistical_significance_language": "unclear",
124 + "magnitude_statements_in_intro": false
125 + },
126 + "robustness": {
127 + "location": "main_text_section",
128 + "types": [
129 + "other"
130 + ]
131 + },
132 + "tables": {
133 + "convention_notes": "There are no tables. Results are stated as propositions, with proofs in Appendix A.",
134 + "units_stated_in_notes": false,
135 + "standard_errors_description": "none"
136 + },
137 + "figures": {
138 + "uncertainty_shown": "no"
139 + },
140 + "conclusion": {
141 + "bounded_claims": true,
142 + "policy_implications": false,
143 + "open_questions": false,
144 + "restates_contribution": true
145 + },
146 + "hedging": {
147 + "level": "moderate",
148 + "typical_devices": [
149 + "modal verbs ('may distort', 'can arise')",
150 + "'suggests' wording for model implications",
151 + "'consistent with' when linking predictions to prior evidence",
152 + "conditional framing ('in some situations')"
153 + ]
154 + },
155 + "claim_evidence": [
156 + {
157 + "claim": "When executives are privately informed, an unconstrained executive tends to overinvest relative to first best, because higher investment signals better prospects and induces more worker effort.",
158 + "evidence_type": "equation_or_proof",
159 + "evidence_ref": "Section II.B.3; Appendix A proofs"
160 + },
161 + {
162 + "claim": "Investment rigidities, implemented as pooling mechanisms, can be an optimal response to the tendency to overinvest.",
163 + "evidence_type": "equation_or_proof",
164 + "evidence_ref": "Section II.B.4-B.5; Appendix A"
165 + },
166 + {
167 + "claim": "Multidivisional firms show investment distortions that differ from those of single-division firms, with partial pooling mechanisms examined.",
168 + "evidence_type": "equation_or_proof",
169 + "evidence_ref": "Section III"
170 + },
171 + {
172 + "claim": "In a continuum-of-types extension, overinvestment persists and the optimal mechanism pools types at the bottom of the distribution.",
173 + "evidence_type": "none_stated",
174 + "evidence_ref": "Section IV"
175 + },
176 + {
177 + "claim": "Minimum investment thresholds, lumpy investment and expenditure caps observed in firms are consistent with the model's predicted rigidities.",
178 + "evidence_type": "prior_literature",
179 + "evidence_ref": "Section V; Cooper and Haltiwanger (2006), Ross (1986)"
180 + }
181 + ],
182 + "structural_summary": "A pure theory paper on top-down capital budgeting.\n- The introduction motivates the question, outlines the principal-agent setting (an executive whose investment choices signal information to a worker), and previews the main results.\n- The model section sets up the game. The analysis section first solves a known-type benchmark, then the private-information case: optimal separating vs. pooling mechanisms and when each is preferred.\n- A separate section extends the analysis to a two-division firm and characterizes partial-pooling mechanisms.\n- Short sections then discuss extensions (e.g., a continuum of types) and empirical implications tied to prior evidence.\n- The conclusion restates the contribution. Proofs are in a long appendix.",
183 + "writing_profile_keywords": [
184 + "mechanism design",
185 + "signaling",
186 + "principal-agent",
187 + "capital budgeting",
188 + "overinvestment",
189 + "pooling vs separating",
190 + "proposition-proof structure",
191 + "empirical implications section"
192 + ],
193 + "confidence": 0.8,
194 + "notes": "The introduction is given as a single truncated paragraph, so moves beyond it (e.g., literature positioning, roadmap) are not visible. The literature-positioning value is assigned on the assumption that it sits inside the introduction, since no separate literature section exists. The Extensions section is coded 'other' because it extends the theory rather than performing empirical robustness checks. Figures are model illustrations without uncertainty bands."
195 + },
196 + "deterministic_checks": {
197 + "roadmap_regex": false,
198 + "contribution_regex_in_intro": false,
199 + "magnitude_in_abstract": false,
200 + "abstract_n_sentences": 4,
201 + "intro_n_paragraphs": 1,
202 + "intro_n_words": 228,
203 + "causal_words_in_abstract": 0,
204 + "hedge_words_in_abstract": 1,
205 + "stars_mentioned": false,
206 + "clustered_se_mentioned": false,
207 + "internet_appendix_mentioned": false,
208 + "section_roles_rule": {
209 + "0": "introduction",
210 + "1": "theory",
211 + "2": "theory",
212 + "3": "other",
213 + "4": "robustness",
214 + "5": "other",
215 + "6": "conclusion",
216 + "7": "appendix"
217 + }
218 + },
219 + "schema_valid": true,
220 + "code_version": "feb6c9e-dirty"
221 +}
\ No newline at end of file
added data/annotations/documents/jof_12530.json +275 −0
@@ -0,0 +1,275 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12530",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:57:47+00:00",
9 + "source_sha256": "e7091a0d4b8826fccbb9cdfd5a1a80d2b93abaa741a5d2db98fb07459b1721af",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21303,
12 + "usage": {
13 + "input": 7476,
14 + "output": 2100,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.072698,
19 + "request_id": "req_011CggpcpmpCGxdqD1gXthDB",
20 + "annotation": {
21 + "genre": "survey_or_address",
22 + "subfield": "financial_econometrics",
23 + "subfield_secondary": "research methodology / meta-science of publication bias in asset pricing",
24 + "primary_method": "other",
25 + "identification": {
26 + "estimand": "not applicable (methodological address); the target quantity is the posterior probability that the null is true given a reported p-value and prior odds",
27 + "strategy": "not applicable; conceptual argument, a hypothetical ticker-symbol data-mining example, and Bayesian conversion of p-values using minimum Bayes factors",
28 + "key_assumptions": [
29 + "prior odds on the null vs. alternative can be specified from economic plausibility",
30 + "minimum Bayes factor bounds the evidence against the null"
31 + ],
32 + "threats_addressed": [
33 + "multiple testing",
34 + "p-hacking",
35 + "publication bias / file drawer effect",
36 + "low prior probability of true effects (rare incidence)"
37 + ],
38 + "causal_language_strength": 2,
39 + "uses_causal_claim": false
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 4
45 + },
46 + "gap": {
47 + "present": true,
48 + "sentence_index": 3
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 5
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 1
57 + },
58 + "magnitude": {
59 + "present": false,
60 + "sentence_index": -1
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 6
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "mechanism_or_theory_sketch"
80 + },
81 + {
82 + "paragraph_index": 2,
83 + "move": "related_literature_positioning"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "policy_or_broader_implication"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "roadmap"
92 + }
93 + ],
94 + "contribution_paragraph_index": 2,
95 + "contribution_types": [
96 + "new_method",
97 + "synthesis"
98 + ],
99 + "roadmap_present": true,
100 + "literature_positioning_location": "both"
101 + },
102 + "section_roles": [
103 + {
104 + "idx": 0,
105 + "role": "introduction"
106 + },
107 + {
108 + "idx": 1,
109 + "role": "literature"
110 + },
111 + {
112 + "idx": 2,
113 + "role": "other"
114 + },
115 + {
116 + "idx": 3,
117 + "role": "other"
118 + },
119 + {
120 + "idx": 4,
121 + "role": "literature"
122 + },
123 + {
124 + "idx": 5,
125 + "role": "theory"
126 + },
127 + {
128 + "idx": 6,
129 + "role": "other"
130 + },
131 + {
132 + "idx": 7,
133 + "role": "theory"
134 + },
135 + {
136 + "idx": 8,
137 + "role": "theory"
138 + },
139 + {
140 + "idx": 9,
141 + "role": "theory"
142 + },
143 + {
144 + "idx": 10,
145 + "role": "theory"
146 + },
147 + {
148 + "idx": 11,
149 + "role": "robustness"
150 + },
151 + {
152 + "idx": 12,
153 + "role": "other"
154 + },
155 + {
156 + "idx": 13,
157 + "role": "conclusion"
158 + }
159 + ],
160 + "results_presentation": {
161 + "economic_significance_discussed": false,
162 + "statistical_significance_language": "mixed",
163 + "magnitude_statements_in_intro": true
164 + },
165 + "robustness": {
166 + "location": "none_seen",
167 + "types": []
168 + },
169 + "tables": {
170 + "convention_notes": "Tables are illustrative computation grids. Rows are significance levels or z-scores/p-values, and columns are prior odds labeled from 'long shot' to 'even odds'. Notes define the MBF and SD-MBF formulas. One table re-expresses p-values from published papers as Bayesianized p-values.",
171 + "units_stated_in_notes": true,
172 + "standard_errors_description": "none"
173 + },
174 + "figures": {
175 + "uncertainty_shown": "no"
176 + },
177 + "conclusion": {
178 + "bounded_claims": true,
179 + "policy_implications": true,
180 + "open_questions": true,
181 + "restates_contribution": false
182 + },
183 + "hedging": {
184 + "level": "moderate",
185 + "typical_devices": [
186 + "modal verbs (may, likely)",
187 + "framing as a conversation starter",
188 + "disclaimer that past research quality is high",
189 + "extended quotations from authorities (Hubble, Russell)"
190 + ]
191 + },
192 + "claim_evidence": [
193 + {
194 + "claim": "Searching many specifications can produce a seemingly significant factor from economically meaningless sorting such as ticker letters.",
195 + "evidence_type": "descriptive_statistic",
196 + "evidence_ref": "Introduction ticker example (t=3.23)"
197 + },
198 + {
199 + "claim": "Published factor t-statistics cluster just above conventional thresholds, consistent with publication bias.",
200 + "evidence_type": "figure",
201 + "evidence_ref": "Figure 1, Figure 6 (HLZ 2016)"
202 + },
203 + {
204 + "claim": "Economics and business publish supportive results at rates close to psychology and above the physical sciences.",
205 + "evidence_type": "figure",
206 + "evidence_ref": "Figure 2 (Fanelli 2010)"
207 + },
208 + {
209 + "claim": "The share of positive results has risen over time across sciences.",
210 + "evidence_type": "figure",
211 + "evidence_ref": "Figure 3 (Fanelli 2012)"
212 + },
213 + {
214 + "claim": "With low prior odds, even strict significance thresholds leave a high probability that the null is true.",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table I, Table II"
217 + },
218 + {
219 + "claim": "The required t-statistic threshold depends on the prior plausibility of the hypothesis.",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table III"
222 + },
223 + {
224 + "claim": "Reported p-values in published factor papers translate into varied Bayesianized p-values depending on prior odds.",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table IV"
227 + }
228 + ],
229 + "structural_summary": "A presidential address that opens with a provocative hypothetical data-mining example and then frames the problem as an agency conflict between editors and authors. It brings in meta-scientific evidence from other fields and reviews what p-values mean and their limitations. It then discusses p-hacking and the low base rates of true effects, and makes the case for Bayesian reasoning. As a compromise, it proposes minimum Bayes factors and illustrates them with tables and applications to published papers. It notes limitations, closes with research-culture recommendations, and ends with a reflective, quotation-rich conclusion.",
230 + "writing_profile_keywords": [
231 + "presidential address",
232 + "first-person voice",
233 + "meta-science",
234 + "p-hacking",
235 + "Bayesian inference",
236 + "minimum Bayes factor",
237 + "publication bias",
238 + "normative recommendations",
239 + "illustrative examples"
240 + ],
241 + "confidence": 0.8,
242 + "notes": "This is not an empirical paper with identification. The figures are mostly reproduced from other studies. Section headings idx 2–4 are artifacts of figure panel labels; idx 4 continues the discussion of Fanelli's explanations. Robustness checks are not applicable, and Section VIII discusses limitations of the method rather than reporting robustness tests."
243 + },
244 + "deterministic_checks": {
245 + "roadmap_regex": false,
246 + "contribution_regex_in_intro": true,
247 + "magnitude_in_abstract": true,
248 + "abstract_n_sentences": 16,
249 + "intro_n_paragraphs": 3,
250 + "intro_n_words": 600,
251 + "causal_words_in_abstract": 0,
252 + "hedge_words_in_abstract": 3,
253 + "stars_mentioned": false,
254 + "clustered_se_mentioned": false,
255 + "internet_appendix_mentioned": false,
256 + "section_roles_rule": {
257 + "0": "introduction",
258 + "1": "results",
259 + "2": "other",
260 + "3": "other",
261 + "4": "other",
262 + "5": "other",
263 + "6": "other",
264 + "7": "other",
265 + "8": "other",
266 + "9": "other",
267 + "10": "other",
268 + "11": "robustness",
269 + "12": "other",
270 + "13": "conclusion"
271 + }
272 + },
273 + "schema_valid": true,
274 + "code_version": "feb6c9e-dirty"
275 +}
\ No newline at end of file
added data/annotations/documents/jof_12539.json +288 −0
@@ -0,0 +1,288 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12539",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:22+00:00",
9 + "source_sha256": "c5aa84514b88774c31f9f8066a127dc669e5c564ba7b7489cfe678b4cd1ac61b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20982,
12 + "usage": {
13 + "input": 7341,
14 + "output": 2304,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.076238,
19 + "request_id": "req_011Cggpp9jdXgcMqcfjA3nWQ",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "macro_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "calibration_simulation",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model; equilibrium effects of nominal rate policy rules on leverage, risk premia, asset prices and growth)",
27 + "strategy": "Two-agent continuous-time general equilibrium model in which a liquidity premium links the nominal rate to the cost of leverage. Solved globally by Chebyshev collocation. Policy rules are compared across equilibria; impulse responses are computed in an extended production model.",
28 + "key_assumptions": [
29 + "Banks face funding shocks that force costly asset liquidation unless they hold liquid buffers",
30 + "Central bank controls the liquidity premium via the nominal rate and the supply of liquid assets",
31 + "Heterogeneous risk aversion between banks and depositors"
32 + ],
33 + "threats_addressed": [],
34 + "causal_language_strength": 4,
35 + "uses_causal_claim": true
36 + },
37 + "abstract_moves": {
38 + "question": {
39 + "present": true,
40 + "sentence_index": 0
41 + },
42 + "gap": {
43 + "present": false,
44 + "sentence_index": -1
45 + },
46 + "method": {
47 + "present": true,
48 + "sentence_index": 1
49 + },
50 + "results": {
51 + "present": true,
52 + "sentence_index": 4
53 + },
54 + "magnitude": {
55 + "present": false,
56 + "sentence_index": -1
57 + },
58 + "contribution": {
59 + "present": true,
60 + "sentence_index": 0
61 + }
62 + },
63 + "introduction": {
64 + "moves": [
65 + {
66 + "paragraph_index": 0,
67 + "move": "motivation"
68 + },
69 + {
70 + "paragraph_index": 0,
71 + "move": "tension_or_puzzle"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "preview_of_approach"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "mechanism_or_theory_sketch"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "other"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "mechanism_or_theory_sketch"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "preview_of_approach"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 4,
99 + "move": "preview_of_results"
100 + },
101 + {
102 + "paragraph_index": 5,
103 + "move": "preview_of_results"
104 + },
105 + {
106 + "paragraph_index": 6,
107 + "move": "preview_of_results"
108 + },
109 + {
110 + "paragraph_index": 6,
111 + "move": "policy_or_broader_implication"
112 + },
113 + {
114 + "paragraph_index": 7,
115 + "move": "preview_of_approach"
116 + },
117 + {
118 + "paragraph_index": 8,
119 + "move": "preview_of_results"
120 + },
121 + {
122 + "paragraph_index": 8,
123 + "move": "related_literature_positioning"
124 + },
125 + {
126 + "paragraph_index": 8,
127 + "move": "roadmap"
128 + }
129 + ],
130 + "contribution_paragraph_index": 0,
131 + "contribution_types": [
132 + "new_theory_or_model"
133 + ],
134 + "roadmap_present": true,
135 + "literature_positioning_location": "separate_section"
136 + },
137 + "section_roles": [
138 + {
139 + "idx": 0,
140 + "role": "introduction"
141 + },
142 + {
143 + "idx": 1,
144 + "role": "literature"
145 + },
146 + {
147 + "idx": 2,
148 + "role": "theory"
149 + },
150 + {
151 + "idx": 3,
152 + "role": "theory"
153 + },
154 + {
155 + "idx": 4,
156 + "role": "results"
157 + },
158 + {
159 + "idx": 5,
160 + "role": "results"
161 + },
162 + {
163 + "idx": 6,
164 + "role": "conclusion"
165 + },
166 + {
167 + "idx": 7,
168 + "role": "appendix"
169 + },
170 + {
171 + "idx": 8,
172 + "role": "appendix"
173 + }
174 + ],
175 + "results_presentation": {
176 + "economic_significance_discussed": true,
177 + "statistical_significance_language": "unclear",
178 + "magnitude_statements_in_intro": false
179 + },
180 + "robustness": {
181 + "location": "none_seen",
182 + "types": []
183 + },
184 + "tables": {
185 + "convention_notes": "Only two tables, both listing calibrated parameters with their symbols and values. Notes link each table to the sections and figures it generates. The extended-model table lists only the parameters that differ from the baseline.",
186 + "units_stated_in_notes": false,
187 + "standard_errors_description": "none"
188 + },
189 + "figures": {
190 + "uncertainty_shown": "no"
191 + },
192 + "conclusion": {
193 + "bounded_claims": false,
194 + "policy_implications": true,
195 + "open_questions": true,
196 + "restates_contribution": true
197 + },
198 + "hedging": {
199 + "level": "low",
200 + "typical_devices": [
201 + "'arguably' when interpreting historical episodes",
202 + "'consistent with' when linking model output to empirical findings",
203 + "'potentially greater instability'"
204 + ]
205 + },
206 + "claim_evidence": [
207 + {
208 + "claim": "Nominal policy rate and the government-bond liquidity premium co-move strongly over time",
209 + "evidence_type": "figure",
210 + "evidence_ref": "Figure 1 (correlation 78%, 1955-2010)"
211 + },
212 + {
213 + "claim": "Lower nominal rates raise bank leverage and lower risk premia and Sharpe ratios",
214 + "evidence_type": "figure",
215 + "evidence_ref": "Figure 2"
216 + },
217 + {
218 + "claim": "High-rate policy lowers valuations despite a lower real rate, and affects volatility",
219 + "evidence_type": "figure",
220 + "evidence_ref": "Figure 3"
221 + },
222 + {
223 + "claim": "Forward guidance raises asset prices ahead of liftoff; small timing changes cause large price responses",
224 + "evidence_type": "figure",
225 + "evidence_ref": "Figure 5"
226 + },
227 + {
228 + "claim": "Greenspan put stabilizes prices locally but leads to larger crashes and volatility later",
229 + "evidence_type": "figure",
230 + "evidence_ref": "Figure 6"
231 + },
232 + {
233 + "claim": "Positive nominal rate shock lowers investment and growth and permanently reduces output",
234 + "evidence_type": "figure",
235 + "evidence_ref": "Figures 7-8"
236 + },
237 + {
238 + "claim": "Model generates an upward-sloping term structure and a term premium that rises after rate shocks",
239 + "evidence_type": "figure",
240 + "evidence_ref": "Figure 9"
241 + },
242 + {
243 + "claim": "Equilibrium characterization and analytical expressions for key mechanisms",
244 + "evidence_type": "equation_or_proof",
245 + "evidence_ref": "Section III; Appendix A"
246 + }
247 + ],
248 + "structural_summary": "The paper opens with stylized facts and policy debates suggesting a risk premium channel of monetary policy, then states that it builds a model of that channel. A single motivating figure supports the key mechanism linking the nominal rate to the liquidity premium. The introduction previews numerically solved results in order: static policy rules, liquidity holdings, forward guidance, Greenspan put, and an extended production model with impulse responses and term structure. A roadmap follows. The literature has its own section. The model and equilibrium sections derive analytical characterizations, with proofs in appendices. The results section calibrates the model and presents results mainly through figures. A further section adds production and nominal rate shocks. The conclusion restates the mechanism and suggests unconventional policy analysis for future work.",
249 + "writing_profile_keywords": [
250 + "continuous-time general equilibrium",
251 + "heterogeneous agents",
252 + "liquidity premium mechanism",
253 + "global numerical solution",
254 + "figure-driven results",
255 + "policy experiments",
256 + "impulse responses",
257 + "motivating stylized fact"
258 + ],
259 + "confidence": 0.85,
260 + "notes": "The genre is theoretical with a calibration/simulation method. The only empirical content is the motivating Figure 1. No robustness or sensitivity analysis is visible in the excerpts. Abstract sentence 0 combines the research question with the contribution."
261 + },
262 + "deterministic_checks": {
263 + "roadmap_regex": true,
264 + "contribution_regex_in_intro": false,
265 + "magnitude_in_abstract": false,
266 + "abstract_n_sentences": 6,
267 + "intro_n_paragraphs": 9,
268 + "intro_n_words": 1288,
269 + "causal_words_in_abstract": 0,
270 + "hedge_words_in_abstract": 0,
271 + "stars_mentioned": false,
272 + "clustered_se_mentioned": false,
273 + "internet_appendix_mentioned": false,
274 + "section_roles_rule": {
275 + "0": "introduction",
276 + "1": "literature",
277 + "2": "theory",
278 + "3": "theory",
279 + "4": "results",
280 + "5": "other",
281 + "6": "conclusion",
282 + "7": "appendix",
283 + "8": "appendix"
284 + }
285 + },
286 + "schema_valid": true,
287 + "code_version": "feb6c9e-dirty"
288 +}
\ No newline at end of file
added data/annotations/documents/jof_12540.json +296 −0
@@ -0,0 +1,296 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('notes' was unexpected) @ figures"
4 + ],
5 + "doc_id": "jof_12540",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:59:18+00:00",
11 + "source_sha256": "fc0b38b62ff10b679833bc183e307748583d05c6ac8784fae1a24a2e2b2a5893",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 24070,
14 + "usage": {
15 + "input": 8517,
16 + "output": 2384,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.082542,
21 + "request_id": "req_011CggpjQVG8CmPPRnu6HP6T",
22 + "annotation": {
23 + "genre": "theoretical",
24 + "subfield": "macro_finance",
25 + "subfield_secondary": "financial_intermediation_banking",
26 + "primary_method": "calibration_simulation",
27 + "identification": {
28 + "estimand": "not applicable (theoretical model; equilibrium effects of uncertainty shocks on shadow banking, asset prices and investment)",
29 + "strategy": "Analytical static model to microfound fragile liquidity, followed by a calibrated dynamic continuous-time model solved with projection methods; impulse responses and counterfactual comparisons (with vs. without shadow banking, with and without policy interventions)",
30 + "key_assumptions": [
31 + "Investors face liquidity events and value liquid securities",
32 + "Producing liquid securities requires scarce collateral",
33 + "Shadow money is liquid in quiet times but becomes illiquid when uncertainty is high",
34 + "Net worth frictions are abstracted from in the baseline"
35 + ],
36 + "threats_addressed": [
37 + "Omission of net worth frictions (an extension with such a friction is placed in the Internet Appendix)"
38 + ],
39 + "causal_language_strength": 4,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 2
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 4
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "preview_of_approach"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "mechanism_or_theory_sketch"
81 + },
82 + {
83 + "paragraph_index": 2,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "motivation"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "related_literature_positioning"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "preview_of_results"
105 + },
106 + {
107 + "paragraph_index": 6,
108 + "move": "preview_of_results"
109 + },
110 + {
111 + "paragraph_index": 7,
112 + "move": "policy_or_broader_implication"
113 + },
114 + {
115 + "paragraph_index": 8,
116 + "move": "policy_or_broader_implication"
117 + },
118 + {
119 + "paragraph_index": 9,
120 + "move": "related_literature_positioning"
121 + },
122 + {
123 + "paragraph_index": 9,
124 + "move": "contribution_statement"
125 + },
126 + {
127 + "paragraph_index": 9,
128 + "move": "robustness_mention"
129 + }
130 + ],
131 + "contribution_paragraph_index": 9,
132 + "contribution_types": [
133 + "new_theory_or_model"
134 + ],
135 + "roadmap_present": false,
136 + "literature_positioning_location": "inside_introduction"
137 + },
138 + "section_roles": [
139 + {
140 + "idx": 0,
141 + "role": "introduction"
142 + },
143 + {
144 + "idx": 1,
145 + "role": "theory"
146 + },
147 + {
148 + "idx": 2,
149 + "role": "theory"
150 + },
151 + {
152 + "idx": 3,
153 + "role": "results"
154 + },
155 + {
156 + "idx": 4,
157 + "role": "other"
158 + },
159 + {
160 + "idx": 5,
161 + "role": "conclusion"
162 + },
163 + {
164 + "idx": 6,
165 + "role": "appendix"
166 + },
167 + {
168 + "idx": 7,
169 + "role": "appendix"
170 + },
171 + {
172 + "idx": 8,
173 + "role": "appendix"
174 + },
175 + {
176 + "idx": 9,
177 + "role": "appendix"
178 + },
179 + {
180 + "idx": 10,
181 + "role": "appendix"
182 + }
183 + ],
184 + "results_presentation": {
185 + "economic_significance_discussed": true,
186 + "statistical_significance_language": "unclear",
187 + "magnitude_statements_in_intro": false
188 + },
189 + "robustness": {
190 + "location": "internet_appendix",
191 + "types": [
192 + "other"
193 + ]
194 + },
195 + "tables": {
196 + "convention_notes": "Only two tables appear: a glossary of model variables and a list of benchmark parameter values. Neither reports regressions. Notes describe the functional forms chosen for the parameterization.",
197 + "units_stated_in_notes": false,
198 + "standard_errors_description": "none"
199 + },
200 + "figures": {
201 + "uncertainty_shown": "no",
202 + "notes": ""
203 + },
204 + "conclusion": {
205 + "bounded_claims": false,
206 + "policy_implications": true,
207 + "open_questions": false,
208 + "restates_contribution": true
209 + },
210 + "hedging": {
211 + "level": "low",
212 + "typical_devices": [
213 + "'can be interpreted as'",
214 + "'consistent with' when linking model dynamics to stylized facts",
215 + "'sheds light on'"
216 + ]
217 + },
218 + "claim_evidence": [
219 + {
220 + "claim": "When uncertainty is low, shadow banking expands liquidity, lowers discount rates and raises asset prices and growth.",
221 + "evidence_type": "figure",
222 + "evidence_ref": "Figures 3-5"
223 + },
224 + {
225 + "claim": "An uncertainty shock collapses shadow money, shrinks liquidity, raises discount rates and lowers investment.",
226 + "evidence_type": "figure",
227 + "evidence_ref": "Figure 6"
228 + },
229 + {
230 + "claim": "Crashes followed by collateral mining and the collateral decelerator produce slow recoveries.",
231 + "evidence_type": "figure",
232 + "evidence_ref": "Figure 7"
233 + },
234 + {
235 + "claim": "LSAP raises asset prices both at announcement and ex ante, by supplying collateral when it is scarce.",
236 + "evidence_type": "figure",
237 + "evidence_ref": "Figure 8"
238 + },
239 + {
240 + "claim": "Operation Twist lowers aggregate collateral and is counterproductive even though long-term bond yields fall.",
241 + "evidence_type": "figure",
242 + "evidence_ref": "Figure 9"
243 + },
244 + {
245 + "claim": "Fragile liquidity creates a trade-off between the size and the stability of liquidity supply.",
246 + "evidence_type": "equation_or_proof",
247 + "evidence_ref": "Section I; Appendix A (Proposition 1)"
248 + },
249 + {
250 + "claim": "The pre-crisis credit expansion was concentrated among riskier borrowers, consistent with the model's boom dynamics.",
251 + "evidence_type": "prior_literature",
252 + "evidence_ref": "Mian and Sufi (2009); Krishnamurthy and Muir (2015); Ranciere et al. (2008)"
253 + }
254 + ],
255 + "structural_summary": "The paper opens with a narrative of boom, bust and slow recovery and frames shadow banking as fragile liquidity transformation. The introduction then walks through the model's cycle phase by phase (boom, shock, collapse, slow recovery, flight to quality), linking each phase to stylized facts. It previews policy results and ends by positioning the paper against net-worth-based macrofinance models. The body moves from a static microfoundation of the core trade-off, to a dynamic continuous-time model, to calibrated numerical results presented through figures and impulse responses, and then to policy counterfactuals. A short conclusion restates the mechanism. Proofs, the numerical method and parameter details are placed in appendices.",
256 + "writing_profile_keywords": [
257 + "macrofinance theory",
258 + "narrative mechanism walkthrough",
259 + "static-to-dynamic model progression",
260 + "calibrated impulse responses",
261 + "policy counterfactuals",
262 + "figure-driven results",
263 + "stylized facts as motivation"
264 + ],
265 + "confidence": 0.85,
266 + "notes": "This is a pure theory and calibration paper with no empirical estimation, so fields such as statistical significance and standard errors are not applicable. The only robustness exercise visible is the net-worth friction extension in the Internet Appendix. The Policy Interventions section is coded 'other' because it is a counterfactual policy-analysis section, not a robustness check."
267 + },
268 + "deterministic_checks": {
269 + "roadmap_regex": false,
270 + "contribution_regex_in_intro": true,
271 + "magnitude_in_abstract": false,
272 + "abstract_n_sentences": 5,
273 + "intro_n_paragraphs": 10,
274 + "intro_n_words": 1468,
275 + "causal_words_in_abstract": 0,
276 + "hedge_words_in_abstract": 0,
277 + "stars_mentioned": false,
278 + "clustered_se_mentioned": false,
279 + "internet_appendix_mentioned": true,
280 + "section_roles_rule": {
281 + "0": "introduction",
282 + "1": "theory",
283 + "2": "theory",
284 + "3": "results",
285 + "4": "other",
286 + "5": "conclusion",
287 + "6": "appendix",
288 + "7": "appendix",
289 + "8": "appendix",
290 + "9": "appendix",
291 + "10": "appendix"
292 + }
293 + },
294 + "schema_valid": false,
295 + "code_version": "feb6c9e-dirty"
296 +}
\ No newline at end of file
added data/annotations/documents/jof_12541.json +230 −0
@@ -0,0 +1,230 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12541",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:35+00:00",
9 + "source_sha256": "afc22649cbfe82287b5e558704eb58c4994341da91b379bd745a9359f47a80c7",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 16597,
12 + "usage": {
13 + "input": 6074,
14 + "output": 1826,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06161,
19 + "request_id": "req_011CggpkwVcEBK8aSNfuickp",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "credit derivatives / auction design",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model of equilibrium prices and allocations)",
27 + "strategy": "not applicable; equilibrium analysis of a stylized two-stage CDS auction compared against a competitive/double-auction benchmark",
28 + "key_assumptions": [
29 + "infinitesimal price-taking traders",
30 + "symmetric distribution of CDS benefits",
31 + "price caps/floors not formally modeled in core analysis"
32 + ],
33 + "threats_addressed": [
34 + "imperfect competition handled in a generalized double auction appendix (finite n traders)"
35 + ],
36 + "causal_language_strength": 1,
37 + "uses_causal_claim": false
38 + },
39 + "abstract_moves": {
40 + "question": {
41 + "present": true,
42 + "sentence_index": 0
43 + },
44 + "gap": {
45 + "present": false,
46 + "sentence_index": -1
47 + },
48 + "method": {
49 + "present": true,
50 + "sentence_index": 1
51 + },
52 + "results": {
53 + "present": true,
54 + "sentence_index": 1
55 + },
56 + "magnitude": {
57 + "present": false,
58 + "sentence_index": -1
59 + },
60 + "contribution": {
61 + "present": true,
62 + "sentence_index": 4
63 + }
64 + },
65 + "introduction": {
66 + "moves": [
67 + {
68 + "paragraph_index": 0,
69 + "move": "motivation"
70 + },
71 + {
72 + "paragraph_index": 0,
73 + "move": "preview_of_approach"
74 + }
75 + ],
76 + "contribution_paragraph_index": -1,
77 + "contribution_types": [
78 + "new_theory_or_model"
79 + ],
80 + "roadmap_present": false,
81 + "literature_positioning_location": "inside_introduction"
82 + },
83 + "section_roles": [
84 + {
85 + "idx": 0,
86 + "role": "introduction"
87 + },
88 + {
89 + "idx": 1,
90 + "role": "other"
91 + },
92 + {
93 + "idx": 2,
94 + "role": "theory"
95 + },
96 + {
97 + "idx": 3,
98 + "role": "theory"
99 + },
100 + {
101 + "idx": 4,
102 + "role": "theory"
103 + },
104 + {
105 + "idx": 5,
106 + "role": "theory"
107 + },
108 + {
109 + "idx": 6,
110 + "role": "other"
111 + },
112 + {
113 + "idx": 7,
114 + "role": "conclusion"
115 + },
116 + {
117 + "idx": 8,
118 + "role": "appendix"
119 + },
120 + {
121 + "idx": 9,
122 + "role": "appendix"
123 + }
124 + ],
125 + "results_presentation": {
126 + "economic_significance_discussed": false,
127 + "statistical_significance_language": "unclear",
128 + "magnitude_statements_in_intro": false
129 + },
130 + "robustness": {
131 + "location": "appendix",
132 + "types": [
133 + "other"
134 + ]
135 + },
136 + "tables": {
137 + "convention_notes": "Tables are qualitative summaries: one maps states and open interest to the direction of price bias and to allocation errors by trader type; the other compares the paper's economic channels with prior models using yes/no entries.",
138 + "units_stated_in_notes": false,
139 + "standard_errors_description": "none"
140 + },
141 + "figures": {
142 + "uncertainty_shown": "no"
143 + },
144 + "conclusion": {
145 + "bounded_claims": true,
146 + "policy_implications": true,
147 + "open_questions": false,
148 + "restates_contribution": true
149 + },
150 + "hedging": {
151 + "level": "moderate",
152 + "typical_devices": [
153 + "explicit model-scope caveats (infinitesimal traders)",
154 + "acknowledging double auction inefficiency under imperfect competition",
155 + "'in principle'/'may' phrasing on practical motivations"
156 + ]
157 + },
158 + "claim_evidence": [
159 + {
160 + "claim": "Current two-stage CDS auction design yields biased prices and inefficient allocations of defaulted bonds",
161 + "evidence_type": "equation_or_proof",
162 + "evidence_ref": "Section IV, Propositions; Appendix A proofs; Table I"
163 + },
164 + {
165 + "claim": "A double auction with unconstrained demand schedules attains the competitive first-best outcome",
166 + "evidence_type": "equation_or_proof",
167 + "evidence_ref": "Section III"
168 + },
169 + {
170 + "claim": "Price caps/floors formed by dealer quotes give dealers incentive and ability to manipulate the final price",
171 + "evidence_type": "none_stated",
172 + "evidence_ref": "Section V (informal discussion)"
173 + },
174 + {
175 + "claim": "With finitely many traders the double auction keeps the competitive price and its allocation inefficiency is O(1/n)",
176 + "evidence_type": "equation_or_proof",
177 + "evidence_ref": "Appendix B"
178 + },
179 + {
180 + "claim": "Auction-generated information matters for postauction price formation",
181 + "evidence_type": "prior_literature",
182 + "evidence_ref": "Gupta and Sundaram (2012)"
183 + },
184 + {
185 + "claim": "Numerical example illustrates CDS positions, physical requests, auction prices and allocations",
186 + "evidence_type": "figure",
187 + "evidence_ref": "Figures 3-6"
188 + }
189 + ],
190 + "structural_summary": "Opens with the size of the CDS market and its policy relevance, then states that recovery value set by the auction matters for pricing and settlement. Institutional background on the two-stage auction follows, illustrated with Lehman figures. A stylized model with infinitesimal traders is set up and benchmarked against competitive equilibrium. Equilibrium analysis shows price bias and misallocation, supported by a numerical example and propositions. A short discussion covers price caps/floors and dealer manipulation, followed by empirical predictions. The paper closes by proposing a double auction design and assessing why the current restrictions exist. Proofs and a finite-trader generalization are in the appendices.",
191 + "writing_profile_keywords": [
192 + "mechanism design",
193 + "auction theory",
194 + "policy-oriented",
195 + "propositions and proofs",
196 + "numerical illustration",
197 + "institutional detail",
198 + "design proposal"
199 + ],
200 + "confidence": 0.75,
201 + "notes": "The introduction is given as a single truncated paragraph, so contribution and roadmap moves are not visible. Section V is labeled 'conclusion' by the rule, but it is analytical, so it is annotated as theory. Section I is institutional background and is annotated as other. Empirical predictions appear in Section VI, but the paper contains no data analysis."
202 + },
203 + "deterministic_checks": {
204 + "roadmap_regex": false,
205 + "contribution_regex_in_intro": false,
206 + "magnitude_in_abstract": false,
207 + "abstract_n_sentences": 5,
208 + "intro_n_paragraphs": 1,
209 + "intro_n_words": 234,
210 + "causal_words_in_abstract": 1,
211 + "hedge_words_in_abstract": 0,
212 + "stars_mentioned": false,
213 + "clustered_se_mentioned": false,
214 + "internet_appendix_mentioned": false,
215 + "section_roles_rule": {
216 + "0": "introduction",
217 + "1": "literature",
218 + "2": "theory",
219 + "3": "theory",
220 + "4": "theory",
221 + "5": "conclusion",
222 + "6": "other",
223 + "7": "conclusion",
224 + "8": "appendix",
225 + "9": "appendix"
226 + }
227 + },
228 + "schema_valid": true,
229 + "code_version": "feb6c9e-dirty"
230 +}
\ No newline at end of file
added data/annotations/documents/jof_12542.json +283 −0
@@ -0,0 +1,283 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('claim_evidence', 'confidence', 'notes', 'structural_summary', 'writing_profile_keywords' were unexpected) @ hedging"
4 + ],
5 + "doc_id": "jof_12542",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:59:34+00:00",
11 + "source_sha256": "b29a4217fefa017878b5e283eb9c6fe598903368d719ca608853fc3664f1368d",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 20662,
14 + "usage": {
15 + "input": 6854,
16 + "output": 2242,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.07305,
21 + "request_id": "req_011CggpkgFa27VHncb9jkV9D",
22 + "annotation": {
23 + "genre": "mixed_theory_empirical",
24 + "subfield": "asset_pricing",
25 + "subfield_secondary": "corporate_finance (theory of the firm, internal capital allocation in mutual fund firms)",
26 + "primary_method": "reduced_form_regression",
27 + "identification": {
28 + "estimand": "Change in manager value added (excess return over benchmark times AUM) following a firm-initiated capital reallocation (promotion/demotion), bounded from below",
29 + "strategy": "Panel regressions of manager value added on firm capital reallocation dummies, with counterfactual adjustments (e.g., excluding poorly performing funds, filling in zeros for fired managers) to produce a lower bound; external-hire comparisons and probit predictability tests support the information channel",
30 + "key_assumptions": [
31 + "Vanguard index fund benchmarks correctly measure the investor's alternative investment opportunity",
32 + "Counterfactual adjustments yield a conservative lower bound",
33 + "Null hypothesis that reallocation decisions add no value is the appropriate test"
34 + ],
35 + "threats_addressed": [
36 + "Endogeneity of manager moves in firm/manager fixed-effects attribution models",
37 + "Survivorship from managers dropping out after demotion",
38 + "Predictability of reallocations from public information",
39 + "Behavioral explanations such as investor inattention (acknowledged, not ruled out)"
40 + ],
41 + "causal_language_strength": 4,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 0
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 1
60 + },
61 + "magnitude": {
62 + "present": true,
63 + "sentence_index": 1
64 + },
65 + "contribution": {
66 + "present": true,
67 + "sentence_index": 0
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "contribution_statement"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "motivation"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "mechanism_or_theory_sketch"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 1,
94 + "move": "related_literature_positioning"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "other"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "roadmap"
103 + }
104 + ],
105 + "contribution_paragraph_index": 0,
106 + "contribution_types": [
107 + "new_evidence",
108 + "new_theory_or_model"
109 + ],
110 + "roadmap_present": true,
111 + "literature_positioning_location": "both"
112 + },
113 + "section_roles": [
114 + {
115 + "idx": 0,
116 + "role": "introduction"
117 + },
118 + {
119 + "idx": 1,
120 + "role": "literature"
121 + },
122 + {
123 + "idx": 2,
124 + "role": "theory"
125 + },
126 + {
127 + "idx": 3,
128 + "role": "data"
129 + },
130 + {
131 + "idx": 4,
132 + "role": "data"
133 + },
134 + {
135 + "idx": 5,
136 + "role": "results"
137 + },
138 + {
139 + "idx": 6,
140 + "role": "results"
141 + },
142 + {
143 + "idx": 7,
144 + "role": "results"
145 + },
146 + {
147 + "idx": 8,
148 + "role": "robustness"
149 + },
150 + {
151 + "idx": 9,
152 + "role": "conclusion"
153 + },
154 + {
155 + "idx": 10,
156 + "role": "appendix"
157 + },
158 + {
159 + "idx": 11,
160 + "role": "appendix"
161 + }
162 + ],
163 + "results_presentation": {
164 + "economic_significance_discussed": true,
165 + "statistical_significance_language": "unclear",
166 + "magnitude_statements_in_intro": true
167 + },
168 + "robustness": {
169 + "location": "mixed",
170 + "types": [
171 + "alternative_measure",
172 + "sensitivity_to_parameters",
173 + "alternative_specification"
174 + ]
175 + },
176 + "tables": {
177 + "convention_notes": "Table notes name the dependent variable and point to its defining equation number; units are given for value added ($ millions/month). Columns are described as alternative regression specifications. Later tables state that they repeat earlier specifications with a modified sample or variable.",
178 + "units_stated_in_notes": true,
179 + "standard_errors_description": "unclear"
180 + },
181 + "figures": {
182 + "uncertainty_shown": "unclear"
183 + },
184 + "conclusion": {
185 + "bounded_claims": true,
186 + "policy_implications": false,
187 + "open_questions": true,
188 + "restates_contribution": true
189 + },
190 + "hedging": {
191 + "level": "moderate",
192 + "typical_devices": [
193 + "'at least' / lower-bound framing",
194 + "'consistent with' the hypothesis",
195 + "explicit acknowledgment that behavioral explanations cannot be ruled out",
196 + "limiting generalization to the mutual fund industry"
197 + ],
198 + "claim_evidence": [],
199 + "structural_summary": "",
200 + "writing_profile_keywords": [],
201 + "confidence": 0,
202 + "notes": ""
203 + },
204 + "claim_evidence": [
205 + {
206 + "claim": "Firm reallocation of capital to managers raises manager value added by at least about $474K per month",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table II, Table III"
209 + },
210 + {
211 + "claim": "The firm accounts for at least 30% of total manager value added under conservative counterfactuals",
212 + "evidence_type": "figure",
213 + "evidence_ref": "Figure 4, Figure 5"
214 + },
215 + {
216 + "claim": "Investors reward firms after reallocation decisions with additional capital",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table IV, Table V"
219 + },
220 + {
221 + "claim": "External hires with AUM changes do not produce detectable value-added changes",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table VI"
224 + },
225 + {
226 + "claim": "Public information such as past performance has weak power to predict firm reallocation decisions",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table VII, Table VIII"
229 + },
230 + {
231 + "claim": "Under decreasing returns to scale, value added rather than gross alpha measures skill",
232 + "evidence_type": "figure",
233 + "evidence_ref": "Figure 1; Section II model"
234 + },
235 + {
236 + "claim": "Firm-driven AUM changes would take investors years to replicate through flows",
237 + "evidence_type": "figure",
238 + "evidence_ref": "Figure 6"
239 + }
240 + ],
241 + "structural_summary": "The paper opens with an assertive contribution statement and the headline magnitude. It then sketches an information-asymmetry mechanism with a numbered list of supporting findings, adds a paragraph acknowledging a behavioral alternative it cannot rule out, and closes the introduction with a roadmap. A short theory section frames a Coasian benchmark against asymmetric information. Data definitions argue for value added over gross alpha. The main results test a null of no value added and bound the firm's contribution from below under counterfactuals. Follow-up sections show investor responses and the information source of firm skill, and a section reports gross alpha results at referees' request. The conclusion ties the findings to the theory of the firm and limits generalization to other industries.",
242 + "writing_profile_keywords": [
243 + "theory of the firm",
244 + "lower bound",
245 + "value added",
246 + "information asymmetry",
247 + "mutual fund firms",
248 + "counterfactuals",
249 + "assertive opening"
250 + ],
251 + "confidence": 0.8,
252 + "notes": "No statistical significance conventions or standard error descriptions are visible in the excerpts. Section VIII (gross alpha) is treated as robustness. The causal reading relies on the bounding approach rather than a quasi-experiment."
253 + },
254 + "deterministic_checks": {
255 + "roadmap_regex": true,
256 + "contribution_regex_in_intro": false,
257 + "magnitude_in_abstract": true,
258 + "abstract_n_sentences": 4,
259 + "intro_n_paragraphs": 4,
260 + "intro_n_words": 658,
261 + "causal_words_in_abstract": 0,
262 + "hedge_words_in_abstract": 0,
263 + "stars_mentioned": false,
264 + "clustered_se_mentioned": false,
265 + "internet_appendix_mentioned": true,
266 + "section_roles_rule": {
267 + "0": "introduction",
268 + "1": "literature",
269 + "2": "theory",
270 + "3": "data",
271 + "4": "data",
272 + "5": "other",
273 + "6": "other",
274 + "7": "other",
275 + "8": "other",
276 + "9": "conclusion",
277 + "10": "appendix",
278 + "11": "appendix"
279 + }
280 + },
281 + "schema_valid": false,
282 + "code_version": "feb6c9e-dirty"
283 +}
\ No newline at end of file
added data/annotations/documents/jof_12544.json +245 −0
@@ -0,0 +1,245 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('convention_notes_placeholder' was unexpected) @ figures"
4 + ],
5 + "doc_id": "jof_12544",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:59:51+00:00",
11 + "source_sha256": "9650ee6f9da03c5eb8b6723eba2ae121b3cf5306df986b99a2d6c5a4768760fa",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 19398,
14 + "usage": {
15 + "input": 6638,
16 + "output": 1929,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.065926,
21 + "request_id": "req_011Cggpn76LRKtVDJEp4cDgQ",
22 + "annotation": {
23 + "genre": "mixed_theory_empirical",
24 + "subfield": "asset_pricing",
25 + "subfield_secondary": "behavioral_finance",
26 + "primary_method": "theoretical_model_analytical",
27 + "identification": {
28 + "estimand": "Model-implied, then empirical, relations between analyst-forecast dispersion and future excess returns, and conditional time-series momentum in high-dispersion periods and recessions.",
29 + "strategy": "Equilibrium model with heterogeneous forecasting models, calibrated to US data. Simulated predictions are compared with the same regressions run on observed data using empirical proxies such as analyst-forecast dispersion.",
30 + "key_assumptions": [
31 + "Investors use different forecasting models (continuous-state vs. discrete-state)",
32 + "Calibrated parameters satisfy restrictions that generate countercyclical disagreement spikes",
33 + "Analyst-forecast dispersion proxies for squared disagreement"
34 + ],
35 + "threats_addressed": [
36 + "Linear trend in dispersion",
37 + "12-month seasonality in the fundamental"
38 + ],
39 + "causal_language_strength": 3,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 3
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 6
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "literature_gap"
81 + },
82 + {
83 + "paragraph_index": 0,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "preview_of_approach"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "mechanism_or_theory_sketch"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "related_literature_positioning"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "roadmap"
109 + }
110 + ],
111 + "contribution_paragraph_index": 0,
112 + "contribution_types": [
113 + "new_theory_or_model",
114 + "new_evidence"
115 + ],
116 + "roadmap_present": true,
117 + "literature_positioning_location": "inside_introduction"
118 + },
119 + "section_roles": [
120 + {
121 + "idx": 0,
122 + "role": "introduction"
123 + },
124 + {
125 + "idx": 1,
126 + "role": "theory"
127 + },
128 + {
129 + "idx": 2,
130 + "role": "theory"
131 + },
132 + {
133 + "idx": 3,
134 + "role": "theory"
135 + },
136 + {
137 + "idx": 4,
138 + "role": "results"
139 + },
140 + {
141 + "idx": 5,
142 + "role": "conclusion"
143 + }
144 + ],
145 + "results_presentation": {
146 + "economic_significance_discussed": false,
147 + "statistical_significance_language": "stars_and_tstats",
148 + "magnitude_statements_in_intro": false
149 + },
150 + "robustness": {
151 + "location": "none_seen",
152 + "types": []
153 + },
154 + "tables": {
155 + "convention_notes": "Tables report calibrated parameters with delta-method standard errors in brackets, a regime definition table, and regression outputs. Regression standard errors appear in parentheses, with stars marking the 10/5/1% significance levels.",
156 + "units_stated_in_notes": false,
157 + "standard_errors_description": "mixed"
158 + },
159 + "figures": {
160 + "uncertainty_shown": "mixed",
161 + "convention_notes_placeholder": null
162 + },
163 + "conclusion": {
164 + "bounded_claims": true,
165 + "policy_implications": false,
166 + "open_questions": true,
167 + "restates_contribution": false
168 + },
169 + "hedging": {
170 + "level": "moderate",
171 + "typical_devices": [
172 + "'we provide empirical support' framing",
173 + "'suggests' wording",
174 + "model-conditional statements ('in our model')"
175 + ]
176 + },
177 + "claim_evidence": [
178 + {
179 + "claim": "Under the calibrated parameters, disagreement between the two agents spikes countercyclically, with beliefs polarizing in bad times.",
180 + "evidence_type": "figure",
181 + "evidence_ref": "Section II, Figure 4, Table I"
182 + },
183 + {
184 + "claim": "Contemporaneous excess returns decrease with squared disagreement because disagreement variation is priced.",
185 + "evidence_type": "equation_or_proof",
186 + "evidence_ref": "Section III.A"
187 + },
188 + {
189 + "claim": "The model generates time-series momentum that is strongest in bad times.",
190 + "evidence_type": "figure",
191 + "evidence_ref": "Figure 8"
192 + },
193 + {
194 + "claim": "Empirically, future excess returns are positively related to analyst-forecast dispersion.",
195 + "evidence_type": "figure",
196 + "evidence_ref": "Figure 10"
197 + },
198 + {
199 + "claim": "Empirical time-series momentum is stronger in high-dispersion periods and in recessions.",
200 + "evidence_type": "figure",
201 + "evidence_ref": "Figures 11-12"
202 + },
203 + {
204 + "claim": "Empirical dispersion is higher in NBER recessions and co-moves with model-implied dispersion.",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table IV"
207 + }
208 + ],
209 + "structural_summary": "The paper opens with the puzzle that return predictability concentrates in bad times and proposes a heterogeneous-model learning mechanism to explain it. It then builds a two-agent equilibrium model and calibrates it to the US business cycle, showing that disagreement spikes countercyclically. From the model it derives predictions on return predictability and momentum. Finally, it tests these predictions by setting simulated statistics beside the same statistics computed on empirical proxies, and closes with avenues for future research.",
210 + "writing_profile_keywords": [
211 + "equilibrium model",
212 + "heterogeneous beliefs",
213 + "disagreement",
214 + "calibration",
215 + "model vs. data simulations",
216 + "time-series momentum",
217 + "business cycle"
218 + ],
219 + "confidence": 0.8,
220 + "notes": "Theory-led paper with an empirical validation section. Robustness checks are not visible in the excerpts. Figures report t-statistics, and Figure 4 shows confidence bands."
221 + },
222 + "deterministic_checks": {
223 + "roadmap_regex": true,
224 + "contribution_regex_in_intro": false,
225 + "magnitude_in_abstract": false,
226 + "abstract_n_sentences": 8,
227 + "intro_n_paragraphs": 6,
228 + "intro_n_words": 870,
229 + "causal_words_in_abstract": 0,
230 + "hedge_words_in_abstract": 0,
231 + "stars_mentioned": true,
232 + "clustered_se_mentioned": false,
233 + "internet_appendix_mentioned": true,
234 + "section_roles_rule": {
235 + "0": "introduction",
236 + "1": "theory",
237 + "2": "other",
238 + "3": "other",
239 + "4": "results",
240 + "5": "conclusion"
241 + }
242 + },
243 + "schema_valid": false,
244 + "code_version": "feb6c9e-dirty"
245 +}
\ No newline at end of file
added data/annotations/documents/jof_12545.json +247 −0
@@ -0,0 +1,247 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12545",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:35+00:00",
9 + "source_sha256": "5a6b2f64e456daafcdf097f3af5f753512c4db8489ea9befd79f64db4fc702d0",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20866,
12 + "usage": {
13 + "input": 7030,
14 + "output": 2096,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.070834,
19 + "request_id": "req_011CggpkpE3CkeqHV3TSvy3N",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "executive compensation and risk-taking",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Overall net effect of an increase in the value of newly granted at-the-money options on CEO risk-taking, measured by changes in equity volatility and in firm outcomes such as leverage and investment",
27 + "strategy": "Two IV designs built on multiyear compensation plans. IV1 instruments option grants with a predicted first year of a new fixed-value cycle, predicted from the length of the previous cycle. IV2 instruments with the interaction of fixed-number plan type and industry returns.",
28 + "key_assumptions": [
29 + "Predicted cycle first years, based only on past cycle length, are unrelated to risk except through option grants.",
30 + "Sorting of firms into fixed-number versus fixed-value plans is quasi-arbitrary.",
31 + "Industry returns are beyond the executive's control and hard to predict, and plan type is not otherwise related to risk."
32 + ],
33 + "threats_addressed": [
34 + "Endogenous renegotiation of when new cycles start (handled by using predicted rather than actual first years)",
35 + "Endogeneity of option grants",
36 + "Nonlinear and interacted effects of industry returns (polynomial and interacted controls)",
37 + "Placebo test on CEOs who are not currently on cycles",
38 + "Validity tests that decompose the predicted first-year indicator"
39 + ],
40 + "causal_language_strength": 4,
41 + "uses_causal_claim": true
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": true,
46 + "sentence_index": 0
47 + },
48 + "gap": {
49 + "present": true,
50 + "sentence_index": 1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 2
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 3
59 + },
60 + "magnitude": {
61 + "present": true,
62 + "sentence_index": 3
63 + },
64 + "contribution": {
65 + "present": false,
66 + "sentence_index": -1
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "tension_or_puzzle"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 2,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 2,
89 + "move": "identification_defense"
90 + },
91 + {
92 + "paragraph_index": 3,
93 + "move": "preview_of_approach"
94 + },
95 + {
96 + "paragraph_index": 3,
97 + "move": "identification_defense"
98 + },
99 + {
100 + "paragraph_index": 4,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 5,
105 + "move": "contribution_statement"
106 + },
107 + {
108 + "paragraph_index": 5,
109 + "move": "related_literature_positioning"
110 + }
111 + ],
112 + "contribution_paragraph_index": 5,
113 + "contribution_types": [
114 + "new_identification",
115 + "new_evidence"
116 + ],
117 + "roadmap_present": false,
118 + "literature_positioning_location": "inside_introduction"
119 + },
120 + "section_roles": [
121 + {
122 + "idx": 0,
123 + "role": "introduction"
124 + },
125 + {
126 + "idx": 1,
127 + "role": "data"
128 + },
129 + {
130 + "idx": 2,
131 + "role": "empirical_design"
132 + },
133 + {
134 + "idx": 3,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 4,
139 + "role": "conclusion"
140 + }
141 + ],
142 + "results_presentation": {
143 + "economic_significance_discussed": true,
144 + "statistical_significance_language": "unclear",
145 + "magnitude_statements_in_intro": true
146 + },
147 + "robustness": {
148 + "location": "mixed",
149 + "types": [
150 + "placebo",
151 + "alternative_specification",
152 + "subsample",
153 + "falsification",
154 + "other"
155 + ]
156 + },
157 + "tables": {
158 + "convention_notes": "Tables are organized by IV strategy, each with first-stage, outcome and validity tables. Panels separate IV estimates from reduced-form estimates. Observations are at the CEO-year level. Notes define the sample restrictions and the variables. One illustrative numerical table is adapted from Hall (1999).",
159 + "units_stated_in_notes": true,
160 + "standard_errors_description": "unclear"
161 + },
162 + "figures": {
163 + "uncertainty_shown": "no"
164 + },
165 + "conclusion": {
166 + "bounded_claims": true,
167 + "policy_implications": true,
168 + "open_questions": false,
169 + "restates_contribution": true
170 + },
171 + "hedging": {
172 + "level": "moderate",
173 + "typical_devices": [
174 + "describing evidence as suggestive",
175 + "restricting findings to the range of grant changes observed in the sample",
176 + "noting that the effect may be nonmonotonic",
177 + "framing results as conditional on average grant levels"
178 + ]
179 + },
180 + "claim_evidence": [
181 + {
182 + "claim": "Predicted fixed-value first years strongly predict roughly a 15% increase in the value, delta and vega of new option grants.",
183 + "evidence_type": "table",
184 + "evidence_ref": "Table II"
185 + },
186 + {
187 + "claim": "A 10% increase in new options leads to a 2.8–4.2% rise in equity volatility.",
188 + "evidence_type": "table",
189 + "evidence_ref": "Tables IV, VIII"
190 + },
191 + {
192 + "claim": "The increase in risk is driven largely by increased leverage, while the effect on investment is modest and generally insignificant.",
193 + "evidence_type": "table",
194 + "evidence_ref": "Tables V, IX"
195 + },
196 + {
197 + "claim": "The placebo instrument for CEOs not currently on cycles yields no comparable effect, supporting IV2 validity.",
198 + "evidence_type": "table",
199 + "evidence_ref": "Table XI"
200 + },
201 + {
202 + "claim": "Effects are larger when new grants are large relative to existing holdings and in the financial and high-tech sectors.",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table XII"
205 + },
206 + {
207 + "claim": "Firms sort into plan types somewhat arbitrarily.",
208 + "evidence_type": "prior_literature",
209 + "evidence_ref": "Hall (1999)"
210 + }
211 + ],
212 + "structural_summary": "The introduction motivates the paper with the prevalence of option pay and frames the net effect of options on risk-taking as theoretically ambiguous. It then lays out two IV strategies based on multiyear plans, defending each against endogeneity concerns, and previews the results with magnitudes. It closes by positioning the paper against the vega–risk literature. Later sections cover data and cycle detection, the empirical strategy for each IV, results split by IV along with heterogeneity and robustness, and a short conclusion that bounds the findings to moderate grant changes.",
213 + "writing_profile_keywords": [
214 + "natural experiment",
215 + "two instruments",
216 + "cross-validation",
217 + "executive compensation",
218 + "institutional detail",
219 + "theoretical ambiguity",
220 + "bounded inference"
221 + ],
222 + "confidence": 0.85,
223 + "notes": "The excerpts do not show how statistical significance or standard errors are reported. Robustness appears both within the Results section (subsection D) and in an internet appendix."
224 + },
225 + "deterministic_checks": {
226 + "roadmap_regex": false,
227 + "contribution_regex_in_intro": true,
228 + "magnitude_in_abstract": true,
229 + "abstract_n_sentences": 5,
230 + "intro_n_paragraphs": 6,
231 + "intro_n_words": 1248,
232 + "causal_words_in_abstract": 2,
233 + "hedge_words_in_abstract": 0,
234 + "stars_mentioned": false,
235 + "clustered_se_mentioned": false,
236 + "internet_appendix_mentioned": false,
237 + "section_roles_rule": {
238 + "0": "introduction",
239 + "1": "data",
240 + "2": "empirical_design",
241 + "3": "results",
242 + "4": "conclusion"
243 + }
244 + },
245 + "schema_valid": true,
246 + "code_version": "feb6c9e-dirty"
247 +}
\ No newline at end of file
added data/annotations/documents/jof_12546.json +251 −0
@@ -0,0 +1,251 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('note' was unexpected) @ figures"
4 + ],
5 + "doc_id": "jof_12546",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T22:59:51+00:00",
11 + "source_sha256": "54a87712bbe3d47ae775768397244d621ba5ae9d1089237b69f32029755574d2",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 19588,
14 + "usage": {
15 + "input": 6759,
16 + "output": 2113,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.07009,
21 + "request_id": "req_011Cggpn37DmwWz9ecfjEuPU",
22 + "annotation": {
23 + "genre": "mixed_theory_empirical",
24 + "subfield": "international_finance",
25 + "subfield_secondary": "asset_pricing",
26 + "primary_method": "theoretical_model_analytical",
27 + "identification": {
28 + "estimand": "Cross-sectional association between countries' trade composition (import ratio of finished vs. basic goods) and currency interest rates, excess returns and global risk exposure. Time-series association between relative productivity and real exchange rates and interest rate differentials.",
29 + "strategy": "A general equilibrium model generates predictions, which are tested with cross-sectional regressions, portfolio sorts, and GMM/Fama-MacBeth asset pricing tests. Time-series regressions relate relative productivity to real exchange rates and rates, and predictive regressions use the Baltic Dry Index.",
30 + "key_assumptions": [
31 + "Covered interest parity holds, so forward discounts approximate interest differentials",
32 + "Model structure: comparative advantage, convex trade costs, inelastic trade channel",
33 + "Import ratio proxies a country's specialization type"
34 + ],
35 + "threats_addressed": [
36 + "Controls for GDP size and inflation in cross-sectional regressions",
37 + "Carry-sorted portfolios compared with import-ratio sorts",
38 + "G10 subsample versus full sample"
39 + ],
40 + "causal_language_strength": 2,
41 + "uses_causal_claim": false
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": false,
46 + "sentence_index": -1
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": true,
54 + "sentence_index": 2
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 3
59 + },
60 + "magnitude": {
61 + "present": false,
62 + "sentence_index": -1
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 2
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "tension_or_puzzle"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 2,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 2,
93 + "move": "mechanism_or_theory_sketch"
94 + },
95 + {
96 + "paragraph_index": 3,
97 + "move": "related_literature_positioning"
98 + }
99 + ],
100 + "contribution_paragraph_index": 0,
101 + "contribution_types": [
102 + "new_theory_or_model",
103 + "new_evidence"
104 + ],
105 + "roadmap_present": false,
106 + "literature_positioning_location": "inside_introduction"
107 + },
108 + "section_roles": [
109 + {
110 + "idx": 0,
111 + "role": "introduction"
112 + },
113 + {
114 + "idx": 1,
115 + "role": "theory"
116 + },
117 + {
118 + "idx": 2,
119 + "role": "results"
120 + },
121 + {
122 + "idx": 3,
123 + "role": "conclusion"
124 + },
125 + {
126 + "idx": 4,
127 + "role": "appendix"
128 + },
129 + {
130 + "idx": 5,
131 + "role": "appendix"
132 + },
133 + {
134 + "idx": 6,
135 + "role": "appendix"
136 + },
137 + {
138 + "idx": 7,
139 + "role": "appendix"
140 + }
141 + ],
142 + "results_presentation": {
143 + "economic_significance_discussed": true,
144 + "statistical_significance_language": "unclear",
145 + "magnitude_statements_in_intro": false
146 + },
147 + "robustness": {
148 + "location": "within_results",
149 + "types": [
150 + "subsample",
151 + "alternative_specification"
152 + ]
153 + },
154 + "tables": {
155 + "convention_notes": "Notes define variable construction (import ratio, relative productivity against an equal-weighted basket of trading partners) and data frequency, and annualize returns. Results are reported for the full sample and the G10 subsample. Asset pricing tests use both GMM and Fama-MacBeth.",
156 + "units_stated_in_notes": true,
157 + "standard_errors_description": "unclear"
158 + },
159 + "figures": {
160 + "uncertainty_shown": "unclear",
161 + "note": "Model figures plot comparative statics against relative productivity. Empirical figures are scatter and time-series plots."
162 + },
163 + "conclusion": {
164 + "bounded_claims": true,
165 + "policy_implications": false,
166 + "open_questions": false,
167 + "restates_contribution": true
168 + },
169 + "hedging": {
170 + "level": "moderate",
171 + "typical_devices": [
172 + "'tend to' generalizations",
173 + "'helps rationalize' framing of the mechanism",
174 + "'consistent with' the data"
175 + ]
176 + },
177 + "claim_evidence": [
178 + {
179 + "claim": "Sorting currencies on import ratios yields large spreads in interest rates and excess returns, largely subsuming the carry trade",
180 + "evidence_type": "table",
181 + "evidence_ref": "Table VI, Table VII"
182 + },
183 + {
184 + "claim": "Import ratios explain the cross-section of forward discounts and FX returns",
185 + "evidence_type": "table",
186 + "evidence_ref": "Table II, Figure 5"
187 + },
188 + {
189 + "claim": "Final-good producers are more exposed to global output shocks",
190 + "evidence_type": "figure",
191 + "evidence_ref": "Figure 6"
192 + },
193 + {
194 + "claim": "Relative productivity tracks real exchange rates and real interest rate differentials",
195 + "evidence_type": "table",
196 + "evidence_ref": "Tables III-IV, Figures 7-8"
197 + },
198 + {
199 + "claim": "The Baltic Dry Index (shipping costs) forecasts carry trade returns",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table VIII"
202 + },
203 + {
204 + "claim": "In the model, the commodity-country currency is risky because domestic production insulates that country from global shocks",
205 + "evidence_type": "equation_or_proof",
206 + "evidence_ref": "Section I, Appendix A"
207 + },
208 + {
209 + "claim": "Unconditional currency risk premia are substantial",
210 + "evidence_type": "prior_literature",
211 + "evidence_ref": "Lustig, Roussanov, and Verdelhan (2011); Hassan and Mano (2014)"
212 + }
213 + ],
214 + "structural_summary": "The introduction opens with the carry trade and the puzzle of heterogeneous currency risk, and proposes trade composition as its source. It then previews model predictions and empirical confirmations, and positions the paper within the UIP-failure literature. A long model section derives exchange rates, risk premia, import ratios, trade costs and conditional implications, ending with a summary of implications. A long empirical section tests each implication in turn: portfolios, asset pricing tests, productivity regressions, and a case study of the financial crisis. A brief conclusion restates the evidence and the mechanism, and appendices hold proofs and derivations.",
215 + "writing_profile_keywords": [
216 + "model-then-test",
217 + "general equilibrium",
218 + "prediction-by-prediction validation",
219 + "portfolio sorts",
220 + "carry trade",
221 + "case study"
222 + ],
223 + "confidence": 0.8,
224 + "notes": "The introduction is truncated, so there may be a roadmap or a more explicit contribution paragraph that is not visible. The type of standard errors is not visible in the excerpts."
225 + },
226 + "deterministic_checks": {
227 + "roadmap_regex": false,
228 + "contribution_regex_in_intro": false,
229 + "magnitude_in_abstract": false,
230 + "abstract_n_sentences": 7,
231 + "intro_n_paragraphs": 4,
232 + "intro_n_words": 708,
233 + "causal_words_in_abstract": 0,
234 + "hedge_words_in_abstract": 1,
235 + "stars_mentioned": false,
236 + "clustered_se_mentioned": false,
237 + "internet_appendix_mentioned": false,
238 + "section_roles_rule": {
239 + "0": "introduction",
240 + "1": "theory",
241 + "2": "results",
242 + "3": "conclusion",
243 + "4": "appendix",
244 + "5": "appendix",
245 + "6": "appendix",
246 + "7": "appendix"
247 + }
248 + },
249 + "schema_valid": false,
250 + "code_version": "feb6c9e-dirty"
251 +}
\ No newline at end of file
added data/annotations/documents/jof_12547.json +310 −0
@@ -0,0 +1,310 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12547",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:36+00:00",
9 + "source_sha256": "828f889d3544eed23d0d9bb805daf0b7f7cbdc3a8050e0fb419ec0f9b8ac5f8a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24085,
12 + "usage": {
13 + "input": 8190,
14 + "output": 2262,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.078794,
19 + "request_id": "req_011CggpknVcEJitikcjo7wVj",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "behavioral_finance",
23 + "subfield_secondary": "household_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Association between investor motives (intrinsic social preferences from a trust game, social signaling, return and risk expectations) and the probability and share of holding SRI equity funds",
27 + "strategy": "Probit and OLS cross-sectional regressions that link administrative brokerage holdings to survey answers and incentivized experimental measures, with a battery of controls",
28 + "key_assumptions": [
29 + "Second-mover behavior in an anonymous one-shot trust game cleanly measures intrinsic social preferences, free of reputation or image concerns",
30 + "Survey responses on expectations and signaling reflect true beliefs and behavior"
31 + ],
32 + "threats_addressed": [
33 + "Strategic reputation or social image contamination of the preference measure, handled by an anonymous one-shot experiment",
34 + "Investors who regard SRI as a marketing trick (excluded in appendix)",
35 + "Inattentive experiment participants who failed practice questions (excluded in appendix)",
36 + "Inconsistent respondents (excluded in appendix)"
37 + ],
38 + "causal_language_strength": 2,
39 + "uses_causal_claim": false
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 0
45 + },
46 + "gap": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 0
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 1
57 + },
58 + "magnitude": {
59 + "present": false,
60 + "sentence_index": -1
61 + },
62 + "contribution": {
63 + "present": false,
64 + "sentence_index": -1
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "literature_gap"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "identification_defense"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 4,
99 + "move": "preview_of_results"
100 + },
101 + {
102 + "paragraph_index": 5,
103 + "move": "policy_or_broader_implication"
104 + },
105 + {
106 + "paragraph_index": 6,
107 + "move": "roadmap"
108 + }
109 + ],
110 + "contribution_paragraph_index": -1,
111 + "contribution_types": [
112 + "new_evidence",
113 + "new_data_or_measure"
114 + ],
115 + "roadmap_present": true,
116 + "literature_positioning_location": "inside_introduction"
117 + },
118 + "section_roles": [
119 + {
120 + "idx": 0,
121 + "role": "introduction"
122 + },
123 + {
124 + "idx": 1,
125 + "role": "data"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "data"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "results"
138 + },
139 + {
140 + "idx": 5,
141 + "role": "results"
142 + },
143 + {
144 + "idx": 6,
145 + "role": "results"
146 + },
147 + {
148 + "idx": 7,
149 + "role": "conclusion"
150 + },
151 + {
152 + "idx": 8,
153 + "role": "appendix"
154 + },
155 + {
156 + "idx": 9,
157 + "role": "appendix"
158 + },
159 + {
160 + "idx": 10,
161 + "role": "appendix"
162 + },
163 + {
164 + "idx": 11,
165 + "role": "appendix"
166 + },
167 + {
168 + "idx": 12,
169 + "role": "appendix"
170 + },
171 + {
172 + "idx": 13,
173 + "role": "robustness"
174 + },
175 + {
176 + "idx": 14,
177 + "role": "robustness"
178 + },
179 + {
180 + "idx": 15,
181 + "role": "robustness"
182 + },
183 + {
184 + "idx": 16,
185 + "role": "robustness"
186 + }
187 + ],
188 + "results_presentation": {
189 + "economic_significance_discussed": true,
190 + "statistical_significance_language": "stars_and_tstats",
191 + "magnitude_statements_in_intro": true
192 + },
193 + "robustness": {
194 + "location": "appendix",
195 + "types": [
196 + "subsample",
197 + "alternative_specification"
198 + ]
199 + },
200 + "tables": {
201 + "convention_notes": "Notes state whether each specification reports probit marginal effects or OLS coefficients and define dependent variables. They point to an appendix table of variable definitions, give robust standard errors in parentheses, and mark significance with stars at the 10/5/1% levels.",
202 + "units_stated_in_notes": false,
203 + "standard_errors_description": "robust"
204 + },
205 + "figures": {
206 + "uncertainty_shown": "no"
207 + },
208 + "conclusion": {
209 + "bounded_claims": true,
210 + "policy_implications": true,
211 + "open_questions": true,
212 + "restates_contribution": true
213 + },
214 + "hedging": {
215 + "level": "moderate",
216 + "typical_devices": [
217 + "suggests",
218 + "indicates",
219 + "perhaps",
220 + "may",
221 + "explicit disclaimer that causal effects cannot be addressed",
222 + "discussion of generalizability beyond Dutch sample"
223 + ]
224 + },
225 + "claim_evidence": [
226 + {
227 + "claim": "Investors with stronger intrinsic social preferences (trust game second-mover returns) are substantially more likely to hold SRI equity funds",
228 + "evidence_type": "table",
229 + "evidence_ref": "Table III"
230 + },
231 + {
232 + "claim": "More frequent talking about investments (social signaling) is associated with SRI holding",
233 + "evidence_type": "table",
234 + "evidence_ref": "Table III; Table IV"
235 + },
236 + {
237 + "claim": "Most SRI investors expect SRI funds to underperform conventional funds",
238 + "evidence_type": "figure",
239 + "evidence_ref": "Figure 1"
240 + },
241 + {
242 + "claim": "Risk perceptions of SRI funds are unrelated to holding them",
243 + "evidence_type": "figure",
244 + "evidence_ref": "Figure 2; Table III"
245 + },
246 + {
247 + "claim": "SRI investors donate more to charity, so SRI is not a substitute for donations",
248 + "evidence_type": "table",
249 + "evidence_ref": "Table VII"
250 + },
251 + {
252 + "claim": "Social preferences show few demographic covariates",
253 + "evidence_type": "table",
254 + "evidence_ref": "Table VI"
255 + },
256 + {
257 + "claim": "Dutch environmental attitudes resemble those in other countries, supporting generalizability",
258 + "evidence_type": "descriptive_statistic",
259 + "evidence_ref": "World Values Survey, ERNOP figures in conclusion"
260 + }
261 + ],
262 + "structural_summary": "Opens with SRI growth framed as a finance puzzle and lists competing motives: social preferences, financial beliefs and signaling. It justifies an experimental measure of preferences, then previews results with magnitudes and links them to asset pricing via a taste-versus-beliefs argument, followed by a roadmap. Data and variable sections precede descriptive characteristics, the main regression analysis of motives, covariates of preferences and the charity-substitution test. The conclusion restates findings, draws asset-pricing implications and discusses generalizability. Appendix sample exclusions serve as robustness checks.",
263 + "writing_profile_keywords": [
264 + "administrative data linked to survey",
265 + "incentivized experiment",
266 + "trust game",
267 + "probit marginal effects",
268 + "social preferences",
269 + "social signaling",
270 + "household portfolio choice",
271 + "associational framing"
272 + ],
273 + "confidence": 0.8,
274 + "notes": "No explicit contribution paragraph is visible, so contribution_paragraph_index is set to -1. The causal-sounding wording ('determining', 'effect') is tempered by an explicit statement that causality cannot be addressed. Robustness checks are carried out as appendix sample exclusions."
275 + },
276 + "deterministic_checks": {
277 + "roadmap_regex": true,
278 + "contribution_regex_in_intro": true,
279 + "magnitude_in_abstract": false,
280 + "abstract_n_sentences": 6,
281 + "intro_n_paragraphs": 7,
282 + "intro_n_words": 1115,
283 + "causal_words_in_abstract": 0,
284 + "hedge_words_in_abstract": 0,
285 + "stars_mentioned": true,
286 + "clustered_se_mentioned": false,
287 + "internet_appendix_mentioned": true,
288 + "section_roles_rule": {
289 + "0": "introduction",
290 + "1": "data",
291 + "2": "data",
292 + "3": "other",
293 + "4": "other",
294 + "5": "other",
295 + "6": "other",
296 + "7": "conclusion",
297 + "8": "appendix",
298 + "9": "appendix",
299 + "10": "appendix",
300 + "11": "appendix",
301 + "12": "appendix",
302 + "13": "appendix",
303 + "14": "appendix",
304 + "15": "appendix",
305 + "16": "appendix"
306 + }
307 + },
308 + "schema_valid": true,
309 + "code_version": "feb6c9e-dirty"
310 +}
\ No newline at end of file
added data/annotations/documents/jof_12548.json +205 −0
@@ -0,0 +1,205 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12548",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:59:48+00:00",
9 + "source_sha256": "d2646afd2f16de6d508d7d313f11c89a4c2bc16d2ab10c15573fe03699cd03b9",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 6566,
12 + "usage": {
13 + "input": 2692,
14 + "output": 1536,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.042282,
19 + "request_id": "req_011Cggpn8ywscmTMrAR9b42d",
20 + "annotation": {
21 + "genre": "replication_or_correction",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "theoretical Kyle-type model with nonfundamental speculation; corrigendum",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical correction of equilibrium characterization)",
27 + "strategy": "not applicable; analytical re-derivation of the original model's equilibrium including the omitted second-order condition",
28 + "key_assumptions": [
29 + "original Madrigal (1996) model setup and notation retained"
30 + ],
31 + "threats_addressed": [],
32 + "causal_language_strength": 1,
33 + "uses_causal_claim": false
34 + },
35 + "abstract_moves": {
36 + "question": {
37 + "present": false,
38 + "sentence_index": -1
39 + },
40 + "gap": {
41 + "present": true,
42 + "sentence_index": 1
43 + },
44 + "method": {
45 + "present": false,
46 + "sentence_index": -1
47 + },
48 + "results": {
49 + "present": true,
50 + "sentence_index": 0
51 + },
52 + "magnitude": {
53 + "present": false,
54 + "sentence_index": -1
55 + },
56 + "contribution": {
57 + "present": true,
58 + "sentence_index": 3
59 + }
60 + },
61 + "introduction": {
62 + "moves": [
63 + {
64 + "paragraph_index": 0,
65 + "move": "motivation"
66 + },
67 + {
68 + "paragraph_index": 0,
69 + "move": "tension_or_puzzle"
70 + },
71 + {
72 + "paragraph_index": 0,
73 + "move": "contribution_statement"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "preview_of_results"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "roadmap"
82 + }
83 + ],
84 + "contribution_paragraph_index": 0,
85 + "contribution_types": [
86 + "replication_or_correction"
87 + ],
88 + "roadmap_present": true,
89 + "literature_positioning_location": "none"
90 + },
91 + "section_roles": [
92 + {
93 + "idx": 0,
94 + "role": "introduction"
95 + },
96 + {
97 + "idx": 1,
98 + "role": "theory"
99 + },
100 + {
101 + "idx": 2,
102 + "role": "results"
103 + },
104 + {
105 + "idx": 3,
106 + "role": "appendix"
107 + },
108 + {
109 + "idx": 4,
110 + "role": "appendix"
111 + },
112 + {
113 + "idx": 5,
114 + "role": "appendix"
115 + }
116 + ],
117 + "results_presentation": {
118 + "economic_significance_discussed": false,
119 + "statistical_significance_language": "unclear",
120 + "magnitude_statements_in_intro": false
121 + },
122 + "robustness": {
123 + "location": "none_seen",
124 + "types": []
125 + },
126 + "tables": {
127 + "convention_notes": "The tables set the original equations and the original numerical table beside corrected versions, giving each original expression next to its correction. They contain no statistical estimates.",
128 + "units_stated_in_notes": false,
129 + "standard_errors_description": "none"
130 + },
131 + "figures": {
132 + "uncertainty_shown": "no_figures"
133 + },
134 + "conclusion": {
135 + "bounded_claims": false,
136 + "policy_implications": false,
137 + "open_questions": false,
138 + "restates_contribution": false
139 + },
140 + "hedging": {
141 + "level": "low",
142 + "typical_devices": [
143 + "'may not exist' conditional existence statements",
144 + "parameter-region qualifiers ('if information is sufficiently precise')"
145 + ]
146 + },
147 + "claim_evidence": [
148 + {
149 + "claim": "A linear pure strategy equilibrium may fail to exist in Madrigal's model when the speculator's noise-trading information is sufficiently precise",
150 + "evidence_type": "equation_or_proof",
151 + "evidence_ref": "Section II.A; Appendix B (Proof of Proposition 2)"
152 + },
153 + {
154 + "claim": "The original characterization omits a second-order condition",
155 + "evidence_type": "equation_or_proof",
156 + "evidence_ref": "Section I; Appendix A"
157 + },
158 + {
159 + "claim": "Calculation errors in the original derivations mislead implications for liquidity and early price discovery",
160 + "evidence_type": "table",
161 + "evidence_ref": "Tables CI and CII; Section II.B; Appendix C"
162 + },
163 + {
164 + "claim": "Order-flow-based speculation remains relevant, e.g., HFT and proprietary trading",
165 + "evidence_type": "none_stated",
166 + "evidence_ref": "Introduction"
167 + }
168 + ],
169 + "structural_summary": "This is a short corrigendum to a published Kyle-type model. A single introductory paragraph motivates current relevance, states that the original solution is partly wrong, and previews two corrections with a section roadmap. Section I restates the original setup in its original notation and gives the correct equilibrium characterization, including the second-order condition. Section II first proves that the equilibrium need not exist and then corrects calculation errors and their implications. Appendices hold the derivations and the proof, and corrected material is placed in tables and an internet appendix. There is no conclusion and essentially no literature positioning.",
170 + "writing_profile_keywords": [
171 + "corrigendum",
172 + "Kyle model",
173 + "equilibrium existence",
174 + "second-order condition",
175 + "calculation errors",
176 + "original notation",
177 + "proof in appendix"
178 + ],
179 + "confidence": 0.88,
180 + "notes": "No conclusion section is visible in the excerpts. The abstract has no explicit question or method sentence. The 'gap' role is assigned to the sentence on the omitted second-order condition, which explains the error."
181 + },
182 + "deterministic_checks": {
183 + "roadmap_regex": false,
184 + "contribution_regex_in_intro": false,
185 + "magnitude_in_abstract": false,
186 + "abstract_n_sentences": 4,
187 + "intro_n_paragraphs": 1,
188 + "intro_n_words": 188,
189 + "causal_words_in_abstract": 0,
190 + "hedge_words_in_abstract": 1,
191 + "stars_mentioned": false,
192 + "clustered_se_mentioned": false,
193 + "internet_appendix_mentioned": true,
194 + "section_roles_rule": {
195 + "0": "introduction",
196 + "1": "theory",
197 + "2": "other",
198 + "3": "appendix",
199 + "4": "appendix",
200 + "5": "appendix"
201 + }
202 + },
203 + "schema_valid": true,
204 + "code_version": "feb6c9e-dirty"
205 +}
\ No newline at end of file
added data/annotations/documents/jof_12583.json +265 −0
@@ -0,0 +1,265 @@
1 +{
2 + "validation_errors": [
3 + "Additional properties are not allowed ('figures' was unexpected) @ figures"
4 + ],
5 + "doc_id": "jof_12583",
6 + "ontology_version": "1.0",
7 + "prompt_version": "1.0",
8 + "model": "claude-opus-5-5",
9 + "effort": "medium",
10 + "annotated_at": "2026-11-03T23:00:27+00:00",
11 + "source_sha256": "074e6a8620dc1531236a29942c3e072a275fc5bc1f428dc8faaa2dfb0d20b1c0",
12 + "parser": "pymupdf-jof-heuristics-1.3",
13 + "payload_chars": 23050,
14 + "usage": {
15 + "input": 7726,
16 + "output": 2176,
17 + "cache_read": 3968,
18 + "cache_write": 0
19 + },
20 + "cost_usd": 0.075218,
21 + "request_id": "req_011CggppYqhVccuPdcDQBtVp",
22 + "annotation": {
23 + "genre": "mixed_theory_empirical",
24 + "subfield": "corporate_finance",
25 + "subfield_secondary": "dynamic contracting / executive compensation and CEO turnover",
26 + "primary_method": "theoretical_model_analytical",
27 + "identification": {
28 + "estimand": "Associations between a firm's growth prospects (initial Q, ratio of Q) and the probability of CEO turnover and the front-loading of CEO pay",
29 + "strategy": "Probit regressions of turnover and panel regressions of log pay on tenure-year interacted with initial Q, run on ExecuComp/Compustat/CRSP 1992-2014, used to test model predictions; no quasi-experimental design visible",
30 + "key_assumptions": [
31 + "Average Q measures growth prospects",
32 + "The model's comparative statics map onto cross-firm variation in observed Q"
33 + ],
34 + "threats_addressed": [
35 + "Acknowledges other theories may explain the same patterns"
36 + ],
37 + "causal_language_strength": 2,
38 + "uses_causal_claim": false
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": true,
43 + "sentence_index": 0
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": true,
51 + "sentence_index": 0
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 2
56 + },
57 + "magnitude": {
58 + "present": false,
59 + "sentence_index": -1
60 + },
61 + "contribution": {
62 + "present": true,
63 + "sentence_index": 1
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "tension_or_puzzle"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "preview_of_approach"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "contribution_statement"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "preview_of_results"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 2,
94 + "move": "mechanism_or_theory_sketch"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "mechanism_or_theory_sketch"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "preview_of_results"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "preview_of_approach"
107 + },
108 + {
109 + "paragraph_index": 5,
110 + "move": "related_literature_positioning"
111 + },
112 + {
113 + "paragraph_index": 6,
114 + "move": "related_literature_positioning"
115 + },
116 + {
117 + "paragraph_index": 6,
118 + "move": "contribution_statement"
119 + },
120 + {
121 + "paragraph_index": 7,
122 + "move": "roadmap"
123 + }
124 + ],
125 + "contribution_paragraph_index": 6,
126 + "contribution_types": [
127 + "new_theory_or_model",
128 + "new_method",
129 + "new_evidence"
130 + ],
131 + "roadmap_present": true,
132 + "literature_positioning_location": "inside_introduction"
133 + },
134 + "section_roles": [
135 + {
136 + "idx": 0,
137 + "role": "introduction"
138 + },
139 + {
140 + "idx": 1,
141 + "role": "theory"
142 + },
143 + {
144 + "idx": 2,
145 + "role": "theory"
146 + },
147 + {
148 + "idx": 3,
149 + "role": "theory"
150 + },
151 + {
152 + "idx": 4,
153 + "role": "results"
154 + },
155 + {
156 + "idx": 5,
157 + "role": "conclusion"
158 + }
159 + ],
160 + "results_presentation": {
161 + "economic_significance_discussed": false,
162 + "statistical_significance_language": "unclear",
163 + "magnitude_statements_in_intro": false
164 + },
165 + "robustness": {
166 + "location": "none_seen",
167 + "types": []
168 + },
169 + "tables": {
170 + "convention_notes": "Table notes name the merged data set and sample period, define each variable, and identify the estimator (probit marginal effects, compensation regressions). The visible excerpts do not describe standard errors.",
171 + "units_stated_in_notes": true,
172 + "standard_errors_description": "unclear"
173 + },
174 + "figures": {
175 + "uncertainty_shown": "no",
176 + "figures": null
177 + },
178 + "conclusion": {
179 + "bounded_claims": true,
180 + "policy_implications": false,
181 + "open_questions": true,
182 + "restates_contribution": true
183 + },
184 + "hedging": {
185 + "level": "moderate",
186 + "typical_devices": [
187 + "'broadly supports' / 'consistent with' for empirical findings",
188 + "'may' for model outcomes",
189 + "explicit acknowledgment that other theories may fit the evidence",
190 + "'we believe' for extensions"
191 + ]
192 + },
193 + "claim_evidence": [
194 + {
195 + "claim": "The risk of growth-induced dismissal raises managers' effective discount rate and therefore agency costs, which leads to front-loaded pay.",
196 + "evidence_type": "equation_or_proof",
197 + "evidence_ref": "Sections I-III"
198 + },
199 + {
200 + "claim": "Low-growth firms optimally forgo growth opportunities after good performance, which amounts to job protection.",
201 + "evidence_type": "equation_or_proof",
202 + "evidence_ref": "Section III.D, Figure 2"
203 + },
204 + {
205 + "claim": "Severance pay is suboptimal, but incoming managers may receive a signing bonus.",
206 + "evidence_type": "equation_or_proof",
207 + "evidence_ref": "Sections II-III"
208 + },
209 + {
210 + "claim": "Firms with better growth prospects have higher CEO turnover.",
211 + "evidence_type": "table",
212 + "evidence_ref": "Tables II-III, Figure 5"
213 + },
214 + {
215 + "claim": "Firms with better growth prospects use more front-loaded CEO compensation.",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table IV, Figure 6"
218 + },
219 + {
220 + "claim": "High-growth firms have shorter tenures and shorter compensation duration in model simulations.",
221 + "evidence_type": "figure",
222 + "evidence_ref": "Figures 3-4"
223 + },
224 + {
225 + "claim": "Managers matter for firm performance, especially in growing environments.",
226 + "evidence_type": "prior_literature",
227 + "evidence_ref": "Bertrand and Schoar (2003); Bennedsen et al. (2012)"
228 + }
229 + ],
230 + "structural_summary": "The paper is theory-led and adds a short empirical test. The introduction motivates a tension between replacing managers to seize growth and the cost of incentive provision. It then previews the mechanism and the main contract results (two firm types, drift adjustment, no severance pay), positions the work against the CEO-match and dynamic-contracting literatures, lists technical contributions, and closes with a roadmap. The body builds a two-period model with derived empirical implications, then a continuous-time model with a state-space representation of contracts. Next it characterizes the optimal contract with numerical illustrations, and finally reports probit and compensation regressions on ExecuComp data. The conclusion restates the insight, suggests extensions, and hedges on the empirical channel.",
231 + "writing_profile_keywords": [
232 + "dynamic moral hazard",
233 + "continuous-time contracting",
234 + "CEO turnover",
235 + "front-loaded compensation",
236 + "model-then-evidence",
237 + "consistent-with empirics"
238 + ],
239 + "confidence": 0.85,
240 + "notes": "The primary contribution is theoretical. The empirical section is supportive and its claims are associational. Standard errors and robustness checks are not visible in the excerpts, and an Internet Appendix is mentioned."
241 + },
242 + "deterministic_checks": {
243 + "roadmap_regex": true,
244 + "contribution_regex_in_intro": true,
245 + "magnitude_in_abstract": false,
246 + "abstract_n_sentences": 5,
247 + "intro_n_paragraphs": 8,
248 + "intro_n_words": 1506,
249 + "causal_words_in_abstract": 0,
250 + "hedge_words_in_abstract": 1,
251 + "stars_mentioned": false,
252 + "clustered_se_mentioned": false,
253 + "internet_appendix_mentioned": false,
254 + "section_roles_rule": {
255 + "0": "introduction",
256 + "1": "theory",
257 + "2": "theory",
258 + "3": "other",
259 + "4": "results",
260 + "5": "conclusion"
261 + }
262 + },
263 + "schema_valid": false,
264 + "code_version": "feb6c9e-dirty"
265 +}
\ No newline at end of file
added data/annotations/documents/jof_12584.json +245 −0
@@ -0,0 +1,245 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12584",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:07+00:00",
9 + "source_sha256": "498956349ed5015ca3ffdf01ff40c4c79fab7d485138261560850dfe12f5a918",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 15952,
12 + "usage": {
13 + "input": 5670,
14 + "output": 1936,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.062194,
19 + "request_id": "req_011CggpoHhLdfXjkQEptAiWg",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "market_microstructure; financial fragility / limits to arbitrage",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theory): comparative statics of equilibrium prices, asset quality, and welfare with respect to the amount of arbitrage capital",
27 + "strategy": "Noisy rational expectations equilibrium with an endogenous lemons problem; arbitrage capital is varied in comparative statics",
28 + "key_assumptions": [
29 + "two-period economy with a risk-free asset in elastic supply",
30 + "informed participants (seller, arbitrageurs) are liquidity- or margin-constrained",
31 + "uninformed hedgers with random exposures generate noise",
32 + "risk-neutral competitive uninformed investors set prices",
33 + "seller chooses between selling the marketable asset and liquidating a nonmarketable asset"
34 + ],
35 + "threats_addressed": [
36 + "exogenous margin constraints relaxed via an endogenous-margin extension"
37 + ],
38 + "causal_language_strength": 2,
39 + "uses_causal_claim": false
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 0
45 + },
46 + "gap": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 1
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 2
57 + },
58 + "magnitude": {
59 + "present": false,
60 + "sentence_index": -1
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 4
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "related_literature_positioning"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "tension_or_puzzle"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "mechanism_or_theory_sketch"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "preview_of_results"
92 + }
93 + ],
94 + "contribution_paragraph_index": 2,
95 + "contribution_types": [
96 + "new_theory_or_model",
97 + "resolution_of_puzzle"
98 + ],
99 + "roadmap_present": false,
100 + "literature_positioning_location": "both"
101 + },
102 + "section_roles": [
103 + {
104 + "idx": 0,
105 + "role": "introduction"
106 + },
107 + {
108 + "idx": 1,
109 + "role": "literature"
110 + },
111 + {
112 + "idx": 2,
113 + "role": "theory"
114 + },
115 + {
116 + "idx": 3,
117 + "role": "theory"
118 + },
119 + {
120 + "idx": 4,
121 + "role": "results"
122 + },
123 + {
124 + "idx": 5,
125 + "role": "robustness"
126 + },
127 + {
128 + "idx": 6,
129 + "role": "other"
130 + },
131 + {
132 + "idx": 7,
133 + "role": "conclusion"
134 + },
135 + {
136 + "idx": 8,
137 + "role": "appendix"
138 + },
139 + {
140 + "idx": 9,
141 + "role": "appendix"
142 + }
143 + ],
144 + "results_presentation": {
145 + "economic_significance_discussed": false,
146 + "statistical_significance_language": "unclear",
147 + "magnitude_statements_in_intro": false
148 + },
149 + "robustness": {
150 + "location": "main_text_section",
151 + "types": [
152 + "alternative_specification",
153 + "sensitivity_to_parameters"
154 + ]
155 + },
156 + "tables": {
157 + "convention_notes": "No tables. Results are shown through numerical comparative-statics figures, with the x-axis scaled as a percentage of baseline arbitrage capital.",
158 + "units_stated_in_notes": false,
159 + "standard_errors_description": "none"
160 + },
161 + "figures": {
162 + "uncertainty_shown": "no"
163 + },
164 + "conclusion": {
165 + "bounded_claims": true,
166 + "policy_implications": true,
167 + "open_questions": false,
168 + "restates_contribution": true
169 + },
170 + "hedging": {
171 + "level": "moderate",
172 + "typical_devices": [
173 + "'plausible mechanism' framing",
174 + "modal verbs (can, may)",
175 + "'shed light on' phrasing"
176 + ]
177 + },
178 + "claim_evidence": [
179 + {
180 + "claim": "When arbitrage capital is ample, prices are informative and there is no price dislocation.",
181 + "evidence_type": "equation_or_proof",
182 + "evidence_ref": "Section IV.A; Appendix B proofs"
183 + },
184 + {
185 + "claim": "Scarce arbitrage capital makes prices less informative and creates an endogenous lemons problem, causing fire sales and market freezes.",
186 + "evidence_type": "equation_or_proof",
187 + "evidence_ref": "Sections III-IV; Figures 3-4"
188 + },
189 + {
190 + "claim": "Reducing arbitrage capital lowers the expected price of the high-quality asset and the average quality of traded assets.",
191 + "evidence_type": "figure",
192 + "evidence_ref": "Figures 5-7"
193 + },
194 + {
195 + "claim": "Welfare declines as arbitrage capital falls, across parameter values.",
196 + "evidence_type": "figure",
197 + "evidence_ref": "Figure 8, Section IV.B"
198 + },
199 + {
200 + "claim": "The results are robust to endogenizing margin constraints.",
201 + "evidence_type": "figure",
202 + "evidence_ref": "Section V, Figure 9"
203 + }
204 + ],
205 + "structural_summary": "The paper opens with a puzzle: why do fire sales occur when well-capitalized nonspecialists exist? It contrasts this with existing explanations based on specialist valuations and limited arbitrage. It then sketches a two-period model combining informed-but-constrained and uninformed-but-unconstrained agents, and previews the capital-multiplier mechanism. A separate literature review follows. The model and its equilibrium are set out formally, and the main results cover fire sales, market freezes, and welfare. An extension endogenizes margin constraints as a robustness check. A discussion section covers fragility, policy, and the 2007-08 crisis, and the conclusion restates the mechanism. Proofs are in the appendix.",
206 + "writing_profile_keywords": [
207 + "puzzle-driven",
208 + "noisy REE",
209 + "lemons problem",
210 + "limits to arbitrage",
211 + "comparative statics figures",
212 + "policy discussion",
213 + "crisis interpretation"
214 + ],
215 + "confidence": 0.85,
216 + "notes": "Pure theory paper. The figures are numerical illustrations, not empirical results. The Discussion section (idx 6) is coded 'other' rather than 'conclusion'; its content is policy implications and an interpretation of the 2007-08 crisis."
217 + },
218 + "deterministic_checks": {
219 + "roadmap_regex": false,
220 + "contribution_regex_in_intro": false,
221 + "magnitude_in_abstract": false,
222 + "abstract_n_sentences": 5,
223 + "intro_n_paragraphs": 3,
224 + "intro_n_words": 624,
225 + "causal_words_in_abstract": 0,
226 + "hedge_words_in_abstract": 0,
227 + "stars_mentioned": false,
228 + "clustered_se_mentioned": false,
229 + "internet_appendix_mentioned": false,
230 + "section_roles_rule": {
231 + "0": "introduction",
232 + "1": "literature",
233 + "2": "theory",
234 + "3": "theory",
235 + "4": "results",
236 + "5": "robustness",
237 + "6": "conclusion",
238 + "7": "conclusion",
239 + "8": "appendix",
240 + "9": "appendix"
241 + }
242 + },
243 + "schema_valid": true,
244 + "code_version": "feb6c9e-dirty"
245 +}
\ No newline at end of file
added data/annotations/documents/jof_12585.json +271 −0
@@ -0,0 +1,271 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12585",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:11+00:00",
9 + "source_sha256": "7347d19ef06d3680c172b9fc5aa92d4f48efe74f95d493e222d4af4d6d1fbae0",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22129,
12 + "usage": {
13 + "input": 7666,
14 + "output": 2450,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.080458,
19 + "request_id": "req_011CggpoHp3AX4XaoHuxAaUo",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "corporate_finance; political economy of government credit in China",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Effect of CDB government credit (industrial SOE loans and infrastructure loans) on private firms' and SOEs' assets, employment, sales and efficiency, in the same industry and in downstream industries",
27 + "strategy": "2SLS using the predicted timing of municipal party-secretary turnover cycles, interacted with each city's predetermined focal SOE industry, as instruments for province-industry CDB loan flows; OLS baseline; input-output matrix used to identify downstream industries",
28 + "key_assumptions": [
29 + "Predicted political turnover timing is exogenous to local economic conditions",
30 + "Exclusion: turnover cycles affect firms only through CDB credit, not through other channels such as other-bank borrowing, land sales, fiscal transfers or tax enforcement",
31 + "Focal industry is predetermined and stable over time",
32 + "Effects of SOE loans are separable from those of infrastructure loans"
33 + ],
34 + "threats_addressed": [
35 + "Turnover timing correlated with local conditions (hazard regressions)",
36 + "Alternative political channels (other banks, land sales, transfers, tax enforcement)",
37 + "Confounding by infrastructure loans",
38 + "Coincidence with national political cycles (off-cycle cities)",
39 + "Unobserved investment-opportunity shocks over the turnover cycle"
40 + ],
41 + "causal_language_strength": 4,
42 + "uses_causal_claim": true
43 + },
44 + "abstract_moves": {
45 + "question": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "gap": {
50 + "present": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 4
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 1
60 + },
61 + "magnitude": {
62 + "present": true,
63 + "sentence_index": 2
64 + },
65 + "contribution": {
66 + "present": false,
67 + "sentence_index": -1
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "tension_or_puzzle"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "literature_gap"
83 + },
84 + {
85 + "paragraph_index": 0,
86 + "move": "preview_of_approach"
87 + },
88 + {
89 + "paragraph_index": 0,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 1,
94 + "move": "preview_of_approach"
95 + },
96 + {
97 + "paragraph_index": 1,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 2,
102 + "move": "preview_of_results"
103 + },
104 + {
105 + "paragraph_index": 2,
106 + "move": "contribution_statement"
107 + },
108 + {
109 + "paragraph_index": 3,
110 + "move": "identification_defense"
111 + },
112 + {
113 + "paragraph_index": 4,
114 + "move": "related_literature_positioning"
115 + },
116 + {
117 + "paragraph_index": 4,
118 + "move": "contribution_statement"
119 + },
120 + {
121 + "paragraph_index": 5,
122 + "move": "roadmap"
123 + }
124 + ],
125 + "contribution_paragraph_index": 2,
126 + "contribution_types": [
127 + "new_evidence",
128 + "new_data_or_measure",
129 + "new_identification",
130 + "resolution_of_puzzle"
131 + ],
132 + "roadmap_present": true,
133 + "literature_positioning_location": "inside_introduction"
134 + },
135 + "section_roles": [
136 + {
137 + "idx": 0,
138 + "role": "introduction"
139 + },
140 + {
141 + "idx": 1,
142 + "role": "other"
143 + },
144 + {
145 + "idx": 2,
146 + "role": "data"
147 + },
148 + {
149 + "idx": 3,
150 + "role": "results"
151 + },
152 + {
153 + "idx": 4,
154 + "role": "conclusion"
155 + },
156 + {
157 + "idx": 5,
158 + "role": "appendix"
159 + }
160 + ],
161 + "results_presentation": {
162 + "economic_significance_discussed": true,
163 + "statistical_significance_language": "unclear",
164 + "magnitude_statements_in_intro": true
165 + },
166 + "robustness": {
167 + "location": "mixed",
168 + "types": [
169 + "subsample",
170 + "falsification",
171 + "alternative_specification"
172 + ]
173 + },
174 + "tables": {
175 + "convention_notes": "Notes give unit of observation, sample coverage (cities, provinces, industries, years), estimator (OLS, 2SLS, Cox hazard, Probit) and instrument definitions. An appendix table defines variables and gives loan units (100 million RMB). Appendix tables report first-stage results and an off-national-cycle subsample.",
176 + "units_stated_in_notes": true,
177 + "standard_errors_description": "unclear"
178 + },
179 + "figures": {
180 + "uncertainty_shown": "unclear"
181 + },
182 + "conclusion": {
183 + "bounded_claims": true,
184 + "policy_implications": true,
185 + "open_questions": false,
186 + "restates_contribution": true
187 + },
188 + "hedging": {
189 + "level": "moderate",
190 + "typical_devices": [
191 + "'suggestive evidence' for the political career-concern channel",
192 + "'alleviate the concern' when defending the exclusion restriction",
193 + "'could explain' when linking results to mixed prior findings",
194 + "'not likely to be confounded'"
195 + ]
196 + },
197 + "claim_evidence": [
198 + {
199 + "claim": "CDB SOE industrial loans crowd out private firms in the same industry, reducing their assets, employment and sales",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table IV, Table V"
202 + },
203 + {
204 + "claim": "CDB SOE loans to upstream focal industries raise assets and sales of downstream private firms",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table VI"
207 + },
208 + {
209 + "claim": "CDB loans to cities rise early in a city secretary's term, giving a zig-zag borrowing pattern",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table III, Figure 3, Table AIII"
212 + },
213 + {
214 + "claim": "Other channels of politician influence are uncorrelated with the turnover cycles",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table VII"
217 + },
218 + {
219 + "claim": "Infrastructure loans crowd in private firms, while industrial SOE loans have the opposite effect",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table IX, Figure A2"
222 + },
223 + {
224 + "claim": "Turnover timing is not predicted by local economic conditions",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table II"
227 + },
228 + {
229 + "claim": "Turnover-cycle effects are stronger in cities with better access to CDB credit",
230 + "evidence_type": "table",
231 + "evidence_ref": "Table VIII"
232 + }
233 + ],
234 + "structural_summary": "The introduction frames a crowd-out versus crowd-in debate and attributes mixed prior evidence to aggregate data. It previews a supply-chain decomposition built on proprietary bank data, with an instrument from political turnover timing, and reports magnitudes for first-stage and 2SLS effects. It then defends the exclusion restriction, positions the paper in the political business cycle literature and gives a roadmap. A background section on the bank and a data section are followed by one long results section. That section moves from politicians' career concerns, to the first stage, to same-industry and downstream effects, to alternative channels, infrastructure versus SOE loans, reduced form and overall effects. A one-paragraph conclusion restates the findings and draws broader implications. Further results are in an internet appendix.",
235 + "writing_profile_keywords": [
236 + "proprietary data",
237 + "political turnover instrument",
238 + "2SLS",
239 + "supply-chain spillovers",
240 + "crowding out versus crowding in",
241 + "China",
242 + "state-owned enterprises",
243 + "exclusion restriction defense"
244 + ],
245 + "confidence": 0.8,
246 + "notes": "Section I is institutional background, so it is coded 'other' rather than literature. The excerpts do not show whether results are reported with stars or standard errors, or how standard errors are computed. Robustness checks (off-cycle cities, other channels, heterogeneity by CDB access) appear within the results section and the appendix tables. The conclusion is truncated, so policy discussion is inferred only from its broader framing of government credit's costs and benefits."
247 + },
248 + "deterministic_checks": {
249 + "roadmap_regex": true,
250 + "contribution_regex_in_intro": false,
251 + "magnitude_in_abstract": false,
252 + "abstract_n_sentences": 5,
253 + "intro_n_paragraphs": 6,
254 + "intro_n_words": 974,
255 + "causal_words_in_abstract": 2,
256 + "hedge_words_in_abstract": 0,
257 + "stars_mentioned": false,
258 + "clustered_se_mentioned": false,
259 + "internet_appendix_mentioned": true,
260 + "section_roles_rule": {
261 + "0": "introduction",
262 + "1": "literature",
263 + "2": "data",
264 + "3": "results",
265 + "4": "conclusion",
266 + "5": "appendix"
267 + }
268 + },
269 + "schema_valid": true,
270 + "code_version": "feb6c9e-dirty"
271 +}
\ No newline at end of file
added data/annotations/documents/jof_12586.json +258 −0
@@ -0,0 +1,258 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12586",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:08+00:00",
9 + "source_sha256": "557693761121cbd71b6db556b38e95779dd6f444bdc5857f247e941adf65c106",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 16123,
12 + "usage": {
13 + "input": 5653,
14 + "output": 1982,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.063046,
19 + "request_id": "req_011CggpoJAs4f1ifBATuiYyU",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "macro_finance",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theory paper with descriptive motivating evidence on predictability of forecast errors and revisions)",
27 + "strategy": "Analytical model of diagnostic expectations embedded in a neoclassical investment model; motivating time-series regressions of forecast errors and revisions on current credit spreads",
28 + "key_assumptions": [
29 + "Beliefs follow diagnostic expectations derived from the representativeness heuristic",
30 + "No financial frictions in the baseline model",
31 + "Fundamentals follow a persistent stochastic process"
32 + ],
33 + "threats_addressed": [
34 + "Lucas critique of mechanical backward-looking belief models",
35 + "Distinguishing which predictions also hold under rational expectations"
36 + ],
37 + "causal_language_strength": 2,
38 + "uses_causal_claim": false
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": false,
43 + "sentence_index": -1
44 + },
45 + "gap": {
46 + "present": false,
47 + "sentence_index": -1
48 + },
49 + "method": {
50 + "present": true,
51 + "sentence_index": 0
52 + },
53 + "results": {
54 + "present": true,
55 + "sentence_index": 4
56 + },
57 + "magnitude": {
58 + "present": false,
59 + "sentence_index": -1
60 + },
61 + "contribution": {
62 + "present": true,
63 + "sentence_index": 3
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "preview_of_approach"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "related_literature_positioning"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "motivation"
83 + },
84 + {
85 + "paragraph_index": 2,
86 + "move": "mechanism_or_theory_sketch"
87 + },
88 + {
89 + "paragraph_index": 3,
90 + "move": "contribution_statement"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "mechanism_or_theory_sketch"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 5,
102 + "move": "preview_of_results"
103 + },
104 + {
105 + "paragraph_index": 5,
106 + "move": "related_literature_positioning"
107 + },
108 + {
109 + "paragraph_index": 5,
110 + "move": "literature_gap"
111 + }
112 + ],
113 + "contribution_paragraph_index": 3,
114 + "contribution_types": [
115 + "new_theory_or_model",
116 + "synthesis",
117 + "new_evidence"
118 + ],
119 + "roadmap_present": false,
120 + "literature_positioning_location": "inside_introduction"
121 + },
122 + "section_roles": [
123 + {
124 + "idx": 0,
125 + "role": "introduction"
126 + },
127 + {
128 + "idx": 1,
129 + "role": "results"
130 + },
131 + {
132 + "idx": 2,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 3,
137 + "role": "results"
138 + },
139 + {
140 + "idx": 4,
141 + "role": "theory"
142 + },
143 + {
144 + "idx": 5,
145 + "role": "theory"
146 + },
147 + {
148 + "idx": 6,
149 + "role": "theory"
150 + },
151 + {
152 + "idx": 7,
153 + "role": "theory"
154 + },
155 + {
156 + "idx": 8,
157 + "role": "conclusion"
158 + }
159 + ],
160 + "results_presentation": {
161 + "economic_significance_discussed": false,
162 + "statistical_significance_language": "unclear",
163 + "magnitude_statements_in_intro": false
164 + },
165 + "robustness": {
166 + "location": "none_seen",
167 + "types": []
168 + },
169 + "tables": {
170 + "convention_notes": "No table notes are visible. The text refers to an econometric test in Table II, but table contents are not shown in the excerpts.",
171 + "units_stated_in_notes": false,
172 + "standard_errors_description": "unclear"
173 + },
174 + "figures": {
175 + "uncertainty_shown": "no"
176 + },
177 + "conclusion": {
178 + "bounded_claims": true,
179 + "policy_implications": false,
180 + "open_questions": false,
181 + "restates_contribution": true
182 + },
183 + "hedging": {
184 + "level": "moderate",
185 + "typical_devices": [
186 + "framing evidence as preliminary or motivating",
187 + "stating that data are insufficient for a stringent test",
188 + "'can account for' phrasing",
189 + "explicitly noting which predictions also arise under rational expectations"
190 + ]
191 + },
192 + "claim_evidence": [
193 + {
194 + "claim": "Credit expansions and low spreads forecast later reversals and weaker real activity",
195 + "evidence_type": "prior_literature",
196 + "evidence_ref": "Intro para 1 (Schularick and Taylor; Greenwood and Hanson; Lopez-Salido et al.)"
197 + },
198 + {
199 + "claim": "Errors in analyst forecasts of the Baa–Treasury spread are predictable from the current spread",
200 + "evidence_type": "figure",
201 + "evidence_ref": "Figure 1, Section I.A"
202 + },
203 + {
204 + "claim": "Forecast revisions of spreads are predictable: upward when current spreads are low, downward when high",
205 + "evidence_type": "figure",
206 + "evidence_ref": "Figure 2 and Table II, Section I.B"
207 + },
208 + {
209 + "claim": "Diagnostic expectations generate both neglect of tail risk and extrapolation",
210 + "evidence_type": "figure",
211 + "evidence_ref": "Figure 3, Section II"
212 + },
213 + {
214 + "claim": "In the model, spreads are excessively volatile, overreact to news and revert predictably",
215 + "evidence_type": "equation_or_proof",
216 + "evidence_ref": "Section IV"
217 + }
218 + ],
219 + "structural_summary": "The paper opens with crisis and Minsky-style motivation and reviews empirical credit-cycle facts. It sketches the representativeness-based belief mechanism and claims to unify extrapolation with risk neglect. Four model predictions are listed, and the authors separate those that also hold under rational expectations from those unique to their mechanism. They then position the work against financial-friction models. The body presents motivating survey-forecast evidence, derives diagnostic expectations from first principles, embeds them in a frictionless macro model and solves for equilibrium spreads. A safe-debt extension follows, and a short conclusion stresses portability and robustness to the Lucas critique.",
220 + "writing_profile_keywords": [
221 + "behavioral macro-finance",
222 + "microfounded beliefs",
223 + "enumerated predictions",
224 + "motivating stylized facts",
225 + "unification claim",
226 + "Lucas critique defense",
227 + "model extension"
228 + ],
229 + "confidence": 0.8,
230 + "notes": "Mostly theoretical; the empirical Section I is explicitly motivating rather than a formal test. Sections idx 2–3 are figure-title artifacts of Section I. No roadmap paragraph is visible in the truncated introduction."
231 + },
232 + "deterministic_checks": {
233 + "roadmap_regex": false,
234 + "contribution_regex_in_intro": false,
235 + "magnitude_in_abstract": false,
236 + "abstract_n_sentences": 6,
237 + "intro_n_paragraphs": 6,
238 + "intro_n_words": 966,
239 + "causal_words_in_abstract": 0,
240 + "hedge_words_in_abstract": 1,
241 + "stars_mentioned": false,
242 + "clustered_se_mentioned": false,
243 + "internet_appendix_mentioned": false,
244 + "section_roles_rule": {
245 + "0": "introduction",
246 + "1": "results",
247 + "2": "other",
248 + "3": "other",
249 + "4": "other",
250 + "5": "theory",
251 + "6": "theory",
252 + "7": "other",
253 + "8": "conclusion"
254 + }
255 + },
256 + "schema_valid": true,
257 + "code_version": "feb6c9e-dirty"
258 +}
\ No newline at end of file
added data/annotations/documents/jof_12587.json +260 −0
@@ -0,0 +1,260 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12587",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:25+00:00",
9 + "source_sha256": "1eff6038eb93684cc903684430706325580f2dda095f3c21e5b684da1b54a38c",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19367,
12 + "usage": {
13 + "input": 6777,
14 + "output": 2179,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.071482,
19 + "request_id": "req_011CggppSbA8GCouf3nWJZXN",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "international_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "The share of time-series variation in bilateral exchange rates explained by systematic currency factors (carry and dollar), plus the risk premia on portfolios of currencies sorted by dollar beta.",
27 + "strategy": "Country-level time-series factor regressions of exchange rate changes on carry, conditional carry and dollar factors, with the dependent currency excluded from the factors. These are combined with portfolio sorts on dollar betas and GMM/Fama-MacBeth asset pricing tests.",
28 + "key_assumptions": [
29 + "Portfolios average out country-specific shocks (law of large numbers).",
30 + "High-minus-low factors are immune to U.S.-specific shocks.",
31 + "Markets are complete, so exchange rate changes equal differences in log SDFs (used in the model interpretation)."
32 + ],
33 + "threats_addressed": [
34 + "Mechanical correlation, handled by excluding the dependent currency from the factors.",
35 + "U.S.-specific shocks, handled with the global dollar component and other base currencies.",
36 + "Data frequency, checked with daily, quarterly and annual data.",
37 + "Sample composition, checked with developed versus emerging countries."
38 + ],
39 + "causal_language_strength": 2,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": true,
61 + "sentence_index": 3
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 4
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "literature_gap"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_approach"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_results"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "contribution_statement"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "tension_or_puzzle"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "mechanism_or_theory_sketch"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "mechanism_or_theory_sketch"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "related_literature_positioning"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "related_literature_positioning"
113 + }
114 + ],
115 + "contribution_paragraph_index": 2,
116 + "contribution_types": [
117 + "new_evidence",
118 + "new_theory_or_model"
119 + ],
120 + "roadmap_present": false,
121 + "literature_positioning_location": "inside_introduction"
122 + },
123 + "section_roles": [
124 + {
125 + "idx": 0,
126 + "role": "introduction"
127 + },
128 + {
129 + "idx": 1,
130 + "role": "theory"
131 + },
132 + {
133 + "idx": 2,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 3,
138 + "role": "results"
139 + },
140 + {
141 + "idx": 4,
142 + "role": "robustness"
143 + },
144 + {
145 + "idx": 5,
146 + "role": "conclusion"
147 + },
148 + {
149 + "idx": 6,
150 + "role": "appendix"
151 + }
152 + ],
153 + "results_presentation": {
154 + "economic_significance_discussed": true,
155 + "statistical_significance_language": "unclear",
156 + "magnitude_statements_in_intro": true
157 + },
158 + "robustness": {
159 + "location": "main_text_section",
160 + "types": [
161 + "alternative_sample_period",
162 + "subsample",
163 + "alternative_measure",
164 + "other"
165 + ]
166 + },
167 + "tables": {
168 + "convention_notes": "Table notes write out the full country-level regression equation and define each variable. Tables are organised by country group and data frequency. One table reports portfolio summary statistics alongside GMM and Fama-MacBeth asset pricing results. Cross-country second-stage regressions relate shares of systematic exchange rate variation to comovement measures.",
169 + "units_stated_in_notes": true,
170 + "standard_errors_description": "unclear"
171 + },
172 + "figures": {
173 + "uncertainty_shown": "unclear"
174 + },
175 + "conclusion": {
176 + "bounded_claims": true,
177 + "policy_implications": true,
178 + "open_questions": true,
179 + "restates_contribution": true
180 + },
181 + "hedging": {
182 + "level": "moderate",
183 + "typical_devices": [
184 + "appear/appears",
185 + "suggest",
186 + "tentative interpretation",
187 + "open question",
188 + "could be interpreted"
189 + ]
190 + },
191 + "claim_evidence": [
192 + {
193 + "claim": "Carry and dollar factors together explain a large share of monthly bilateral exchange rate changes in developed countries.",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table I"
196 + },
197 + {
198 + "claim": "Sorting currencies on dollar beta produces a cross section of excess returns that is priced by a dollar slope factor.",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table IV; Figure 1"
201 + },
202 + {
203 + "claim": "Dollar-neutral global factors explain 18% to 83% of exchange rate variation and also describe yen- and pound-based rates.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Tables V and VI"
206 + },
207 + {
208 + "claim": "At the daily frequency, the dollar factor explains much more variation than the carry factor.",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table III"
211 + },
212 + {
213 + "claim": "Cross-country shares of systematic exchange rate variation are related to the comovement of capital flows.",
214 + "evidence_type": "figure",
215 + "evidence_ref": "Table VII; Figure 2"
216 + },
217 + {
218 + "claim": "SDFs must contain at least two global shocks to generate two independent cross sections of currency risk premia.",
219 + "evidence_type": "equation_or_proof",
220 + "evidence_ref": "Section I; Internet Appendix"
221 + }
222 + ],
223 + "structural_summary": "The introduction contrasts purely statistical principal components with interpretable risk factors. It then previews the factor regressions and their R2 ranges, names the dollar-neutral regressions as the key result, and treats the economic source of the shocks as an open question. A reduced-form SDF model is sketched, followed by in-intro literature positioning. The body moves from the model framework to time-series factor regressions across frequencies and samples, then to dollar-beta portfolio pricing. A section of implications and additional tests covers global shocks, other base currencies, conditional carry and links to macro comovement. A short conclusion draws implications for practitioners and researchers.",
224 + "writing_profile_keywords": [
225 + "factor regressions",
226 + "R2 decomposition",
227 + "currency portfolios",
228 + "reduced-form SDF model",
229 + "first-person singular",
230 + "dollar factor",
231 + "global shocks"
232 + ],
233 + "confidence": 0.8,
234 + "notes": "Single-author paper. Section IV's first paragraph announces 'three novel facts' but then lists four. Section II is classified as results. Significance conventions and standard errors are not visible in the excerpts."
235 + },
236 + "deterministic_checks": {
237 + "roadmap_regex": false,
238 + "contribution_regex_in_intro": true,
239 + "magnitude_in_abstract": true,
240 + "abstract_n_sentences": 5,
241 + "intro_n_paragraphs": 7,
242 + "intro_n_words": 1230,
243 + "causal_words_in_abstract": 0,
244 + "hedge_words_in_abstract": 1,
245 + "stars_mentioned": false,
246 + "clustered_se_mentioned": false,
247 + "internet_appendix_mentioned": true,
248 + "section_roles_rule": {
249 + "0": "introduction",
250 + "1": "theory",
251 + "2": "other",
252 + "3": "other",
253 + "4": "robustness",
254 + "5": "conclusion",
255 + "6": "appendix"
256 + }
257 + },
258 + "schema_valid": true,
259 + "code_version": "feb6c9e-dirty"
260 +}
\ No newline at end of file
added data/annotations/documents/jof_12588.json +236 −0
@@ -0,0 +1,236 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12588",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:03+00:00",
9 + "source_sha256": "96955ce449181b41a7f9ec8ea1db99ea16aa30f3e06f90b7caa3d9b889f0ebb3",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19917,
12 + "usage": {
13 + "input": 6832,
14 + "output": 1800,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.064122,
19 + "request_id": "req_011Cggpo3gyrkHX2tNkBYVUv",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model of shareholder incentives over leverage adjustments)",
27 + "strategy": "not applicable; analytical propositions derived in reduced-form static and dynamic trade-off models",
28 + "key_assumptions": [
29 + "shareholders cannot commit to future funding choices",
30 + "standard trade-off frictions: taxes, bankruptcy costs, agency costs",
31 + "creditors anticipate shareholder-controlled funding decisions"
32 + ],
33 + "threats_addressed": [],
34 + "causal_language_strength": 1,
35 + "uses_causal_claim": false
36 + },
37 + "abstract_moves": {
38 + "question": {
39 + "present": false,
40 + "sentence_index": -1
41 + },
42 + "gap": {
43 + "present": false,
44 + "sentence_index": -1
45 + },
46 + "method": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "results": {
51 + "present": true,
52 + "sentence_index": 1
53 + },
54 + "magnitude": {
55 + "present": false,
56 + "sentence_index": -1
57 + },
58 + "contribution": {
59 + "present": true,
60 + "sentence_index": 3
61 + }
62 + },
63 + "introduction": {
64 + "moves": [
65 + {
66 + "paragraph_index": 0,
67 + "move": "contribution_statement"
68 + },
69 + {
70 + "paragraph_index": 0,
71 + "move": "preview_of_approach"
72 + },
73 + {
74 + "paragraph_index": 1,
75 + "move": "mechanism_or_theory_sketch"
76 + },
77 + {
78 + "paragraph_index": 2,
79 + "move": "related_literature_positioning"
80 + },
81 + {
82 + "paragraph_index": 2,
83 + "move": "preview_of_results"
84 + },
85 + {
86 + "paragraph_index": 3,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "preview_of_approach"
92 + },
93 + {
94 + "paragraph_index": 4,
95 + "move": "policy_or_broader_implication"
96 + }
97 + ],
98 + "contribution_paragraph_index": 0,
99 + "contribution_types": [
100 + "new_theory_or_model"
101 + ],
102 + "roadmap_present": false,
103 + "literature_positioning_location": "both"
104 + },
105 + "section_roles": [
106 + {
107 + "idx": 0,
108 + "role": "introduction"
109 + },
110 + {
111 + "idx": 1,
112 + "role": "literature"
113 + },
114 + {
115 + "idx": 2,
116 + "role": "theory"
117 + },
118 + {
119 + "idx": 3,
120 + "role": "theory"
121 + },
122 + {
123 + "idx": 4,
124 + "role": "theory"
125 + },
126 + {
127 + "idx": 5,
128 + "role": "conclusion"
129 + },
130 + {
131 + "idx": 6,
132 + "role": "appendix"
133 + }
134 + ],
135 + "results_presentation": {
136 + "economic_significance_discussed": false,
137 + "statistical_significance_language": "unclear",
138 + "magnitude_statements_in_intro": false
139 + },
140 + "robustness": {
141 + "location": "none_seen",
142 + "types": []
143 + },
144 + "tables": {
145 + "convention_notes": "No tables. Results are stated as propositions with proofs, and numerical illustrations are given in figures.",
146 + "units_stated_in_notes": false,
147 + "standard_errors_description": "none"
148 + },
149 + "figures": {
150 + "uncertainty_shown": "no"
151 + },
152 + "conclusion": {
153 + "bounded_claims": true,
154 + "policy_implications": true,
155 + "open_questions": false,
156 + "restates_contribution": true
157 + },
158 + "hedging": {
159 + "level": "low",
160 + "typical_devices": [
161 + "conditional statements (if shareholders cannot commit...)",
162 + "'generally'/'typically'",
163 + "'may' in policy suggestions",
164 + "'consistent with' when linking to evidence"
165 + ]
166 + },
167 + "claim_evidence": [
168 + {
169 + "claim": "Shareholders lose from any leverage reduction once debt is in place, because the reduction transfers wealth to existing creditors",
170 + "evidence_type": "equation_or_proof",
171 + "evidence_ref": "Intro paragraph 1 algebra; Section II.B"
172 + },
173 + {
174 + "claim": "Resistance to leverage reduction persists even when the reduction alleviates bankruptcy or agency costs and raises firm value",
175 + "evidence_type": "equation_or_proof",
176 + "evidence_ref": "Section II.B-C"
177 + },
178 + {
179 + "claim": "Shareholders gain from issuing junior debt even when it destroys firm value",
180 + "evidence_type": "figure",
181 + "evidence_ref": "Section II.D; Figures 1-2"
182 + },
183 + {
184 + "claim": "Without commitment, equilibrium leverage takes stable levels that ratchet up with each shock",
185 + "evidence_type": "figure",
186 + "evidence_ref": "Section III; Figures 3-4"
187 + },
188 + {
189 + "claim": "When forced to delever, shareholders prefer asset sales over pure recapitalization",
190 + "evidence_type": "figure",
191 + "evidence_ref": "Section IV; Figures 6-7"
192 + },
193 + {
194 + "claim": "Firms respond asymmetrically to tax changes, consistent with the ratchet effect",
195 + "evidence_type": "prior_literature",
196 + "evidence_ref": "Conclusion citing Heider and Ljungqvist (2015)"
197 + }
198 + ],
199 + "structural_summary": "A theory paper. The abstract states the core mechanism and names it. The introduction states the contribution and builds intuition with simple algebra. It positions the paper against the Black-Scholes result, previews static, dynamic and asset-transaction results, and ends with banking and policy implications. A separate literature section follows. The body has three model sections: a one-time adjustment with frictions, dynamic equilibrium with shocks, maturity and covenants, and alternative deleveraging modes. The conclusion lists bulleted testable predictions and discusses welfare and policy. Proofs are in the appendix.",
200 + "writing_profile_keywords": [
201 + "analytical theory",
202 + "named effect",
203 + "propositions and proofs",
204 + "intuition after results",
205 + "testable predictions",
206 + "policy implications",
207 + "banking application"
208 + ],
209 + "confidence": 0.85,
210 + "notes": "Section IV was rule-labeled as robustness but is an extension of the model. The intro appears truncated, so no roadmap is visible."
211 + },
212 + "deterministic_checks": {
213 + "roadmap_regex": false,
214 + "contribution_regex_in_intro": false,
215 + "magnitude_in_abstract": false,
216 + "abstract_n_sentences": 5,
217 + "intro_n_paragraphs": 5,
218 + "intro_n_words": 830,
219 + "causal_words_in_abstract": 1,
220 + "hedge_words_in_abstract": 1,
221 + "stars_mentioned": false,
222 + "clustered_se_mentioned": false,
223 + "internet_appendix_mentioned": false,
224 + "section_roles_rule": {
225 + "0": "introduction",
226 + "1": "literature",
227 + "2": "other",
228 + "3": "theory",
229 + "4": "robustness",
230 + "5": "conclusion",
231 + "6": "appendix"
232 + }
233 + },
234 + "schema_valid": true,
235 + "code_version": "feb6c9e-dirty"
236 +}
\ No newline at end of file
added data/annotations/documents/jof_12589.json +284 −0
@@ -0,0 +1,284 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12589",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:54:20+00:00",
9 + "source_sha256": "0afa415dfc13f9e4335bc0a6cb82c5bcb2786606520cf17328e1466aa3d56e7e",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24501,
12 + "usage": {
13 + "input": 7870,
14 + "output": 2347,
15 + "cache_read": 0,
16 + "cache_write": 3968
17 + },
18 + "cost_usd": 0.09826,
19 + "request_id": "req_011CggpMRmJ56eBjjbFYjXy9",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "corporate_finance",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Effect of executives' personal lending relationships on lender choice and loan terms (spreads, amounts), and downstream effects on investment and employment during downturns",
27 + "strategy": "Executive deaths and quasi-exogenous retirements at other firms instrument for own-firm executive turnover, which shocks the firm's set of personal bank connections. Within-firm comparison across lenders, before and after turnover, using LPM/OLS/IV regressions.",
28 + "key_assumptions": [
29 + "Exclusion restriction: deaths and retirements at other firms do not directly affect a firm's loan terms across connected versus unconnected lenders",
30 + "Relevance: a geographically constrained executive market with industry-specific expertise links outside deaths to own-firm turnover",
31 + "Relationship-intensity measures capture the ex ante expected number of transactions between borrower and lender"
32 + ],
33 + "threats_addressed": [
34 + "Executive-firm matching on lending relationships (effects absent before turnover)",
35 + "Endogeneity of turnover",
36 + "Firm-level confounds (within-firm, across-lender comparison)",
37 + "Compensating differentials in other loan terms"
38 + ],
39 + "causal_language_strength": 4,
40 + "uses_causal_claim": true
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": true,
61 + "sentence_index": 1
62 + },
63 + "contribution": {
64 + "present": false,
65 + "sentence_index": -1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "mechanism_or_theory_sketch"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "identification_defense"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "identification_defense"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "preview_of_approach"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "preview_of_results"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "preview_of_results"
109 + },
110 + {
111 + "paragraph_index": 5,
112 + "move": "policy_or_broader_implication"
113 + },
114 + {
115 + "paragraph_index": 6,
116 + "move": "contribution_statement"
117 + },
118 + {
119 + "paragraph_index": 7,
120 + "move": "related_literature_positioning"
121 + },
122 + {
123 + "paragraph_index": 8,
124 + "move": "roadmap"
125 + }
126 + ],
127 + "contribution_paragraph_index": 6,
128 + "contribution_types": [
129 + "new_evidence",
130 + "new_data_or_measure",
131 + "new_identification",
132 + "resolution_of_puzzle"
133 + ],
134 + "roadmap_present": true,
135 + "literature_positioning_location": "inside_introduction"
136 + },
137 + "section_roles": [
138 + {
139 + "idx": 0,
140 + "role": "introduction"
141 + },
142 + {
143 + "idx": 1,
144 + "role": "data"
145 + },
146 + {
147 + "idx": 2,
148 + "role": "results"
149 + },
150 + {
151 + "idx": 3,
152 + "role": "results"
153 + },
154 + {
155 + "idx": 4,
156 + "role": "results"
157 + },
158 + {
159 + "idx": 5,
160 + "role": "other"
161 + },
162 + {
163 + "idx": 6,
164 + "role": "robustness"
165 + },
166 + {
167 + "idx": 7,
168 + "role": "conclusion"
169 + },
170 + {
171 + "idx": 8,
172 + "role": "appendix"
173 + }
174 + ],
175 + "results_presentation": {
176 + "economic_significance_discussed": true,
177 + "statistical_significance_language": "unclear",
178 + "magnitude_statements_in_intro": true
179 + },
180 + "robustness": {
181 + "location": "mixed",
182 + "types": [
183 + "placebo",
184 + "alternative_specification",
185 + "other"
186 + ]
187 + },
188 + "tables": {
189 + "convention_notes": "Table notes describe the estimator (fixed-effect LPM, OLS vs. IV), the dependent variable definition, sample size and period (9,458 Dealscan loans, 1994-2012), and the panel structure. Variable construction is referred to the text and the Internet Appendix. The aggregation table explicitly states its partial-equilibrium ceteris paribus assumption.",
190 + "units_stated_in_notes": false,
191 + "standard_errors_description": "unclear"
192 + },
193 + "figures": {
194 + "uncertainty_shown": "unclear"
195 + },
196 + "conclusion": {
197 + "bounded_claims": false,
198 + "policy_implications": false,
199 + "open_questions": false,
200 + "restates_contribution": true
201 + },
202 + "hedging": {
203 + "level": "moderate",
204 + "typical_devices": [
205 + "plausibly exogenous / quasi-exogenous",
206 + "suggests",
207 + "most consistent with",
208 + "may alleviate",
209 + "under certain conditions",
210 + "back-of-the-envelope / partial equilibrium caveat"
211 + ]
212 + },
213 + "claim_evidence": [
214 + {
215 + "claim": "Outside executive deaths and retirements predict own-firm executive turnover",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table III"
218 + },
219 + {
220 + "claim": "After turnover, firms favor lenders tied to new executives and get lower spreads and larger loans, with no pre-turnover effect",
221 + "evidence_type": "table",
222 + "evidence_ref": "Tables IV, V; Figure 2"
223 + },
224 + {
225 + "claim": "Personal relationship intensity has a larger, more precise effect on lender choice than institutional intensity",
226 + "evidence_type": "table",
227 + "evidence_ref": "Table VI"
228 + },
229 + {
230 + "claim": "Personal relationship effects on loan terms are stronger in NBER recessions and the 2008-09 crisis",
231 + "evidence_type": "table",
232 + "evidence_ref": "Table VII"
233 + },
234 + {
235 + "claim": "Firms with intense personal relationships invest more and cut fewer jobs in downturns",
236 + "evidence_type": "table",
237 + "evidence_ref": "Table VIII; Figure 3"
238 + },
239 + {
240 + "claim": "Aggregate crisis investment would be at least $32.6bn lower without intense personal relationships",
241 + "evidence_type": "table",
242 + "evidence_ref": "Table IX"
243 + }
244 + ],
245 + "structural_summary": "The introduction opens with Coasean relationship theory and the mixed evidence on firm-bank relationships, then proposes that relationships sit with executives rather than firms. It previews an instrument based on executive deaths and retirements at other firms and defends the exclusion restriction. Results follow with magnitudes, a within-firm design illustrated by a figure, and evidence on downturns. Contributions and positioning against the social-ties papers come next, then a roadmap. The body proceeds from measurement and sample, to turnover-based lender choice and loan terms, to downturn heterogeneity and real effects, then a back-of-the-envelope aggregation. It closes with polynomial, placebo and robustness tests and a short conclusion restating the findings with a dollar figure.",
246 + "writing_profile_keywords": [
247 + "first-person singular",
248 + "instrumental variables",
249 + "executive turnover shock",
250 + "economic magnitudes in dollars",
251 + "within-firm across-lender comparison",
252 + "financial crisis real effects",
253 + "theory-motivated puzzle"
254 + ],
255 + "confidence": 0.8,
256 + "notes": "How significance is reported (stars, t-stats, clustering) is not visible in the excerpts. Section V's opening paragraph refers to 'Sections IV and V', apparently a typo for Sections III and IV."
257 + },
258 + "deterministic_checks": {
259 + "roadmap_regex": true,
260 + "contribution_regex_in_intro": true,
261 + "magnitude_in_abstract": true,
262 + "abstract_n_sentences": 4,
263 + "intro_n_paragraphs": 9,
264 + "intro_n_words": 1408,
265 + "causal_words_in_abstract": 3,
266 + "hedge_words_in_abstract": 1,
267 + "stars_mentioned": false,
268 + "clustered_se_mentioned": false,
269 + "internet_appendix_mentioned": false,
270 + "section_roles_rule": {
271 + "0": "introduction",
272 + "1": "data",
273 + "2": "other",
274 + "3": "other",
275 + "4": "other",
276 + "5": "other",
277 + "6": "robustness",
278 + "7": "conclusion",
279 + "8": "appendix"
280 + }
281 + },
282 + "schema_valid": true,
283 + "code_version": "feb6c9e-dirty"
284 +}
\ No newline at end of file
added data/annotations/documents/jof_12590.json +285 −0
@@ -0,0 +1,285 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12590",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T22:54:38+00:00",
9 + "source_sha256": "39d3f8cbc07db81cec73da58f942316b5f734e7f2b2803471963dec58163e4dc",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23181,
12 + "usage": {
13 + "input": 8022,
14 + "output": 2281,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.078502,
19 + "request_id": "req_011CggpNrJwyPzEQPVYETQou",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "capital structure / dynamic trade-off theory",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model); comparative statics of optimal debt and leverage with respect to current and expected future profitability",
27 + "strategy": "analytical characterization of a stochastic dynamic model with zero-maturity debt, Poisson regime arrivals for EBIT, endogenous default cost and a borrowing constraint",
28 + "key_assumptions": [
29 + "EBIT i.i.d. across Poisson-arriving regimes (relaxed in Section VII)",
30 + "zero-maturity debt valued at par",
31 + "risk-neutral lenders with the same information as shareholders",
32 + "no renegotiation and no flotation costs",
33 + "nondecreasing density of EBIT, which ensures concavity"
34 + ],
35 + "threats_addressed": [
36 + "i.i.d. assumption relaxed by allowing persistence across regimes",
37 + "future profitability examined via rightward translation of the distribution"
38 + ],
39 + "causal_language_strength": 1,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": false,
45 + "sentence_index": -1
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 3
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 4
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "related_literature_positioning"
81 + },
82 + {
83 + "paragraph_index": 0,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "preview_of_approach"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "preview_of_approach"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "preview_of_approach"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "mechanism_or_theory_sketch"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "preview_of_results"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "mechanism_or_theory_sketch"
113 + },
114 + {
115 + "paragraph_index": 6,
116 + "move": "preview_of_results"
117 + },
118 + {
119 + "paragraph_index": 7,
120 + "move": "roadmap"
121 + }
122 + ],
123 + "contribution_paragraph_index": 0,
124 + "contribution_types": [
125 + "new_theory_or_model",
126 + "resolution_of_puzzle"
127 + ],
128 + "roadmap_present": true,
129 + "literature_positioning_location": "inside_introduction"
130 + },
131 + "section_roles": [
132 + {
133 + "idx": 0,
134 + "role": "introduction"
135 + },
136 + {
137 + "idx": 1,
138 + "role": "theory"
139 + },
140 + {
141 + "idx": 2,
142 + "role": "theory"
143 + },
144 + {
145 + "idx": 3,
146 + "role": "theory"
147 + },
148 + {
149 + "idx": 4,
150 + "role": "theory"
151 + },
152 + {
153 + "idx": 5,
154 + "role": "theory"
155 + },
156 + {
157 + "idx": 6,
158 + "role": "theory"
159 + },
160 + {
161 + "idx": 7,
162 + "role": "theory"
163 + },
164 + {
165 + "idx": 8,
166 + "role": "conclusion"
167 + },
168 + {
169 + "idx": 9,
170 + "role": "appendix"
171 + },
172 + {
173 + "idx": 10,
174 + "role": "appendix"
175 + }
176 + ],
177 + "results_presentation": {
178 + "economic_significance_discussed": false,
179 + "statistical_significance_language": "unclear",
180 + "magnitude_statements_in_intro": false
181 + },
182 + "robustness": {
183 + "location": "main_text_section",
184 + "types": [
185 + "other"
186 + ]
187 + },
188 + "tables": {
189 + "convention_notes": "A single qualitative summary table maps the sign of profitability effects on optimal debt across scenarios, noting whether the borrowing constraint binds. It contains no estimates.",
190 + "units_stated_in_notes": false,
191 + "standard_errors_description": "none"
192 + },
193 + "figures": {
194 + "uncertainty_shown": "no"
195 + },
196 + "conclusion": {
197 + "bounded_claims": true,
198 + "policy_implications": false,
199 + "open_questions": true,
200 + "restates_contribution": true
201 + },
202 + "hedging": {
203 + "level": "moderate",
204 + "typical_devices": [
205 + "conditioning results on whether the trade-off theory is operative",
206 + "acknowledging that results depend on the i.i.d. assumption",
207 + "qualifiers such as 'potentially' and 'at least in the example'",
208 + "framing results as an alternative explanation"
209 + ]
210 + },
211 + "claim_evidence": [
212 + {
213 + "claim": "When the borrowing constraint is slack, optimal debt equates the marginal interest tax shield with the marginal expected default cost.",
214 + "evidence_type": "equation_or_proof",
215 + "evidence_ref": "Section IV.A; Appendix A"
216 + },
217 + {
218 + "claim": "Under i.i.d. EBIT, optimal debt is invariant to current EBIT where the trade-off theory holds, so leverage falls as EBIT rises.",
219 + "evidence_type": "equation_or_proof",
220 + "evidence_ref": "Sections II-IV; Figure 1"
221 + },
222 + {
223 + "claim": "The tax rate determines which of the three scenarios prevails, via explicit threshold tax rates.",
224 + "evidence_type": "equation_or_proof",
225 + "evidence_ref": "Section V; Figure 2"
226 + },
227 + {
228 + "claim": "With a uniform distribution, a rightward shift of the profitability distribution reduces optimal debt in the trade-off region.",
229 + "evidence_type": "equation_or_proof",
230 + "evidence_ref": "Section VI; Figure 3; Table I"
231 + },
232 + {
233 + "claim": "The negative profitability-leverage relation matches empirical findings.",
234 + "evidence_type": "prior_literature",
235 + "evidence_ref": "Introduction footnotes 3-4"
236 + },
237 + {
238 + "claim": "With persistence across regimes in a uniform example, both debt and leverage decrease in EBIT.",
239 + "evidence_type": "equation_or_proof",
240 + "evidence_ref": "Section VII"
241 + }
242 + ],
243 + "structural_summary": "A single-author theory paper opens with a puzzle: the conventional reading of trade-off theory conflicts with the empirical negative relation between profitability and leverage. The introduction motivates the modeling choices (zero-maturity debt, Poisson EBIT regimes, endogenous borrowing limit), previews the analytical results, and gives a roadmap. The body builds the environment, derives the Bellman problem, partitions outcomes into three scenarios, and analyzes the interior scenario as the trade-off case. It then derives tax-rate thresholds, examines shifts in future profitability, and extends the model to persistent EBIT. The conclusion restates which conditions make the trade-off theory operative and lists extensions. Proofs are placed in appendices.",
244 + "writing_profile_keywords": [
245 + "analytical dynamic model",
246 + "trade-off theory",
247 + "puzzle resolution",
248 + "scenario-based characterization",
249 + "comparative statics",
250 + "proofs in appendix",
251 + "single author",
252 + "modeling-assumption justification"
253 + ],
254 + "confidence": 0.88,
255 + "notes": "Purely theoretical paper with no empirical estimation; empirical facts are cited only as motivation. The section on persistence across regimes is treated as a robustness-type extension within the main text."
256 + },
257 + "deterministic_checks": {
258 + "roadmap_regex": true,
259 + "contribution_regex_in_intro": false,
260 + "magnitude_in_abstract": false,
261 + "abstract_n_sentences": 5,
262 + "intro_n_paragraphs": 8,
263 + "intro_n_words": 1553,
264 + "causal_words_in_abstract": 0,
265 + "hedge_words_in_abstract": 1,
266 + "stars_mentioned": false,
267 + "clustered_se_mentioned": false,
268 + "internet_appendix_mentioned": false,
269 + "section_roles_rule": {
270 + "0": "introduction",
271 + "1": "other",
272 + "2": "other",
273 + "3": "other",
274 + "4": "other",
275 + "5": "other",
276 + "6": "other",
277 + "7": "other",
278 + "8": "conclusion",
279 + "9": "appendix",
280 + "10": "appendix"
281 + }
282 + },
283 + "schema_valid": true,
284 + "code_version": "feb6c9e-dirty"
285 +}
\ No newline at end of file
added data/annotations/documents/jof_12591.json +251 −0
@@ -0,0 +1,251 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12591",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:27+00:00",
9 + "source_sha256": "2743beca9bf99d3d89e8544bcb65a6927073ab432d7a11fef527df1bfea02d55",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20247,
12 + "usage": {
13 + "input": 7420,
14 + "output": 2013,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.070734,
19 + "request_id": "req_011CggppjiYreP4Y1q26AeVm",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Predictive association between ex-ante bank LMI and crisis outcomes (stock crash probability, borrowing from Fed facilities/TARP); aggregate liquidity mismatch over time",
27 + "strategy": "Construct a theory-based measure from balance sheets weighted by market-based liquidity factors; validate it through time-series aggregates, a stress test, and cross-sectional probability regressions on pre-crisis LMI with controls; compare against Basel LCR/NSFR and the Berger-Bouwman measure",
28 + "key_assumptions": [
29 + "Repo haircuts and the OIS-Tbill spread proxy asset and funding liquidity",
30 + "Calibrated liquidity weights (e.g., kappa) are appropriate",
31 + "Measuring LMI ex ante makes it predetermined relative to crisis outcomes"
32 + ],
33 + "threats_addressed": [
34 + "Fixed versus time-varying weights",
35 + "Comparison with alternative liquidity measures",
36 + "Bank characteristic controls"
37 + ],
38 + "causal_language_strength": 1,
39 + "uses_causal_claim": false
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 0
45 + },
46 + "gap": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 0
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 3
57 + },
58 + "magnitude": {
59 + "present": true,
60 + "sentence_index": 1
61 + },
62 + "contribution": {
63 + "present": true,
64 + "sentence_index": 4
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "literature_gap"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "related_literature_positioning"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 2,
95 + "move": "policy_or_broader_implication"
96 + },
97 + {
98 + "paragraph_index": 3,
99 + "move": "related_literature_positioning"
100 + },
101 + {
102 + "paragraph_index": 3,
103 + "move": "contribution_statement"
104 + },
105 + {
106 + "paragraph_index": 3,
107 + "move": "preview_of_results"
108 + },
109 + {
110 + "paragraph_index": 4,
111 + "move": "roadmap"
112 + }
113 + ],
114 + "contribution_paragraph_index": 3,
115 + "contribution_types": [
116 + "new_data_or_measure",
117 + "new_theory_or_model",
118 + "new_evidence"
119 + ],
120 + "roadmap_present": true,
121 + "literature_positioning_location": "inside_introduction"
122 + },
123 + "section_roles": [
124 + {
125 + "idx": 0,
126 + "role": "introduction"
127 + },
128 + {
129 + "idx": 1,
130 + "role": "theory"
131 + },
132 + {
133 + "idx": 2,
134 + "role": "empirical_design"
135 + },
136 + {
137 + "idx": 3,
138 + "role": "results"
139 + },
140 + {
141 + "idx": 4,
142 + "role": "results"
143 + },
144 + {
145 + "idx": 5,
146 + "role": "conclusion"
147 + }
148 + ],
149 + "results_presentation": {
150 + "economic_significance_discussed": true,
151 + "statistical_significance_language": "unclear",
152 + "magnitude_statements_in_intro": true
153 + },
154 + "robustness": {
155 + "location": "within_results",
156 + "types": [
157 + "alternative_measure",
158 + "sensitivity_to_parameters",
159 + "alternative_sample_period"
160 + ]
161 + },
162 + "tables": {
163 + "convention_notes": "Table notes state the sample, the regression equation in probability form, dependent-variable definitions, and the quarters covered. Stress-test table reports scenarios at 1σ, 2σ and 3σ shocks. Standard errors are not visible in the excerpts.",
164 + "units_stated_in_notes": true,
165 + "standard_errors_description": "unclear"
166 + },
167 + "figures": {
168 + "uncertainty_shown": "no"
169 + },
170 + "conclusion": {
171 + "bounded_claims": true,
172 + "policy_implications": true,
173 + "open_questions": true,
174 + "restates_contribution": true
175 + },
176 + "hedging": {
177 + "level": "moderate",
178 + "typical_devices": [
179 + "acknowledging the measure is not a finished product",
180 + "noting calibration choices drive results",
181 + "suggesting future data to pin down weights"
182 + ]
183 + },
184 + "claim_evidence": [
185 + {
186 + "claim": "Aggregate LMI falls from about +$4 trillion before the crisis to about -$6 trillion in 2008",
187 + "evidence_type": "figure",
188 + "evidence_ref": "Figure 4"
189 + },
190 + {
191 + "claim": "A stress test based on LMI signals banking fragility in early 2007",
192 + "evidence_type": "table",
193 + "evidence_ref": "Table III, Figure 9"
194 + },
195 + {
196 + "claim": "Banks with lower pre-crisis LMI were more likely to borrow from Fed facilities and TARP",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table VI"
199 + },
200 + {
201 + "claim": "Lower ex-ante LMI predicts higher stock crash probability during the crisis",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table VII"
204 + },
205 + {
206 + "claim": "With fixed liquidity weights, LMI varies little between normal and crisis periods",
207 + "evidence_type": "figure",
208 + "evidence_ref": "Figure 8"
209 + },
210 + {
211 + "claim": "Federal Reserve liquidity injections correlate with the increase in LMI in 2008",
212 + "evidence_type": "figure",
213 + "evidence_ref": "Figure 6"
214 + }
215 + ],
216 + "structural_summary": "The paper opens with liquidity crises and post-crisis regulation, then identifies the lack of an agreed liquidity measurement framework. It derives the LMI from a recursive theoretical framework and builds the empirical measure using haircut-based and spread-based liquidity weights. It then validates the measure on macro criteria (aggregate time series, decomposition, stress test) and on micro criteria (cross-sectional prediction of crash risk and government borrowing), benchmarking against the Basel ratios and Berger-Bouwman. The conclusion restates the measure's advantages and notes that the calibration is still open to refinement.",
217 + "writing_profile_keywords": [
218 + "measurement paper",
219 + "macroprudential",
220 + "stress test",
221 + "benchmark comparison",
222 + "theory-grounded index",
223 + "validation criteria"
224 + ],
225 + "confidence": 0.8,
226 + "notes": "Validation is predictive rather than causal. How statistical significance is reported is not visible in the excerpts."
227 + },
228 + "deterministic_checks": {
229 + "roadmap_regex": true,
230 + "contribution_regex_in_intro": true,
231 + "magnitude_in_abstract": false,
232 + "abstract_n_sentences": 7,
233 + "intro_n_paragraphs": 5,
234 + "intro_n_words": 884,
235 + "causal_words_in_abstract": 0,
236 + "hedge_words_in_abstract": 0,
237 + "stars_mentioned": false,
238 + "clustered_se_mentioned": false,
239 + "internet_appendix_mentioned": false,
240 + "section_roles_rule": {
241 + "0": "introduction",
242 + "1": "theory",
243 + "2": "empirical_design",
244 + "3": "other",
245 + "4": "other",
246 + "5": "conclusion"
247 + }
248 + },
249 + "schema_valid": true,
250 + "code_version": "feb6c9e-dirty"
251 +}
\ No newline at end of file
added data/annotations/documents/jof_12592.json +224 −0
@@ -0,0 +1,292 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12592",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:00:48+00:00",
9 + "source_sha256": "0e2fc7d71bc2b1c1a2fac09202838766bf389f07a9ff4be06abcd4af4ca67be6",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 25279,
12 + "usage": {
13 + "input": 8752,
14 + "output": 2321,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.082222,
19 + "request_id": "req_011CggpqwzFcFqqL9gY1AXYW",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Whether bank-specific CD funding dry-ups predict future bank performance conditional on public information, and whether funds shift toward high-quality banks under stress, as a test of adverse-selection versus informed-lender theories",
27 + "strategy": "Predictive panel regressions of future changes in ROA, impaired loans and size on dry-up dummies with current controls, plus event studies of stock returns, maturity shortening and rating downgrades, and probit models of reallocation",
28 + "key_assumptions": [
29 + "Future performance conditional on current public information proxies for quality unobserved by the market",
30 + "Impaired loans reflect loans extended before the dry-up and so are less prone to reverse causality",
31 + "CDs are unsecured and lenders are mostly money market funds, so asymmetric information concerns only issuer creditworthiness and liquidity hoarding by lenders is unlikely"
32 + ],
33 + "threats_addressed": [
34 + "reverse causality from dry-ups to performance via fire sales or forgone lending",
35 + "self-fulfilling runs",
36 + "demand-driven switching to central bank funding",
37 + "heavy reliance on CD funding driving the results"
38 + ],
39 + "causal_language_strength": 2,
40 + "uses_causal_claim": false
41 + },
42 + "abstract_moves": {
43 + "question": {
44 + "present": true,
45 + "sentence_index": 0
46 + },
47 + "gap": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 1
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 5
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "policy_or_broader_implication"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_approach"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "tension_or_puzzle"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "identification_defense"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "mechanism_or_theory_sketch"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "contribution_statement"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "contribution_statement"
113 + },
114 + {
115 + "paragraph_index": 6,
116 + "move": "related_literature_positioning"
117 + },
118 + {
119 + "paragraph_index": 7,
120 + "move": "related_literature_positioning"
121 + },
122 + {
123 + "paragraph_index": 8,
124 + "move": "roadmap"
125 + }
126 + ],
127 + "contribution_paragraph_index": 6,
128 + "contribution_types": [
129 + "new_evidence",
130 + "new_data_or_measure"
131 + ],
132 + "roadmap_present": true,
133 + "literature_positioning_location": "inside_introduction"
134 + },
135 + "section_roles": [
136 + {
137 + "idx": 0,
138 + "role": "introduction"
139 + },
140 + {
141 + "idx": 1,
142 + "role": "theory"
143 + },
144 + {
145 + "idx": 2,
146 + "role": "data"
147 + },
148 + {
149 + "idx": 3,
150 + "role": "results"
151 + },
152 + {
153 + "idx": 4,
154 + "role": "results"
155 + },
156 + {
157 + "idx": 5,
158 + "role": "results"
159 + },
160 + {
161 + "idx": 6,
162 + "role": "results"
163 + },
164 + {
165 + "idx": 7,
166 + "role": "conclusion"
167 + },
168 + {
169 + "idx": 8,
170 + "role": "appendix"
171 + }
172 + ],
173 + "results_presentation": {
174 + "economic_significance_discussed": true,
175 + "statistical_significance_language": "unclear",
176 + "magnitude_statements_in_intro": false
177 + },
178 + "robustness": {
179 + "location": "mixed",
180 + "types": [
181 + "alternative_measure",
182 + "subsample",
183 + "falsification"
184 + ]
185 + },
186 + "tables": {
187 + "convention_notes": "Notes open by naming the equation estimated and the dependent variable. Panels split partial and full dry-ups. Timing of observability (t-1 vs t) is spelled out, along with fixed effects and event windows. Descriptive tables report pooled means and quantiles, and an appendix table defines all variables with their data sources.",
188 + "units_stated_in_notes": true,
189 + "standard_errors_description": "unclear"
190 + },
191 + "figures": {
192 + "uncertainty_shown": "unclear"
193 + },
194 + "conclusion": {
195 + "bounded_claims": true,
196 + "policy_implications": true,
197 + "open_questions": true,
198 + "restates_contribution": true
199 + },
200 + "hedging": {
201 + "level": "moderate",
202 + "typical_devices": [
203 + "consistent with",
204 + "suggesting",
205 + "we fail to find evidence",
206 + "cannot draw any definite conclusion",
207 + "unlikely to explain"
208 + ]
209 + },
210 + "claim_evidence": [
211 + {
212 + "claim": "No market-wide freeze occurred in the European CD market from 2008 to 2014.",
213 + "evidence_type": "figure",
214 + "evidence_ref": "Figure 1; Section III.A"
215 + },
216 + {
217 + "claim": "Bank-specific dry-ups predict future declines in ROA, controlling for current performance.",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table V"
220 + },
221 + {
222 + "claim": "Dry-ups predict rising impaired loans, which helps address reverse causality.",
223 + "evidence_type": "table",
224 + "evidence_ref": "Table VI"

Diff truncated — file too large.

Menu

Theme
↑↓ navigate ↵ open esc close / or ⌘K anywhere