SPB Git
spboucher.ai

spb/wp45_uqo

Public

WP45: full annotation (276/276 valid), profiles, index, fixtures, tests (28), docs, lit review (58 verified refs); benchmark heldout_v1 launched

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Simon-Pierre Boucher committed (Nov 3, 2026) parent 7ef3b9a Browse files

227 changed files +19,201 −282

227 changed files
Unified Split
modified .gitignore +4 −0
@@ -41,3 +41,7 @@ data/derived/embeddings/*.npy
41 41 experiments/outputs/raw/
42 42 .env
43 43 *.duckdb
44 +
45 +# evaluation fixtures contain copyrighted excerpts
46 +experiments/fixtures/heldout_v1/*/
47 +experiments/outputs/*/outputs/
added README.md +66 −0
@@ -0,0 +1,66 @@
1 +<!-- Author: Simon-Pierre Boucher — simon-pierre.boucher@uqo.ca -->
2 +# 📝 JoF-250: Learning Scientific Writing Conventions from Pre-LLM Financial Research
3 +
4 +**UQO Working Paper No. 45** — *A measured-convention writing engine for Claude Code, built from 276 Journal of Finance articles (2017–2020), with a preregistered evaluation*
5 +
6 +[![Paper](https://img.shields.io/badge/paper-PDF-b31b1b?logo=latex&logoColor=white)](paper/main.pdf)
7 +[![Python](https://img.shields.io/badge/python-3.12-3776AB?logo=python&logoColor=white)](pyproject.toml)
8 +[![Data](https://img.shields.io/badge/data-276%20JoF%20articles%20·%20private-4051b5)](docs/RIGHTS_AND_ETHICS.md)
9 +[![Reproducible](https://img.shields.io/badge/reproducible-make%20pipeline-2e7d32)](docs/REPRODUCIBILITY.md)
10 +[![Institution](https://img.shields.io/badge/UQO-D%C3%A9pt.%20des%20sciences%20administratives-16365c)](https://uqo.ca)
11 +[![Author](https://img.shields.io/badge/author-Simon--Pierre%20Boucher-16365c)](mailto:simon-pierre.boucher@uqo.ca)
12 +
13 +> **TL;DR** — We parse 276 pre-ChatGPT *Journal of Finance* articles (vol. 72–75), annotate their rhetorical structure with a schema-validated LLM ontology, turn the train split into measured writing profiles (medians, IQRs, bootstrap CIs), index structural units for hybrid retrieval with provenance, and expose the result as a Claude Code writing/review/audit engine (`/jof-write`, `/jof-review`, `/jof-audit`, …). A preregistered held-out experiment compares five conditions (baseline → full engine) with a blind LLM judge and deterministic fidelity audits. See `docs/DELIVERY_REPORT.md` for measured results.
14 +
15 +## 📌 Status
16 +| Item | Value |
17 +|---|---|
18 +| Version | 1.0 (November 2026) |
19 +| SSRN | not yet submitted |
20 +| Journal | — |
21 +| Code & data | pipeline reproducible from `make` (see Quick start); PDFs not redistributed |
22 +
23 +## 🎯 What this paper does
24 +It asks whether *abstract, domain-specific writing conventions* extracted from a pre-LLM corpus improve LLM-assisted scholarly writing and review beyond strong prompting and conventional RAG, while preserving originality and factual fidelity. It delivers both the study and the tool (the JoF-250 Writing Engine), and the paper studies the tool.
25 +
26 +## 🏆 Headline results
27 +See `docs/DELIVERY_REPORT.md` and `experiments/analysis/results_heldout_v1.json` (generated from recorded runs only).
28 +
29 +## 🗂 Repository layout
30 +`CLAUDE.md` (mandate) · `config/` · `src/jof250/` · `prompts/` · `profiles/` · `.claude/{commands,agents,skills}` · `experiments/{preregistration.md,protocols,fixtures,outputs,analysis}` · `paper/` · `tests/` · `scripts/` · `docs/`.
31 +
32 +## 🚀 Quick start
33 +```bash
34 +uv venv --python 3.12 .venv && uv pip install --python .venv/bin/python -e ".[dev]"
35 +cp .env.example .env # API key file
36 +make audit parse profiles index test
37 +.venv/bin/jof250 write --paper /path/to/paper --section introduction --output /tmp/intro.tex
38 +cd paper && make # compiles main.pdf
39 +```
40 +
41 +## 📊 Data
42 +277 PDFs under `~/Desktop/Journal-of-Finance-2018-2020` (UQO library access, 2026-11-03); 276 research documents included; metadata verified against Crossref. Derived full texts are private (`docs/RIGHTS_AND_ETHICS.md`).
43 +
44 +## 🧪 Methodology
45 +Phases 1–9 of `CLAUDE.md`; preregistration in `experiments/preregistration.md`; limitations in `docs/LIMITATIONS.md`.
46 +
47 +## 📈 Figures & tables inventory
48 +`figures/` (regenerated by `make profiles` and `scripts/05_analyze_benchmark.py`), `paper/tables/` (generated LaTeX fragments).
49 +
50 +## ⚠️ Limitations
51 +`docs/LIMITATIONS.md`.
52 +
53 +## 📚 Citation
54 +```bibtex
55 +@techreport{boucher2026jof250,
56 + title = {JoF-250: Learning Scientific Writing Conventions from Pre-LLM Financial Research},
57 + author = {Boucher, Simon-Pierre}, institution = {Université du Québec en Outaouais, Département des sciences administratives},
58 + type = {UQO Working Paper}, number = {45}, year = {2026}, month = {11}
59 +}
60 +```
61 +
62 +## 👤 Author & contact
63 +Simon-Pierre Boucher — simon-pierre.boucher@uqo.ca
64 +
65 +## 📝 Notes
66 +Folder named `wp45-jof250` as required by the mandate; corpus scope amended to volumes 72–75 (A-002); global Claude Code install authorized by the author (A-003).
added data/annotations/_llm_cache/02ecec4fe3bdb05c21ba0c4e7462575ada873e71c06c8395e12708d5d1fc63c4.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"macro_finance (housing markets, fiscal policy evaluation)\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of the First-Time Homebuyer Credit on home sales, house prices, and individual transitions into homeownership; aggregate induced sales and stimulus effects\",\"strategy\":\"Cross-ZIP difference-in-differences using ex ante exposure (year-2000 share of first-time buyers) with CBSA fixed effects, combined with a sharp regression kink design at the income phase-out, using administrative tax data\",\"key_assumptions\":[\"No time-varying ZIP-specific shocks correlated with program exposure (parallel trends)\",\"Low-exposure ZIPs serve as a valid counterfactual\",\"Smoothness of household outcomes in income around the phase-out kink apart from the policy-induced slope change\"],\"threats_addressed\":[\"Time-varying ZIP-level shocks correlated with exposure\",\"Exposure to the subprime bust\",\"Contemporaneous policies\",\"Specification choices: weighting schemes and sample definitions\",\"ZIP-by-time trends, addressed through age-distribution shifts and a within-time starter-home placebo\"],\"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\":1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"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\":\"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\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"placebo\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes give the regression equation, CBSA fixed effects, and weighting by 2007 home sales. Some variables are normalized to standard-deviation units. Significance is marked with stars. Detailed variable construction is deferred to the Internet Appendix.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"likely\",\"appears to\",\"suggests\",\"back-of-the-envelope\",\"explicit statement of research-design limits (cannot determine permanence)\"]},\"claim_evidence\":[{\"claim\":\"The FTHC raised home sales during the policy period by about 490,000 units nationally\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table VII, Figure 5\"},{\"claim\":\"Induced sales did not reverse immediately after expiration; demand was pulled forward from future years\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5, Figure 6\"},{\"claim\":\"Receiving the credit raised the first-time homebuying transition rate by roughly 53%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Figure 8\"},{\"claim\":\"Program exposure raised market-level house prices\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 9\"},{\"claim\":\"Indirect stabilization effects via housing wealth likely exceeded the direct stimulus effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Induced buyers did not default more than other cohorts\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 10\"}],\"structural_summary\":\"The introduction moves from crisis motivation to a gap in existing policy targeting, then introduces the evaluated program, data and designs. It previews five numbered findings with magnitudes, defends identification point by point, and positions against close papers before a roadmap. A policy background section and a data section precede the empirical design. Four results sections follow, covering sales, homeownership via the regression kink design, prices and reallocation, and aggregate effects. The conclusion restates findings with explicit design caveats and policy framing.\",\"writing_profile_keywords\":[\"policy evaluation\",\"quasi-experimental\",\"difference-in-differences\",\"regression kink\",\"administrative tax data\",\"numbered findings\",\"back-of-the-envelope aggregation\",\"graphical evidence\"],\"confidence\":0.85,\"notes\":\"Section VI is mislabelled as 'Section B' in the roadmap. Robustness and placebo tests appear as a subsection within the results (IV.C) and in the Internet Appendix. The policy background section is classified as other rather than literature.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "macro_finance (housing markets, fiscal policy evaluation)", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of the First-Time Homebuyer Credit on home sales, house prices, and individual transitions into homeownership; aggregate induced sales and stimulus effects", "strategy": "Cross-ZIP difference-in-differences using ex ante exposure (year-2000 share of first-time buyers) with CBSA fixed effects, combined with a sharp regression kink design at the income phase-out, using administrative tax data", "key_assumptions": ["No time-varying ZIP-specific shocks correlated with program exposure (parallel trends)", "Low-exposure ZIPs serve as a valid counterfactual", "Smoothness of household outcomes in income around the phase-out kink apart from the policy-induced slope change"], "threats_addressed": ["Time-varying ZIP-level shocks correlated with exposure", "Exposure to the subprime bust", "Contemporaneous policies", "Specification choices: weighting schemes and sample definitions", "ZIP-by-time trends, addressed through age-distribution shifts and a within-time starter-home placebo"], "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": 1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 6, "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": "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": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "placebo", "subsample", "other"]}, "tables": {"convention_notes": "Notes give the regression equation, CBSA fixed effects, and weighting by 2007 home sales. Some variables are normalized to standard-deviation units. Significance is marked with stars. Detailed variable construction is deferred to the Internet Appendix.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["likely", "appears to", "suggests", "back-of-the-envelope", "explicit statement of research-design limits (cannot determine permanence)"]}, "claim_evidence": [{"claim": "The FTHC raised home sales during the policy period by about 490,000 units nationally", "evidence_type": "table", "evidence_ref": "Table III, Table VII, Figure 5"}, {"claim": "Induced sales did not reverse immediately after expiration; demand was pulled forward from future years", "evidence_type": "figure", "evidence_ref": "Figure 5, Figure 6"}, {"claim": "Receiving the credit raised the first-time homebuying transition rate by roughly 53%", "evidence_type": "table", "evidence_ref": "Table V, Figure 8"}, {"claim": "Program exposure raised market-level house prices", "evidence_type": "table", "evidence_ref": "Table VI, Figure 9"}, {"claim": "Indirect stabilization effects via housing wealth likely exceeded the direct stimulus effects", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Induced buyers did not default more than other cohorts", "evidence_type": "figure", "evidence_ref": "Figure 10"}], "structural_summary": "The introduction moves from crisis motivation to a gap in existing policy targeting, then introduces the evaluated program, data and designs. It previews five numbered findings with magnitudes, defends identification point by point, and positions against close papers before a roadmap. A policy background section and a data section precede the empirical design. Four results sections follow, covering sales, homeownership via the regression kink design, prices and reallocation, and aggregate effects. The conclusion restates findings with explicit design caveats and policy framing.", "writing_profile_keywords": ["policy evaluation", "quasi-experimental", "difference-in-differences", "regression kink", "administrative tax data", "numbered findings", "back-of-the-envelope aggregation", "graphical evidence"], "confidence": 0.85, "notes": "Section VI is mislabelled as 'Section B' in the roadmap. Robustness and placebo tests appear as a subsection within the results (IV.C) and in the Internet Appendix. The policy background section is classified as other rather than literature."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.45, "usage": {"input": 8168, "output": 2094, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075346, "request_id": "req_011Cggqd75DEFJ99j2rFhYPr", "cached": false, "ts": "2026-11-03T23:10:57+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/05a1bd71a760dcd19473dddacb6fd47a8073eae0e374a5526ff6409526922f8c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"executive compensation / dynamic contracting\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; continuous-time principal-agent model solved via stochastic maximum principle and HJB, supplemented by numerical solutions\",\"key_assumptions\":[\"CARA risk-averse CEO with private savings\",\"hidden effort and manipulation with persistent effect on performance\",\"finite tenure T with postretirement window tau\"],\"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\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"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\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"theory\"},{\"idx\":7,\"role\":\"theory\"},{\"idx\":8,\"role\":\"other\"},{\"idx\":9,\"role\":\"conclusion\"},{\"idx\":10,\"role\":\"appendix\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"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. Figures are numerical solutions, and each caption lists the parameter values used.\",\"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\":[\"'may' for nonmonotonic effort\",\"'generally' qualifying results\",\"'might not' regarding firms' choices\",\"noting limits of postretirement compensation\"]},\"claim_evidence\":[{\"claim\":\"Optimal contracts defer compensation, and vesting accelerates near the end of tenure.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections V-VI, Appendix C\"},{\"claim\":\"Manipulation escalates over the CEO's tenure, producing an endogenous horizon problem.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Without manipulation, vesting is deterministic; with manipulation, vesting is performance-based.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections IV, VI\"},{\"claim\":\"Short-termism is prevalent: most surveyed CEOs would sacrifice long-term value to meet expectations.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Graham, Harvey, and Rajgopal (2005)\"},{\"claim\":\"Under deterministic contracts, long-term incentives can be reduced only by distorting effort.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section VII, Figure 4\"},{\"claim\":\"Incentive compatibility conditions are necessary and sufficient.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 1, Appendices A-B\"}],\"structural_summary\":\"The introduction motivates the paper with survey and empirical evidence of CEO short-termism. It then identifies a gap: prior models either take incentives as given or study contracts that rule out manipulation entirely. The paper builds a continuous-time dynamic agency model with hidden effort and manipulation. The analysis proceeds in steps: the agent's incentive-compatibility conditions, the principal's control problem, an infinite-tenure benchmark, finite tenure solved by backward induction, the general optimal contract, and a deterministic-contract subclass. Numerical examples illustrate the results and link them to empirical implications. The conclusion restates the findings, and proofs are placed in the appendices.\",\"writing_profile_keywords\":[\"dynamic contracting\",\"continuous time\",\"CEO horizon\",\"short-termism\",\"vesting\",\"numerical illustration\",\"proofs in appendix\"],\"confidence\":0.85,\"notes\":\"Theory paper. Section VIII presents numerical comparative statics and relates them to existing evidence; it contains no estimation of its own. No tables are present.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "executive compensation / dynamic contracting", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; continuous-time principal-agent model solved via stochastic maximum principle and HJB, supplemented by numerical solutions", "key_assumptions": ["CARA risk-averse CEO with private savings", "hidden effort and manipulation with persistent effect on performance", "finite tenure T with postretirement window tau"], "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": 5}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 3, "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": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "theory"}, {"idx": 7, "role": "theory"}, {"idx": 8, "role": "other"}, {"idx": 9, "role": "conclusion"}, {"idx": 10, "role": "appendix"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "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. Figures are numerical solutions, and each caption lists the parameter values used.", "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": ["'may' for nonmonotonic effort", "'generally' qualifying results", "'might not' regarding firms' choices", "noting limits of postretirement compensation"]}, "claim_evidence": [{"claim": "Optimal contracts defer compensation, and vesting accelerates near the end of tenure.", "evidence_type": "equation_or_proof", "evidence_ref": "Sections V-VI, Appendix C"}, {"claim": "Manipulation escalates over the CEO's tenure, producing an endogenous horizon problem.", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Without manipulation, vesting is deterministic; with manipulation, vesting is performance-based.", "evidence_type": "equation_or_proof", "evidence_ref": "Sections IV, VI"}, {"claim": "Short-termism is prevalent: most surveyed CEOs would sacrifice long-term value to meet expectations.", "evidence_type": "prior_literature", "evidence_ref": "Graham, Harvey, and Rajgopal (2005)"}, {"claim": "Under deterministic contracts, long-term incentives can be reduced only by distorting effort.", "evidence_type": "figure", "evidence_ref": "Section VII, Figure 4"}, {"claim": "Incentive compatibility conditions are necessary and sufficient.", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 1, Appendices A-B"}], "structural_summary": "The introduction motivates the paper with survey and empirical evidence of CEO short-termism. It then identifies a gap: prior models either take incentives as given or study contracts that rule out manipulation entirely. The paper builds a continuous-time dynamic agency model with hidden effort and manipulation. The analysis proceeds in steps: the agent's incentive-compatibility conditions, the principal's control problem, an infinite-tenure benchmark, finite tenure solved by backward induction, the general optimal contract, and a deterministic-contract subclass. Numerical examples illustrate the results and link them to empirical implications. The conclusion restates the findings, and proofs are placed in the appendices.", "writing_profile_keywords": ["dynamic contracting", "continuous time", "CEO horizon", "short-termism", "vesting", "numerical illustration", "proofs in appendix"], "confidence": 0.85, "notes": "Theory paper. Section VIII presents numerical comparative statics and relates them to existing evidence; it contains no estimation of its own. No tables are present."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.34, "usage": {"input": 8199, "output": 1961, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07281, "request_id": "req_011CggqSZshvXubj6wdBKhxF", "cached": false, "ts": "2026-11-03T23:08:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/078ed694a74223e3094ba6a11b5417aaa554f1d52bf8a873af15d4fb42771785.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Latent continuous and jump factor structure (number of factors, loadings, explained variation) of individual stock returns, and intraday vs overnight risk premia of these factors\",\"strategy\":\"PCA on locally estimated continuous and jump covariance matrices from five-minute high-frequency returns, with truncation to separate jumps; comparison with observable proxy factors via generalized correlations\",\"key_assumptions\":[\"approximate local factor model with locally constant loadings\",\"truncation threshold correctly separates continuous and jump increments\",\"five-minute sampling limits microstructure noise and asynchronicity\"],\"threats_addressed\":[\"microstructure noise and asynchronous trading\",\"factor misspecification from prespecified factors\",\"dependence on choice of characteristics\",\"panel composition (balanced vs unbalanced)\"],\"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\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"literature_gap\"},{\"paragraph_index\":3,\"move\":\"literature_gap\"},{\"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_evidence\",\"new_method\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"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\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Notes open with 'This table presents...' and report year-by-year descriptive quantities (jump fractions, generalized correlations, Sharpe ratios). Sample sizes (N) and jump thresholds are given in the notes. Main comparison metric is generalized correlation between factor spaces. No regression coefficient tables with standard errors are visible.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"approximately\",\"can be well approximated\",\"subtle\",\"roughly\"]},\"claim_evidence\":[{\"claim\":\"Four stable continuous systematic factors and one stable jump factor explain individual stock comovement\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-2; Table II\"},{\"claim\":\"Continuous PCA factors are approximated by market, oil, finance and electricity portfolios\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figures 3-4\"},{\"claim\":\"Factor loadings and explained variation vary over time while the loading span is stable\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 6-10; Table IV\"},{\"claim\":\"Fama-French-Carhart factors have time-varying span and explain less variation\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 9, 11; Table III\"},{\"claim\":\"Continuous factors earn an intraday risk premium that reverses overnight\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figures 12-15\"}],\"structural_summary\":\"The introduction poses three questions about systematic risk, then outlines a three-part approach: latent factors, high-frequency time variation, and a continuous/jump split. It positions this against constant-factor and characteristic-based studies and closes with a roadmap. Methodology and estimation sections set up a local factor model and PCA on high-frequency covariance matrices. Results cover the data, number of factors, balanced-panel checks, factor interpretation, time variation, and intraday versus overnight pricing. The conclusion enumerates the contributions.\",\"writing_profile_keywords\":[\"high-frequency econometrics\",\"latent factors\",\"PCA\",\"jump vs continuous\",\"time-varying loadings\",\"intraday vs overnight\",\"first-person singular\",\"figure-heavy\"],\"confidence\":0.8,\"notes\":\"The primary method is statistical factor analysis (PCA on high-frequency data), which has no exact schema category, so it is coded as 'other'. Section idx 4 appears to be a mis-parsed table heading that continues the results section. The estimation theory comes from the author's companion work (Pelger 2019).\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "other", "identification": {"estimand": "Latent continuous and jump factor structure (number of factors, loadings, explained variation) of individual stock returns, and intraday vs overnight risk premia of these factors", "strategy": "PCA on locally estimated continuous and jump covariance matrices from five-minute high-frequency returns, with truncation to separate jumps; comparison with observable proxy factors via generalized correlations", "key_assumptions": ["approximate local factor model with locally constant loadings", "truncation threshold correctly separates continuous and jump increments", "five-minute sampling limits microstructure noise and asynchronicity"], "threats_addressed": ["microstructure noise and asynchronous trading", "factor misspecification from prespecified factors", "dependence on choice of characteristics", "panel composition (balanced vs unbalanced)"], "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": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "other"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "literature_gap"}, {"paragraph_index": 3, "move": "literature_gap"}, {"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_evidence", "new_method", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "empirical_design"}, {"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": false}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Notes open with 'This table presents...' and report year-by-year descriptive quantities (jump fractions, generalized correlations, Sharpe ratios). Sample sizes (N) and jump thresholds are given in the notes. Main comparison metric is generalized correlation between factor spaces. No regression coefficient tables with standard errors are visible.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["approximately", "can be well approximated", "subtle", "roughly"]}, "claim_evidence": [{"claim": "Four stable continuous systematic factors and one stable jump factor explain individual stock comovement", "evidence_type": "figure", "evidence_ref": "Figures 1-2; Table II"}, {"claim": "Continuous PCA factors are approximated by market, oil, finance and electricity portfolios", "evidence_type": "table", "evidence_ref": "Table III; Figures 3-4"}, {"claim": "Factor loadings and explained variation vary over time while the loading span is stable", "evidence_type": "figure", "evidence_ref": "Figures 6-10; Table IV"}, {"claim": "Fama-French-Carhart factors have time-varying span and explain less variation", "evidence_type": "figure", "evidence_ref": "Figures 9, 11; Table III"}, {"claim": "Continuous factors earn an intraday risk premium that reverses overnight", "evidence_type": "table", "evidence_ref": "Table V; Figures 12-15"}], "structural_summary": "The introduction poses three questions about systematic risk, then outlines a three-part approach: latent factors, high-frequency time variation, and a continuous/jump split. It positions this against constant-factor and characteristic-based studies and closes with a roadmap. Methodology and estimation sections set up a local factor model and PCA on high-frequency covariance matrices. Results cover the data, number of factors, balanced-panel checks, factor interpretation, time variation, and intraday versus overnight pricing. The conclusion enumerates the contributions.", "writing_profile_keywords": ["high-frequency econometrics", "latent factors", "PCA", "jump vs continuous", "time-varying loadings", "intraday vs overnight", "first-person singular", "figure-heavy"], "confidence": 0.8, "notes": "The primary method is statistical factor analysis (PCA on high-frequency data), which has no exact schema category, so it is coded as 'other'. Section idx 4 appears to be a mis-parsed table heading that continues the results section. The estimation theory comes from the author's companion work (Pelger 2019)."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.44, "usage": {"input": 6825, "output": 1911, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066314, "request_id": "req_011CggqtyUFnrkgoaGZqPnyW", "cached": false, "ts": "2026-11-03T23:14:31+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/07fb5481ac8e3909608462ba9c7774a92d927980ab5b24118f903f0e696b77f3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Degree of overreaction in analysts' long-term growth expectations (representativeness parameter theta) and its implications for return predictability across LTG-sorted portfolios\",\"strategy\":\"Portfolio sorts on analyst LTG; forecast-error regressions of the Coibion-Gorodnichenko type; estimation of a diagnostic Kalman filter model by simulated method of moments; cross-sectional tests of kernel-of-truth predictions\",\"key_assumptions\":[\"Analyst LTG forecasts proxy for market beliefs\",\"Earnings follow the assumed latent-fundamentals process\",\"Diagnostic expectations are the correct belief-formation model\"],\"threats_addressed\":[\"Bayesian learning as an alternative explanation\",\"Adaptive/mechanical extrapolation as an alternative explanation\"],\"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\":2}},\"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\":1,\"move\":\"roadmap\"},{\"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\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":1,\"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\":\"theory\"},{\"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\":\"Table notes describe the decile portfolio formation on LTG each December from 1981 to 2015, equal weighting, and the variable definitions. Significance is marked with a letter superscript (e.g., 'a' for 1%). Year fixed effects are included but not shown. Model parameters are defined at the quarterly scale.\",\"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\":[\"'does a reasonable job' fit language\",\"'this is just a start' framing\",\"'in line with' phrasing for predictions\"]},\"claim_evidence\":[{\"claim\":\"With 20 extra years of data, low-LTG stocks still outperform high-LTG stocks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"High-LTG firms show fast past EPS growth that slows afterward\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Analyst forecast errors are predictable, consistent with overreaction\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The SMM-estimated diagnostic model reproduces key data patterns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 6\"},{\"claim\":\"The LLTG-minus-HLTG return spread widens for firms with more volatile or persistent fundamentals\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"High-LTG group has a fat right tail of top performers, and analysts overstate its frequency\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 7-8\"},{\"claim\":\"Expectations revert even without news, reflecting true mean reversion in earnings growth\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"}],\"structural_summary\":\"Opens by revisiting a known return anomaly and frames it as a puzzle. Lists three innovations: a portable psychological model, a joint account of fundamentals, expectations and returns, and new predictions that set the model apart from adaptive expectations. Data and stylized facts come first. Next, a diagnostic Kalman filter model is developed and mapped to those facts. Overreaction is then quantified with forecast-error regressions and SMM estimation. Kernel-of-truth predictions are tested cross-sectionally. The conclusion restates portability and points to extensions.\",\"writing_profile_keywords\":[\"behavioral asset pricing\",\"representativeness\",\"diagnostic expectations\",\"portfolio sorts\",\"kernel of truth\",\"SMM estimation\",\"analyst forecasts\"],\"confidence\":0.8,\"notes\":\"Primary method is ambiguous: the paper mixes portfolio-sort evidence with SMM estimation. Bootstrapped means are mentioned in the figure captions, but whether confidence bands are plotted is not visible.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "structural_estimation", "identification": {"estimand": "Degree of overreaction in analysts' long-term growth expectations (representativeness parameter theta) and its implications for return predictability across LTG-sorted portfolios", "strategy": "Portfolio sorts on analyst LTG; forecast-error regressions of the Coibion-Gorodnichenko type; estimation of a diagnostic Kalman filter model by simulated method of moments; cross-sectional tests of kernel-of-truth predictions", "key_assumptions": ["Analyst LTG forecasts proxy for market beliefs", "Earnings follow the assumed latent-fundamentals process", "Diagnostic expectations are the correct belief-formation model"], "threats_addressed": ["Bayesian learning as an alternative explanation", "Adaptive/mechanical extrapolation as an alternative explanation"], "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": 2}}, "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": 1, "move": "roadmap"}, {"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": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}], "contribution_paragraph_index": 1, "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": "theory"}, {"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": "Table notes describe the decile portfolio formation on LTG each December from 1981 to 2015, equal weighting, and the variable definitions. Significance is marked with a letter superscript (e.g., 'a' for 1%). Year fixed effects are included but not shown. Model parameters are defined at the quarterly scale.", "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": ["'does a reasonable job' fit language", "'this is just a start' framing", "'in line with' phrasing for predictions"]}, "claim_evidence": [{"claim": "With 20 extra years of data, low-LTG stocks still outperform high-LTG stocks", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "High-LTG firms show fast past EPS growth that slows afterward", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Analyst forecast errors are predictable, consistent with overreaction", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The SMM-estimated diagnostic model reproduces key data patterns", "evidence_type": "table", "evidence_ref": "Table III; Figure 6"}, {"claim": "The LLTG-minus-HLTG return spread widens for firms with more volatile or persistent fundamentals", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "High-LTG group has a fat right tail of top performers, and analysts overstate its frequency", "evidence_type": "figure", "evidence_ref": "Figures 7-8"}, {"claim": "Expectations revert even without news, reflecting true mean reversion in earnings growth", "evidence_type": "figure", "evidence_ref": "Figure 9"}], "structural_summary": "Opens by revisiting a known return anomaly and frames it as a puzzle. Lists three innovations: a portable psychological model, a joint account of fundamentals, expectations and returns, and new predictions that set the model apart from adaptive expectations. Data and stylized facts come first. Next, a diagnostic Kalman filter model is developed and mapped to those facts. Overreaction is then quantified with forecast-error regressions and SMM estimation. Kernel-of-truth predictions are tested cross-sectionally. The conclusion restates portability and points to extensions.", "writing_profile_keywords": ["behavioral asset pricing", "representativeness", "diagnostic expectations", "portfolio sorts", "kernel of truth", "SMM estimation", "analyst forecasts"], "confidence": 0.8, "notes": "Primary method is ambiguous: the paper mixes portfolio-sort evidence with SMM estimation. Bootstrapped means are mentioned in the figure captions, but whether confidence bands are plotted is not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.98, "usage": {"input": 6904, "output": 2073, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06987, "request_id": "req_011CggqYdTS7WFZxfDL1WwFb", "cached": false, "ts": "2026-11-03T23:09:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/0984ff4d68525b141809efdcbc389662955e8d6cbb598831b0d5e231de49487d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"financial_econometrics\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"not applicable (Bayesian model comparison: posterior model probabilities / marginal likelihoods for factor models)\",\"strategy\":\"Analytical derivation that priors on nuisance parameters must satisfy an across-model change-of-variable property, followed by Monte Carlo simulations measuring how often each method selects the true DGP\",\"key_assumptions\":[\"Nuisance parameters across models are connected by invertible mappings\",\"Simulated DGPs calibrated to real factor data are representative\"],\"threats_addressed\":[\"Results being specific to a single DGP (33 DGPs used)\",\"Dependence on sample size (multiple T values)\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":true,\"sentence_index\":3},\"results\":{\"present\":true,\"sentence_index\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"replication_or_correction\",\"new_method\"],\"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\":\"theory\"},{\"idx\":6,\"role\":\"theory\"},{\"idx\":7,\"role\":\"theory\"},{\"idx\":8,\"role\":\"theory\"},{\"idx\":9,\"role\":\"theory\"},{\"idx\":10,\"role\":\"theory\"},{\"idx\":11,\"role\":\"results\"},{\"idx\":12,\"role\":\"results\"},{\"idx\":13,\"role\":\"conclusion\"},{\"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\"},{\"idx\":21,\"role\":\"appendix\"},{\"idx\":22,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Simulation tables, each row a true DGP labelled by its risk factors, columns are sample sizes T, and entries are the percentage of replications in which the true model is selected. BS and CZZ results are reported in parallel tables for the 8- and 12-factor experiments.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"categorical verdicts such as 'unsound' and 'invalidates'\",\"self-description of the method as a notable advance\"]},\"claim_evidence\":[{\"claim\":\"The Jeffreys priors used in the BS method violate a required across-model change-of-variable property, so the resulting marginal likelihoods depend on arbitrary constants\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Propositions 2-3, Sections II-IV, Appendix B\"},{\"claim\":\"The BS permutation method from their online appendix is also unsound\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV.B\"},{\"claim\":\"In simulations, the BS method rarely selects the true DGP, even with very large samples\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-II\"},{\"claim\":\"The CZZ priors yield valid marginal likelihoods that select the true model far more often, even at small T\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-IV\"},{\"claim\":\"A model-specific prior on alpha gives performance gains of up to 20% in small samples (results not reported)\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section III.C\"}],\"structural_summary\":\"A critique of a prior JF paper. The introduction immediately states that the target method is flawed and previews the simulation design. The paper then reconstructs the original framework and proves that improper priors must satisfy a change-of-variable condition. It derives a corrected class of priors with closed-form marginal likelihoods and adds further arguments against the original method, including its permutation variant. Monte Carlo experiments with 8 and 12 factors compare the selection accuracy of both methods. A brief conclusion restates the result, and appendices cover DGP selection and the Jacobian proof.\",\"writing_profile_keywords\":[\"critique\",\"Bayesian model comparison\",\"improper priors\",\"propositions\",\"Monte Carlo\",\"factor models\",\"assertive tone\"],\"confidence\":0.8,\"notes\":\"Article type is a correction/comment. Many section headings are artifacts of equation fragments and are assigned the theory role. There is no separate literature section and no visible roadmap. The genre could arguably be methodological_econometric.\"}", "json": {"genre": "replication_or_correction", "subfield": "financial_econometrics", "subfield_secondary": "asset_pricing", "primary_method": "calibration_simulation", "identification": {"estimand": "not applicable (Bayesian model comparison: posterior model probabilities / marginal likelihoods for factor models)", "strategy": "Analytical derivation that priors on nuisance parameters must satisfy an across-model change-of-variable property, followed by Monte Carlo simulations measuring how often each method selects the true DGP", "key_assumptions": ["Nuisance parameters across models are connected by invertible mappings", "Simulated DGPs calibrated to real factor data are representative"], "threats_addressed": ["Results being specific to a single DGP (33 DGPs used)", "Dependence on sample size (multiple T values)"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": true, "sentence_index": 3}, "results": {"present": true, "sentence_index": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_approach"}], "contribution_paragraph_index": 0, "contribution_types": ["replication_or_correction", "new_method"], "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": "theory"}, {"idx": 6, "role": "theory"}, {"idx": 7, "role": "theory"}, {"idx": 8, "role": "theory"}, {"idx": 9, "role": "theory"}, {"idx": 10, "role": "theory"}, {"idx": 11, "role": "results"}, {"idx": 12, "role": "results"}, {"idx": 13, "role": "conclusion"}, {"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"}, {"idx": 21, "role": "appendix"}, {"idx": 22, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "Simulation tables, each row a true DGP labelled by its risk factors, columns are sample sizes T, and entries are the percentage of replications in which the true model is selected. BS and CZZ results are reported in parallel tables for the 8- and 12-factor experiments.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["categorical verdicts such as 'unsound' and 'invalidates'", "self-description of the method as a notable advance"]}, "claim_evidence": [{"claim": "The Jeffreys priors used in the BS method violate a required across-model change-of-variable property, so the resulting marginal likelihoods depend on arbitrary constants", "evidence_type": "equation_or_proof", "evidence_ref": "Propositions 2-3, Sections II-IV, Appendix B"}, {"claim": "The BS permutation method from their online appendix is also unsound", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV.B"}, {"claim": "In simulations, the BS method rarely selects the true DGP, even with very large samples", "evidence_type": "table", "evidence_ref": "Tables I-II"}, {"claim": "The CZZ priors yield valid marginal likelihoods that select the true model far more often, even at small T", "evidence_type": "table", "evidence_ref": "Tables III-IV"}, {"claim": "A model-specific prior on alpha gives performance gains of up to 20% in small samples (results not reported)", "evidence_type": "none_stated", "evidence_ref": "Section III.C"}], "structural_summary": "A critique of a prior JF paper. The introduction immediately states that the target method is flawed and previews the simulation design. The paper then reconstructs the original framework and proves that improper priors must satisfy a change-of-variable condition. It derives a corrected class of priors with closed-form marginal likelihoods and adds further arguments against the original method, including its permutation variant. Monte Carlo experiments with 8 and 12 factors compare the selection accuracy of both methods. A brief conclusion restates the result, and appendices cover DGP selection and the Jacobian proof.", "writing_profile_keywords": ["critique", "Bayesian model comparison", "improper priors", "propositions", "Monte Carlo", "factor models", "assertive tone"], "confidence": 0.8, "notes": "Article type is a correction/comment. Many section headings are artifacts of equation fragments and are assigned the theory role. There is no separate literature section and no visible roadmap. The genre could arguably be methodological_econometric."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.65, "usage": {"input": 7108, "output": 1959, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068406, "request_id": "req_011CggqgTbaG6ecUo7n2ka3q", "cached": false, "ts": "2026-11-03T23:11:41+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/0b589c503b552bf39df5549af335d894130707cd4c67422ba146a0b7b163cb62.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"market design / regulation of high-frequency trading\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model; comparative statics of spread and research intensity with respect to HFT speed and trading mechanism)\",\"strategy\":\"not applicable; equilibrium analysis and mechanism-design characterization of the feasible frontier\",\"key_assumptions\":[\"random latencies prevent synchronized order processing across exchanges\",\"informed trading rents drive endogenous research intensity\",\"multiple fragmented exchanges with LOB trading in baseline\"],\"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\":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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":2,\"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\":\"theory\"},{\"idx\":6,\"role\":\"conclusion\"}],\"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 conveyed through propositions/corollaries and schematic timeline 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\":\"moderate\",\"typical_devices\":[\"'in the model' qualifiers\",\"'generally' does not achieve the frontier\",\"'potentially' for unmodeled distortions\",\"recommendation contingent on regulator's weights\"]},\"claim_evidence\":[{\"claim\":\"Faster HFT speeds reduce information production but narrow bid-ask spreads under LOB trading\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C, Corollary 1\"},{\"claim\":\"LOB outcomes generally lie inside the liquidity–information frontier due to aggressive-side order anticipation\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV.C\"},{\"claim\":\"Noncancellation delays and frequent batch auctions implement outcomes on the frontier\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section V.C\"},{\"claim\":\"Model predictions are consistent with many empirical HFT studies\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction paragraph 1\"},{\"claim\":\"Exchange latencies have fallen sharply and trading is fragmented across many venues\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Introduction paragraph 0, footnotes 1-2\"}],\"structural_summary\":\"Opens with stylized facts on falling latencies and fragmentation, defines two forms of order anticipation, then previews comparative statics: faster speed lowers research but narrows spreads. Uses mechanism design to characterize the feasible frontier, shows LOB is inside it, and shows two alternative mechanisms reach it. Positions against Budish-Cramton-Shim, with a separate literature section, baseline model, equilibrium, frontier and alternative-mechanism sections, ending in a policy-oriented conclusion.\",\"writing_profile_keywords\":[\"theory model\",\"market design\",\"mechanism design\",\"order anticipation\",\"liquidity–information tradeoff\",\"policy evaluation\",\"stylized facts motivation\"],\"confidence\":0.85,\"notes\":\"Theory paper with no tables or empirical estimation. Section V's automatic 'robustness' role is reassigned to theory. Section IV.D is labeled Robustness, which motivates the robustness location field. Causal language here refers to model comparative statics.\"}", "json": {"genre": "theoretical", "subfield": "market_microstructure", "subfield_secondary": "market design / regulation of high-frequency trading", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model; comparative statics of spread and research intensity with respect to HFT speed and trading mechanism)", "strategy": "not applicable; equilibrium analysis and mechanism-design characterization of the feasible frontier", "key_assumptions": ["random latencies prevent synchronized order processing across exchanges", "informed trading rents drive endogenous research intensity", "multiple fragmented exchanges with LOB trading in baseline"], "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": 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": "mechanism_or_theory_sketch"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 2, "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": "theory"}, {"idx": 6, "role": "conclusion"}], "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 conveyed through propositions/corollaries and schematic timeline 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": "moderate", "typical_devices": ["'in the model' qualifiers", "'generally' does not achieve the frontier", "'potentially' for unmodeled distortions", "recommendation contingent on regulator's weights"]}, "claim_evidence": [{"claim": "Faster HFT speeds reduce information production but narrow bid-ask spreads under LOB trading", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C, Corollary 1"}, {"claim": "LOB outcomes generally lie inside the liquidity–information frontier due to aggressive-side order anticipation", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV.C"}, {"claim": "Noncancellation delays and frequent batch auctions implement outcomes on the frontier", "evidence_type": "equation_or_proof", "evidence_ref": "Section V.C"}, {"claim": "Model predictions are consistent with many empirical HFT studies", "evidence_type": "prior_literature", "evidence_ref": "Introduction paragraph 1"}, {"claim": "Exchange latencies have fallen sharply and trading is fragmented across many venues", "evidence_type": "descriptive_statistic", "evidence_ref": "Introduction paragraph 0, footnotes 1-2"}], "structural_summary": "Opens with stylized facts on falling latencies and fragmentation, defines two forms of order anticipation, then previews comparative statics: faster speed lowers research but narrows spreads. Uses mechanism design to characterize the feasible frontier, shows LOB is inside it, and shows two alternative mechanisms reach it. Positions against Budish-Cramton-Shim, with a separate literature section, baseline model, equilibrium, frontier and alternative-mechanism sections, ending in a policy-oriented conclusion.", "writing_profile_keywords": ["theory model", "market design", "mechanism design", "order anticipation", "liquidity–information tradeoff", "policy evaluation", "stylized facts motivation"], "confidence": 0.85, "notes": "Theory paper with no tables or empirical estimation. Section V's automatic 'robustness' role is reassigned to theory. Section IV.D is labeled Robustness, which motivates the robustness location field. Causal language here refers to model comparative statics."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.92, "usage": {"input": 5319, "output": 1789, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.05785, "request_id": "req_011CggqoqFGSEiPyNuE7dLf4", "cached": false, "ts": "2026-11-03T23:13:20+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/0d5a25ec8d5728fda0842f54eaccb4d20c4b17439266bb45769b8d3761fd530d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Effect of the expected delay until public disclosure on informed trading intensity, time to first trade, order type choice, and the speed of price discovery during the private window.\",\"strategy\":\"Quasi-natural experiment: SEC filings were accidentally leaked to PDS subscribers before public EDGAR release. The delay is treated as exogenous, and an expected delay is predicted mainly from hour of day. Trading at the security-second and filing level is related to expected versus unexpected delay through panel regressions, tobit and multinomial logit models.\",\"key_assumptions\":[\"The length of the delay until public release is exogenous to filing content and firm characteristics.\",\"Informed traders can form expectations about the delay from observable factors such as time of day.\",\"Trades during the private window reflect trading by those with access to the leaked filings.\"],\"threats_addressed\":[\"Endogenous timing of disclosure, addressed by the accidental leak and the exogenous delay.\",\"Censoring of time to first trade, addressed with a two-sided tobit model.\",\"Expected and unexpected delay components separated.\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":true,\"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\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\",\"new_identification\"],\"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\":\"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 define the dependent variable, its unit (seconds, $1,000, basis points), the sample (Trimmed Sample defined in Figure 2), the unit of observation (security-second or filing) and the estimator (panel, tobit, multinomial logit). Standard errors 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' the model's predictions or hypotheses\",\"'appears to' when describing the cause of the leak\",\"'quasi-natural experiment' framing\"]},\"claim_evidence\":[{\"claim\":\"The delay until public release can be predicted mainly by the hour of the trading day.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 2\"},{\"claim\":\"A longer expected delay reduces informed trading intensity in both dollar value and number of trades.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"A longer expected delay increases the time until the first trade after the leak.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Limit orders are preferred over market orders when the expected delay is longer and market conditions are favorable.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Information disseminates faster in large-cap, high-turnover stocks and when market activity is high.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Optimal insider strategy and price dynamics under a random, exponentially distributed stopping time.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II\"}],\"structural_summary\":\"The introduction opens with an institutional anecdote (an accidental leak of SEC filings) and frames it as a lab-like natural experiment. It then sketches a model adapted from Caldentey and Stacchetti, previews the empirical measures and Kyle-consistent findings, and positions the paper against the information-transmission and leakage literatures before giving a roadmap. The body proceeds from institutional background and data, to the model, to empirical tests. A short conclusion restates the findings as consistent with the hypotheses.\",\"writing_profile_keywords\":[\"natural experiment\",\"informed trading\",\"random stopping time\",\"Kyle model\",\"trade-level data\",\"model-guided empirics\",\"institutional anecdote\"],\"confidence\":0.8,\"notes\":\"Section I is labeled 'literature' by the rule, but its content is institutional background and data. Robustness checks and standard-error conventions are not visible; an Internet Appendix exists.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "market_microstructure", "subfield_secondary": "asset_pricing", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Effect of the expected delay until public disclosure on informed trading intensity, time to first trade, order type choice, and the speed of price discovery during the private window.", "strategy": "Quasi-natural experiment: SEC filings were accidentally leaked to PDS subscribers before public EDGAR release. The delay is treated as exogenous, and an expected delay is predicted mainly from hour of day. Trading at the security-second and filing level is related to expected versus unexpected delay through panel regressions, tobit and multinomial logit models.", "key_assumptions": ["The length of the delay until public release is exogenous to filing content and firm characteristics.", "Informed traders can form expectations about the delay from observable factors such as time of day.", "Trades during the private window reflect trading by those with access to the leaked filings."], "threats_addressed": ["Endogenous timing of disclosure, addressed by the accidental leak and the exogenous delay.", "Censoring of time to first trade, addressed with a two-sided tobit model.", "Expected and unexpected delay components separated."], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": true, "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": "identification_defense"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"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": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 4, "contribution_types": ["new_evidence", "new_theory_or_model", "new_identification"], "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": "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 define the dependent variable, its unit (seconds, $1,000, basis points), the sample (Trimmed Sample defined in Figure 2), the unit of observation (security-second or filing) and the estimator (panel, tobit, multinomial logit). Standard errors 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' the model's predictions or hypotheses", "'appears to' when describing the cause of the leak", "'quasi-natural experiment' framing"]}, "claim_evidence": [{"claim": "The delay until public release can be predicted mainly by the hour of the trading day.", "evidence_type": "table", "evidence_ref": "Table II; Figure 2"}, {"claim": "A longer expected delay reduces informed trading intensity in both dollar value and number of trades.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "A longer expected delay increases the time until the first trade after the leak.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Limit orders are preferred over market orders when the expected delay is longer and market conditions are favorable.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Information disseminates faster in large-cap, high-turnover stocks and when market activity is high.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Optimal insider strategy and price dynamics under a random, exponentially distributed stopping time.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II"}], "structural_summary": "The introduction opens with an institutional anecdote (an accidental leak of SEC filings) and frames it as a lab-like natural experiment. It then sketches a model adapted from Caldentey and Stacchetti, previews the empirical measures and Kyle-consistent findings, and positions the paper against the information-transmission and leakage literatures before giving a roadmap. The body proceeds from institutional background and data, to the model, to empirical tests. A short conclusion restates the findings as consistent with the hypotheses.", "writing_profile_keywords": ["natural experiment", "informed trading", "random stopping time", "Kyle model", "trade-level data", "model-guided empirics", "institutional anecdote"], "confidence": 0.8, "notes": "Section I is labeled 'literature' by the rule, but its content is institutional background and data. Robustness checks and standard-error conventions are not visible; an Internet Appendix exists."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.85, "usage": {"input": 5311, "output": 1985, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061738, "request_id": "req_011CggqfDDcDsDd4KhZx1ghi", "cached": false, "ts": "2026-11-03T23:11:25+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/10b5cc1686c6b22206a5f90408ba4bc1c86973ffdbfa137c87ac6f5475a102cf.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Change in the disposition effect (gain-minus-loss selling propensity) for internet investors relative to noninternet investors after a brokerage display change that made purchase price/capital gain more salient\",\"strategy\":\"Difference-in-differences around the October 1, 2004 change in the online trading screen. Internet investors are treated and phone/in-branch investors are controls. Supplemented with fixed effects, propensity score matching, hazard models, data from additional branches, and quarterly placebo tests.\",\"key_assumptions\":[\"Parallel trends in the disposition effect between internet and noninternet investors absent the display change\",\"Display change altered salience but not the investor's information set\",\"Treatment and control assignment based on preperiod order channel is not confounded with other contemporaneous shocks\"],\"threats_addressed\":[\"Correlation of information display with other information (information set unchanged)\",\"Differences in investor characteristics between groups (propensity score matching)\",\"Spurious time variation (multiple placebo tests on adjacent quarters)\",\"Weighting by large portfolios (individual-level estimates)\",\"Reference point definition (break-even price)\",\"Sampling method (Odean 1998 criterion)\"],\"causal_language_strength\":5,\"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\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"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\":\"other\"},{\"idx\":5,\"role\":\"robustness\"},{\"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\",\"placebo\",\"alternative_measure\",\"alternative_sample_period\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable (sale dummy) and the Gain/Post/Internet indicators, and give the estimation method (OLS, Cox hazard). Coefficients are multiplied by 100, and some notes state two-way clustering. Later tables refer back to earlier table captions for variable definitions.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'evidence consistent with' for links to attention models\",\"explicit admission that the mechanism cannot be identified\",\"'likely to be more applicable today' for external validity\"]},\"claim_evidence\":[{\"claim\":\"The salience shock causally raised the disposition effect by about 17%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III\"},{\"claim\":\"Result holds in a propensity-score-matched sample\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Figure 5\"},{\"claim\":\"Result holds in hazard model specifications\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"No comparable effects arise at placebo quarter breaks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Treatment effect varies substantially across individual investors\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6, Table VI\"},{\"claim\":\"Individual rank effect explains heterogeneity in the response to the salience shock\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Table VII\"},{\"claim\":\"Main result robust to extended postperiod, break-even reference price, and Odean sampling\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-X\"},{\"claim\":\"Lab studies show information display affects investor choices\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Gneezy and Potters 1997; Frydman and Rangel 2014\"}],\"structural_summary\":\"The paper contrasts models in which display is irrelevant with lab evidence that it matters, then notes that field evidence is missing. It introduces a brokerage screen change as a natural experiment and frames the outcome through the disposition effect. A DiD headline estimate is followed by layered validation: fixed effects, other branches, matching, hazard models, and placebo tests. It then moves to individual-level heterogeneity linked to the rank effect, discusses candidate mechanisms while acknowledging they cannot be identified, and adds short robustness checks. The conclusion restates the causal magnitude and connects it to work on correlated biases and salience theories.\",\"writing_profile_keywords\":[\"natural experiment\",\"difference-in-differences\",\"disposition effect\",\"salience\",\"brokerage microdata\",\"heterogeneity\",\"field versus lab\"],\"confidence\":0.85,\"notes\":\"Introduction paragraph 1 is truncated where the three identification challenges are listed. The mechanisms section is labeled robustness by the rule but is a discussion section. The form of the significance reporting is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "household_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Change in the disposition effect (gain-minus-loss selling propensity) for internet investors relative to noninternet investors after a brokerage display change that made purchase price/capital gain more salient", "strategy": "Difference-in-differences around the October 1, 2004 change in the online trading screen. Internet investors are treated and phone/in-branch investors are controls. Supplemented with fixed effects, propensity score matching, hazard models, data from additional branches, and quarterly placebo tests.", "key_assumptions": ["Parallel trends in the disposition effect between internet and noninternet investors absent the display change", "Display change altered salience but not the investor's information set", "Treatment and control assignment based on preperiod order channel is not confounded with other contemporaneous shocks"], "threats_addressed": ["Correlation of information display with other information (information set unchanged)", "Differences in investor characteristics between groups (propensity score matching)", "Spurious time variation (multiple placebo tests on adjacent quarters)", "Weighting by large portfolios (individual-level estimates)", "Reference point definition (break-even price)", "Sampling method (Odean 1998 criterion)"], "causal_language_strength": 5, "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": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 3, "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": "other"}, {"idx": 5, "role": "robustness"}, {"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", "placebo", "alternative_measure", "alternative_sample_period", "subsample", "other"]}, "tables": {"convention_notes": "Table notes define the dependent variable (sale dummy) and the Gain/Post/Internet indicators, and give the estimation method (OLS, Cox hazard). Coefficients are multiplied by 100, and some notes state two-way clustering. Later tables refer back to earlier table captions for variable definitions.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'evidence consistent with' for links to attention models", "explicit admission that the mechanism cannot be identified", "'likely to be more applicable today' for external validity"]}, "claim_evidence": [{"claim": "The salience shock causally raised the disposition effect by about 17%", "evidence_type": "table", "evidence_ref": "Table II, Table III"}, {"claim": "Result holds in a propensity-score-matched sample", "evidence_type": "table", "evidence_ref": "Table IV, Figure 5"}, {"claim": "Result holds in hazard model specifications", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "No comparable effects arise at placebo quarter breaks", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Treatment effect varies substantially across individual investors", "evidence_type": "figure", "evidence_ref": "Figure 6, Table VI"}, {"claim": "Individual rank effect explains heterogeneity in the response to the salience shock", "evidence_type": "table", "evidence_ref": "Table VI, Table VII"}, {"claim": "Main result robust to extended postperiod, break-even reference price, and Odean sampling", "evidence_type": "table", "evidence_ref": "Tables VIII-X"}, {"claim": "Lab studies show information display affects investor choices", "evidence_type": "prior_literature", "evidence_ref": "Gneezy and Potters 1997; Frydman and Rangel 2014"}], "structural_summary": "The paper contrasts models in which display is irrelevant with lab evidence that it matters, then notes that field evidence is missing. It introduces a brokerage screen change as a natural experiment and frames the outcome through the disposition effect. A DiD headline estimate is followed by layered validation: fixed effects, other branches, matching, hazard models, and placebo tests. It then moves to individual-level heterogeneity linked to the rank effect, discusses candidate mechanisms while acknowledging they cannot be identified, and adds short robustness checks. The conclusion restates the causal magnitude and connects it to work on correlated biases and salience theories.", "writing_profile_keywords": ["natural experiment", "difference-in-differences", "disposition effect", "salience", "brokerage microdata", "heterogeneity", "field versus lab"], "confidence": 0.85, "notes": "Introduction paragraph 1 is truncated where the three identification challenges are listed. The mechanisms section is labeled robustness by the rule but is a discussion section. The form of the significance reporting is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.29, "usage": {"input": 6546, "output": 2195, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070878, "request_id": "req_011CggqcusGDKrgnTTLAnHxk", "cached": false, "ts": "2026-11-03T23:10:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/137c40d9432146deb120ef991664163035d2441483944a8c83ee16393b77e1fc.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"law and finance / credit access and financial development\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of France's 2006 collateral reform, which allowed nonpossessory pledges of physical assets, on the borrowing, real outcomes, start-up formation and survival, and credit-access inequality of private French firms.\",\"strategy\":\"Matched difference-in-differences comparing high- versus low-fixed-assets-intensity firms before and after 2006. Firms are propensity-score matched on predetermined traits, with firm and industry-year fixed effects. Triple-difference splits by size, age and financial constraints; hazard models for bankruptcy; spatial and Gini analyses; Wurgler-style allocation-efficiency regressions.\",\"key_assumptions\":[\"Absent the reform, high- and low-fixed-assets firms would have followed parallel trends.\",\"Pre-reform fixed-asset intensity is a valid proxy for exposure to newly pledgeable assets.\",\"No contemporaneous shocks differentially affect high-fixed-assets firms; other law changes are discussed in the institutional section.\"],\"threats_addressed\":[\"Pre-reform differences between treated and control firms (matching; Table I balance)\",\"Pre-existing differential trends (parallel trends tests)\",\"Generic shocks to asset-intensive firms (placebo regressions on other Napoleonic-Code countries)\",\"Non-collateral channels (short-term unsecured debt used as a falsification outcome)\",\"Rural effect driven by industry or size composition (spatial distribution checks)\"],\"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\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":4,\"move\":\"motivation\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":9,\"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\":\"robustness\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"falsification\",\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes write out the estimating equation (firm and industry-year fixed effects, Post×Treated), define outcome columns and panels, and describe the matched sample. Hazard-model specifications are given in the notes where used. Descriptive tables report means before and after the reform plus differences. 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\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests' / 'consistent with' for interpretive claims\",\"'may lead' for extrapolations\",\"confident 'we show/find' for core estimates\"]},\"claim_evidence\":[{\"claim\":\"The reform raised long-term leverage and cut zero-debt incidence mainly for high-fixed-assets private firms; short-term debt was unaffected.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figures 2-4\"},{\"claim\":\"Treated firms increased investment, employment, sales and profitability and had lower profit volatility and bankruptcy hazard.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Small, young and financially constrained firms gained the most credit access.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV-V\"},{\"claim\":\"Rural firms gained most, and departmental credit Gini indices fell sharply, especially in rural departments.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6\"},{\"claim\":\"Start-ups had higher at-inception leverage, located farther from large cities, and had lower exit and bankruptcy rates after the reform.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VII; Figure 7\"},{\"claim\":\"Cross-industry capital allocation efficiency improved, more so in fixed-asset-heavy and externally dependent sectors.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-IX\"},{\"claim\":\"Treated and control firms show parallel pre-reform trends.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"Opens with a political epigraph, then motivates the topic with the Napoleonic possessory-ownership regime and an understanding gap. The introduction previews results in sequence: credit, real outcomes, democratization by firm type and geography, then macro allocation efficiency, closing with law-and-finance positioning. The body describes the institutional setting and identification logic, then data and the matched DID design. Results sections cover the average firm, distributional and spatial effects, start-ups, and capital allocation efficiency. A dedicated robustness section follows, and the conclusion restates findings and draws policy implications.\",\"writing_profile_keywords\":[\"natural policy experiment\",\"matched difference-in-differences\",\"law and finance\",\"collateral\",\"credit democratization\",\"spatial analysis\",\"intro magnitude previews\",\"policy-oriented\"],\"confidence\":0.8,\"notes\":\"Abstract text includes author footnotes, which are ignored for sentence indexing. Introduction paragraphs 0-3 are epigraph fragments. Section I is labeled theory by rule but is an institutional and identification discussion, so it is coded 'other'. Standard-error and significance conventions are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "law and finance / credit access and financial development", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of France's 2006 collateral reform, which allowed nonpossessory pledges of physical assets, on the borrowing, real outcomes, start-up formation and survival, and credit-access inequality of private French firms.", "strategy": "Matched difference-in-differences comparing high- versus low-fixed-assets-intensity firms before and after 2006. Firms are propensity-score matched on predetermined traits, with firm and industry-year fixed effects. Triple-difference splits by size, age and financial constraints; hazard models for bankruptcy; spatial and Gini analyses; Wurgler-style allocation-efficiency regressions.", "key_assumptions": ["Absent the reform, high- and low-fixed-assets firms would have followed parallel trends.", "Pre-reform fixed-asset intensity is a valid proxy for exposure to newly pledgeable assets.", "No contemporaneous shocks differentially affect high-fixed-assets firms; other law changes are discussed in the institutional section."], "threats_addressed": ["Pre-reform differences between treated and control firms (matching; Table I balance)", "Pre-existing differential trends (parallel trends tests)", "Generic shocks to asset-intensive firms (placebo regressions on other Napoleonic-Code countries)", "Non-collateral channels (short-term unsecured debt used as a falsification outcome)", "Rural effect driven by industry or size composition (spatial distribution checks)"], "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": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "other"}, {"paragraph_index": 4, "move": "motivation"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "policy_or_broader_implication"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "related_literature_positioning"}], "contribution_paragraph_index": 9, "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": "robustness"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "falsification", "subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes write out the estimating equation (firm and industry-year fixed effects, Post×Treated), define outcome columns and panels, and describe the matched sample. Hazard-model specifications are given in the notes where used. Descriptive tables report means before and after the reform plus differences. 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": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests' / 'consistent with' for interpretive claims", "'may lead' for extrapolations", "confident 'we show/find' for core estimates"]}, "claim_evidence": [{"claim": "The reform raised long-term leverage and cut zero-debt incidence mainly for high-fixed-assets private firms; short-term debt was unaffected.", "evidence_type": "table", "evidence_ref": "Table II; Figures 2-4"}, {"claim": "Treated firms increased investment, employment, sales and profitability and had lower profit volatility and bankruptcy hazard.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Small, young and financially constrained firms gained the most credit access.", "evidence_type": "table", "evidence_ref": "Tables IV-V"}, {"claim": "Rural firms gained most, and departmental credit Gini indices fell sharply, especially in rural departments.", "evidence_type": "figure", "evidence_ref": "Figures 5-6"}, {"claim": "Start-ups had higher at-inception leverage, located farther from large cities, and had lower exit and bankruptcy rates after the reform.", "evidence_type": "table", "evidence_ref": "Tables VI-VII; Figure 7"}, {"claim": "Cross-industry capital allocation efficiency improved, more so in fixed-asset-heavy and externally dependent sectors.", "evidence_type": "table", "evidence_ref": "Tables VIII-IX"}, {"claim": "Treated and control firms show parallel pre-reform trends.", "evidence_type": "table", "evidence_ref": "Table X"}], "structural_summary": "Opens with a political epigraph, then motivates the topic with the Napoleonic possessory-ownership regime and an understanding gap. The introduction previews results in sequence: credit, real outcomes, democratization by firm type and geography, then macro allocation efficiency, closing with law-and-finance positioning. The body describes the institutional setting and identification logic, then data and the matched DID design. Results sections cover the average firm, distributional and spatial effects, start-ups, and capital allocation efficiency. A dedicated robustness section follows, and the conclusion restates findings and draws policy implications.", "writing_profile_keywords": ["natural policy experiment", "matched difference-in-differences", "law and finance", "collateral", "credit democratization", "spatial analysis", "intro magnitude previews", "policy-oriented"], "confidence": 0.8, "notes": "Abstract text includes author footnotes, which are ignored for sentence indexing. Introduction paragraphs 0-3 are epigraph fragments. Section I is labeled theory by rule but is an institutional and identification discussion, so it is coded 'other'. Standard-error and significance conventions are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 27.97, "usage": {"input": 10292, "output": 3314, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.108242, "request_id": "req_011CggqePv1tCxVYwG3KSGDU", "cached": false, "ts": "2026-11-03T23:11:25+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/15431e3aa08ae0fd1f99b60fbd4a4a722783db3954ccbe5479963cbfe98cd419.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; analytical characterization of equilibrium in a two-date labor-search model with limited-liability household debt\",\"key_assumptions\":[\"households protected by limited liability\",\"Diamond-Mortensen-Pissarides random matching with bilateral wage bargaining\",\"square-root matching probabilities\",\"perfectly competitive banks\",\"repayment schedules nondecreasing in wages\"],\"threats_addressed\":[\"relaxing equal bargaining power\",\"alternative matching function\",\"imperfectly competitive banks\",\"nonincreasing financial contracts\"],\"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\":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\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"other\"},{\"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_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\":\"other\"},{\"idx\":4,\"role\":\"robustness\"},{\"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\":\"main_text_section\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"No tables; results are delivered as propositions and equations, with proofs collected in the Appendix.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'our model is stylized, but may'\",\"'our model suggests'\",\"'may contribute' / 'may exacerbate' in extension previews\",\"'could be a bad thing'\"]},\"claim_evidence\":[{\"claim\":\"Limited-liability household debt creates a debt overhang that makes indebted households demand high wages to work\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.A-B; proofs in Appendix\"},{\"claim\":\"Debt is the optimal financing contract within the class of repayment schedules increasing in wages\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.A; Section IV.D\"},{\"claim\":\"Firms respond with high wages and few vacancies, so high household debt raises unemployment (vacancy posting effect)\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B-C\"},{\"claim\":\"Equilibrium household debt is inefficiently high because of a household debt externality; debt caps can mitigate it\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.D\"},{\"claim\":\"Lower default penalties can worsen the vacancy posting effect\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C\"},{\"claim\":\"Household debt reducing labor supply is consistent with recent empirical studies\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Bernstein (2018); Herkenhoff (2012); Dobbie and Goldsmith-Pinkham (2015)\"},{\"claim\":\"The model matches macro stylized facts on household leverage, wage rigidity and collateral shocks\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Mian and Sufi (2010, 2014b); Bewley (1999); Daly and Hobijn (2015)\"}],\"structural_summary\":\"A theory paper. The introduction opens with the prevalence of personal bankruptcy and limited liability, then poses explicit research questions. It previews a two-date search model and its headline results (debt overhang, optimal debt contract, vacancy posting effect, externality). It then sketches the extensions, draws policy implications, links predictions to existing empirical findings and stylized facts, positions the paper against the aggregate-demand and debt-overhang literatures, and closes with a roadmap. The body proceeds in order: model setup with assumptions, analytical results (including the constrained-efficient benchmark and multiple equilibria), reduced-form extensions, and robustness to relaxed modeling assumptions. Proofs are in the Appendix. The conclusion restates the mechanism and gives a regulatory and monetary-policy implication.\",\"writing_profile_keywords\":[\"labor search model\",\"household debt overhang\",\"limited liability\",\"externality\",\"macroprudential policy\",\"propositions with appendix proofs\",\"corporate finance analogy\",\"labeled intro paragraphs\"],\"confidence\":0.85,\"notes\":\"No tables or figures. The intro uses inline labels (Model preview, Results preview, Policy, Empirical content, Related literature, Layout). The Extensions section was coded 'other' because it adds new results rather than serving as robustness. Causal language about the model's mechanism is strong.\\n\"}", "json": {"genre": "theoretical", "subfield": "macro_finance", "subfield_secondary": "household_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; analytical characterization of equilibrium in a two-date labor-search model with limited-liability household debt", "key_assumptions": ["households protected by limited liability", "Diamond-Mortensen-Pissarides random matching with bilateral wage bargaining", "square-root matching probabilities", "perfectly competitive banks", "repayment schedules nondecreasing in wages"], "threats_addressed": ["relaxing equal bargaining power", "alternative matching function", "imperfectly competitive banks", "nonincreasing financial contracts"], "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": 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": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "other"}, {"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_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": "other"}, {"idx": 4, "role": "robustness"}, {"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": "main_text_section", "types": ["sensitivity_to_parameters", "alternative_specification"]}, "tables": {"convention_notes": "No tables; results are delivered as propositions and equations, with proofs collected in the Appendix.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'our model is stylized, but may'", "'our model suggests'", "'may contribute' / 'may exacerbate' in extension previews", "'could be a bad thing'"]}, "claim_evidence": [{"claim": "Limited-liability household debt creates a debt overhang that makes indebted households demand high wages to work", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.A-B; proofs in Appendix"}, {"claim": "Debt is the optimal financing contract within the class of repayment schedules increasing in wages", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.A; Section IV.D"}, {"claim": "Firms respond with high wages and few vacancies, so high household debt raises unemployment (vacancy posting effect)", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B-C"}, {"claim": "Equilibrium household debt is inefficiently high because of a household debt externality; debt caps can mitigate it", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.D"}, {"claim": "Lower default penalties can worsen the vacancy posting effect", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C"}, {"claim": "Household debt reducing labor supply is consistent with recent empirical studies", "evidence_type": "prior_literature", "evidence_ref": "Bernstein (2018); Herkenhoff (2012); Dobbie and Goldsmith-Pinkham (2015)"}, {"claim": "The model matches macro stylized facts on household leverage, wage rigidity and collateral shocks", "evidence_type": "prior_literature", "evidence_ref": "Mian and Sufi (2010, 2014b); Bewley (1999); Daly and Hobijn (2015)"}], "structural_summary": "A theory paper. The introduction opens with the prevalence of personal bankruptcy and limited liability, then poses explicit research questions. It previews a two-date search model and its headline results (debt overhang, optimal debt contract, vacancy posting effect, externality). It then sketches the extensions, draws policy implications, links predictions to existing empirical findings and stylized facts, positions the paper against the aggregate-demand and debt-overhang literatures, and closes with a roadmap. The body proceeds in order: model setup with assumptions, analytical results (including the constrained-efficient benchmark and multiple equilibria), reduced-form extensions, and robustness to relaxed modeling assumptions. Proofs are in the Appendix. The conclusion restates the mechanism and gives a regulatory and monetary-policy implication.", "writing_profile_keywords": ["labor search model", "household debt overhang", "limited liability", "externality", "macroprudential policy", "propositions with appendix proofs", "corporate finance analogy", "labeled intro paragraphs"], "confidence": 0.85, "notes": "No tables or figures. The intro uses inline labels (Model preview, Results preview, Policy, Empirical content, Related literature, Layout). The Extensions section was coded 'other' because it adds new results rather than serving as robustness. Causal language about the model's mechanism is strong.\n"}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.24, "usage": {"input": 4963, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064086, "request_id": "req_011CggqNPAg7Jb6TyY4G9qb8", "cached": false, "ts": "2026-11-03T23:07:38+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/1cd0ccf36aca0c97f076fbee14ddb1e59c5c7c7fc4dfc47cf2979b0302fa7ea1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"corporate_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Difference in the predictive relation between insider trading strength and future stock returns across short- versus long-horizon insiders.\",\"strategy\":\"Two-way portfolio sorts on trading strength and horizon, plus panel regressions of future returns on trading strength, horizon and their interaction with controls; supplemented by earnings-surprise, large-price-change and SEC-enforcement tests.\",\"key_assumptions\":[\"Past net order flow proxies for an insider's investment horizon\",\"Horizon-based expectations about continued trading identify unexpected trades\",\"Markets are efficient, so unexpected trades should carry more information\"],\"threats_addressed\":[\"Differences in insider and firm attributes between SH and LH insiders (additional controls)\",\"Overlap with the routine/nonroutine classification of CMP\",\"Firm-level aggregation of the return relation\"],\"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\":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\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"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_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\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"robustness\"},{\"idx\":9,\"role\":\"conclusion\"},{\"idx\":10,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define variables, describe the sorting procedure or write out the full regression equation, and state the dependent variable and sample construction. Tables are organized in panels, e.g., buys vs. sells or all trades vs. subsamples.\",\"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\":[\"conjecture framing\",\"'consistent with'\",\"'on average'\",\"'suggests'\",\"'tend to'\",\"'is associated with'\"]},\"claim_evidence\":[{\"claim\":\"Short-horizon insiders' trades predict returns better than long-horizon insiders' trades for up to 12 months, on both buys and sells.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III\"},{\"claim\":\"Unexpected insider trades carry more information about future returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Typical SH trades are more unexpected than LH trades.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"SH insiders and their firms display characteristics associated with short-termism, such as more earnings management and less R&D.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"SH insider trades better predict earnings surprises and large price changes.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"SH insiders trade less after intensified SEC enforcement.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"The horizon measure adds information beyond the routine/nonroutine classification.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"}],\"structural_summary\":\"The paper defines a new insider-horizon measure based on past net order flow and builds a conjecture about expected versus unexpected trades, from which it derives two hypotheses. It then documents baseline return predictability with sorts and regressions. Next, it tests the unexpected-trade mechanism, profiles SH/LH insiders and short-termism, traces sources of the information advantage through earnings and price events, and examines responses to SEC enforcement. It differentiates the measure from the routine/nonroutine concept, runs robustness tests, and closes with practitioner and regulatory implications and future research.\",\"writing_profile_keywords\":[\"new measure\",\"conjecture-driven hypotheses\",\"portfolio sorts\",\"panel regressions\",\"mechanism tests\",\"anecdotal examples\",\"policy implications\"],\"confidence\":0.8,\"notes\":\"The abstract appears to merge the start of the introduction. Standard error and significance conventions are not visible in the excerpts. The robustness section lists subsections A and C, so B is missing from the excerpt.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "corporate_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Difference in the predictive relation between insider trading strength and future stock returns across short- versus long-horizon insiders.", "strategy": "Two-way portfolio sorts on trading strength and horizon, plus panel regressions of future returns on trading strength, horizon and their interaction with controls; supplemented by earnings-surprise, large-price-change and SEC-enforcement tests.", "key_assumptions": ["Past net order flow proxies for an insider's investment horizon", "Horizon-based expectations about continued trading identify unexpected trades", "Markets are efficient, so unexpected trades should carry more information"], "threats_addressed": ["Differences in insider and firm attributes between SH and LH insiders (additional controls)", "Overlap with the routine/nonroutine classification of CMP", "Firm-level aggregation of the return relation"], "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": 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": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"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_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": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "robustness"}, {"idx": 9, "role": "conclusion"}, {"idx": 10, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes define variables, describe the sorting procedure or write out the full regression equation, and state the dependent variable and sample construction. Tables are organized in panels, e.g., buys vs. sells or all trades vs. subsamples.", "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": ["conjecture framing", "'consistent with'", "'on average'", "'suggests'", "'tend to'", "'is associated with'"]}, "claim_evidence": [{"claim": "Short-horizon insiders' trades predict returns better than long-horizon insiders' trades for up to 12 months, on both buys and sells.", "evidence_type": "table", "evidence_ref": "Tables II, III"}, {"claim": "Unexpected insider trades carry more information about future returns.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Typical SH trades are more unexpected than LH trades.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "SH insiders and their firms display characteristics associated with short-termism, such as more earnings management and less R&D.", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "SH insider trades better predict earnings surprises and large price changes.", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "SH insiders trade less after intensified SEC enforcement.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "The horizon measure adds information beyond the routine/nonroutine classification.", "evidence_type": "table", "evidence_ref": "Table XI"}], "structural_summary": "The paper defines a new insider-horizon measure based on past net order flow and builds a conjecture about expected versus unexpected trades, from which it derives two hypotheses. It then documents baseline return predictability with sorts and regressions. Next, it tests the unexpected-trade mechanism, profiles SH/LH insiders and short-termism, traces sources of the information advantage through earnings and price events, and examines responses to SEC enforcement. It differentiates the measure from the routine/nonroutine concept, runs robustness tests, and closes with practitioner and regulatory implications and future research.", "writing_profile_keywords": ["new measure", "conjecture-driven hypotheses", "portfolio sorts", "panel regressions", "mechanism tests", "anecdotal examples", "policy implications"], "confidence": 0.8, "notes": "The abstract appears to merge the start of the introduction. Standard error and significance conventions are not visible in the excerpts. The robustness section lists subsections A and C, so B is missing from the excerpt."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.25, "usage": {"input": 9312, "output": 2112, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080282, "request_id": "req_011CggqpVokERFULLTx3f2Dj", "cached": false, "ts": "2026-11-03T23:13:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/1d90fd10a16c50f99e6d5152d7fa6bc981b6dd1b30553546753f49dfcf569b13.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\":\"Differential likelihood of predatory selling (versus liquidity provision) by clients of brokers who intermediate fire-sale trades, especially the brokers' best clients; the resulting profits; and the added price impact borne by the liquidating fund\",\"strategy\":\"Ticket- and stock-day-manager-broker-level regressions that compare clients of aware and unaware brokers, and best and other clients, during versus before fire sales, supplemented by a natural experiment from the 2003 late-trading scandal outflows\",\"key_assumptions\":[\"Fire-sale events identified from volume Z-scores reflect liquidity-motivated rather than information-motivated selling\",\"Followers are not trading on a common information signal shared with the originator\",\"Outflows from scandal-implicated fund families are exogenous to stock fundamentals\"],\"threats_addressed\":[\"common information signals such as recessions, earnings announcements, analyst recommendation changes and negative press news\",\"shared negative-signal strategies such as momentum or high short interest\",\"endogeneity of the liquidation source, addressed through the late-trading scandal\",\"alternative explanation that followers are providing liquidity rather than predating\"],\"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\":3},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"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\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"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\":\"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\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"placebo\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the level at which each regression is run, define the dependent variables and event windows, and explain how the sample is constructed (e.g., excluding the originating managers' own trades). Standard-error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'evidence suggests' framing\",\"'plausibly exogenous' qualifier\",\"modal verbs such as 'might' and 'can'\",\"'increases our confidence' phrasing\"]},\"claim_evidence\":[{\"claim\":\"Clients of brokers intermediating fire sales are more likely to predate on the liquidating fund than to provide liquidity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Best clients of aware brokers have a net predation probability more than twice that of small clients\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"A substantial fraction (30-42%) of positions sold by other managers during fire sales is bought back within 10 days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"After the 2003 scandal, clients of the implicated funds' brokers sold the same stocks on the same days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Predation earns roughly 25 bps over about 10 days for best clients of aware brokers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 4\"},{\"claim\":\"Information leakage raises the distressed fund's liquidation costs by about 40%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII; Figure 5\"},{\"claim\":\"Predators pay higher commissions to aware brokers after events, consistent with a quid pro quo\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The paper motivates the problem through predatory trading around slowly executed large orders and fire sales, and argues that brokers are privileged observers of client order flow. It identifies fire sales in institutional trade-level data and then shows that clients of intermediating brokers, especially the best clients, sell alongside the liquidator and later reverse their positions. Robustness checks exclude news- and signal-driven episodes, and a natural experiment based on the scandal outflows supports the main result. A separate section quantifies predator profits, the price impact on the originator, the persistence of broker-manager relationships, and commission payments. The conclusion discusses implications for execution practice and regulation.\",\"writing_profile_keywords\":[\"trade-level institutional data\",\"fire sales\",\"predatory trading\",\"broker information leakage\",\"natural experiment\",\"economic magnitudes\",\"policy implications\"],\"confidence\":0.8,\"notes\":\"The introduction excerpt is truncated in places. Standard-error conventions are not visible in the excerpts. Figure 2 shows confidence bands; whether the other figures show uncertainty is not stated. The intro mentions 'clients of the relevant brokers' selling after the scandal; the claim wording paraphrases this.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Differential likelihood of predatory selling (versus liquidity provision) by clients of brokers who intermediate fire-sale trades, especially the brokers' best clients; the resulting profits; and the added price impact borne by the liquidating fund", "strategy": "Ticket- and stock-day-manager-broker-level regressions that compare clients of aware and unaware brokers, and best and other clients, during versus before fire sales, supplemented by a natural experiment from the 2003 late-trading scandal outflows", "key_assumptions": ["Fire-sale events identified from volume Z-scores reflect liquidity-motivated rather than information-motivated selling", "Followers are not trading on a common information signal shared with the originator", "Outflows from scandal-implicated fund families are exogenous to stock fundamentals"], "threats_addressed": ["common information signals such as recessions, earnings announcements, analyst recommendation changes and negative press news", "shared negative-signal strategies such as momentum or high short interest", "endogeneity of the liquidation source, addressed through the late-trading scandal", "alternative explanation that followers are providing liquidity rather than predating"], "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": 3}, "contribution": {"present": true, "sentence_index": 4}}, "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": "robustness_mention"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"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": "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": true}, "robustness": {"location": "within_results", "types": ["subsample", "placebo", "other"]}, "tables": {"convention_notes": "Table notes state the level at which each regression is run, define the dependent variables and event windows, and explain how the sample is constructed (e.g., excluding the originating managers' own trades). Standard-error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'evidence suggests' framing", "'plausibly exogenous' qualifier", "modal verbs such as 'might' and 'can'", "'increases our confidence' phrasing"]}, "claim_evidence": [{"claim": "Clients of brokers intermediating fire sales are more likely to predate on the liquidating fund than to provide liquidity", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Best clients of aware brokers have a net predation probability more than twice that of small clients", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "A substantial fraction (30-42%) of positions sold by other managers during fire sales is bought back within 10 days", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "After the 2003 scandal, clients of the implicated funds' brokers sold the same stocks on the same days", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Predation earns roughly 25 bps over about 10 days for best clients of aware brokers", "evidence_type": "table", "evidence_ref": "Table VII; Figure 4"}, {"claim": "Information leakage raises the distressed fund's liquidation costs by about 40%", "evidence_type": "table", "evidence_ref": "Table VIII; Figure 5"}, {"claim": "Predators pay higher commissions to aware brokers after events, consistent with a quid pro quo", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The paper motivates the problem through predatory trading around slowly executed large orders and fire sales, and argues that brokers are privileged observers of client order flow. It identifies fire sales in institutional trade-level data and then shows that clients of intermediating brokers, especially the best clients, sell alongside the liquidator and later reverse their positions. Robustness checks exclude news- and signal-driven episodes, and a natural experiment based on the scandal outflows supports the main result. A separate section quantifies predator profits, the price impact on the originator, the persistence of broker-manager relationships, and commission payments. The conclusion discusses implications for execution practice and regulation.", "writing_profile_keywords": ["trade-level institutional data", "fire sales", "predatory trading", "broker information leakage", "natural experiment", "economic magnitudes", "policy implications"], "confidence": 0.8, "notes": "The introduction excerpt is truncated in places. Standard-error conventions are not visible in the excerpts. Figure 2 shows confidence bands; whether the other figures show uncertainty is not stated. The intro mentions 'clients of the relevant brokers' selling after the scandal; the claim wording paraphrases this."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.29, "usage": {"input": 6561, "output": 2209, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071218, "request_id": "req_011CggqXjPqPp19BLMSZMvX8", "cached": false, "ts": "2026-11-03T23:09:45+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2191044f4ec7264189d44b7d68a27bbcbfe72a8e1a410bd3349b46c1f0043856.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Difference in well time-to-completion and completion rates between high- and low-leverage firms during contango versus before it, and around prescheduled asset-based lending renegotiations\",\"strategy\":\"Two difference-in-differences designs: one comparing the contango period with the period before it across leverage groups, and one using the staggered timing of prescheduled ABL renegotiations. Both include granular township fixed effects to control for differences in project quality.\",\"key_assumptions\":[\"Renegotiation timing is prescheduled and plausibly exogenous to the contango period\",\"Township fixed effects absorb differences in investment opportunities between high- and low-leverage firms\",\"The futures curve gives a valid counterfactual value for delaying completion\"],\"threats_addressed\":[\"endogeneity of leverage and differences in investment opportunities\",\"firm heterogeneity in profitability, size, and market-to-book\",\"presence of ABL itself rather than leverage\",\"outlier high-leverage firms driving results\",\"binding covenants and immediate cash needs as an alternative channel\",\"alternative leverage definitions\"],\"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\":1},\"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\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"identification_defense\"},{\"paragraph_index\":8,\"move\":\"identification_defense\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":13,\"move\":\"preview_of_results\"},{\"paragraph_index\":14,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":15,\"move\":\"preview_of_results\"},{\"paragraph_index\":16,\"move\":\"robustness_mention\"},{\"paragraph_index\":17,\"move\":\"contribution_statement\"},{\"paragraph_index\":18,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":19,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":20,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":17,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"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\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"placebo\",\"alternative_sample_period\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes define the unit of observation (well-firm-year or well-firm-month), the dependent variable, the leverage threshold (top quintile), and the sample construction. The pattern is univariate tables paired with regression DiD 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\":[\"results suggest\",\"consistent with\",\"likely aim to\",\"plausibly exogenous\",\"make progress on challenges\"]},\"claim_evidence\":[{\"claim\":\"High-leverage firms complete wells about one month earlier during contango, losing roughly 4.8% of project NPV\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table IV\"},{\"claim\":\"The contango effect survives controls for profitability, size and Tobin's Q\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Contango made delaying completion value-maximizing at the project level\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Figure 2\"},{\"claim\":\"High-leverage firms complete far more wells in the month before ABL renegotiation than after it\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Table VII\"},{\"claim\":\"Results are not driven by a few high-leverage outlier firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII, Table IX\"},{\"claim\":\"Wells completed before renegotiation have higher initial production, consistent with a collateral motive\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Before renegotiation, firms favor wells with high collateral impact on undeveloped leases\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI, Table XII\"},{\"claim\":\"Investment opportunities are comparable across leverage groups\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"}],\"structural_summary\":\"The introduction opens with the debt-agency literature and claims a new distortion: accelerated investment. It lists four identification challenges and answers each with features of the shale setting. It previews results from two DiD designs with magnitudes, then rules out traditional distortions. Two candidate channels follow: covenants and cash flow, which are rejected, and collateral, which is supported. Robustness tests are mentioned, contributions are positioned against three literatures, and a roadmap closes the introduction. The body has a single data and institutional section, a long results section with subsections on alternative explanations, robustness, and channels, and a short conclusion that restates results and implications.\",\"writing_profile_keywords\":[\"project-level data\",\"shale oil setting\",\"contango counterfactual\",\"two DiD designs\",\"channel tests\",\"collateral\",\"hidden cost of secured debt\",\"moderate hedging\"],\"confidence\":0.8,\"notes\":\"Section I is labeled 'literature' by the role rule but is actually institutional background and data. The excerpts do not state the standard error type or the significance notation.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "difference_in_differences", "identification": {"estimand": "Difference in well time-to-completion and completion rates between high- and low-leverage firms during contango versus before it, and around prescheduled asset-based lending renegotiations", "strategy": "Two difference-in-differences designs: one comparing the contango period with the period before it across leverage groups, and one using the staggered timing of prescheduled ABL renegotiations. Both include granular township fixed effects to control for differences in project quality.", "key_assumptions": ["Renegotiation timing is prescheduled and plausibly exogenous to the contango period", "Township fixed effects absorb differences in investment opportunities between high- and low-leverage firms", "The futures curve gives a valid counterfactual value for delaying completion"], "threats_addressed": ["endogeneity of leverage and differences in investment opportunities", "firm heterogeneity in profitability, size, and market-to-book", "presence of ABL itself rather than leverage", "outlier high-leverage firms driving results", "binding covenants and immediate cash needs as an alternative channel", "alternative leverage definitions"], "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": 1}, "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": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "identification_defense"}, {"paragraph_index": 8, "move": "identification_defense"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "preview_of_results"}, {"paragraph_index": 11, "move": "related_literature_positioning"}, {"paragraph_index": 12, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 13, "move": "preview_of_results"}, {"paragraph_index": 14, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 15, "move": "preview_of_results"}, {"paragraph_index": 16, "move": "robustness_mention"}, {"paragraph_index": 17, "move": "contribution_statement"}, {"paragraph_index": 18, "move": "related_literature_positioning"}, {"paragraph_index": 19, "move": "policy_or_broader_implication"}, {"paragraph_index": 20, "move": "roadmap"}], "contribution_paragraph_index": 17, "contribution_types": ["new_evidence", "new_data_or_measure", "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": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "placebo", "alternative_sample_period", "alternative_specification"]}, "tables": {"convention_notes": "Notes define the unit of observation (well-firm-year or well-firm-month), the dependent variable, the leverage threshold (top quintile), and the sample construction. The pattern is univariate tables paired with regression DiD 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": ["results suggest", "consistent with", "likely aim to", "plausibly exogenous", "make progress on challenges"]}, "claim_evidence": [{"claim": "High-leverage firms complete wells about one month earlier during contango, losing roughly 4.8% of project NPV", "evidence_type": "table", "evidence_ref": "Table II, Table IV"}, {"claim": "The contango effect survives controls for profitability, size and Tobin's Q", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Contango made delaying completion value-maximizing at the project level", "evidence_type": "figure", "evidence_ref": "Figure 1, Figure 2"}, {"claim": "High-leverage firms complete far more wells in the month before ABL renegotiation than after it", "evidence_type": "table", "evidence_ref": "Table VI, Table VII"}, {"claim": "Results are not driven by a few high-leverage outlier firms", "evidence_type": "table", "evidence_ref": "Table VIII, Table IX"}, {"claim": "Wells completed before renegotiation have higher initial production, consistent with a collateral motive", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Before renegotiation, firms favor wells with high collateral impact on undeveloped leases", "evidence_type": "table", "evidence_ref": "Table XI, Table XII"}, {"claim": "Investment opportunities are comparable across leverage groups", "evidence_type": "table", "evidence_ref": "Table III"}], "structural_summary": "The introduction opens with the debt-agency literature and claims a new distortion: accelerated investment. It lists four identification challenges and answers each with features of the shale setting. It previews results from two DiD designs with magnitudes, then rules out traditional distortions. Two candidate channels follow: covenants and cash flow, which are rejected, and collateral, which is supported. Robustness tests are mentioned, contributions are positioned against three literatures, and a roadmap closes the introduction. The body has a single data and institutional section, a long results section with subsections on alternative explanations, robustness, and channels, and a short conclusion that restates results and implications.", "writing_profile_keywords": ["project-level data", "shale oil setting", "contango counterfactual", "two DiD designs", "channel tests", "collateral", "hidden cost of secured debt", "moderate hedging"], "confidence": 0.8, "notes": "Section I is labeled 'literature' by the role rule but is actually institutional background and data. The excerpts do not state the standard error type or the significance notation."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.55, "usage": {"input": 9580, "output": 2417, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.087454, "request_id": "req_011Cggqn7N5CSq1mWF9ZZBDr", "cached": false, "ts": "2026-11-03T23:13:02+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/21a1bbe9dd87153e56403f7016f29c8f78dced1668577681ea52dc547a8def80.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"derivatives/option-implied moments; cross-section of equity returns\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Factor-model alphas and residual coskewness of beta- and volatility-sorted long-short strategies, before and after controlling for coskewness factor-mimicking portfolios; common variation across low-risk anomaly returns\",\"strategy\":\"A skew-aware quadratic SDF and a simulated Merton-type firm model generate predictions. These are tested with monthly decile portfolio sorts, time-series factor regressions (CAPM, FF3, FF4) augmented with option-implied skewness factors, cross-sectional regressions of alphas on residual coskewness, and principal component analysis of anomaly returns.\",\"key_assumptions\":[\"Pricing kernel is quadratic in the market excess return\",\"Option-implied ex ante skewness is a valid forward-looking proxy for residual coskewness\",\"Skew factor-mimicking portfolios capture coskewness risk\"],\"threats_addressed\":[\"Results driven by standard factors (FF3/FF4 controls)\",\"Sensitivity to skewness factor construction (alternative specifications)\",\"Portfolio weighting (equal vs value-weighted)\"],\"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\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\",\"resolution_of_puzzle\",\"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\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes open by stating what the table reports. They then describe the monthly decile sorting procedure, the weighting scheme (value- vs equal-weighted panels) and the factor models used. Panels split by weighting or by analysis step.\",\"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\":[\"'suggest' when interpreting findings\",\"'may not' / 'do not necessarily' for the puzzle-resolution claim\",\"'consistent with' the model\",\"'largely' qualifiers in the abstract\"]},\"claim_evidence\":[{\"claim\":\"In a skew-aware SDF, CAPM alphas reflect compensation for residual coskewness that the CAPM ignores\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.A\"},{\"claim\":\"In simulated Merton-type firms, alphas line up with residual coskewness under a skew-aware pricing kernel\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 3-4\"},{\"claim\":\"Option-implied ex ante skewness is associated with residual coskewness and CAPM alphas in the data\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Controlling for skew factors cuts the BaB CAPM alpha from about 125 to 33 bp per month and makes it insignificant\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 6\"},{\"claim\":\"The first principal component explains over 90% of variation in low-risk anomaly excess returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 8\"},{\"claim\":\"Skew factor returns explain up to 95% of the first PC of anomaly returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"}],\"structural_summary\":\"The introduction frames beta and volatility anomalies and claims they reflect coskewness compensation. It sketches a quadratic-SDF theory and then previews three empirical blocks with magnitudes before positioning against the skewness literature. Section I derives the theory and runs a simulation. Section II covers data and variable construction. Section III presents three empirical steps: ex ante skewness as a proxy for coskewness, skew-adjusted alphas, and PCA showing a common skew-related driver. Section IV adds robustness and extensions. A short conclusion restates the contribution, and a simulation appendix follows.\",\"writing_profile_keywords\":[\"theory-guided empirics\",\"portfolio sorts\",\"factor-mimicking portfolios\",\"option-implied moments\",\"anomaly explanation\",\"principal components\",\"simulation\",\"figure-heavy\"],\"confidence\":0.8,\"notes\":\"Section II is rule-labeled theory but is clearly data and variable construction. Section IV mixes robustness checks with extensions and future research. Standard error conventions and significance notation are not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "derivatives/option-implied moments; cross-section of equity returns", "primary_method": "reduced_form_regression", "identification": {"estimand": "Factor-model alphas and residual coskewness of beta- and volatility-sorted long-short strategies, before and after controlling for coskewness factor-mimicking portfolios; common variation across low-risk anomaly returns", "strategy": "A skew-aware quadratic SDF and a simulated Merton-type firm model generate predictions. These are tested with monthly decile portfolio sorts, time-series factor regressions (CAPM, FF3, FF4) augmented with option-implied skewness factors, cross-sectional regressions of alphas on residual coskewness, and principal component analysis of anomaly returns.", "key_assumptions": ["Pricing kernel is quadratic in the market excess return", "Option-implied ex ante skewness is a valid forward-looking proxy for residual coskewness", "Skew factor-mimicking portfolios capture coskewness risk"], "threats_addressed": ["Results driven by standard factors (FF3/FF4 controls)", "Sensitivity to skewness factor construction (alternative specifications)", "Portfolio weighting (equal vs value-weighted)"], "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": "contribution_statement"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"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": "preview_of_results"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 4, "move": "related_literature_positioning"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_theory_or_model", "resolution_of_puzzle", "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": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Table notes open by stating what the table reports. They then describe the monthly decile sorting procedure, the weighting scheme (value- vs equal-weighted panels) and the factor models used. Panels split by weighting or by analysis step.", "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": ["'suggest' when interpreting findings", "'may not' / 'do not necessarily' for the puzzle-resolution claim", "'consistent with' the model", "'largely' qualifiers in the abstract"]}, "claim_evidence": [{"claim": "In a skew-aware SDF, CAPM alphas reflect compensation for residual coskewness that the CAPM ignores", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.A"}, {"claim": "In simulated Merton-type firms, alphas line up with residual coskewness under a skew-aware pricing kernel", "evidence_type": "figure", "evidence_ref": "Figures 3-4"}, {"claim": "Option-implied ex ante skewness is associated with residual coskewness and CAPM alphas in the data", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Controlling for skew factors cuts the BaB CAPM alpha from about 125 to 33 bp per month and makes it insignificant", "evidence_type": "table", "evidence_ref": "Table III; Figure 6"}, {"claim": "The first principal component explains over 90% of variation in low-risk anomaly excess returns", "evidence_type": "table", "evidence_ref": "Table IV; Figure 8"}, {"claim": "Skew factor returns explain up to 95% of the first PC of anomaly returns", "evidence_type": "table", "evidence_ref": "Table V"}], "structural_summary": "The introduction frames beta and volatility anomalies and claims they reflect coskewness compensation. It sketches a quadratic-SDF theory and then previews three empirical blocks with magnitudes before positioning against the skewness literature. Section I derives the theory and runs a simulation. Section II covers data and variable construction. Section III presents three empirical steps: ex ante skewness as a proxy for coskewness, skew-adjusted alphas, and PCA showing a common skew-related driver. Section IV adds robustness and extensions. A short conclusion restates the contribution, and a simulation appendix follows.", "writing_profile_keywords": ["theory-guided empirics", "portfolio sorts", "factor-mimicking portfolios", "option-implied moments", "anomaly explanation", "principal components", "simulation", "figure-heavy"], "confidence": 0.8, "notes": "Section II is rule-labeled theory but is clearly data and variable construction. Section IV mixes robustness checks with extensions and future research. Standard error conventions and significance notation are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 25.17, "usage": {"input": 8508, "output": 3062, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.096066, "request_id": "req_011CggqxmZzxhCNyonDuGrQr", "cached": false, "ts": "2026-11-03T23:15:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/21d16eff8739ebf0b6e86bebade33e5c9f7ed8c6f3ef746efa167d288940a884.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"asset_pricing (investor skill / performance evaluation)\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Cross-sectional variance of a latent LP-specific skill component in private equity fund returns, and the IRR change from a one-standard-deviation increase in that skill.\",\"strategy\":\"Model-free bootstrap tests compare the dispersion of LP persistence and returns with chance. An extended Korteweg–Sorensen Bayesian random-effects model, estimated by MCMC, then quantifies skill. Alternative explanations are probed with LP-type subsamples, public-LP controls, return-distribution quartiles and first-time or follow-on fund samples.\",\"key_assumptions\":[\"Under the null of no skill, LPs draw returns from a common distribution; the bootstrap restrictions mimic actual investment opportunities.\",\"GP abilities are not competed away by fundraising (no strong Berk–Green effect), so GP quality is persistent.\",\"Within an LP type, risk preferences are similar.\",\"The parametric latent-skill structure and its priors are adequate.\"],\"threats_addressed\":[\"Differences in risk tolerance across LPs\",\"Political pressure on public pension funds\",\"Differential access to high-quality GPs (first-time funds, reinvestment)\",\"Chance variation in returns\"],\"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\":2},\"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\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_method\"],\"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\"},{\"idx\":6,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables run from summary statistics to bootstrap comparisons of actual versus simulated distributions, then to Bayesian posterior estimates by period and LP type. Notes describe sample restrictions, panels and model variants (odd versus even columns for Model 1/2), and cross-reference variable definitions in Table V.\",\"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\":[\"suggest/suggesting\",\"potentially reflects\",\"it does not appear that\",\"unlikely to be the main factor\",\"explicit limitations subsection\"]},\"claim_evidence\":[{\"claim\":\"Dispersion of LP persistence in top-half funds exceeds what bootstrapped no-skill samples would produce.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III, Figure 1\"},{\"claim\":\"A one-SD increase in LP skill raises annual IRR by roughly 1-2 pp, and by more in venture funds.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Figure 2\"},{\"claim\":\"The skill effect is stronger in earlier vintages and declines over the sample period.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V panels B-D\"},{\"claim\":\"Risk preferences do not explain the results: variation persists within LP types, and high-skill LPs outperform across the return distribution.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Table VIII, Table X\"},{\"claim\":\"Public pension funds do not earn systematically different returns, which argues against political pressure as the driver.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Skill estimates hold in first-time-fund and reinvestment samples, addressing the access explanation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI, Table XII\"}],\"structural_summary\":\"The paper motivates the question with the economic importance of institutional investors and notes a gap in measuring their skill, focusing on private equity fund selection. After describing a merged LP-investment dataset, it runs model-free bootstrap tests of persistence and return dispersion. It then quantifies skill with an extended Bayesian latent-skill model estimated by MCMC, with a variance decomposition and individual skill estimates. A separate section weighs alternative interpretations (risk tolerance, political pressure, access) and limitations. The conclusion restates findings with hedged interpretation, and an appendix details the MCMC algorithm and priors.\",\"writing_profile_keywords\":[\"investor skill\",\"bootstrap null distribution\",\"Bayesian MCMC\",\"persistence\",\"alternative explanations\",\"private equity LPs\",\"variance decomposition\"],\"confidence\":0.8,\"notes\":\"Primary method is coded 'other' because the design combines bootstrap tests with a Bayesian hierarchical random-effects model (extended Korteweg–Sorensen); structural_estimation is a defensible alternative. Section II is labelled 'theory' by the rule but contains empirical model-free tests, so it is coded as results. No explicit roadmap is visible in the excerpt.\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "asset_pricing (investor skill / performance evaluation)", "primary_method": "other", "identification": {"estimand": "Cross-sectional variance of a latent LP-specific skill component in private equity fund returns, and the IRR change from a one-standard-deviation increase in that skill.", "strategy": "Model-free bootstrap tests compare the dispersion of LP persistence and returns with chance. An extended Korteweg–Sorensen Bayesian random-effects model, estimated by MCMC, then quantifies skill. Alternative explanations are probed with LP-type subsamples, public-LP controls, return-distribution quartiles and first-time or follow-on fund samples.", "key_assumptions": ["Under the null of no skill, LPs draw returns from a common distribution; the bootstrap restrictions mimic actual investment opportunities.", "GP abilities are not competed away by fundraising (no strong Berk–Green effect), so GP quality is persistent.", "Within an LP type, risk preferences are similar.", "The parametric latent-skill structure and its priors are adequate."], "threats_addressed": ["Differences in risk tolerance across LPs", "Political pressure on public pension funds", "Differential access to high-quality GPs (first-time funds, reinvestment)", "Chance variation in returns"], "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": 2}, "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": "identification_defense"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_method"], "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"}, {"idx": 6, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Tables run from summary statistics to bootstrap comparisons of actual versus simulated distributions, then to Bayesian posterior estimates by period and LP type. Notes describe sample restrictions, panels and model variants (odd versus even columns for Model 1/2), and cross-reference variable definitions in Table V.", "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": ["suggest/suggesting", "potentially reflects", "it does not appear that", "unlikely to be the main factor", "explicit limitations subsection"]}, "claim_evidence": [{"claim": "Dispersion of LP persistence in top-half funds exceeds what bootstrapped no-skill samples would produce.", "evidence_type": "table", "evidence_ref": "Table II, Table III, Figure 1"}, {"claim": "A one-SD increase in LP skill raises annual IRR by roughly 1-2 pp, and by more in venture funds.", "evidence_type": "table", "evidence_ref": "Table V, Figure 2"}, {"claim": "The skill effect is stronger in earlier vintages and declines over the sample period.", "evidence_type": "table", "evidence_ref": "Table V panels B-D"}, {"claim": "Risk preferences do not explain the results: variation persists within LP types, and high-skill LPs outperform across the return distribution.", "evidence_type": "table", "evidence_ref": "Table VII, Table VIII, Table X"}, {"claim": "Public pension funds do not earn systematically different returns, which argues against political pressure as the driver.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Skill estimates hold in first-time-fund and reinvestment samples, addressing the access explanation.", "evidence_type": "table", "evidence_ref": "Table XI, Table XII"}], "structural_summary": "The paper motivates the question with the economic importance of institutional investors and notes a gap in measuring their skill, focusing on private equity fund selection. After describing a merged LP-investment dataset, it runs model-free bootstrap tests of persistence and return dispersion. It then quantifies skill with an extended Bayesian latent-skill model estimated by MCMC, with a variance decomposition and individual skill estimates. A separate section weighs alternative interpretations (risk tolerance, political pressure, access) and limitations. The conclusion restates findings with hedged interpretation, and an appendix details the MCMC algorithm and priors.", "writing_profile_keywords": ["investor skill", "bootstrap null distribution", "Bayesian MCMC", "persistence", "alternative explanations", "private equity LPs", "variance decomposition"], "confidence": 0.8, "notes": "Primary method is coded 'other' because the design combines bootstrap tests with a Bayesian hierarchical random-effects model (extended Korteweg–Sorensen); structural_estimation is a defensible alternative. Section II is labelled 'theory' by the rule but contains empirical model-free tests, so it is coded as results. No explicit roadmap is visible in the excerpt."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 26.29, "usage": {"input": 7232, "output": 3008, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.089882, "request_id": "req_011CggqbAHP9NWTM8AiVrTbS", "cached": false, "ts": "2026-11-03T23:10:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/251ceef6deb5062f6ed1c78c28cf0cae1cb8155ec24061e0a15f11980f809994.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of locally experienced house price changes and volatility, and of personal unemployment, on individuals' expectations about national house prices and unemployment\",\"strategy\":\"Cross-sectional regressions of survey expectations on local (ZIP/MSA/state) house price experiences with time fixed effects and demographic controls; within-individual panel variation in employment status for the unemployment experience\",\"key_assumptions\":[\"Local house price variation is not correlated with unobserved determinants of national expectations beyond controls and time fixed effects\",\"Job transitions within the panel are not driven by contemporaneous changes in beliefs about aggregate unemployment\"],\"threats_addressed\":[\"common aggregate shocks absorbed by time fixed effects\",\"measurement error in local house price experience (simulation)\",\"multicollinearity across geographic levels and lags (lasso)\",\"optimal use of limited information (informativeness tests)\",\"risk-adjusted probability reporting (homeownership heterogeneity)\",\"recall versus actual price changes\"],\"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\":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\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"}],\"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\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"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\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes refer each specification to the paper's baseline regression equation. They state the dependent variable and its units (percentage points), the geographic level of the experience measure in each column, the fixed effects and controls included, and the clustering level. Significance stars are used, with standard errors shown in parentheses.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\",\"notes\":\"\",\"uncertainty_note\":\"\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests' for mechanism interpretations\",\"'plausible foundation' and 'one possible explanation' for broader implications\",\"'unlikely to explain' when ruling out alternatives\"]},\"claim_evidence\":[{\"claim\":\"Prior-year local house price changes raise expected U.S. house price changes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 1\"},{\"claim\":\"Personally experiencing unemployment raises the stated likelihood of higher U.S. unemployment by about 1.44 pp\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Higher experienced local price volatility is associated with wider expected national price distributions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The extent of extrapolation does not depend on how informative local prices are for national prices\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, V; Figure 2\"},{\"claim\":\"Less sophisticated respondents extrapolate more\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, X\"},{\"claim\":\"Domain-specific experiences do not shift expectations about other aggregate outcomes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Elicited expectations predict job loss and attitudes toward housing investment\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XII, XIII\"}],\"structural_summary\":\"The introduction motivates the paper with prior work on experience-based expectations and poses questions about what counts as a relevant personal experience. It then previews two designs: local house-price variation and within-person job transitions. Data are described next, followed by a framework separating full-information from limited-information interpretations. House-price results come first, covering prior-year effects, informativeness, geographic levels, weighted histories and volatility. Unemployment results follow, then heterogeneity used to adjudicate between mechanisms, then cross-domain expectations, and finally a check that expectations predict outcomes. The conclusion restates the findings and three contributions. Appendices provide a measurement-error simulation, lasso estimation and recall robustness.\",\"writing_profile_keywords\":[\"survey expectations\",\"extrapolation\",\"experience effects\",\"mechanism discrimination\",\"panel within-person variation\",\"heterogeneity by sophistication\",\"house prices\",\"unemployment\"],\"confidence\":0.8,\"notes\":\"The abstract does not report magnitudes; the intro reports the 1.44 pp unemployment effect. Figure captions are truncated, so whether figures show uncertainty is unclear. No roadmap paragraph appears in the visible intro excerpts.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of locally experienced house price changes and volatility, and of personal unemployment, on individuals' expectations about national house prices and unemployment", "strategy": "Cross-sectional regressions of survey expectations on local (ZIP/MSA/state) house price experiences with time fixed effects and demographic controls; within-individual panel variation in employment status for the unemployment experience", "key_assumptions": ["Local house price variation is not correlated with unobserved determinants of national expectations beyond controls and time fixed effects", "Job transitions within the panel are not driven by contemporaneous changes in beliefs about aggregate unemployment"], "threats_addressed": ["common aggregate shocks absorbed by time fixed effects", "measurement error in local house price experience (simulation)", "multicollinearity across geographic levels and lags (lasso)", "optimal use of limited information (informativeness tests)", "risk-adjusted probability reporting (homeownership heterogeneity)", "recall versus actual price changes"], "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": 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": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}], "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": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"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"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Table notes refer each specification to the paper's baseline regression equation. They state the dependent variable and its units (percentage points), the geographic level of the experience measure in each column, the fixed effects and controls included, and the clustering level. Significance stars are used, with standard errors shown in parentheses.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear", "notes": "", "uncertainty_note": ""}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests' for mechanism interpretations", "'plausible foundation' and 'one possible explanation' for broader implications", "'unlikely to explain' when ruling out alternatives"]}, "claim_evidence": [{"claim": "Prior-year local house price changes raise expected U.S. house price changes", "evidence_type": "table", "evidence_ref": "Table II, Figure 1"}, {"claim": "Personally experiencing unemployment raises the stated likelihood of higher U.S. unemployment by about 1.44 pp", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Higher experienced local price volatility is associated with wider expected national price distributions", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The extent of extrapolation does not depend on how informative local prices are for national prices", "evidence_type": "table", "evidence_ref": "Tables III, V; Figure 2"}, {"claim": "Less sophisticated respondents extrapolate more", "evidence_type": "table", "evidence_ref": "Tables VIII, X"}, {"claim": "Domain-specific experiences do not shift expectations about other aggregate outcomes", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Elicited expectations predict job loss and attitudes toward housing investment", "evidence_type": "table", "evidence_ref": "Tables XII, XIII"}], "structural_summary": "The introduction motivates the paper with prior work on experience-based expectations and poses questions about what counts as a relevant personal experience. It then previews two designs: local house-price variation and within-person job transitions. Data are described next, followed by a framework separating full-information from limited-information interpretations. House-price results come first, covering prior-year effects, informativeness, geographic levels, weighted histories and volatility. Unemployment results follow, then heterogeneity used to adjudicate between mechanisms, then cross-domain expectations, and finally a check that expectations predict outcomes. The conclusion restates the findings and three contributions. Appendices provide a measurement-error simulation, lasso estimation and recall robustness.", "writing_profile_keywords": ["survey expectations", "extrapolation", "experience effects", "mechanism discrimination", "panel within-person variation", "heterogeneity by sophistication", "house prices", "unemployment"], "confidence": 0.8, "notes": "The abstract does not report magnitudes; the intro reports the 1.44 pp unemployment effect. Figure captions are truncated, so whether figures show uncertainty is unclear. No roadmap paragraph appears in the visible intro excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.81, "usage": {"input": 9421, "output": 2272, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.083918, "request_id": "req_011CggqVTdFnMLSuv1RpheyD", "cached": false, "ts": "2026-11-03T23:09:15+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2ae051de4a22914d2a665565c0e97a3971c0e5829b2ff14ee6168ba05caa3043.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Whether the MFFlow measure of mutual fund outflow-induced price pressure is a valid exogenous, nonfundamental shock to stock prices, and its true price impact\",\"strategy\":\"Algebraic decomposition of MFFlow into flow, return and volume components, followed by replication of published regressions and CAAR event paths using each component separately\",\"key_assumptions\":[\"Only the flow-to-stock component reflects true outflow exposure\",\"Return and volume components capture variation from sources other than fund flows\"],\"threats_addressed\":[\"mechanical correlation with realized returns\",\"volume scaling sorting on past returns\",\"small-cap sorting bias in extreme decile\",\"equal-weighting artifacts in apparent reversals\"],\"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\":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\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"other\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":8,\"contribution_types\":[\"replication_or_correction\",\"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\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Tables replicate the original papers' specifications, with the source paper named in the table title. Notes state the estimator (OLS, probit, calendar-time three-factor model) and the clustering scheme, and t-statistics appear in parentheses. The takeover table reports marginal effects.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'may not be entirely mechanical' qualifications\",\"'perhaps' phrasing\",\"acknowledging the original authors' reasonable motivation\",\"'fairly negligible'\"]},\"claim_evidence\":[{\"claim\":\"MFFlow decomposes into three multiplicative terms, only one of which is related to fund flows.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.A\"},{\"claim\":\"Most of the correlation between MFFlow and returns comes from the return and volume components, not from outflow exposure.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Established Discount and takeover results weaken once the components are separated.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-III\"},{\"claim\":\"The apparent postevent reversal reflects constant outperformance from portfolio sorting rather than true reversal.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-3, Table IV\"},{\"claim\":\"The extreme MFFlow decile is disproportionately made up of small-cap stocks by construction.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"}],\"structural_summary\":\"A critique and replication note. It motivates the use of mutual fund flows as an exogenous price shock, then decomposes the standard measure algebraically to show that it mechanically embeds returns and volume. It replicates key published results using the separated components and reinterprets the evidence on postevent reversals. It ends by calling for reevaluation and offering methodological suggestions.\",\"writing_profile_keywords\":[\"critique\",\"decomposition\",\"replication\",\"measurement validity\",\"instrument contamination\",\"mechanical correlation\"],\"confidence\":0.8,\"notes\":\"Article type is corrigendum/replication. Figure uncertainty bands are not visible in the captions. No robustness section is visible; Table V uses simulated outflows.\"}", "json": {"genre": "replication_or_correction", "subfield": "corporate_finance", "subfield_secondary": "asset_pricing", "primary_method": "reduced_form_regression", "identification": {"estimand": "Whether the MFFlow measure of mutual fund outflow-induced price pressure is a valid exogenous, nonfundamental shock to stock prices, and its true price impact", "strategy": "Algebraic decomposition of MFFlow into flow, return and volume components, followed by replication of published regressions and CAAR event paths using each component separately", "key_assumptions": ["Only the flow-to-stock component reflects true outflow exposure", "Return and volume components capture variation from sources other than fund flows"], "threats_addressed": ["mechanical correlation with realized returns", "volume scaling sorting on past returns", "small-cap sorting bias in extreme decile", "equal-weighting artifacts in apparent reversals"], "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": 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": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "tension_or_puzzle"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "other"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 8, "contribution_types": ["replication_or_correction", "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": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Tables replicate the original papers' specifications, with the source paper named in the table title. Notes state the estimator (OLS, probit, calendar-time three-factor model) and the clustering scheme, and t-statistics appear in parentheses. The takeover table reports marginal effects.", "units_stated_in_notes": true, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'may not be entirely mechanical' qualifications", "'perhaps' phrasing", "acknowledging the original authors' reasonable motivation", "'fairly negligible'"]}, "claim_evidence": [{"claim": "MFFlow decomposes into three multiplicative terms, only one of which is related to fund flows.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.A"}, {"claim": "Most of the correlation between MFFlow and returns comes from the return and volume components, not from outflow exposure.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Established Discount and takeover results weaken once the components are separated.", "evidence_type": "table", "evidence_ref": "Tables I-III"}, {"claim": "The apparent postevent reversal reflects constant outperformance from portfolio sorting rather than true reversal.", "evidence_type": "figure", "evidence_ref": "Figures 2-3, Table IV"}, {"claim": "The extreme MFFlow decile is disproportionately made up of small-cap stocks by construction.", "evidence_type": "table", "evidence_ref": "Table V"}], "structural_summary": "A critique and replication note. It motivates the use of mutual fund flows as an exogenous price shock, then decomposes the standard measure algebraically to show that it mechanically embeds returns and volume. It replicates key published results using the separated components and reinterprets the evidence on postevent reversals. It ends by calling for reevaluation and offering methodological suggestions.", "writing_profile_keywords": ["critique", "decomposition", "replication", "measurement validity", "instrument contamination", "mechanical correlation"], "confidence": 0.8, "notes": "Article type is corrigendum/replication. Figure uncertainty bands are not visible in the captions. No robustness section is visible; Table V uses simulated outflows."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.4, "usage": {"input": 6078, "output": 1799, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061086, "request_id": "req_011Cggr3Fzu93A4EptHLDviQ", "cached": false, "ts": "2026-11-03T23:16:23+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2c0624905e39be8e60eab94902a3ad6d2310ca7a6600fd8beeb3e14e31491840.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"asset_pricing (cost of capital measurement); industrial organization / labor share\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Time trends in the labor, capital, and pure-profit shares of gross value added for the U.S. nonfinancial corporate sector over 1984–2014, plus the cross-industry association between changes in concentration and changes in labor share.\",\"strategy\":\"Measurement-based accounting: capital costs are built as the required rate of return times the capital stock, following Hall–Jorgenson, and pure profits are taken as the residual. This is complemented by cross-sectional first-difference regressions of industry labor share changes on concentration changes.\",\"key_assumptions\":[\"Required rate of return is correctly measured from debt and equity costs, including the equity risk premium.\",\"BEA capital stocks, depreciation, and capital inflation are accurately measured.\",\"Increases in concentration proxy for declines in competition and increases in pure profits.\"],\"threats_addressed\":[\"Mismeasured depreciation and capital inflation\",\"Alternative capital measures (inventories, land/real estate)\",\"Omitted or unobserved intangible capital\",\"Alternative equity risk premium assumptions\"],\"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\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_data_or_measure\",\"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\":\"results\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"literature\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state the sample, data sources, unit of observation (six-digit NAICS industry), and sales weighting. Regression tables report standard errors clustered by three-digit NAICS industry. Trend tables report fitted percentage-point changes.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'indicative of'\",\"'consistent with'\",\"'associated with'\",\"'suggest that'\",\"scenario analysis framing\"]},\"claim_evidence\":[{\"claim\":\"Capital share declines over 1984–2014, by about 22% in percentage terms versus 11% for the labor share.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table I\"},{\"claim\":\"Pure profit share rises by about 13.5 pp, worth over $1.2 trillion in 2014.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Section III.C\"},{\"claim\":\"Required rate of return on capital fell substantially, driven by declining debt and equity costs.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1–2\"},{\"claim\":\"Existing measures and plausible scenarios of omitted intangible capital cannot fully explain the rise in pure profits.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Section IV.C\"},{\"claim\":\"Industries with larger concentration increases experience larger labor share declines.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV–VI\"},{\"claim\":\"Pure profit trend is robust to alternative depreciation, capital inflation, and equity risk premium values.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 4–5\"}],\"structural_summary\":\"The paper opens with the declining labor share and the prevailing labor-for-capital substitution explanations, then states the counterclaim up front. It distinguishes capital costs from pure profits and explains why capital costs are hard to measure. A framework section builds Hall–Jorgenson capital costs, followed by a data section on national accounts and debt and equity costs of capital. Results report fitted trends showing falling capital and labor shares and rising pure profits, with a magnitude discussion. An extensive robustness section varies depreciation and inflation, capital measures, intangible-capital scenarios, and the equity risk premium. Industry concentration regressions supply supporting reduced-form evidence. A long discussion and literature section precedes a conclusion that restates the findings and attributes them to declining competition.\",\"writing_profile_keywords\":[\"measurement-driven\",\"accounting decomposition\",\"counter-narrative to prevailing view\",\"dollar-magnitude framing\",\"scenario analysis\",\"single-author first person\",\"macro-finance\"],\"confidence\":0.8,\"notes\":\"The core contribution is measurement rather than causal identification; the concentration regressions are explicitly described as reduced-form and associational. The conclusion attributes the trends to declining competition with 'suggest' wording. Statistical significance conventions are not visible in the excerpts. No roadmap is visible in the truncated introduction.\"}", "json": {"genre": "empirical", "subfield": "macro_finance", "subfield_secondary": "asset_pricing (cost of capital measurement); industrial organization / labor share", "primary_method": "other", "identification": {"estimand": "Time trends in the labor, capital, and pure-profit shares of gross value added for the U.S. nonfinancial corporate sector over 1984–2014, plus the cross-industry association between changes in concentration and changes in labor share.", "strategy": "Measurement-based accounting: capital costs are built as the required rate of return times the capital stock, following Hall–Jorgenson, and pure profits are taken as the residual. This is complemented by cross-sectional first-difference regressions of industry labor share changes on concentration changes.", "key_assumptions": ["Required rate of return is correctly measured from debt and equity costs, including the equity risk premium.", "BEA capital stocks, depreciation, and capital inflation are accurately measured.", "Increases in concentration proxy for declines in competition and increases in pure profits."], "threats_addressed": ["Mismeasured depreciation and capital inflation", "Alternative capital measures (inventories, land/real estate)", "Omitted or unobserved intangible capital", "Alternative equity risk premium assumptions"], "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": "motivation"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"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": "identification_defense"}, {"paragraph_index": 5, "move": "robustness_mention"}], "contribution_paragraph_index": 1, "contribution_types": ["new_data_or_measure", "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": "results"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "literature"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Notes state the sample, data sources, unit of observation (six-digit NAICS industry), and sales weighting. Regression tables report standard errors clustered by three-digit NAICS industry. Trend tables report fitted percentage-point changes.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'indicative of'", "'consistent with'", "'associated with'", "'suggest that'", "scenario analysis framing"]}, "claim_evidence": [{"claim": "Capital share declines over 1984–2014, by about 22% in percentage terms versus 11% for the labor share.", "evidence_type": "figure", "evidence_ref": "Figure 3; Table I"}, {"claim": "Pure profit share rises by about 13.5 pp, worth over $1.2 trillion in 2014.", "evidence_type": "table", "evidence_ref": "Table I; Section III.C"}, {"claim": "Required rate of return on capital fell substantially, driven by declining debt and equity costs.", "evidence_type": "figure", "evidence_ref": "Figures 1–2"}, {"claim": "Existing measures and plausible scenarios of omitted intangible capital cannot fully explain the rise in pure profits.", "evidence_type": "table", "evidence_ref": "Table II; Section IV.C"}, {"claim": "Industries with larger concentration increases experience larger labor share declines.", "evidence_type": "table", "evidence_ref": "Tables IV–VI"}, {"claim": "Pure profit trend is robust to alternative depreciation, capital inflation, and equity risk premium values.", "evidence_type": "figure", "evidence_ref": "Figures 4–5"}], "structural_summary": "The paper opens with the declining labor share and the prevailing labor-for-capital substitution explanations, then states the counterclaim up front. It distinguishes capital costs from pure profits and explains why capital costs are hard to measure. A framework section builds Hall–Jorgenson capital costs, followed by a data section on national accounts and debt and equity costs of capital. Results report fitted trends showing falling capital and labor shares and rising pure profits, with a magnitude discussion. An extensive robustness section varies depreciation and inflation, capital measures, intangible-capital scenarios, and the equity risk premium. Industry concentration regressions supply supporting reduced-form evidence. A long discussion and literature section precedes a conclusion that restates the findings and attributes them to declining competition.", "writing_profile_keywords": ["measurement-driven", "accounting decomposition", "counter-narrative to prevailing view", "dollar-magnitude framing", "scenario analysis", "single-author first person", "macro-finance"], "confidence": 0.8, "notes": "The core contribution is measurement rather than causal identification; the concentration regressions are explicitly described as reduced-form and associational. The conclusion attributes the trends to declining competition with 'suggest' wording. Statistical significance conventions are not visible in the excerpts. No roadmap is visible in the truncated introduction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.9, "usage": {"input": 6730, "output": 2182, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071354, "request_id": "req_011Cggqw94mEByYhVxRLQPCg", "cached": false, "ts": "2026-11-03T23:15:04+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/2e7f7ee4ec3461e390ee7997fe53115493f9a24844b6d088ca5ec88892f731a8.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"How well parameter-free earnings surprise measures (CE vs FOM) approximate an ideal bias-adjusted surprise and explain announcement and post-announcement returns\",\"strategy\":\"Analytical model that derives relative-efficiency bounds, followed by comparison of explanatory power in canonical event-study regressions of CAR/POSTCAR on CE, Rank(CE), FOM and indicators, including model selection\",\"key_assumptions\":[\"A subset of analyst forecasts is biased\",\"Market expectation is a filter of individual forecasts\",\"Abnormal returns reflect earnings surprise\"],\"threats_addressed\":[\"outliers in forecasts\",\"stale forecasts / walk-down effects\",\"earnings persistence\",\"forecast dispersion\",\"international external validity\"],\"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\":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\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_data_or_measure\",\"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\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"other\"},{\"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\":\"within_results\",\"types\":[\"alternative_measure\",\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variables (CAR over the three-day window, POSTCAR over trading days 2–126) and the surprise measures. Regressions include year dummies. Results are reported as coefficients with t-statistics and R2. The visible notes do not state the standard-error method. The first table reports numerical efficiency bounds from the model.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'potentially superior'\",\"'where bias is potentially large'\",\"conditional framing ('if a subset of forecasts can be biased')\"]},\"claim_evidence\":[{\"claim\":\"CE is a poor parameter-free approximation of the ideal bias-adjusted surprise measure; FOM approximates it far better\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendix A\"},{\"claim\":\"Lower bound on FOM's relative efficiency versus the infeasible ideal can reach about 50% under plausible parameters\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"FOM explains announcement returns much better than CE; a 1-SD rise in FOM raises CAR by about 1.5%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"FOM also outperforms in post-announcement drift regressions and in international stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V\"},{\"claim\":\"Model selection favors FOM-based specifications\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"FOM beats parameter-dependent bias-adjusted CE in explaining CAR\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"The paper opens with the standard use of the consensus error in event studies and argues that its validity rests on unbiased forecasts, which analyst bias undermines. It then proposes the fraction-of-misses measure and builds a model showing that the ideal surprise measure is a nonlinear filter downweighting large errors. From the model it derives efficiency bounds and comparative statics. Empirically, it compares CE and FOM in canonical CAR/POSTCAR regressions on I/B/E/S data, with international, model-selection, timing, subsample and parameter-dependent checks inside the results section. Proofs are in the appendix, and the conclusion restates the contribution and its broader applicability to other forecasts.\",\"writing_profile_keywords\":[\"measurement\",\"earnings surprise\",\"event study\",\"analyst bias\",\"theory-then-evidence\",\"parameter-free measure\",\"relative efficiency\"],\"confidence\":0.8,\"notes\":\"Sections idx 4 and 5 appear to be equation fragments mis-parsed as headings. The introduction excerpt is fragmented, with paragraphs 1–2 being parts of the FOM formula and its explanation. No explicit roadmap is visible.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "How well parameter-free earnings surprise measures (CE vs FOM) approximate an ideal bias-adjusted surprise and explain announcement and post-announcement returns", "strategy": "Analytical model that derives relative-efficiency bounds, followed by comparison of explanatory power in canonical event-study regressions of CAR/POSTCAR on CE, Rank(CE), FOM and indicators, including model selection", "key_assumptions": ["A subset of analyst forecasts is biased", "Market expectation is a filter of individual forecasts", "Abnormal returns reflect earnings surprise"], "threats_addressed": ["outliers in forecasts", "stale forecasts / walk-down effects", "earnings persistence", "forecast dispersion", "international external validity"], "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": 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": "related_literature_positioning"}, {"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 2, "contribution_types": ["new_data_or_measure", "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": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "other"}, {"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": "within_results", "types": ["alternative_measure", "subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes define the dependent variables (CAR over the three-day window, POSTCAR over trading days 2–126) and the surprise measures. Regressions include year dummies. Results are reported as coefficients with t-statistics and R2. The visible notes do not state the standard-error method. The first table reports numerical efficiency bounds from the model.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'potentially superior'", "'where bias is potentially large'", "conditional framing ('if a subset of forecasts can be biased')"]}, "claim_evidence": [{"claim": "CE is a poor parameter-free approximation of the ideal bias-adjusted surprise measure; FOM approximates it far better", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendix A"}, {"claim": "Lower bound on FOM's relative efficiency versus the infeasible ideal can reach about 50% under plausible parameters", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "FOM explains announcement returns much better than CE; a 1-SD rise in FOM raises CAR by about 1.5%", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "FOM also outperforms in post-announcement drift regressions and in international stocks", "evidence_type": "table", "evidence_ref": "Tables IV, V"}, {"claim": "Model selection favors FOM-based specifications", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "FOM beats parameter-dependent bias-adjusted CE in explaining CAR", "evidence_type": "table", "evidence_ref": "Table X"}], "structural_summary": "The paper opens with the standard use of the consensus error in event studies and argues that its validity rests on unbiased forecasts, which analyst bias undermines. It then proposes the fraction-of-misses measure and builds a model showing that the ideal surprise measure is a nonlinear filter downweighting large errors. From the model it derives efficiency bounds and comparative statics. Empirically, it compares CE and FOM in canonical CAR/POSTCAR regressions on I/B/E/S data, with international, model-selection, timing, subsample and parameter-dependent checks inside the results section. Proofs are in the appendix, and the conclusion restates the contribution and its broader applicability to other forecasts.", "writing_profile_keywords": ["measurement", "earnings surprise", "event study", "analyst bias", "theory-then-evidence", "parameter-free measure", "relative efficiency"], "confidence": 0.8, "notes": "Sections idx 4 and 5 appear to be equation fragments mis-parsed as headings. The introduction excerpt is fragmented, with paragraphs 1–2 being parts of the FOM formula and its explanation. No explicit roadmap is visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.66, "usage": {"input": 5907, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064662, "request_id": "req_011CggqJcWV9Jv5pNpJxSFhb", "cached": false, "ts": "2026-11-03T23:06:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/31f696fc4109ce47f84d1b10256f9c21c5bf19871d1f54cb252c9ec66c3819d8.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\":\"Structural preference and refinancing transaction-cost parameters of a household liquidity-management model, and the model's ability to reproduce untargeted cross-sectional and time-series patterns.\",\"strategy\":\"Simulated method of moments (indirect inference). Exogenous state dynamics are estimated from data, institutional parameters are calibrated, and preference and cost parameters are estimated by matching household balance-sheet and refinancing moments. The model is then evaluated on untargeted moments, with historical paths of aggregate shocks fed in.\",\"key_assumptions\":[\"House prices, aggregate income and interest rates are exogenous (partial equilibrium)\",\"LTV and LTI limits follow conforming lending standards\",\"Households have no ex ante heterogeneity\",\"Idiosyncratic labor income risk is countercyclical\"],\"threats_addressed\":[\"Interest-rate-driven refinancing as an alternative explanation for countercyclical cash-outs\",\"Competing explanations: self-control problems, optimistic expectations, lender moral hazard\",\"Lending standards looser than the baseline (model treated as a lower bound; constraints relaxed)\"],\"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\":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\":\"preview_of_results\"},{\"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\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"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\":\"empirical_design\"},{\"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\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Time-series regressions report standard errors in parentheses. Parameter tables are split into exogenously set and estimated panels. The moment-matching table shows data, model and SE columns side by side. Counterfactual tables vary one model feature per column.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'consistent with' / 'suggests'\",\"model framed as a lower bound\",\"partial-equilibrium caveat\",\"explicit list of model shortcomings in the conclusion\"]},\"claim_evidence\":[{\"claim\":\"Cash-out refinancing has a strongly countercyclical component not explained by interest rate movements.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Figure 1\"},{\"claim\":\"The estimated model matches targeted household balance-sheet and consumption moments.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Income risk and collateral vs debt-service constraints materially change model consumption volatility and homeownership.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"High-leverage households experience larger consumption declines after the crash, in line with PSID estimates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 8\"},{\"claim\":\"The model reproduces the cross-sectional refinancing and balance-sheet patterns seen in SCF and PSID data.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 3-5, 7\"},{\"claim\":\"The model replicates the Mian and Sufi (2011) debt growth patterns across regions.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"}],\"structural_summary\":\"The paper opens with a stylized-facts section on aggregate refinancing and equity extraction. It then builds a household model with income risk, long- and short-term mortgages and LTV/LTI constraints. Parameters are estimated by simulated moments, followed by counterfactual sensitivity exercises. The model is then evaluated on untargeted cross-sectional and boom-bust patterns. The conclusion draws policy lessons and lists where the model misfits the data.\",\"writing_profile_keywords\":[\"structural estimation\",\"household liquidity\",\"mortgage refinancing\",\"model-vs-data comparison\",\"untargeted moments\",\"housing boom-bust\",\"partial equilibrium\"],\"confidence\":0.8,\"notes\":\"Section IV is flagged 'other' by the rule but functions as model validation and results. Statistical-significance language is unclear beyond the Newey-West SEs reported in Table I.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "macro_finance", "primary_method": "structural_estimation", "identification": {"estimand": "Structural preference and refinancing transaction-cost parameters of a household liquidity-management model, and the model's ability to reproduce untargeted cross-sectional and time-series patterns.", "strategy": "Simulated method of moments (indirect inference). Exogenous state dynamics are estimated from data, institutional parameters are calibrated, and preference and cost parameters are estimated by matching household balance-sheet and refinancing moments. The model is then evaluated on untargeted moments, with historical paths of aggregate shocks fed in.", "key_assumptions": ["House prices, aggregate income and interest rates are exogenous (partial equilibrium)", "LTV and LTI limits follow conforming lending standards", "Households have no ex ante heterogeneity", "Idiosyncratic labor income risk is countercyclical"], "threats_addressed": ["Interest-rate-driven refinancing as an alternative explanation for countercyclical cash-outs", "Competing explanations: self-control problems, optimistic expectations, lender moral hazard", "Lending standards looser than the baseline (model treated as a lower bound; constraints relaxed)"], "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": 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": "preview_of_results"}, {"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": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 0, "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": "empirical_design"}, {"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": "within_results", "types": ["sensitivity_to_parameters", "alternative_specification"]}, "tables": {"convention_notes": "Time-series regressions report standard errors in parentheses. Parameter tables are split into exogenously set and estimated panels. The moment-matching table shows data, model and SE columns side by side. Counterfactual tables vary one model feature per column.", "units_stated_in_notes": true, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'consistent with' / 'suggests'", "model framed as a lower bound", "partial-equilibrium caveat", "explicit list of model shortcomings in the conclusion"]}, "claim_evidence": [{"claim": "Cash-out refinancing has a strongly countercyclical component not explained by interest rate movements.", "evidence_type": "table", "evidence_ref": "Table I; Figure 1"}, {"claim": "The estimated model matches targeted household balance-sheet and consumption moments.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Income risk and collateral vs debt-service constraints materially change model consumption volatility and homeownership.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "High-leverage households experience larger consumption declines after the crash, in line with PSID estimates.", "evidence_type": "table", "evidence_ref": "Table VI; Figure 8"}, {"claim": "The model reproduces the cross-sectional refinancing and balance-sheet patterns seen in SCF and PSID data.", "evidence_type": "figure", "evidence_ref": "Figures 3-5, 7"}, {"claim": "The model replicates the Mian and Sufi (2011) debt growth patterns across regions.", "evidence_type": "figure", "evidence_ref": "Figure 9"}], "structural_summary": "The paper opens with a stylized-facts section on aggregate refinancing and equity extraction. It then builds a household model with income risk, long- and short-term mortgages and LTV/LTI constraints. Parameters are estimated by simulated moments, followed by counterfactual sensitivity exercises. The model is then evaluated on untargeted cross-sectional and boom-bust patterns. The conclusion draws policy lessons and lists where the model misfits the data.", "writing_profile_keywords": ["structural estimation", "household liquidity", "mortgage refinancing", "model-vs-data comparison", "untargeted moments", "housing boom-bust", "partial equilibrium"], "confidence": 0.8, "notes": "Section IV is flagged 'other' by the rule but functions as model validation and results. Statistical-significance language is unclear beyond the Newey-West SEs reported in Table I."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.15, "usage": {"input": 8125, "output": 2148, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076254, "request_id": "req_011Cggqdu7ee5y7fCCxfDcH1", "cached": false, "ts": "2026-11-03T23:11:08+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/36a4aa112068cd189c63c919702409b72f4a3ee5fa862e4a8ef19c8b36e040c2.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"machine_learning_textual\",\"identification\":{\"estimand\":\"Predictive relation between year-over-year textual similarity of firms' 10-K/10-Q filings and their subsequent abnormal returns and real outcomes.\",\"strategy\":\"Textual similarity measures (cosine, Jaccard, minimum edit distance, simple comparison) feed calendar-time quintile portfolio sorts with factor alphas and Fama-MacBeth regressions with known return predictors as controls. Interactions with SEC EDGAR download-based attention proxies probe the inattention mechanism.\",\"key_assumptions\":[\"Factor models and controls adequately capture risk\",\"Joint downloading of current and prior 10-Ks proxies for investors comparing filings\",\"Return predictability is not due to coincident news or firm events\"],\"threats_addressed\":[\"traditional risk factors\",\"known return predictors\",\"earnings surprises and coincident news\",\"special firm events such as M&A and SEOs\",\"document length and length changes\",\"transaction costs and limits to arbitrage\",\"small-firm concentration\",\"overreaction and reversal\"],\"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\":0},\"magnitude\":{\"present\":true,\"sentence_index\":1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"motivation\"},{\"paragraph_index\":4,\"move\":\"other\"},{\"paragraph_index\":5,\"move\":\"other\"},{\"paragraph_index\":6,\"move\":\"other\"},{\"paragraph_index\":7,\"move\":\"other\"},{\"paragraph_index\":8,\"move\":\"other\"},{\"paragraph_index\":9,\"move\":\"preview_of_approach\"},{\"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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":15,\"move\":\"robustness_mention\"},{\"paragraph_index\":16,\"move\":\"robustness_mention\"},{\"paragraph_index\":17,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":18,\"move\":\"preview_of_approach\"},{\"paragraph_index\":19,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":2,\"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\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define variables and similarity measures, state that returns are multiplied by 100, and name the factor models used. Market value of equity is reported in $1,000. Panels cover portfolio alphas and Fama-MacBeth estimates.\",\"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\":[\"suggesting\",\"consistent with\",\"perhaps\",\"plausibly\",\"appears to\",\"suggestive evidence\"]},\"claim_evidence\":[{\"claim\":\"Average 10-K length and the number of year-over-year textual changes grew sharply from 1995 to 2014.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Baxter's 2009 10-K changes foreshadowed recall news; the price did not react at filing but fell later.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-5\"},{\"claim\":\"A long-nonchangers, short-changers portfolio earns significant value-weighted abnormal returns that persist up to 18 months without reversal.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 7\"},{\"claim\":\"Similarity predicts returns after controlling for known return predictors.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, XI\"},{\"claim\":\"Changes in the Risk Factors section yield five-factor alphas over 188 bp per month.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Figure 9\"},{\"claim\":\"Return predictability is attenuated when more investors jointly download current and prior 10-Ks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Filing changes predict lower future operating income, net income and sales.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Changer firms are larger and cheaper to short than nonchangers, arguing against limits to arbitrage.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"}],\"structural_summary\":\"Opens with a Grossman-Stiglitz framing of rising information-processing costs. Challenges the prior conclusion that 10-Ks have become less informative. Uses a detailed single-firm case (Baxter) to illustrate the pattern before generalizing to the full cross-section. Previews portfolio alphas, the absence of an announcement effect, section-level mechanism results, real effects, an asymmetry explanation, robustness checks and novel SEC download-based attention data. The body runs literature, then data, then main return results (portfolio sorts and Fama-MacBeth regressions), then an extended mechanism section that also contains robustness checks. The conclusion restates the inattention interpretation and suggests other document types for future research.\",\"writing_profile_keywords\":[\"textual analysis\",\"investor inattention\",\"return predictability\",\"anecdote-driven motivation\",\"calendar-time portfolios\",\"Fama-MacBeth\",\"novel FOIA data\",\"no announcement effect\"],\"confidence\":0.85,\"notes\":\"The intro and conclusion report portfolio returns of roughly 30 to 58 bp per month, while the abstract cites 'up to 188' bp, which comes from the Risk Factors section sort. Section IV, labeled 'Mechanism', also contains robustness checks.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "machine_learning_textual", "identification": {"estimand": "Predictive relation between year-over-year textual similarity of firms' 10-K/10-Q filings and their subsequent abnormal returns and real outcomes.", "strategy": "Textual similarity measures (cosine, Jaccard, minimum edit distance, simple comparison) feed calendar-time quintile portfolio sorts with factor alphas and Fama-MacBeth regressions with known return predictors as controls. Interactions with SEC EDGAR download-based attention proxies probe the inattention mechanism.", "key_assumptions": ["Factor models and controls adequately capture risk", "Joint downloading of current and prior 10-Ks proxies for investors comparing filings", "Return predictability is not due to coincident news or firm events"], "threats_addressed": ["traditional risk factors", "known return predictors", "earnings surprises and coincident news", "special firm events such as M&A and SEOs", "document length and length changes", "transaction costs and limits to arbitrage", "small-firm concentration", "overreaction and reversal"], "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": 0}, "magnitude": {"present": true, "sentence_index": 1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "motivation"}, {"paragraph_index": 4, "move": "other"}, {"paragraph_index": 5, "move": "other"}, {"paragraph_index": 6, "move": "other"}, {"paragraph_index": 7, "move": "other"}, {"paragraph_index": 8, "move": "other"}, {"paragraph_index": 9, "move": "preview_of_approach"}, {"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": "mechanism_or_theory_sketch"}, {"paragraph_index": 15, "move": "robustness_mention"}, {"paragraph_index": 16, "move": "robustness_mention"}, {"paragraph_index": 17, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 18, "move": "preview_of_approach"}, {"paragraph_index": 19, "move": "roadmap"}], "contribution_paragraph_index": 2, "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": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Table notes define variables and similarity measures, state that returns are multiplied by 100, and name the factor models used. Market value of equity is reported in $1,000. Panels cover portfolio alphas and Fama-MacBeth estimates.", "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": ["suggesting", "consistent with", "perhaps", "plausibly", "appears to", "suggestive evidence"]}, "claim_evidence": [{"claim": "Average 10-K length and the number of year-over-year textual changes grew sharply from 1995 to 2014.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Baxter's 2009 10-K changes foreshadowed recall news; the price did not react at filing but fell later.", "evidence_type": "figure", "evidence_ref": "Figures 2-5"}, {"claim": "A long-nonchangers, short-changers portfolio earns significant value-weighted abnormal returns that persist up to 18 months without reversal.", "evidence_type": "table", "evidence_ref": "Table II, Figure 7"}, {"claim": "Similarity predicts returns after controlling for known return predictors.", "evidence_type": "table", "evidence_ref": "Tables IV, XI"}, {"claim": "Changes in the Risk Factors section yield five-factor alphas over 188 bp per month.", "evidence_type": "table", "evidence_ref": "Table VII, Figure 9"}, {"claim": "Return predictability is attenuated when more investors jointly download current and prior 10-Ks.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Filing changes predict lower future operating income, net income and sales.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Changer firms are larger and cheaper to short than nonchangers, arguing against limits to arbitrage.", "evidence_type": "table", "evidence_ref": "Table III"}], "structural_summary": "Opens with a Grossman-Stiglitz framing of rising information-processing costs. Challenges the prior conclusion that 10-Ks have become less informative. Uses a detailed single-firm case (Baxter) to illustrate the pattern before generalizing to the full cross-section. Previews portfolio alphas, the absence of an announcement effect, section-level mechanism results, real effects, an asymmetry explanation, robustness checks and novel SEC download-based attention data. The body runs literature, then data, then main return results (portfolio sorts and Fama-MacBeth regressions), then an extended mechanism section that also contains robustness checks. The conclusion restates the inattention interpretation and suggests other document types for future research.", "writing_profile_keywords": ["textual analysis", "investor inattention", "return predictability", "anecdote-driven motivation", "calendar-time portfolios", "Fama-MacBeth", "novel FOIA data", "no announcement effect"], "confidence": 0.85, "notes": "The intro and conclusion report portfolio returns of roughly 30 to 58 bp per month, while the abstract cites 'up to 188' bp, which comes from the Risk Factors section sort. Section IV, labeled 'Mechanism', also contains robustness checks."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.43, "usage": {"input": 11785, "output": 2435, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.096634, "request_id": "req_011Cggqo4KnaQ5nnwqRUkZHQ", "cached": false, "ts": "2026-11-03T23:13:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3a15c72ab0f87b59a3fa6b1dd0c38cc64d0c7cd083ff40d473ecbb8aaeaacd8c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"institutional investors / high-frequency trading\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between HFT signed net flow and the presence, duration and information content of institutional meta orders, plus investor execution costs and HFT revenue\",\"strategy\":\"Proprietary institutional child-order data linked to HFT trades identified by name; descriptive event-time patterns, a matched placebo sample with similar market conditions, and panel regressions with controls\",\"key_assumptions\":[\"Placebo periods matched on volume, returns, volatility and order imbalance capture common market conditions\",\"Permanent price impact proxies the information content of orders\",\"NASDAQ OMX flow is representative because offsetting positions in other venues are unlikely\"],\"threats_addressed\":[\"Common response to market conditions, addressed with the matched placebo sample\",\"Offsetting positions in other markets or correlated securities\",\"Volume-clock vs wall-clock time\",\"Market share of the venue\",\"HFT positioning before the first child trade\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"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\":3},\"magnitude\":{\"present\":false,\"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\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":19,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":20,\"move\":\"contribution_statement\"},{\"paragraph_index\":21,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":22,\"move\":\"contribution_statement\"},{\"paragraph_index\":23,\"move\":\"other\"}],\"contribution_paragraph_index\":20,\"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\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"internet_appendix\",\"types\":[\"alternative_measure\",\"placebo\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the sample and observation unit (stock-period meta orders), how variables are constructed and signed (sell orders converted to pseudo-buys), and their standardization and units (dollars, % of shares outstanding). Subsample splits are by order size and duration; odd hours are reported for brevity. 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\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"appear to / seem to\",\"consistent with\",\"explicit 'Some Words of Caution' paragraph disclaiming causality\",\"we believe / we suspect\",\"suggests\"]},\"claim_evidence\":[{\"claim\":\"HFTs initially trade against institutional orders, then trade with long-lasting orders after about seven hours, building speculative positions\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2, Table III\"},{\"claim\":\"With-wind trading is absent in matched placebo periods, linking it to the presence of institutional orders\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3, Table II\"},{\"claim\":\"Institutional orders are mostly information-motivated; institutions trade few names per day\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4, Figure 5, Table V\"},{\"claim\":\"Half of investors' private signal is revealed in prices after roughly three hours\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"},{\"claim\":\"HFTs trade with-wind most strongly on the most informed long-lasting orders\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"Investors cut their volume participation by about 50% after HFTs turn with-wind\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"A one-SD rise in HFT with-wind flow is associated with $4,480 higher execution cost and $850 higher HFT gross profit\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Figure 8\"}],\"structural_summary\":\"The paper opens with HFT background and a gap: little is known about HFT interaction with institutional investors. It sets up three theoretical benchmarks (market-making, predatory trading, back-running) and then describes the linked proprietary and public data. The core results show against-wind then with-wind HFT flow, validated with a placebo match. Information content of orders is used to discriminate between predatory trading and back-running. A final section derives four predictions from an economic narrative and tests each one. Robustness checks are placed in the internet appendix. The introduction contains explicit literature differentiation and a caution paragraph disclaiming causality. The conclusion reframes HFTs as high-tech traders and calls for more granular regulatory data.\",\"writing_profile_keywords\":[\"theory-benchmark comparison\",\"proprietary data\",\"placebo matching\",\"hypothesis discrimination\",\"explicit caveats\",\"bulleted findings\",\"economic narrative\",\"policy discussion\"],\"confidence\":0.82,\"notes\":\"Paragraphs 6-18 of the introduction are fragments of a bullet list (findings). The excerpts do not show standard errors or how statistical significance is reported.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "institutional investors / high-frequency trading", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between HFT signed net flow and the presence, duration and information content of institutional meta orders, plus investor execution costs and HFT revenue", "strategy": "Proprietary institutional child-order data linked to HFT trades identified by name; descriptive event-time patterns, a matched placebo sample with similar market conditions, and panel regressions with controls", "key_assumptions": ["Placebo periods matched on volume, returns, volatility and order imbalance capture common market conditions", "Permanent price impact proxies the information content of orders", "NASDAQ OMX flow is representative because offsetting positions in other venues are unlikely"], "threats_addressed": ["Common response to market conditions, addressed with the matched placebo sample", "Offsetting positions in other markets or correlated securities", "Volume-clock vs wall-clock time", "Market share of the venue", "HFT positioning before the first child trade"], "causal_language_strength": 2, "uses_causal_claim": false}, "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": 3}, "magnitude": {"present": false, "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": 3, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 19, "move": "related_literature_positioning"}, {"paragraph_index": 20, "move": "contribution_statement"}, {"paragraph_index": 21, "move": "related_literature_positioning"}, {"paragraph_index": 22, "move": "contribution_statement"}, {"paragraph_index": 23, "move": "other"}], "contribution_paragraph_index": 20, "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": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "internet_appendix", "types": ["alternative_measure", "placebo", "other"]}, "tables": {"convention_notes": "Table notes define the sample and observation unit (stock-period meta orders), how variables are constructed and signed (sell orders converted to pseudo-buys), and their standardization and units (dollars, % of shares outstanding). Subsample splits are by order size and duration; odd hours are reported for brevity. 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": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["appear to / seem to", "consistent with", "explicit 'Some Words of Caution' paragraph disclaiming causality", "we believe / we suspect", "suggests"]}, "claim_evidence": [{"claim": "HFTs initially trade against institutional orders, then trade with long-lasting orders after about seven hours, building speculative positions", "evidence_type": "figure", "evidence_ref": "Figure 2, Table III"}, {"claim": "With-wind trading is absent in matched placebo periods, linking it to the presence of institutional orders", "evidence_type": "figure", "evidence_ref": "Figure 3, Table II"}, {"claim": "Institutional orders are mostly information-motivated; institutions trade few names per day", "evidence_type": "figure", "evidence_ref": "Figure 4, Figure 5, Table V"}, {"claim": "Half of investors' private signal is revealed in prices after roughly three hours", "evidence_type": "figure", "evidence_ref": "Figure 6"}, {"claim": "HFTs trade with-wind most strongly on the most informed long-lasting orders", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "Investors cut their volume participation by about 50% after HFTs turn with-wind", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "A one-SD rise in HFT with-wind flow is associated with $4,480 higher execution cost and $850 higher HFT gross profit", "evidence_type": "table", "evidence_ref": "Table VII, Figure 8"}], "structural_summary": "The paper opens with HFT background and a gap: little is known about HFT interaction with institutional investors. It sets up three theoretical benchmarks (market-making, predatory trading, back-running) and then describes the linked proprietary and public data. The core results show against-wind then with-wind HFT flow, validated with a placebo match. Information content of orders is used to discriminate between predatory trading and back-running. A final section derives four predictions from an economic narrative and tests each one. Robustness checks are placed in the internet appendix. The introduction contains explicit literature differentiation and a caution paragraph disclaiming causality. The conclusion reframes HFTs as high-tech traders and calls for more granular regulatory data.", "writing_profile_keywords": ["theory-benchmark comparison", "proprietary data", "placebo matching", "hypothesis discrimination", "explicit caveats", "bulleted findings", "economic narrative", "policy discussion"], "confidence": 0.82, "notes": "Paragraphs 6-18 of the introduction are fragments of a bullet list (findings). The excerpts do not show standard errors or how statistical significance is reported."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.69, "usage": {"input": 9380, "output": 2241, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.083134, "request_id": "req_011CggqKjnp2Bz2RepMqRRRb", "cached": false, "ts": "2026-11-03T23:07:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3b2893a637e63e56da9722ba3fb0c87c0fad1ef4d6b8ec1d1059c7df77d69a95.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"financial_intermediation_banking; OTC derivatives regulation\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Differences in transaction costs (effective half-spreads), price impacts and dealer profits between dealer-to-client and interdealer index CDS trades, and the information content of each segment for price discovery.\",\"strategy\":\"Descriptive decomposition of effective spreads into realized spread and price impact. OLS regressions control for trade characteristics and market conditions. A cointegrated VAR/VECM in the spirit of Hasbrouck measures permanent price impact and price discovery. Comparisons are made across trading protocols and against contemporaneous interdealer quotes.\",\"key_assumptions\":[\"Mid-quotes (Markit composites for D2C, GFI executable quotes for D2D) proxy for efficient prices\",\"The SEF on which a trade occurs correctly identifies whether it is D2C or D2D\",\"Algorithms correctly identify package transactions and SEFs\",\"A 15-minute horizon captures price impact\"],\"threats_addressed\":[\"Selection bias from differing trade characteristics and market conditions across segments\",\"Contamination from package trades (rolls, curve trades, delta hedges)\"],\"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\":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\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":9,\"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\":\"results\"},{\"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\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes define variables with abbreviations (EffcSprd, RlzdSprd, PrcImp), state the sample period and sample sizes, and split panels by index (CDX.IG vs CDX.HY). Means are reported by trade-size quartile.\",\"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' / 'appear to'\",\"'consistent with' a theoretical model\",\"explicit statement that welfare effects cannot be addressed empirically\",\"'suggestive evidence'\"]},\"claim_evidence\":[{\"claim\":\"Client order flow comprises few, large trades, with D2C volume far exceeding D2D volume.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"D2C transaction costs exceed D2D costs, but the gap is fully explained by higher price impact.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II and III\"},{\"claim\":\"Price impact of client trades is largely permanent and price discovery occurs mainly in the D2C segment.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV, Figure 4\"},{\"claim\":\"Mid-market matches have lower costs and price impacts than limit order book trades, consistent with Zhu's (2014) venue selection model.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V and VI, Figure 5\"},{\"claim\":\"D2C trades almost always improve on contemporaneous interdealer inside quotes.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Client transaction costs are widely dispersed, consistent with dealers price discriminating on perceived price impact.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"}],\"structural_summary\":\"The introduction motivates the paper with the importance of the index CDS market and the Dodd-Frank SEF mandate as a test of market structure. It then previews data and algorithms and walks sequentially through findings with numeric magnitudes, closing with related literature. The body describes institutional detail and data/identification algorithms, then builds the empirical results in sequence: a spread decomposition with regression controls, a VECM for price discovery, interdealer protocol segmentation, price improvement, and dispersion of client costs. The conclusion summarizes and interprets the two-tiered structure as a consequence of client trade characteristics.\",\"writing_profile_keywords\":[\"market microstructure\",\"OTC derivatives\",\"Dodd-Frank\",\"spread decomposition\",\"VECM price discovery\",\"institutional detail\",\"descriptive-heavy\"],\"confidence\":0.8,\"notes\":\"Robustness checks are not visible in the excerpts. Significance language is mentioned in the introduction ('statistically significant'), but the format in tables is not visible. No explicit roadmap paragraph is seen.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "financial_intermediation_banking; OTC derivatives regulation", "primary_method": "reduced_form_regression", "identification": {"estimand": "Differences in transaction costs (effective half-spreads), price impacts and dealer profits between dealer-to-client and interdealer index CDS trades, and the information content of each segment for price discovery.", "strategy": "Descriptive decomposition of effective spreads into realized spread and price impact. OLS regressions control for trade characteristics and market conditions. A cointegrated VAR/VECM in the spirit of Hasbrouck measures permanent price impact and price discovery. Comparisons are made across trading protocols and against contemporaneous interdealer quotes.", "key_assumptions": ["Mid-quotes (Markit composites for D2C, GFI executable quotes for D2D) proxy for efficient prices", "The SEF on which a trade occurs correctly identifies whether it is D2C or D2D", "Algorithms correctly identify package transactions and SEFs", "A 15-minute horizon captures price impact"], "threats_addressed": ["Selection bias from differing trade characteristics and market conditions across segments", "Contamination from package trades (rolls, curve trades, delta hedges)"], "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": 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": "preview_of_results"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"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": "preview_of_results"}, {"paragraph_index": 7, "move": "policy_or_broader_implication"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}], "contribution_paragraph_index": 9, "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": "results"}, {"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": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes define variables with abbreviations (EffcSprd, RlzdSprd, PrcImp), state the sample period and sample sizes, and split panels by index (CDX.IG vs CDX.HY). Means are reported by trade-size quartile.", "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' / 'appear to'", "'consistent with' a theoretical model", "explicit statement that welfare effects cannot be addressed empirically", "'suggestive evidence'"]}, "claim_evidence": [{"claim": "Client order flow comprises few, large trades, with D2C volume far exceeding D2D volume.", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "D2C transaction costs exceed D2D costs, but the gap is fully explained by higher price impact.", "evidence_type": "table", "evidence_ref": "Tables II and III"}, {"claim": "Price impact of client trades is largely permanent and price discovery occurs mainly in the D2C segment.", "evidence_type": "table", "evidence_ref": "Table IV, Figure 4"}, {"claim": "Mid-market matches have lower costs and price impacts than limit order book trades, consistent with Zhu's (2014) venue selection model.", "evidence_type": "table", "evidence_ref": "Tables V and VI, Figure 5"}, {"claim": "D2C trades almost always improve on contemporaneous interdealer inside quotes.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Client transaction costs are widely dispersed, consistent with dealers price discriminating on perceived price impact.", "evidence_type": "table", "evidence_ref": "Table VIII"}], "structural_summary": "The introduction motivates the paper with the importance of the index CDS market and the Dodd-Frank SEF mandate as a test of market structure. It then previews data and algorithms and walks sequentially through findings with numeric magnitudes, closing with related literature. The body describes institutional detail and data/identification algorithms, then builds the empirical results in sequence: a spread decomposition with regression controls, a VECM for price discovery, interdealer protocol segmentation, price improvement, and dispersion of client costs. The conclusion summarizes and interprets the two-tiered structure as a consequence of client trade characteristics.", "writing_profile_keywords": ["market microstructure", "OTC derivatives", "Dodd-Frank", "spread decomposition", "VECM price discovery", "institutional detail", "descriptive-heavy"], "confidence": 0.8, "notes": "Robustness checks are not visible in the excerpts. Significance language is mentioned in the introduction ('statistically significant'), but the format in tables is not visible. No explicit roadmap paragraph is seen."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.85, "usage": {"input": 8854, "output": 2199, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08019, "request_id": "req_011CggqyHYZJhdkNNtTh1ZE3", "cached": false, "ts": "2026-11-03T23:15:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3ca11a3f110cb7c48365ada659a1f7c247b5f7f7a2dcfe24674affc8800825b6.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\":\"Determinants of CMR inclusion in loan contracts, and the association of CMR presence with CEO turnover and future firm performance\",\"strategy\":\"Probit regressions of CMR inclusion on human-capital and contracting-friction proxies. Lee (1978) simultaneous-equation selection model linking loan pricing to covenant choice. Probit models of CEO turnover over the life of the CMR. Within-firm timing analysis of CMR addition and removal. Change-in-performance regressions run in subsamples.\",\"key_assumptions\":[\"Proxies for founder status, internal recruiting, heir apparent, noncompete enforcement and retirement age capture human capital risk and retention frictions\",\"The Lee (1978) structure correctly adjusts for the joint determination of covenants and yields\",\"Controls absorb confounding firm characteristics in the turnover and performance regressions\"],\"threats_addressed\":[\"simultaneity of covenant structure and loan pricing\",\"alternative explanations: debt overhang, large blockholders, takeover threats\",\"other loan features such as change-in-control covenants\",\"timing of CMR addition and removal\"],\"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\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":6,\"move\":\"literature_gap\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"motivation\"},{\"paragraph_index\":9,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":15,\"move\":\"preview_of_results\"},{\"paragraph_index\":16,\"move\":\"preview_of_results\"},{\"paragraph_index\":17,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":18,\"move\":\"robustness_mention\"},{\"paragraph_index\":19,\"move\":\"contribution_statement\"},{\"paragraph_index\":19,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":20,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":21,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":19,\"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\":\"robustness\"},{\"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\":[\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Mostly probit tables with an indicator dependent variable. Fama-French 12-industry fixed effects. z-statistics reported in parentheses, with significance stars at 1/5/10%. Variables are defined in the Appendix, and spec numbering is cross-referenced across tables. Performance outcomes are measured as two-year post-minus-pre averages.\",\"units_stated_in_notes\":false,\"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\":[\"suggesting that\",\"tend to\",\"consistent with\",\"may\",\"is related to / association language\",\"some evidence\"]},\"claim_evidence\":[{\"claim\":\"About 8.5% of sample firms, and 17.2% of small firms, have a CMR in at least one loan.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Section I; Table I\"},{\"claim\":\"CMRs are used more for small, risky, low-tangibility borrowers and for those with founder CEOs.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-III\"},{\"claim\":\"CMR use rises when CEO retention frictions are higher, e.g., weak noncompete enforcement or a CEO near retirement age.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"CMR inclusion relates to a lower predicted loan yield, which is inconsistent with the collusion hypothesis.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"CEO turnover probability falls by about 55% relative to the unconditional rate during a CMR loan term.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 1\"},{\"claim\":\"When firms face retention frictions, CMRs are positively related to future Tobin's Q and operating performance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"}],\"structural_summary\":\"The article opens with an epigraph quoting a CMR clause and states two gaps: how debtholders handle human capital risk, and how creditors affect CEO turnover outside default. It poses two questions and introduces hand-collected contract data. It then contrasts a risk hypothesis with a collusion hypothesis and previews findings for each, followed by turnover and performance results. The introduction closes with robustness checks, contributions to theory (Hart-Moore) and to creditor-governance work, and a roadmap. The body covers data, then the determinants of CMR use (risk and collusion tests), then turnover and performance implications, then additional analysis and robustness. The conclusion restates the findings and calls for further theory.\",\"writing_profile_keywords\":[\"hand-collected contract data\",\"competing hypotheses\",\"question-driven sections\",\"probit regressions\",\"creditor governance\",\"human capital\",\"epigraph opening\"],\"confidence\":0.85,\"notes\":\"Causal wording is mixed: the authors use 'association' language but also say lenders 'can influence' turnover. Figure 1 plots abnormal turnover, and it is unclear from the caption whether uncertainty bands are shown.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Determinants of CMR inclusion in loan contracts, and the association of CMR presence with CEO turnover and future firm performance", "strategy": "Probit regressions of CMR inclusion on human-capital and contracting-friction proxies. Lee (1978) simultaneous-equation selection model linking loan pricing to covenant choice. Probit models of CEO turnover over the life of the CMR. Within-firm timing analysis of CMR addition and removal. Change-in-performance regressions run in subsamples.", "key_assumptions": ["Proxies for founder status, internal recruiting, heir apparent, noncompete enforcement and retirement age capture human capital risk and retention frictions", "The Lee (1978) structure correctly adjusts for the joint determination of covenants and yields", "Controls absorb confounding firm characteristics in the turnover and performance regressions"], "threats_addressed": ["simultaneity of covenant structure and loan pricing", "alternative explanations: debt overhang, large blockholders, takeover threats", "other loan features such as change-in-control covenants", "timing of CMR addition and removal"], "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": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 6, "move": "literature_gap"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "motivation"}, {"paragraph_index": 9, "move": "mechanism_or_theory_sketch"}, {"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": "preview_of_results"}, {"paragraph_index": 15, "move": "preview_of_results"}, {"paragraph_index": 16, "move": "preview_of_results"}, {"paragraph_index": 17, "move": "policy_or_broader_implication"}, {"paragraph_index": 18, "move": "robustness_mention"}, {"paragraph_index": 19, "move": "contribution_statement"}, {"paragraph_index": 19, "move": "related_literature_positioning"}, {"paragraph_index": 20, "move": "related_literature_positioning"}, {"paragraph_index": 21, "move": "roadmap"}], "contribution_paragraph_index": 19, "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": "robustness"}, {"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": ["alternative_specification", "other"]}, "tables": {"convention_notes": "Mostly probit tables with an indicator dependent variable. Fama-French 12-industry fixed effects. z-statistics reported in parentheses, with significance stars at 1/5/10%. Variables are defined in the Appendix, and spec numbering is cross-referenced across tables. Performance outcomes are measured as two-year post-minus-pre averages.", "units_stated_in_notes": false, "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": ["suggesting that", "tend to", "consistent with", "may", "is related to / association language", "some evidence"]}, "claim_evidence": [{"claim": "About 8.5% of sample firms, and 17.2% of small firms, have a CMR in at least one loan.", "evidence_type": "descriptive_statistic", "evidence_ref": "Section I; Table I"}, {"claim": "CMRs are used more for small, risky, low-tangibility borrowers and for those with founder CEOs.", "evidence_type": "table", "evidence_ref": "Tables I-III"}, {"claim": "CMR use rises when CEO retention frictions are higher, e.g., weak noncompete enforcement or a CEO near retirement age.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "CMR inclusion relates to a lower predicted loan yield, which is inconsistent with the collusion hypothesis.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "CEO turnover probability falls by about 55% relative to the unconditional rate during a CMR loan term.", "evidence_type": "table", "evidence_ref": "Table VI; Figure 1"}, {"claim": "When firms face retention frictions, CMRs are positively related to future Tobin's Q and operating performance.", "evidence_type": "table", "evidence_ref": "Table VII"}], "structural_summary": "The article opens with an epigraph quoting a CMR clause and states two gaps: how debtholders handle human capital risk, and how creditors affect CEO turnover outside default. It poses two questions and introduces hand-collected contract data. It then contrasts a risk hypothesis with a collusion hypothesis and previews findings for each, followed by turnover and performance results. The introduction closes with robustness checks, contributions to theory (Hart-Moore) and to creditor-governance work, and a roadmap. The body covers data, then the determinants of CMR use (risk and collusion tests), then turnover and performance implications, then additional analysis and robustness. The conclusion restates the findings and calls for further theory.", "writing_profile_keywords": ["hand-collected contract data", "competing hypotheses", "question-driven sections", "probit regressions", "creditor governance", "human capital", "epigraph opening"], "confidence": 0.85, "notes": "Causal wording is mixed: the authors use 'association' language but also say lenders 'can influence' turnover. Figure 1 plots abnormal turnover, and it is unclear from the caption whether uncertainty bands are shown."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 21.07, "usage": {"input": 9895, "output": 2389, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088154, "request_id": "req_011CggqtHtkgXUVBLjQotgL8", "cached": false, "ts": "2026-11-03T23:14:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/3ffce7e2b1d262c25bdd53e011ded0f92152e560e0456f5d71118c6aed6d845f.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\":\"Association between a model's share of captive (manufacturer) financing and its resale price depreciation, LTVs and production quantities; also the effect of a rise in captive share after an acquisition\",\"strategy\":\"Cross-sectional regressions of model-level depreciation on captive share. Depreciation is decomposed into new-price versus productivity-driven components to separate the commitment and information-asymmetry hypotheses. A natural experiment uses Volvo's 2007 acquisition of an Ingersoll Rand unit. Further tests use within-manufacturer regressions of quantities and time-series variation in loan-officer collateral tightening.\",\"key_assumptions\":[\"Information asymmetry about quality would show up in productivity-driven (cross-sectional age) depreciation rather than in new-price depreciation\",\"The acquisition shifted captive share for reasons unrelated to the machines' ex-ante quality\",\"Used machine prices measured at auction reflect market resale values\"],\"threats_addressed\":[\"ex-ante machine quality / signaling alternative\",\"market power confound\",\"bank credit pull-back as an alternative explanation for machine choice\"],\"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\":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\":\"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\":\"identification_defense\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":7,\"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\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes describe the sample, the unit of analysis and how depreciation rates are estimated (model-level log price on log age). Variable definitions such as the down payment ratio are given in the notes. Excerpts are truncated, so the standard-error treatment is not visible.\",\"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' framing\",\"'strongly suggestive evidence'\",\"'move closer to causal inference'\",\"'may be reasonable to think'\",\"explicit statement that the paper does not rule out other motives\"]},\"claim_evidence\":[{\"claim\":\"Manufacturer-financed shares of new durable investment are large across equipment industries\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Models with more captive financing depreciate less in resale markets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Figure 2\"},{\"claim\":\"Captive finance is unrelated to productivity-driven depreciation, undermining the quality-signaling view\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Resale performance of Ingersoll Rand machines improved after the Volvo acquisition raised captive share\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 4\"},{\"claim\":\"Equipment types with a history of captive backing show lower future production within a manufacturer\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Captive-backed models carry higher LTVs, even when financed by banks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Bank-financed buyers shift toward captive-backed models when collateral requirements tighten\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 5\"}],\"structural_summary\":\"The paper opens with stylized facts on the scale of manufacturer lending and asks why manufacturers finance their own sales. It proposes a Coase-commitment explanation and contrasts it with an information-asymmetry alternative, then separates the two by decomposing depreciation. A background and hypothesis section precedes a brief data section. A long results section moves from correlations to the alternative-hypothesis test, market power, an acquisition-based natural experiment, quantity effects, and pledgeability spillovers. A short discussion restates the findings and frames captive finance as a costly hedge.\",\"writing_profile_keywords\":[\"captive finance\",\"Coase conjecture\",\"durable goods\",\"hypothesis discrimination\",\"natural experiment\",\"pledgeability\",\"heavy equipment auction data\",\"competing hypotheses\"],\"confidence\":0.8,\"notes\":\"No explicit robustness section appears in the excerpts. An internet appendix is mentioned but its contents are not visible. Statistical significance conventions and standard-error treatment are not visible.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between a model's share of captive (manufacturer) financing and its resale price depreciation, LTVs and production quantities; also the effect of a rise in captive share after an acquisition", "strategy": "Cross-sectional regressions of model-level depreciation on captive share. Depreciation is decomposed into new-price versus productivity-driven components to separate the commitment and information-asymmetry hypotheses. A natural experiment uses Volvo's 2007 acquisition of an Ingersoll Rand unit. Further tests use within-manufacturer regressions of quantities and time-series variation in loan-officer collateral tightening.", "key_assumptions": ["Information asymmetry about quality would show up in productivity-driven (cross-sectional age) depreciation rather than in new-price depreciation", "The acquisition shifted captive share for reasons unrelated to the machines' ex-ante quality", "Used machine prices measured at auction reflect market resale values"], "threats_addressed": ["ex-ante machine quality / signaling alternative", "market power confound", "bank credit pull-back as an alternative explanation for machine choice"], "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": 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": "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": "identification_defense"}, {"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": "related_literature_positioning"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 7, "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": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes describe the sample, the unit of analysis and how depreciation rates are estimated (model-level log price on log age). Variable definitions such as the down payment ratio are given in the notes. Excerpts are truncated, so the standard-error treatment is not visible.", "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' framing", "'strongly suggestive evidence'", "'move closer to causal inference'", "'may be reasonable to think'", "explicit statement that the paper does not rule out other motives"]}, "claim_evidence": [{"claim": "Manufacturer-financed shares of new durable investment are large across equipment industries", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Models with more captive financing depreciate less in resale markets", "evidence_type": "table", "evidence_ref": "Table I; Figure 2"}, {"claim": "Captive finance is unrelated to productivity-driven depreciation, undermining the quality-signaling view", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Resale performance of Ingersoll Rand machines improved after the Volvo acquisition raised captive share", "evidence_type": "table", "evidence_ref": "Table IV; Figure 4"}, {"claim": "Equipment types with a history of captive backing show lower future production within a manufacturer", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Captive-backed models carry higher LTVs, even when financed by banks", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Bank-financed buyers shift toward captive-backed models when collateral requirements tighten", "evidence_type": "table", "evidence_ref": "Table VII; Figure 5"}], "structural_summary": "The paper opens with stylized facts on the scale of manufacturer lending and asks why manufacturers finance their own sales. It proposes a Coase-commitment explanation and contrasts it with an information-asymmetry alternative, then separates the two by decomposing depreciation. A background and hypothesis section precedes a brief data section. A long results section moves from correlations to the alternative-hypothesis test, market power, an acquisition-based natural experiment, quantity effects, and pledgeability spillovers. A short discussion restates the findings and frames captive finance as a costly hedge.", "writing_profile_keywords": ["captive finance", "Coase conjecture", "durable goods", "hypothesis discrimination", "natural experiment", "pledgeability", "heavy equipment auction data", "competing hypotheses"], "confidence": 0.8, "notes": "No explicit robustness section appears in the excerpts. An internet appendix is mentioned but its contents are not visible. Statistical significance conventions and standard-error treatment are not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.1, "usage": {"input": 6612, "output": 2195, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071142, "request_id": "req_011CggqHH9dcjTCJ3Jp2ymKg", "cached": false, "ts": "2026-11-03T23:06:29+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/41743a0f5b51bbf404eb9b3835720b29545ea858acf37f807d7f5ac257725d28.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"securitization and credit ratings; information economics\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model; comparative statics of equilibrium lending standards, retention and welfare with respect to rating accuracy and screening precision)\",\"strategy\":\"not applicable; signaling game with D1 refinement, equilibrium characterization and comparative statics\",\"key_assumptions\":[\"risk-neutral agents\",\"competitive investors who break even\",\"banks have private screening information at origination\",\"retention is costly\",\"ratings are an imperfect public signal\"],\"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\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":false,\"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\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"other\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"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 illustrated with numerical-example figures; proofs are collected in the appendix.\",\"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\":[\"conditional statements (provided that, when)\",\"'may' and 'can' for oversupply outcomes\",\"'typically' for welfare gains\",\"'suggests' when linking to the crisis\"]},\"claim_evidence\":[{\"claim\":\"Without ratings, separation through costly retention is the unique stable outcome, leading to undersupply of credit\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.B; Appendix C\"},{\"claim\":\"Ratings shift the equilibrium toward OTD with zero retention and lower lending standards, possibly oversupplying credit\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 3; Section IV; Figure 3\"},{\"claim\":\"Lower rating accuracy under OTD expands credit to negative-NPV borrowers\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 4; Figures 4-5\"},{\"claim\":\"More precise screening with ratings lowers lending standards, so some bad loans are deliberately originated\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section V.B-C; Figure 6\"},{\"claim\":\"Model predictions are consistent with evidence of declining pre-crisis lending standards\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section I.A (Mian and Sufi 2009; Purnanandam 2010; Rajan, Seru, and Vig 2015)\"},{\"claim\":\"Skin-in-the-game, disclosure and liquidity policies improve welfare only under specific conditions\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section VI; Figures 8-9; Appendix F\"}],\"structural_summary\":\"The introduction motivates the paper with securitization growth and post-crisis regulation. It then sketches a no-ratings signaling benchmark and previews how ratings shift the economy to an originate-to-distribute equilibrium, along with the efficiency trade-off and screening interactions. A separate literature section links model implications to empirical evidence, states testable implications and positions the paper in theory. The model, benchmarks and equilibrium characterization come next, followed by comparative statics on rating accuracy and screening and a policy analysis section. A brief conclusion restates the findings, and an appendix collects the proofs.\",\"writing_profile_keywords\":[\"signaling game\",\"originate-to-distribute\",\"comparative statics\",\"proposition-driven\",\"policy analysis\",\"welfare trade-off\",\"numerical illustrations\"],\"confidence\":0.85,\"notes\":\"The introduction excerpt is truncated, and no explicit contribution or roadmap paragraph is visible. The literature section includes a testable-implications subsection.\"}", "json": {"genre": "theoretical", "subfield": "financial_intermediation_banking", "subfield_secondary": "securitization and credit ratings; information economics", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model; comparative statics of equilibrium lending standards, retention and welfare with respect to rating accuracy and screening precision)", "strategy": "not applicable; signaling game with D1 refinement, equilibrium characterization and comparative statics", "key_assumptions": ["risk-neutral agents", "competitive investors who break even", "banks have private screening information at origination", "retention is costly", "ratings are an imperfect public signal"], "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": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": -1, "contribution_types": ["new_theory_or_model"], "roadmap_present": false, "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": "theory"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "other"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "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 illustrated with numerical-example figures; proofs are collected in the appendix.", "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": ["conditional statements (provided that, when)", "'may' and 'can' for oversupply outcomes", "'typically' for welfare gains", "'suggests' when linking to the crisis"]}, "claim_evidence": [{"claim": "Without ratings, separation through costly retention is the unique stable outcome, leading to undersupply of credit", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.B; Appendix C"}, {"claim": "Ratings shift the equilibrium toward OTD with zero retention and lower lending standards, possibly oversupplying credit", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 3; Section IV; Figure 3"}, {"claim": "Lower rating accuracy under OTD expands credit to negative-NPV borrowers", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 4; Figures 4-5"}, {"claim": "More precise screening with ratings lowers lending standards, so some bad loans are deliberately originated", "evidence_type": "equation_or_proof", "evidence_ref": "Section V.B-C; Figure 6"}, {"claim": "Model predictions are consistent with evidence of declining pre-crisis lending standards", "evidence_type": "prior_literature", "evidence_ref": "Section I.A (Mian and Sufi 2009; Purnanandam 2010; Rajan, Seru, and Vig 2015)"}, {"claim": "Skin-in-the-game, disclosure and liquidity policies improve welfare only under specific conditions", "evidence_type": "equation_or_proof", "evidence_ref": "Section VI; Figures 8-9; Appendix F"}], "structural_summary": "The introduction motivates the paper with securitization growth and post-crisis regulation. It then sketches a no-ratings signaling benchmark and previews how ratings shift the economy to an originate-to-distribute equilibrium, along with the efficiency trade-off and screening interactions. A separate literature section links model implications to empirical evidence, states testable implications and positions the paper in theory. The model, benchmarks and equilibrium characterization come next, followed by comparative statics on rating accuracy and screening and a policy analysis section. A brief conclusion restates the findings, and an appendix collects the proofs.", "writing_profile_keywords": ["signaling game", "originate-to-distribute", "comparative statics", "proposition-driven", "policy analysis", "welfare trade-off", "numerical illustrations"], "confidence": 0.85, "notes": "The introduction excerpt is truncated, and no explicit contribution or roadmap paragraph is visible. The literature section includes a testable-implications subsection."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.2, "usage": {"input": 5931, "output": 1909, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062698, "request_id": "req_011CggqgpCou84kuePWsqUKJ", "cached": false, "ts": "2026-11-03T23:11:45+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/41db738e96ed032bfbaa7890b26fb1610e63ddf2625613b1570ab125100caf5e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"methodological_econometric\",\"subfield\":\"financial_econometrics\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Type I and Type II error rates (false discovery rate, miss rate) of multiple-testing procedures applied to fund/strategy selection\",\"strategy\":\"Double-bootstrap simulation on data adjusted so that a fraction p0 of tests are true (set at in-sample estimates) and the rest have zero alpha\",\"key_assumptions\":[\"p0 acts as a sufficient statistic for the alternative\",\"in-sample estimates are reasonable values for true effects\",\"bootstrap preserves the dependence structure of the data\"],\"threats_addressed\":[\"multiple testing / luck\",\"unknown p0, handled by evaluating performance over a range of p0 values\",\"high-dimensional alternative parameter vector\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":true,\"sentence_index\":2},\"results\":{\"present\":true,\"sentence_index\":4},\"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\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":2,\"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\":\"results\"},{\"idx\":3,\"role\":\"conclusion\"},{\"idx\":4,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes describe the simulation design in detail, including bootstrap iteration counts (I, J, M) and the grid of p0 values. Simulated Type I and Type II error rates are tabulated across methods and p0 values. Fund-test tables report results for percentile-based test statistics.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"may lead us to miss\",\"likely more consistent\",\"conditioning results on assumed p0\",\"we expect our method to be useful\"]},\"claim_evidence\":[{\"claim\":\"Existing multiple-testing methods rank in a fairly stable order of performance regardless of p0\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III\"},{\"claim\":\"Error rates of Yan and Zheng's application of the Fama-French approach are inconsistent with a widespread-mispricing narrative\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Fama and French (2010) bootstrap test lacks power to detect outperforming funds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VIII\"},{\"claim\":\"Optimal t-statistic cutoffs vary with the assumed fraction of true strategies p0\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"}],\"structural_summary\":\"Opens by framing manager and factor selection as a two-error problem and noting that current tools ignore Type II error and differential error costs. It then explains why Type II error is hard to compute under multiple tests and proposes a double bootstrap that fixes the fraction of true effects p0. The method is positioned against single-bootstrap predecessors (KTWW, Fama-French). A method section develops the error-rate estimation. A long applications section covers backtested strategies, 18,000 anomalies (critiquing Yan-Zheng), a simulation study, and mutual-fund performance evaluation (critiquing Fama-French). A short conclusion restates the contribution and its expected usefulness.\",\"writing_profile_keywords\":[\"multiple testing\",\"double bootstrap\",\"Type II error\",\"false discovery rate\",\"methodological critique\",\"simulation-based\"],\"confidence\":0.8,\"notes\":\"Zero equations are recorded despite the methodological content, so formulas may not have been captured in the extraction. Robustness checks appear to be embedded in the applications (simulation study, subsamples, alternative test statistics) rather than in a separate section.\"}", "json": {"genre": "methodological_econometric", "subfield": "financial_econometrics", "subfield_secondary": "asset_pricing", "primary_method": "calibration_simulation", "identification": {"estimand": "Type I and Type II error rates (false discovery rate, miss rate) of multiple-testing procedures applied to fund/strategy selection", "strategy": "Double-bootstrap simulation on data adjusted so that a fraction p0 of tests are true (set at in-sample estimates) and the rest have zero alpha", "key_assumptions": ["p0 acts as a sufficient statistic for the alternative", "in-sample estimates are reasonable values for true effects", "bootstrap preserves the dependence structure of the data"], "threats_addressed": ["multiple testing / luck", "unknown p0, handled by evaluating performance over a range of p0 values", "high-dimensional alternative parameter vector"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 0}, "method": {"present": true, "sentence_index": 2}, "results": {"present": true, "sentence_index": 4}, "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": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_results"}], "contribution_paragraph_index": 2, "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": "results"}, {"idx": 3, "role": "conclusion"}, {"idx": 4, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["sensitivity_to_parameters", "subsample", "other"]}, "tables": {"convention_notes": "Table notes describe the simulation design in detail, including bootstrap iteration counts (I, J, M) and the grid of p0 values. Simulated Type I and Type II error rates are tabulated across methods and p0 values. Fund-test tables report results for percentile-based test statistics.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["may lead us to miss", "likely more consistent", "conditioning results on assumed p0", "we expect our method to be useful"]}, "claim_evidence": [{"claim": "Existing multiple-testing methods rank in a fairly stable order of performance regardless of p0", "evidence_type": "table", "evidence_ref": "Tables II, III"}, {"claim": "Error rates of Yan and Zheng's application of the Fama-French approach are inconsistent with a widespread-mispricing narrative", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Fama and French (2010) bootstrap test lacks power to detect outperforming funds", "evidence_type": "table", "evidence_ref": "Tables VI-VIII"}, {"claim": "Optimal t-statistic cutoffs vary with the assumed fraction of true strategies p0", "evidence_type": "figure", "evidence_ref": "Figure 3"}], "structural_summary": "Opens by framing manager and factor selection as a two-error problem and noting that current tools ignore Type II error and differential error costs. It then explains why Type II error is hard to compute under multiple tests and proposes a double bootstrap that fixes the fraction of true effects p0. The method is positioned against single-bootstrap predecessors (KTWW, Fama-French). A method section develops the error-rate estimation. A long applications section covers backtested strategies, 18,000 anomalies (critiquing Yan-Zheng), a simulation study, and mutual-fund performance evaluation (critiquing Fama-French). A short conclusion restates the contribution and its expected usefulness.", "writing_profile_keywords": ["multiple testing", "double bootstrap", "Type II error", "false discovery rate", "methodological critique", "simulation-based"], "confidence": 0.8, "notes": "Zero equations are recorded despite the methodological content, so formulas may not have been captured in the extraction. Robustness checks appear to be embedded in the applications (simulation study, subsamples, alternative test statistics) rather than in a separate section."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.6, "usage": {"input": 6346, "output": 1751, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061198, "request_id": "req_011CggqwgtCH5zGkkKAEoqYZ", "cached": false, "ts": "2026-11-03T23:15:07+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/42db43fd3bfced6b13f02ef74b8c1a12b7e94599990df16c6465e0fc78ffd467.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Variance shares of cash flow and discount rate news in anomaly, market and MVE portfolio returns, and the correlations of these components with each other and with aggregate shocks\",\"strategy\":\"Campbell (1991) log-linear return decomposition. A panel VAR(1) for firm-level variables and an aggregate VAR are estimated, and the resulting firm-level news is aggregated to the portfolio level.\",\"key_assumptions\":[\"VAR(1) dynamics adequately capture firm-level expected returns and cash flows\",\"Short-run firm-level predictive regressions extrapolate correctly to long-run anomaly expected returns\",\"Log-linear approximation is valid\"],\"threats_addressed\":[\"Misspecification of VAR assumptions\",\"Overfitting and misspecified expected returns\",\"Influence of microcaps\",\"Influence of the Great Depression sample period\",\"Discrepancies with prior studies (V02, CPV)\"],\"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\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"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\",\"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\":\"empirical_design\"},{\"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_specification\",\"alternative_sample_period\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes define variables in detail, state the VAR source of each decomposition, and specify portfolio construction (quintile long-short) and the 1929-2017 sample. Alternative specifications are reported relative to a repeated baseline row. 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\":[\"cast doubt on\",\"difficult to reconcile\",\"largely uncorrelated\",\"notable caveat\",\"most promising theories\"]},\"claim_evidence\":[{\"claim\":\"CF news accounts for most of the anomaly MVE portfolio return variance, while DR news dominates market return variance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V; Figures 1, 2\"},{\"claim\":\"Anomaly CF and DR shocks are nearly uncorrelated with market CF and DR shocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figures 3, 4\"},{\"claim\":\"Anomaly return components are weakly related to business cycle measures and to proxies for risk aversion\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"CF and DR shocks are negatively correlated for most anomalies when microcaps are excluded; the correlation turns positive when they are included\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, X\"},{\"claim\":\"VAR-implied expected anomaly returns match realized averages\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Firm-level results reconcile with V02 and prior studies\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The paper opens by framing anomalies as central to the SDF and noting disagreement among competing theories. It introduces a CF/DR decomposition and uses extreme models to show how the decomposition discriminates among them. It previews three findings and uses them to rule out classes of theories, then positions itself against prior decomposition studies. The body has a theory section linking the decomposition to anomalies and the SDF, followed by data and panel plus aggregate VAR estimation. It then presents firm, anomaly and MVE decompositions and their correlations, an extensive robustness section that tests VAR assumptions and reconciles prior findings, and an interpretation section. The conclusion restates three facts with magnitudes and names the most promising theory classes.\",\"writing_profile_keywords\":[\"variance decomposition\",\"panel VAR\",\"stylized facts\",\"theory discrimination\",\"anomalies\",\"MVE portfolio\",\"reconciliation with prior literature\"],\"confidence\":0.8,\"notes\":\"This is a descriptive decomposition paper with no causal identification. The intro excerpt omits the paragraphs stating the first and second findings. Magnitudes (73%, 74%, 69%) are visible in the conclusion; the only quantitative figure visible in the intro excerpts is the 85% SDF variance share in paragraph 0. The decomposition section was labeled as results.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "reduced_form_regression", "identification": {"estimand": "Variance shares of cash flow and discount rate news in anomaly, market and MVE portfolio returns, and the correlations of these components with each other and with aggregate shocks", "strategy": "Campbell (1991) log-linear return decomposition. A panel VAR(1) for firm-level variables and an aggregate VAR are estimated, and the resulting firm-level news is aggregated to the portfolio level.", "key_assumptions": ["VAR(1) dynamics adequately capture firm-level expected returns and cash flows", "Short-run firm-level predictive regressions extrapolate correctly to long-run anomaly expected returns", "Log-linear approximation is valid"], "threats_addressed": ["Misspecification of VAR assumptions", "Overfitting and misspecified expected returns", "Influence of microcaps", "Influence of the Great Depression sample period", "Discrepancies with prior studies (V02, CPV)"], "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": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"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", "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": "empirical_design"}, {"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_specification", "alternative_sample_period", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Notes define variables in detail, state the VAR source of each decomposition, and specify portfolio construction (quintile long-short) and the 1929-2017 sample. Alternative specifications are reported relative to a repeated baseline row. 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": ["cast doubt on", "difficult to reconcile", "largely uncorrelated", "notable caveat", "most promising theories"]}, "claim_evidence": [{"claim": "CF news accounts for most of the anomaly MVE portfolio return variance, while DR news dominates market return variance", "evidence_type": "table", "evidence_ref": "Tables IV, V; Figures 1, 2"}, {"claim": "Anomaly CF and DR shocks are nearly uncorrelated with market CF and DR shocks", "evidence_type": "table", "evidence_ref": "Table VI; Figures 3, 4"}, {"claim": "Anomaly return components are weakly related to business cycle measures and to proxies for risk aversion", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "CF and DR shocks are negatively correlated for most anomalies when microcaps are excluded; the correlation turns positive when they are included", "evidence_type": "table", "evidence_ref": "Tables V, X"}, {"claim": "VAR-implied expected anomaly returns match realized averages", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Firm-level results reconcile with V02 and prior studies", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The paper opens by framing anomalies as central to the SDF and noting disagreement among competing theories. It introduces a CF/DR decomposition and uses extreme models to show how the decomposition discriminates among them. It previews three findings and uses them to rule out classes of theories, then positions itself against prior decomposition studies. The body has a theory section linking the decomposition to anomalies and the SDF, followed by data and panel plus aggregate VAR estimation. It then presents firm, anomaly and MVE decompositions and their correlations, an extensive robustness section that tests VAR assumptions and reconciles prior findings, and an interpretation section. The conclusion restates three facts with magnitudes and names the most promising theory classes.", "writing_profile_keywords": ["variance decomposition", "panel VAR", "stylized facts", "theory discrimination", "anomalies", "MVE portfolio", "reconciliation with prior literature"], "confidence": 0.8, "notes": "This is a descriptive decomposition paper with no causal identification. The intro excerpt omits the paragraphs stating the first and second findings. Magnitudes (73%, 74%, 69%) are visible in the conclusion; the only quantitative figure visible in the intro excerpts is the 85% SDF variance share in paragraph 0. The decomposition section was labeled as results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.81, "usage": {"input": 7614, "output": 2207, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07539, "request_id": "req_011Cggqo7yKSS5fC9wE3GLMq", "cached": false, "ts": "2026-11-03T23:13:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/43b8aad31f3cc75f46a190612d80f4a883d5816862639025d6c2b4b3f0654b91.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"How savings and dissavings behavior relates to survival optimism (subjective minus actuarial survival probability), alongside model-implied effects of subjective versus actuarial mortality beliefs on life-cycle saving.\",\"strategy\":\"Survey elicitation of beliefs; a canonical life-cycle model calibrated to the elicited beliefs and compared with an actuarial benchmark, with fit checked by SMM; ordered logit regressions with controls; Altonji-Elder-Taber/Oster coefficient bounding; a salience-based belief measure presumed orthogonal to confounders.\",\"key_assumptions\":[\"Elicited survey beliefs reflect true subjective survival probabilities\",\"Controls capture sophistication and other confounders\",\"Weight placed on salient risk factors is orthogonal to health status and general optimism\",\"Selection on observables is informative about selection on unobservables (Oster bounds)\"],\"threats_addressed\":[\"omitted variable bias from sophistication or financial literacy\",\"confounding by income, education and demographics\",\"general optimism or health status\",\"risk preferences\"],\"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\":3},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"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_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"identification_defense\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":10,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":9,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"resolution_of_puzzle\"],\"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\":\"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 define key variables, e.g., survival optimism as subjective minus SSA actuarial survival probability. Regression tables report ordered logit log-odds. Model tables list calibration parameters and their survey 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\":[\"explicit caution about causal interpretation of reduced-form regressions\",\"'suggests' / 'gets closer to understanding'\",\"'presumably orthogonal'\",\"'associated with' wording for regression results\"],\"typical_devices_note\":null},\"claim_evidence\":[{\"claim\":\"Young respondents underestimate survival and older respondents overestimate it relative to actuarial data.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Salience of cohort-specific causes of death shapes mortality belief distortions.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Table II; Figures 2-3\"},{\"claim\":\"With subjective beliefs, the young undersave by about 26% and retirees draw down assets about 27% slower than under actuarial calibration.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 5\"},{\"claim\":\"Calibrating to subjective beliefs substantially improves the model's fit to consumption and net worth data.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"},{\"claim\":\"Survival optimism correlates with savings rates after controlling for literacy, risk tolerance and socioeconomic factors.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figure 7\"},{\"claim\":\"Survival beliefs relate to retirement dissavings, with an association comparable to that of bequest motives.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The paper opens with behavioral motivation about overweighting rare, salient events. It then documents a new survey of mortality beliefs and their salience-based formation. Next, it presents a canonical life-cycle model calibrated to the elicited beliefs and compares its results with an actuarial benchmark, including SMM fit to data and comparison with alternative mechanisms. Reduced-form ordered logit regressions on the survey then validate the model's predictions, with an explicit omitted-variable defense. The conclusion restates the findings, discusses policy relevance and points to general-equilibrium and insurance questions for future work.\",\"writing_profile_keywords\":[\"new survey\",\"subjective expectations\",\"life-cycle calibration\",\"salience\",\"household finance puzzles\",\"model-plus-regression validation\",\"cautious causality\"],\"confidence\":0.8,\"notes\":\"Section IV is labeled as validating model predictions; it is assigned the results role. Standard error type is not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "calibration_simulation", "identification": {"estimand": "How savings and dissavings behavior relates to survival optimism (subjective minus actuarial survival probability), alongside model-implied effects of subjective versus actuarial mortality beliefs on life-cycle saving.", "strategy": "Survey elicitation of beliefs; a canonical life-cycle model calibrated to the elicited beliefs and compared with an actuarial benchmark, with fit checked by SMM; ordered logit regressions with controls; Altonji-Elder-Taber/Oster coefficient bounding; a salience-based belief measure presumed orthogonal to confounders.", "key_assumptions": ["Elicited survey beliefs reflect true subjective survival probabilities", "Controls capture sophistication and other confounders", "Weight placed on salient risk factors is orthogonal to health status and general optimism", "Selection on observables is informative about selection on unobservables (Oster bounds)"], "threats_addressed": ["omitted variable bias from sophistication or financial literacy", "confounding by income, education and demographics", "general optimism or health status", "risk preferences"], "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": 3}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "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_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "identification_defense"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 10, "move": "roadmap"}], "contribution_paragraph_index": 9, "contribution_types": ["new_evidence", "new_data_or_measure", "resolution_of_puzzle"], "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": "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 define key variables, e.g., survival optimism as subjective minus SSA actuarial survival probability. Regression tables report ordered logit log-odds. Model tables list calibration parameters and their survey 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": ["explicit caution about causal interpretation of reduced-form regressions", "'suggests' / 'gets closer to understanding'", "'presumably orthogonal'", "'associated with' wording for regression results"], "typical_devices_note": null}, "claim_evidence": [{"claim": "Young respondents underestimate survival and older respondents overestimate it relative to actuarial data.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Salience of cohort-specific causes of death shapes mortality belief distortions.", "evidence_type": "figure", "evidence_ref": "Table II; Figures 2-3"}, {"claim": "With subjective beliefs, the young undersave by about 26% and retirees draw down assets about 27% slower than under actuarial calibration.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 5"}, {"claim": "Calibrating to subjective beliefs substantially improves the model's fit to consumption and net worth data.", "evidence_type": "figure", "evidence_ref": "Figure 6"}, {"claim": "Survival optimism correlates with savings rates after controlling for literacy, risk tolerance and socioeconomic factors.", "evidence_type": "table", "evidence_ref": "Table V; Figure 7"}, {"claim": "Survival beliefs relate to retirement dissavings, with an association comparable to that of bequest motives.", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The paper opens with behavioral motivation about overweighting rare, salient events. It then documents a new survey of mortality beliefs and their salience-based formation. Next, it presents a canonical life-cycle model calibrated to the elicited beliefs and compares its results with an actuarial benchmark, including SMM fit to data and comparison with alternative mechanisms. Reduced-form ordered logit regressions on the survey then validate the model's predictions, with an explicit omitted-variable defense. The conclusion restates the findings, discusses policy relevance and points to general-equilibrium and insurance questions for future work.", "writing_profile_keywords": ["new survey", "subjective expectations", "life-cycle calibration", "salience", "household finance puzzles", "model-plus-regression validation", "cautious causality"], "confidence": 0.8, "notes": "Section IV is labeled as validating model predictions; it is assigned the results role. Standard error type is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.38, "usage": {"input": 7557, "output": 2199, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075002, "request_id": "req_011CggqZshU4Ub4QVJn4becS", "cached": false, "ts": "2026-11-03T23:10:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/447c7f5143d62deb469bb9b69fae8b495ec354b4554d9cf399312289b4920a65.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"methodological_econometric\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Identification of risk premia (rank of the beta matrix) and confidence sets for risk premia and SDF parameters such as relative risk aversion\",\"strategy\":\"Identification-robust tests: a rank test with F bounding distribution, and a GRS-FAR statistic (an extension of GRS/Anderson-Rubin) inverted to give confidence sets; GMM-AR for nonlinear SDFs; Monte Carlo simulations calibrated to the data\",\"key_assumptions\":[\"Exact F distribution of the GRS-type statistic under normality\",\"Model specification (correct versus misspecified) affects the shape of confidence sets\",\"GMM-AR requires T much larger than N\"],\"threats_addressed\":[\"Weak or no identification from small factor–return correlations\",\"Small T relative to N\",\"Spurious t-test rejections in Fama-MacBeth\",\"Model misspecification\"],\"causal_language_strength\":1,\"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\":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\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"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\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"resolution_of_puzzle\" ,\"move\":\"other\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_method\",\"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\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_measure\",\"alternative_sample_period\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes identify the test assets, the sample period and frequency, and the data source (Kroencke 2017; Lettau-Ludvigson 2001). They explain how estimates are produced (FM two-pass, GMM) and define fit measures such as the pseudo-R2 and cross-sectional R2. Test p-values are reported.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"cannot reject\",\"might seem odd\",\"typically\",\"conditional statements on data quality and T versus N\"]},\"claim_evidence\":[{\"claim\":\"Identification test cannot reject unidentified risk premia for five consumption measures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"FM t-test is unreliable under near-constant betas in calibrated simulations\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"GRS-FAR 95% confidence sets for consumption risk premia are unbounded\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"With larger T, confidence sets move from unbounded to bounded to empty under misspecification\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 6-7; Table V\"},{\"claim\":\"GMM-AR gives unbounded confidence sets for relative risk aversion\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9; Table VI\"},{\"claim\":\"GRS-FAR has exact F distribution regardless of identification\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.D\"}],\"structural_summary\":\"Opens with the puzzle that different consumption measures yield conflicting asset pricing results and places this within the weak-identification literature. Proposes a rank test and the GRS-FAR test, then applies them to the Kroencke data, previewing unbounded confidence sets and their reconciling role. Section I develops identification conditions and the tests for a single factor, alternating simulations with applications. Section II covers multifactor models, larger T and misspecification. Section III extends the analysis to nonlinear SDFs via GMM-AR. The conclusion restates the findings, gives practitioner guidance and points to extensions.\",\"writing_profile_keywords\":[\"weak identification\",\"robust inference\",\"Monte Carlo calibration\",\"confidence set inversion\",\"econometric method applied to asset pricing\",\"GRS test\"],\"confidence\":0.8,\"notes\":\"The method is an econometric testing procedure, so primary_method is coded as other. A placeholder move was inadvertently entered for paragraph 6; the intended move there is contribution_statement only. Section idx 2 is a parsing artifact: its heading is an equation and it continues Section I.\"}", "json": {"genre": "methodological_econometric", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "other", "identification": {"estimand": "Identification of risk premia (rank of the beta matrix) and confidence sets for risk premia and SDF parameters such as relative risk aversion", "strategy": "Identification-robust tests: a rank test with F bounding distribution, and a GRS-FAR statistic (an extension of GRS/Anderson-Rubin) inverted to give confidence sets; GMM-AR for nonlinear SDFs; Monte Carlo simulations calibrated to the data", "key_assumptions": ["Exact F distribution of the GRS-type statistic under normality", "Model specification (correct versus misspecified) affects the shape of confidence sets", "GMM-AR requires T much larger than N"], "threats_addressed": ["Weak or no identification from small factor–return correlations", "Small T relative to N", "Spurious t-test rejections in Fama-MacBeth", "Model misspecification"], "causal_language_strength": 1, "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": 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": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "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": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "other"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_method", "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": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_measure", "alternative_sample_period", "alternative_specification"]}, "tables": {"convention_notes": "Table notes identify the test assets, the sample period and frequency, and the data source (Kroencke 2017; Lettau-Ludvigson 2001). They explain how estimates are produced (FM two-pass, GMM) and define fit measures such as the pseudo-R2 and cross-sectional R2. Test p-values are reported.", "units_stated_in_notes": false, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["cannot reject", "might seem odd", "typically", "conditional statements on data quality and T versus N"]}, "claim_evidence": [{"claim": "Identification test cannot reject unidentified risk premia for five consumption measures", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "FM t-test is unreliable under near-constant betas in calibrated simulations", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "GRS-FAR 95% confidence sets for consumption risk premia are unbounded", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "With larger T, confidence sets move from unbounded to bounded to empty under misspecification", "evidence_type": "figure", "evidence_ref": "Figures 6-7; Table V"}, {"claim": "GMM-AR gives unbounded confidence sets for relative risk aversion", "evidence_type": "figure", "evidence_ref": "Figure 9; Table VI"}, {"claim": "GRS-FAR has exact F distribution regardless of identification", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.D"}], "structural_summary": "Opens with the puzzle that different consumption measures yield conflicting asset pricing results and places this within the weak-identification literature. Proposes a rank test and the GRS-FAR test, then applies them to the Kroencke data, previewing unbounded confidence sets and their reconciling role. Section I develops identification conditions and the tests for a single factor, alternating simulations with applications. Section II covers multifactor models, larger T and misspecification. Section III extends the analysis to nonlinear SDFs via GMM-AR. The conclusion restates the findings, gives practitioner guidance and points to extensions.", "writing_profile_keywords": ["weak identification", "robust inference", "Monte Carlo calibration", "confidence set inversion", "econometric method applied to asset pricing", "GRS test"], "confidence": 0.8, "notes": "The method is an econometric testing procedure, so primary_method is coded as other. A placeholder move was inadvertently entered for paragraph 6; the intended move there is contribution_statement only. Section idx 2 is a parsing artifact: its heading is an equation and it continues Section I."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.7, "usage": {"input": 8930, "output": 2093, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078374, "request_id": "req_011CggqfW2tNvDBH3UXmzuj5", "cached": false, "ts": "2026-11-03T23:11:30+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/44b97a75d78ecde77bedbf36dc8cd24b25993c91a380fa77dd63152dd17331ac.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Bias and coverage of FDR estimates of the proportions of zero-, negative- and positive-alpha mutual funds, compared with known true proportions\",\"strategy\":\"Monte Carlo simulation of fund returns from a factor model with data-informed DGPs for alpha; the FDR estimator is applied to each sample and its estimates are compared with the true proportions\",\"key_assumptions\":[\"Simulated DGPs (alpha distributions, sample sizes, error correlation) are representative of real fund data\",\"Linear factor model of returns as in BSW\"],\"threats_addressed\":[\"Discrete alpha distribution with point masses far from zero\",\"Sample size, both number of funds and observations per fund\",\"Unbalanced panel\",\"Cross-sectional error correlation\"],\"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\":true,\"sentence_index\":2},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"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\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"replication_or_correction\",\"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\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"confidence_intervals\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Each table reports simulated FDR estimates against true proportions. Rows vary the true group proportions and columns vary the alpha distance or dispersion. Alphas are given as annualized percentages. Results are averaged across 1,000 simulations, and one table reports actual coverage of 95% confidence intervals.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'can materially change' phrasing\",\"'potentially unrealistic' qualifiers\",\"'under certain conditions' scoping\",\"'question applicability' framing\"]},\"claim_evidence\":[{\"claim\":\"The FDR estimator overestimates the proportion of zero-alpha funds and underestimates nonzero-alpha funds under data-informed DGPs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I, III\"},{\"claim\":\"FDR 95% confidence intervals almost never contain the true proportions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Accuracy does not improve with more funds; bias stems from low signal-to-noise ratio and low test power\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Low-power tests spread alternative p-values across [0,1], so the null proportion is overestimated\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Bias persists in unbalanced panels and with correlated errors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI\"},{\"claim\":\"FDR and 'no-luck' approaches compared across true alphas and proportions\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"}],\"structural_summary\":\"Short comment on an influential prior method. The introduction summarizes the original findings and their wide adoption, then explains intuitively how the method's assumption of high test power can fail in fund data, and previews the expanded simulations. Section I sets out the FDR mechanics. Section II runs simulations that vary the alpha distribution, sample size and error correlation, documenting bias and poor interval coverage. The conclusion attributes the original method's accuracy to unrealistic simulation assumptions and questions its use in low-power settings.\",\"writing_profile_keywords\":[\"comment\",\"simulation-based critique\",\"statistical power\",\"false discovery rate\",\"mutual fund alpha\",\"replication\"],\"confidence\":0.85,\"notes\":\"Labelled as a corrigendum/replication-type comment. No empirical identification; the evidence is entirely Monte Carlo. The original BSW results are cited from that paper rather than re-estimated.\"}", "json": {"genre": "replication_or_correction", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "calibration_simulation", "identification": {"estimand": "Bias and coverage of FDR estimates of the proportions of zero-, negative- and positive-alpha mutual funds, compared with known true proportions", "strategy": "Monte Carlo simulation of fund returns from a factor model with data-informed DGPs for alpha; the FDR estimator is applied to each sample and its estimates are compared with the true proportions", "key_assumptions": ["Simulated DGPs (alpha distributions, sample sizes, error correlation) are representative of real fund data", "Linear factor model of returns as in BSW"], "threats_addressed": ["Discrete alpha distribution with point masses far from zero", "Sample size, both number of funds and observations per fund", "Unbalanced panel", "Cross-sectional error correlation"], "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": true, "sentence_index": 2}, "contribution": {"present": true, "sentence_index": 4}}, "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": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_approach"}], "contribution_paragraph_index": 0, "contribution_types": ["replication_or_correction", "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"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "confidence_intervals", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["sensitivity_to_parameters", "alternative_specification"]}, "tables": {"convention_notes": "Each table reports simulated FDR estimates against true proportions. Rows vary the true group proportions and columns vary the alpha distance or dispersion. Alphas are given as annualized percentages. Results are averaged across 1,000 simulations, and one table reports actual coverage of 95% confidence intervals.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'can materially change' phrasing", "'potentially unrealistic' qualifiers", "'under certain conditions' scoping", "'question applicability' framing"]}, "claim_evidence": [{"claim": "The FDR estimator overestimates the proportion of zero-alpha funds and underestimates nonzero-alpha funds under data-informed DGPs", "evidence_type": "table", "evidence_ref": "Tables I, III"}, {"claim": "FDR 95% confidence intervals almost never contain the true proportions", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Accuracy does not improve with more funds; bias stems from low signal-to-noise ratio and low test power", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Low-power tests spread alternative p-values across [0,1], so the null proportion is overestimated", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Bias persists in unbalanced panels and with correlated errors", "evidence_type": "table", "evidence_ref": "Tables V, VI"}, {"claim": "FDR and 'no-luck' approaches compared across true alphas and proportions", "evidence_type": "figure", "evidence_ref": "Figure 2"}], "structural_summary": "Short comment on an influential prior method. The introduction summarizes the original findings and their wide adoption, then explains intuitively how the method's assumption of high test power can fail in fund data, and previews the expanded simulations. Section I sets out the FDR mechanics. Section II runs simulations that vary the alpha distribution, sample size and error correlation, documenting bias and poor interval coverage. The conclusion attributes the original method's accuracy to unrealistic simulation assumptions and questions its use in low-power settings.", "writing_profile_keywords": ["comment", "simulation-based critique", "statistical power", "false discovery rate", "mutual fund alpha", "replication"], "confidence": 0.85, "notes": "Labelled as a corrigendum/replication-type comment. No empirical identification; the evidence is entirely Monte Carlo. The original BSW results are cited from that paper rather than re-estimated."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.16, "usage": {"input": 4705, "output": 1858, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.056774, "request_id": "req_011CggqWrPF5w93NWWWPqFq8", "cached": false, "ts": "2026-11-03T23:09:30+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/47138d806fe22becc97f0e9adca4d53513dc2dc0cca0f2ad28d07de979cd9b2e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of the French PARE unemployment-insurance reform, which insured unemployed workers starting businesses, on industry-level firm creation, entrant quality, and employment of new and incumbent firms.\",\"strategy\":\"Difference-in-differences comparing industries with high versus low prereform shares of sole proprietorships (treatment intensity, by quartile) before and after the 2002 reform, with industry and month fixed effects.\",\"key_assumptions\":[\"Absent the reform, industries with high and low sole-proprietorship shares would have followed parallel trends in firm creation and outcomes\",\"Treatment intensity captures exposure to the reform rather than differential business-cycle sensitivity\"],\"threats_addressed\":[\"industry exposure to business cycle shocks\",\"alternative definitions of treatment intensity\",\"alternative event windows\",\"link between the creation increase and entry by unemployed entrepreneurs (ACCRE data)\"],\"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\":5},\"contribution\":{\"present\":true,\"sentence_index\":6}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":1,\"move\":\"other\"},{\"paragraph_index\":2,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"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\":\"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\":\"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\":[\"alternative_measure\",\"alternative_sample_period\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variables, POST dummy timing, treatment-quartile dummies, and observation counts. Most tables report difference-in-differences regressions at the industry or firm level, preceded by descriptive industry-composition and summary-statistics tables. 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\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'in line with' / 'supports' the experimentation view\",\"'(observable)' qualifier on quality\",\"'tentative' cost-benefit analysis\",\"'no significant' framing of null effects\",\"'we note that several factors' caveat\"]},\"claim_evidence\":[{\"claim\":\"The reform raised firm creation by about 14 pp in top- versus bottom-quartile treatment-intensity industries.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 3\"},{\"claim\":\"The cross-industry increase in firm creation correlates with the rise in entry by unemployed entrepreneurs.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figure 1\"},{\"claim\":\"Postreform firms show no significant change in failure, hiring, or growth rates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Entrepreneurs' education is unchanged and their growth expectations rise slightly after the reform.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Employment in small incumbents fell by 2.6 pp, roughly offsetting the direct job creation by new firms.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"New entrants are more productive than the incumbents they crowd out.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Reform benefits are about €350M against €100M in transfers, a net cost of about €10,000 per job.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Internet Appendix II (cost-benefit analysis)\"}],\"structural_summary\":\"The paper opens with an epigraph, then frames a tension over whether lowering entry barriers worsens entrant quality. It previews a quasi-experiment using a French unemployment-insurance reform and summarizes results on firm creation, quality, and crowding out. Positioning against the unemployment-insurance literature and a roadmap close the introduction. Later sections cover institutional details, a Lucas-style framework motivating treatment intensity, three administrative data sources, and the difference-in-differences design with a defense of its identifying assumption. Results on entry and quality come first, with robustness embedded, followed by aggregate employment and productivity effects. The conclusion restates the findings, stresses general-equilibrium crowding out, and gives a tentative cost-benefit estimate.\",\"writing_profile_keywords\":[\"policy reform quasi-experiment\",\"treatment intensity DiD\",\"experimentation view\",\"selection and quality\",\"crowding out\",\"administrative data\",\"cost-benefit\"],\"confidence\":0.85,\"notes\":\"Standard-error type and significance notation are not visible in the excerpts. The theory section is a guiding framework rather than a standalone contribution. The cost-benefit analysis is in the Internet Appendix and is not shown in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "household_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of the French PARE unemployment-insurance reform, which insured unemployed workers starting businesses, on industry-level firm creation, entrant quality, and employment of new and incumbent firms.", "strategy": "Difference-in-differences comparing industries with high versus low prereform shares of sole proprietorships (treatment intensity, by quartile) before and after the 2002 reform, with industry and month fixed effects.", "key_assumptions": ["Absent the reform, industries with high and low sole-proprietorship shares would have followed parallel trends in firm creation and outcomes", "Treatment intensity captures exposure to the reform rather than differential business-cycle sensitivity"], "threats_addressed": ["industry exposure to business cycle shocks", "alternative definitions of treatment intensity", "alternative event windows", "link between the creation increase and entry by unemployed entrepreneurs (ACCRE data)"], "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": 5}, "contribution": {"present": true, "sentence_index": 6}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "other"}, {"paragraph_index": 1, "move": "other"}, {"paragraph_index": 2, "move": "motivation"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"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": "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": "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": ["alternative_measure", "alternative_sample_period", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes define the dependent variables, POST dummy timing, treatment-quartile dummies, and observation counts. Most tables report difference-in-differences regressions at the industry or firm level, preceded by descriptive industry-composition and summary-statistics tables. 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": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'in line with' / 'supports' the experimentation view", "'(observable)' qualifier on quality", "'tentative' cost-benefit analysis", "'no significant' framing of null effects", "'we note that several factors' caveat"]}, "claim_evidence": [{"claim": "The reform raised firm creation by about 14 pp in top- versus bottom-quartile treatment-intensity industries.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 3"}, {"claim": "The cross-industry increase in firm creation correlates with the rise in entry by unemployed entrepreneurs.", "evidence_type": "table", "evidence_ref": "Table V; Figure 1"}, {"claim": "Postreform firms show no significant change in failure, hiring, or growth rates.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Entrepreneurs' education is unchanged and their growth expectations rise slightly after the reform.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Employment in small incumbents fell by 2.6 pp, roughly offsetting the direct job creation by new firms.", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "New entrants are more productive than the incumbents they crowd out.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Reform benefits are about €350M against €100M in transfers, a net cost of about €10,000 per job.", "evidence_type": "none_stated", "evidence_ref": "Internet Appendix II (cost-benefit analysis)"}], "structural_summary": "The paper opens with an epigraph, then frames a tension over whether lowering entry barriers worsens entrant quality. It previews a quasi-experiment using a French unemployment-insurance reform and summarizes results on firm creation, quality, and crowding out. Positioning against the unemployment-insurance literature and a roadmap close the introduction. Later sections cover institutional details, a Lucas-style framework motivating treatment intensity, three administrative data sources, and the difference-in-differences design with a defense of its identifying assumption. Results on entry and quality come first, with robustness embedded, followed by aggregate employment and productivity effects. The conclusion restates the findings, stresses general-equilibrium crowding out, and gives a tentative cost-benefit estimate.", "writing_profile_keywords": ["policy reform quasi-experiment", "treatment intensity DiD", "experimentation view", "selection and quality", "crowding out", "administrative data", "cost-benefit"], "confidence": 0.85, "notes": "Standard-error type and significance notation are not visible in the excerpts. The theory section is a guiding framework rather than a standalone contribution. The cost-benefit analysis is in the Internet Appendix and is not shown in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.6, "usage": {"input": 6826, "output": 2299, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074078, "request_id": "req_011CggqmkDaycHQHUhbeVViL", "cached": false, "ts": "2026-11-03T23:12:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/47257b571304bb2e171c60026012a1a3442993d967783d8bbf4d54bdc96d279a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"asset_pricing; institutional investors\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between changes in investor sentiment metrics and cross-sectional institutional versus individual demand shocks for volatile versus safe stocks.\",\"strategy\":\"Time-series correlations between orthogonalized sentiment changes and 13(f)-inferred institutional demand shocks across volatility deciles. These are supplemented by panel regressions, decomposition of demand into flow-induced versus manager-driven components, and sorts by client type and institution type.\",\"key_assumptions\":[\"The complement of 13(f) institutional demand captures individual investors' demand.\",\"Sentiment metrics such as the Baker–Wurgler (BW) index proxy for sentiment.\",\"Stock volatility proxies speculative versus safe stocks.\"],\"threats_addressed\":[\"Mismeasurement of retail demand via the inverse of 13(f) demand, checked against Odean retail broker data.\",\"Retail flows driving institutional trades.\",\"Changes in risk and alternative risk metrics.\",\"Choice of sentiment metric, using 17 alternatives.\"],\"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\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":0,\"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\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"p_values\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Notes state the sample period (1980–2010 quarterly), how variables are constructed (volatility from the prior 12 months of returns), and the panel structure. Results are reported as correlations with p-values and as panel regression coefficients.\",\"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\":[\"conditional framing ('if these metrics capture sentiment, then...')\",\"explicit correlation-not-causation caveat\",\"'largely consistent' qualifiers\",\"'some evidence'\"]},\"claim_evidence\":[{\"claim\":\"When sentiment rises, institutions buy volatile stocks from individuals and sell safe stocks to them.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Table II\"},{\"claim\":\"The result is not explained by changes in risk or by the choice of risk metric.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"None of 17 sentiment metrics captures individual demand shocks; 10 relate to institutional demand, and only these predict returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Underlying investor flows do not drive the aggregate relation between institutional demand and sentiment.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Institutional investment styles such as risk management and momentum explain much of institutional sentiment trading.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"},{\"claim\":\"Institutions with retail versus institutional client bases differ in sentiment trading.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"}],\"structural_summary\":\"The introduction opens by challenging the long-standing assumption that individuals are sentiment traders, then states the counterclaim and previews results. It follows with tests ruling out retail-driven explanations and closes on implications for market efficiency and for the prior sentiment literature. The data section is followed by core evidence on whose demand the BW metric captures, then a robustness section on 17 alternative metrics. A long mechanism section examines retail measurement, client base, flows, stock-level panels, institution types, and individual institutions. The conclusion restates findings with an explicit causation caveat.\",\"writing_profile_keywords\":[\"conventional-wisdom reversal\",\"13(f) holdings\",\"investor sentiment\",\"mechanism exploration\",\"correlation evidence\",\"reinterpretation of literature\"],\"confidence\":0.8,\"notes\":\"No figures or equations. A roadmap is not visible in the truncated introduction. Significance appears to be reported via p-values in correlation tables; regression tables may also report t-statistics, which is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "asset_pricing; institutional investors", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between changes in investor sentiment metrics and cross-sectional institutional versus individual demand shocks for volatile versus safe stocks.", "strategy": "Time-series correlations between orthogonalized sentiment changes and 13(f)-inferred institutional demand shocks across volatility deciles. These are supplemented by panel regressions, decomposition of demand into flow-induced versus manager-driven components, and sorts by client type and institution type.", "key_assumptions": ["The complement of 13(f) institutional demand captures individual investors' demand.", "Sentiment metrics such as the Baker–Wurgler (BW) index proxy for sentiment.", "Stock volatility proxies speculative versus safe stocks."], "threats_addressed": ["Mismeasurement of retail demand via the inverse of 13(f) demand, checked against Odean retail broker data.", "Retail flows driving institutional trades.", "Changes in risk and alternative risk metrics.", "Choice of sentiment metric, using 17 alternatives."], "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": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "related_literature_positioning"}], "contribution_paragraph_index": 0, "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": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "p_values", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample"]}, "tables": {"convention_notes": "Notes state the sample period (1980–2010 quarterly), how variables are constructed (volatility from the prior 12 months of returns), and the panel structure. Results are reported as correlations with p-values and as panel regression coefficients.", "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": ["conditional framing ('if these metrics capture sentiment, then...')", "explicit correlation-not-causation caveat", "'largely consistent' qualifiers", "'some evidence'"]}, "claim_evidence": [{"claim": "When sentiment rises, institutions buy volatile stocks from individuals and sell safe stocks to them.", "evidence_type": "table", "evidence_ref": "Table I, Table II"}, {"claim": "The result is not explained by changes in risk or by the choice of risk metric.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "None of 17 sentiment metrics captures individual demand shocks; 10 relate to institutional demand, and only these predict returns.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Underlying investor flows do not drive the aggregate relation between institutional demand and sentiment.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Institutional investment styles such as risk management and momentum explain much of institutional sentiment trading.", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}, {"claim": "Institutions with retail versus institutional client bases differ in sentiment trading.", "evidence_type": "table", "evidence_ref": "Table V"}], "structural_summary": "The introduction opens by challenging the long-standing assumption that individuals are sentiment traders, then states the counterclaim and previews results. It follows with tests ruling out retail-driven explanations and closes on implications for market efficiency and for the prior sentiment literature. The data section is followed by core evidence on whose demand the BW metric captures, then a robustness section on 17 alternative metrics. A long mechanism section examines retail measurement, client base, flows, stock-level panels, institution types, and individual institutions. The conclusion restates findings with an explicit causation caveat.", "writing_profile_keywords": ["conventional-wisdom reversal", "13(f) holdings", "investor sentiment", "mechanism exploration", "correlation evidence", "reinterpretation of literature"], "confidence": 0.8, "notes": "No figures or equations. A roadmap is not visible in the truncated introduction. Significance appears to be reported via p-values in correlation tables; regression tables may also report t-statistics, which is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.76, "usage": {"input": 6179, "output": 2018, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06587, "request_id": "req_011CggqKAV5xYLQsCH9gaujz", "cached": false, "ts": "2026-11-03T23:06:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/49b8df40aa601f891206b51c199e0b8765acd0a3df0cc3f92104a9bb879d7b97.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between past factor-related returns (and the flows they attract) and funds' subsequent alphas, holding AUM or factor-adjusted expected returns fixed\",\"strategy\":\"Annual conditional portfolio sorts: funds are sorted on AUM or seven-factor alpha, then split into terciles of past factor-related returns, and future benchmark-adjusted returns are compared. Cross-sectional variation in trading costs and style-level flow-induced trading is used to support a scale-effect mechanism.\",\"key_assumptions\":[\"Factor-related returns reflect no managerial skill, so flows they attract are uninformative\",\"The seven-factor model adequately separates factor exposure from skill\",\"Diminishing returns to scale operate through trading costs\"],\"threats_addressed\":[\"Differences in prior real alpha\",\"Differences in fund size alone\",\"Crisis periods and extreme past returns\",\"Choice of factor model and estimation window\",\"Cap-based vs style-based fund classification\",\"Style-level crowding\"],\"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\":3}},\"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\":\"robustness_mention\"},{\"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\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"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\",\"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\":\"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\":\"appendix\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Table notes open with the sorting procedure ('For each calendar year of data...'), then describe the conditional double sorts into terciles or quintiles, the estimation windows, and AUM-weighted or equal-weighted annualized returns. Most tables are portfolio-sort tables; some report time-series regressions on style returns.\",\"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' framing\",\"'likely due to' for the mechanism\",\"'in this sense' qualifiers\",\"'can be explained by'\"]},\"claim_evidence\":[{\"claim\":\"Fund flows rise with factor-related returns after controlling for factor-adjusted expected returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Within AUM quintiles, top-tercile FRR funds underperform bottom-tercile FRR funds by about 300-400 bps per year\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The underperformance persists after controlling for prior seven-factor alpha\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Underperformance of high-FRR funds is concentrated among funds with high trading costs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Styles with high flow-induced trading earn lower future returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table BIII\"},{\"claim\":\"Negative aggregate active-fund performance is driven mainly by high-FRR funds; investors lose about $15bn per year\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Results are robust to alternative windows, factor models, subsamples, and controlling for AUM and alpha jointly\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables AI-AIII, Figure A1\"}],\"structural_summary\":\"The introduction opens with a puzzle: Berk–Green predicts zero net alpha, yet aggregate alpha is negative. It challenges BG's assumption of rational investors and previews the data and factor model. It then walks through the results sequentially: flows respond to factor returns, high-FRR funds underperform, the mechanism runs through trading costs, and aggregate performance follows. It closes with the contribution, contrast with a contemporaneous study, and a roadmap. The body moves from data, to flows, to future alphas, to mechanism (fund-level scale and style crowding), to aggregate implications. A brief conclusion restates the findings, and appendices hold the robustness checks.\",\"writing_profile_keywords\":[\"portfolio sorts\",\"mutual fund flows\",\"decreasing returns to scale\",\"sequential results narrative\",\"first-person singular\",\"magnitudes in bps\",\"appendix robustness\"],\"confidence\":0.82,\"notes\":\"Single-author paper written in the first person. The causal claim about flows causing underperformance is framed as 'consistent with' and 'likely'. Statistical significance conventions (stars, t-stats, p-values) are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between past factor-related returns (and the flows they attract) and funds' subsequent alphas, holding AUM or factor-adjusted expected returns fixed", "strategy": "Annual conditional portfolio sorts: funds are sorted on AUM or seven-factor alpha, then split into terciles of past factor-related returns, and future benchmark-adjusted returns are compared. Cross-sectional variation in trading costs and style-level flow-induced trading is used to support a scale-effect mechanism.", "key_assumptions": ["Factor-related returns reflect no managerial skill, so flows they attract are uninformative", "The seven-factor model adequately separates factor exposure from skill", "Diminishing returns to scale operate through trading costs"], "threats_addressed": ["Differences in prior real alpha", "Differences in fund size alone", "Crisis periods and extreme past returns", "Choice of factor model and estimation window", "Cap-based vs style-based fund classification", "Style-level crowding"], "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": 3}}, "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": "robustness_mention"}, {"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": "identification_defense"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"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", "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": "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": "appendix", "types": ["alternative_specification", "subsample", "alternative_measure", "alternative_sample_period"]}, "tables": {"convention_notes": "Table notes open with the sorting procedure ('For each calendar year of data...'), then describe the conditional double sorts into terciles or quintiles, the estimation windows, and AUM-weighted or equal-weighted annualized returns. Most tables are portfolio-sort tables; some report time-series regressions on style returns.", "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' framing", "'likely due to' for the mechanism", "'in this sense' qualifiers", "'can be explained by'"]}, "claim_evidence": [{"claim": "Fund flows rise with factor-related returns after controlling for factor-adjusted expected returns", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Within AUM quintiles, top-tercile FRR funds underperform bottom-tercile FRR funds by about 300-400 bps per year", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The underperformance persists after controlling for prior seven-factor alpha", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Underperformance of high-FRR funds is concentrated among funds with high trading costs", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Styles with high flow-induced trading earn lower future returns", "evidence_type": "table", "evidence_ref": "Table V, Table BIII"}, {"claim": "Negative aggregate active-fund performance is driven mainly by high-FRR funds; investors lose about $15bn per year", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Results are robust to alternative windows, factor models, subsamples, and controlling for AUM and alpha jointly", "evidence_type": "table", "evidence_ref": "Tables AI-AIII, Figure A1"}], "structural_summary": "The introduction opens with a puzzle: Berk–Green predicts zero net alpha, yet aggregate alpha is negative. It challenges BG's assumption of rational investors and previews the data and factor model. It then walks through the results sequentially: flows respond to factor returns, high-FRR funds underperform, the mechanism runs through trading costs, and aggregate performance follows. It closes with the contribution, contrast with a contemporaneous study, and a roadmap. The body moves from data, to flows, to future alphas, to mechanism (fund-level scale and style crowding), to aggregate implications. A brief conclusion restates the findings, and appendices hold the robustness checks.", "writing_profile_keywords": ["portfolio sorts", "mutual fund flows", "decreasing returns to scale", "sequential results narrative", "first-person singular", "magnitudes in bps", "appendix robustness"], "confidence": 0.82, "notes": "Single-author paper written in the first person. The causal claim about flows causing underperformance is framed as 'consistent with' and 'likely'. Statistical significance conventions (stars, t-stats, p-values) are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.68, "usage": {"input": 8520, "output": 2362, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082114, "request_id": "req_011CggqwuLnwHm462RUKEHnj", "cached": false, "ts": "2026-11-03T23:15:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4bbab0c1538c5cd05b140f4d34c878f3b7905731a29e2768cec70a6d992e0791.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"delegated portfolio management / mutual funds / managerial compensation\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How a fund's mid-year distance from its benchmark affects second-half risk-shifting (relative tracking-error volatility), and whether this differs for managers with versus without performance-based contracts\",\"strategy\":\"Median quantile regressions of the risk adjustment ratio on Distance and excess return. Contract types are hand-collected from SAI filings. Treated funds (performance contracts) are matched to control funds (no performance contracts) on observables. A placebo test randomly reassigns benchmarks.\",\"key_assumptions\":[\"After matching on observables, contract type is uncorrelated with unobservables\",\"The self-designated prospectus benchmark is the relevant compensation benchmark\",\"The ratio of tracking-error volatility captures managerial risk-shifting\"],\"threats_addressed\":[\"selection into contract type on observables\",\"spurious benchmark effects (placebo randomization)\",\"competing flow-performance tournament incentives\",\"managerial ownership, active share and tenure as confounds\",\"alternative risk-shifting measures\"],\"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\":3}},\"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\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"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\":4,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"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\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"alternative_measure\",\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes define dependent variables (risk adjustment ratio as a second-half versus first-half tracking-error volatility ratio) and give the quantile-regression specification. Pooled OLS and median quantile regressions are reported side by side. A bootstrap placebo table summarizes 500 benchmark randomizations.\",\"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\":[\"'our results suggest'\",\"stating that the contract and tournament channels are not mutually exclusive\",\"declaring agnosticism on contract optimality\",\"explicitly stating the matching assumption\"],\"typical_devices_note\":null},\"claim_evidence\":[{\"claim\":\"An asymmetric bonus contract is valued like an exchange option whose vega peaks when the portfolio's return is close to the benchmark's\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendix A (Margrabe 1978)\"},{\"claim\":\"Funds performing closer to the benchmark at mid-year increase relative risk more in the second half\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Risk-shifting is concentrated among funds with a clear compensation benchmark and is absent without performance pay\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Matched treated-control comparison supports a causal effect of contracts on risk-shifting\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Manager ownership reduces risk-shifting, active share increases it, and tenure mitigates it; Distance remains important\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Random benchmark assignment does not reproduce the effect\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"The contract-based incentive dominates flow-performance tournament incentives\",\"evidence_type\":\"table\",\"evidence_ref\":\"Section III.G\"}],\"structural_summary\":\"The paper opens with an agency-and-contracting motivation, then sketches an option-based theory of the manager's bonus that predicts a hypothesis, priced in the appendix. After describing the data, one long empirical section moves through baseline quantile regressions, hand-collected contract heterogeneity, a matched-sample causal analysis, cross-sectional moderators, a placebo, a horse race against flow incentives, and robustness checks. A brief conclusion bounds the interpretation.\",\"writing_profile_keywords\":[\"option-pricing hypothesis\",\"quantile regression\",\"hand-collected compensation data\",\"matching for causality\",\"placebo bootstrap\",\"tournament versus contract\"],\"confidence\":0.8,\"notes\":\"The empirical section is a single large section with robustness checks as subsections, so robustness location is coded as mixed. The theory is light: one hypothesis, with an appendix derivation.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "delegated portfolio management / mutual funds / managerial compensation", "primary_method": "reduced_form_regression", "identification": {"estimand": "How a fund's mid-year distance from its benchmark affects second-half risk-shifting (relative tracking-error volatility), and whether this differs for managers with versus without performance-based contracts", "strategy": "Median quantile regressions of the risk adjustment ratio on Distance and excess return. Contract types are hand-collected from SAI filings. Treated funds (performance contracts) are matched to control funds (no performance contracts) on observables. A placebo test randomly reassigns benchmarks.", "key_assumptions": ["After matching on observables, contract type is uncorrelated with unobservables", "The self-designated prospectus benchmark is the relevant compensation benchmark", "The ratio of tracking-error volatility captures managerial risk-shifting"], "threats_addressed": ["selection into contract type on observables", "spurious benchmark effects (placebo randomization)", "competing flow-performance tournament incentives", "managerial ownership, active share and tenure as confounds", "alternative risk-shifting measures"], "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": 3}}, "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": "related_literature_positioning"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 4, "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": 4, "contribution_types": ["new_evidence", "new_data_or_measure", "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": "results"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["placebo", "alternative_measure", "subsample", "alternative_specification"]}, "tables": {"convention_notes": "Table notes define dependent variables (risk adjustment ratio as a second-half versus first-half tracking-error volatility ratio) and give the quantile-regression specification. Pooled OLS and median quantile regressions are reported side by side. A bootstrap placebo table summarizes 500 benchmark randomizations.", "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": ["'our results suggest'", "stating that the contract and tournament channels are not mutually exclusive", "declaring agnosticism on contract optimality", "explicitly stating the matching assumption"], "typical_devices_note": null}, "claim_evidence": [{"claim": "An asymmetric bonus contract is valued like an exchange option whose vega peaks when the portfolio's return is close to the benchmark's", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendix A (Margrabe 1978)"}, {"claim": "Funds performing closer to the benchmark at mid-year increase relative risk more in the second half", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Risk-shifting is concentrated among funds with a clear compensation benchmark and is absent without performance pay", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Matched treated-control comparison supports a causal effect of contracts on risk-shifting", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Manager ownership reduces risk-shifting, active share increases it, and tenure mitigates it; Distance remains important", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Random benchmark assignment does not reproduce the effect", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "The contract-based incentive dominates flow-performance tournament incentives", "evidence_type": "table", "evidence_ref": "Section III.G"}], "structural_summary": "The paper opens with an agency-and-contracting motivation, then sketches an option-based theory of the manager's bonus that predicts a hypothesis, priced in the appendix. After describing the data, one long empirical section moves through baseline quantile regressions, hand-collected contract heterogeneity, a matched-sample causal analysis, cross-sectional moderators, a placebo, a horse race against flow incentives, and robustness checks. A brief conclusion bounds the interpretation.", "writing_profile_keywords": ["option-pricing hypothesis", "quantile regression", "hand-collected compensation data", "matching for causality", "placebo bootstrap", "tournament versus contract"], "confidence": 0.8, "notes": "The empirical section is a single large section with robustness checks as subsections, so robustness location is coded as mixed. The theory is light: one hypothesis, with an appendix derivation."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.4, "usage": {"input": 6147, "output": 2203, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069442, "request_id": "req_011CggqWCDpcyqTTXrKNNrps", "cached": false, "ts": "2026-11-03T23:09:25+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4d250a68e2dae22f82812846220efc3f0079c6faec898686434b88e0520563dc.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"bank supervision and regulation\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of greater supervisory attention (hours) on bank risk, profitability and growth\",\"strategy\":\"Top size-rank status within a Federal Reserve district as a source of plausibly exogenous variation in supervisory attention; nearest-neighbor matching of Top BHCs to similar non-top BHCs in other districts; district-quarter fixed effects; 2SLS with Top as an instrument for supervisory hours (2006-2014)\",\"key_assumptions\":[\"Being among the largest BHCs in a district is not related to unobserved factors that also affect bank performance\",\"Matched BHCs are otherwise similar to Top BHCs\",\"Top status raises supervisory hours conditional on size, complexity, risk and other controls\"],\"threats_addressed\":[\"Unobserved district-level differences in risk\",\"Differences in competitiveness and franchise value\",\"Size and complexity confounds\"],\"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\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"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\",\"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\":\"empirical_design\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"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\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the sample (BHC-quarters, period), the treatment definition (top five in district-quarter or within 25% of the top five), matching criteria, and how variables are constructed (rolling eight-quarter forward standard deviations, one-year forward growth). Some notes refer readers to the Internet Appendix for variable details. Clustering by BHC is stated in the results text rather than in the visible note excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"suggest\",\"plausibly exogenous\",\"explicit caveats on natural experiment assumptions\",\"acknowledging mechanisms not identified\",\"may\"]},\"claim_evidence\":[{\"claim\":\"Top size-ranked BHCs in a district get more supervisory hours, net of size, complexity and risk\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III, Figure 1\"},{\"claim\":\"Top BHCs hold less risky loans and are less volatile than matched peers\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Top BHCs are less sensitive to industry downturns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"No meaningful difference in profitability, growth or Sharpe ratio between Top BHCs and matches\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Results hold with district-quarter fixed effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"2SLS using Top as an instrument for supervisory hours confirms the effects for 2006-2014\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"The introduction motivates supervision as distinct from regulation, flags the gap that its effect is understudied, and previews the new hours data, the Top-rank identification strategy and the risk-down/performance-unchanged results. It then defends identification, sketches possible mechanisms, positions the paper in the literature, and closes with a roadmap. The body describes the institutional setting, shows that top rank predicts supervisory hours, builds a matched sample, and presents differences in means, downturn tests and district fixed-effects results. Further analyses and 2SLS follow, and the conclusion states caveats.\",\"writing_profile_keywords\":[\"bank supervision\",\"supervisory hours data\",\"matched sample\",\"district rank identification\",\"institutional detail\",\"risk vs performance tradeoff\",\"caveated causal claims\",\"2SLS\"],\"confidence\":0.82,\"notes\":\"Section I is institutional background. Statistical-significance reporting conventions are not visible in the excerpts. The introduction gives no numeric magnitudes, and the abstract contains none either.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "bank supervision and regulation", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of greater supervisory attention (hours) on bank risk, profitability and growth", "strategy": "Top size-rank status within a Federal Reserve district as a source of plausibly exogenous variation in supervisory attention; nearest-neighbor matching of Top BHCs to similar non-top BHCs in other districts; district-quarter fixed effects; 2SLS with Top as an instrument for supervisory hours (2006-2014)", "key_assumptions": ["Being among the largest BHCs in a district is not related to unobserved factors that also affect bank performance", "Matched BHCs are otherwise similar to Top BHCs", "Top status raises supervisory hours conditional on size, complexity, risk and other controls"], "threats_addressed": ["Unobserved district-level differences in risk", "Differences in competitiveness and franchise value", "Size and complexity confounds"], "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": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"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", "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": "empirical_design"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"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", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Table notes define the sample (BHC-quarters, period), the treatment definition (top five in district-quarter or within 25% of the top five), matching criteria, and how variables are constructed (rolling eight-quarter forward standard deviations, one-year forward growth). Some notes refer readers to the Internet Appendix for variable details. Clustering by BHC is stated in the results text rather than in the visible note excerpts.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "suggest", "plausibly exogenous", "explicit caveats on natural experiment assumptions", "acknowledging mechanisms not identified", "may"]}, "claim_evidence": [{"claim": "Top size-ranked BHCs in a district get more supervisory hours, net of size, complexity and risk", "evidence_type": "table", "evidence_ref": "Table II, Table III, Figure 1"}, {"claim": "Top BHCs hold less risky loans and are less volatile than matched peers", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Top BHCs are less sensitive to industry downturns", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "No meaningful difference in profitability, growth or Sharpe ratio between Top BHCs and matches", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Results hold with district-quarter fixed effects", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "2SLS using Top as an instrument for supervisory hours confirms the effects for 2006-2014", "evidence_type": "table", "evidence_ref": "Table X"}], "structural_summary": "The introduction motivates supervision as distinct from regulation, flags the gap that its effect is understudied, and previews the new hours data, the Top-rank identification strategy and the risk-down/performance-unchanged results. It then defends identification, sketches possible mechanisms, positions the paper in the literature, and closes with a roadmap. The body describes the institutional setting, shows that top rank predicts supervisory hours, builds a matched sample, and presents differences in means, downturn tests and district fixed-effects results. Further analyses and 2SLS follow, and the conclusion states caveats.", "writing_profile_keywords": ["bank supervision", "supervisory hours data", "matched sample", "district rank identification", "institutional detail", "risk vs performance tradeoff", "caveated causal claims", "2SLS"], "confidence": 0.82, "notes": "Section I is institutional background. Statistical-significance reporting conventions are not visible in the excerpts. The introduction gives no numeric magnitudes, and the abstract contains none either."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.26, "usage": {"input": 6964, "output": 2133, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07131, "request_id": "req_011CggqypdppARXsUNAuEEjC", "cached": false, "ts": "2026-11-03T23:15:38+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/4fd1a8fbc5fb0f8a2a98a053acf07d33242f93e61efbf736db11a872927484a1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Return predictability and risk-exposure associations of the market-to-value and value-to-book components of market-to-book\",\"strategy\":\"RRV valuation-model decomposition of market-to-book, followed by portfolio sorts, Fama-MacBeth regressions, beta-spread comparisons and second-pass pricing tests on component-sorted portfolios\",\"key_assumptions\":[\"Industry multiples-based valuation model captures fundamental value\",\"Covariances and characteristics are estimated with precision (joint hypothesis)\"],\"threats_addressed\":[\"joint hypothesis problem acknowledged\",\"microstructure noise in equal-weighted returns (RW/VW weighting)\",\"small-stock dominance (size-conditional sorts)\",\"industry classification choice (alternatives for robustness)\"],\"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\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"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\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"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\":\"contribution_statement\"},{\"paragraph_index\":13,\"move\":\"other\"},{\"paragraph_index\":14,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\"],\"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\":\"results\"},{\"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\",\"alternative_measure\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Notes state the sample period, the estimation method (Fama-MacBeth, ordered logit, second-pass OLS), the weighting scheme (RW/VW), units (monthly returns in percent) and the panel structure. P-values appear in parentheses for some 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\":[\"joint hypothesis caveat\",\"'unlikely explanation' phrasing\",\"'consistent with' language\",\"explicit disclaimer of not taking a stance on whether characteristics are priced\",\"'mixed evidence' labeling\"]},\"claim_evidence\":[{\"claim\":\"Market-to-value drives the entire value strategy return; value-to-book has insignificant return spread\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III, Figure 1\"},{\"claim\":\"IST shock exposure of value strategy stems mainly from value-to-book\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Assets-in-place/operating leverage differences arise from value-to-book only\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Cash flow duration differences derive from unpriced value-to-book\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Analyst risk rating association with market-to-book driven by value-to-book\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Cash flow risk evidence is mixed depending on approach\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Ultimate consumption risk in returns is linked to market-to-value; Parker-Julliard model prices test assets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table X, Figure 2\"},{\"claim\":\"Earnings announcement surprises reverse around portfolio formation for market-to-value, consistent with overextrapolation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Market-to-value premium concentrates where limits to arbitrage bind and when arbitrage capital is low\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XII-XIV\"}],\"structural_summary\":\"The paper motivates with the value premium and the unresolved debate over its source. It applies the RRV decomposition to split market-to-book into a priced component and an unpriced component. It then systematically tests whether characteristics and risk exposures proposed by prominent theories load on the priced component. Results proceed in three blocks: return predictability, aggregate risk and production-based explanations, and mispricing/limits to arbitrage. The intro previews each theory's test in turn, and the paper closes with a joint-hypothesis caveat.\",\"writing_profile_keywords\":[\"decomposition\",\"theory horse-race\",\"value premium\",\"portfolio sorts\",\"replication of prior tests\",\"limits to arbitrage\",\"joint hypothesis caveat\"],\"confidence\":0.85,\"notes\":\"Standard errors are not described in the visible table notes. Section IV is labeled 'theory' by the role rule, but it consists of empirical tests of theories, so it is annotated as results.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Return predictability and risk-exposure associations of the market-to-value and value-to-book components of market-to-book", "strategy": "RRV valuation-model decomposition of market-to-book, followed by portfolio sorts, Fama-MacBeth regressions, beta-spread comparisons and second-pass pricing tests on component-sorted portfolios", "key_assumptions": ["Industry multiples-based valuation model captures fundamental value", "Covariances and characteristics are estimated with precision (joint hypothesis)"], "threats_addressed": ["joint hypothesis problem acknowledged", "microstructure noise in equal-weighted returns (RW/VW weighting)", "small-stock dominance (size-conditional sorts)", "industry classification choice (alternatives for robustness)"], "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": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "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": "preview_of_results"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"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": "contribution_statement"}, {"paragraph_index": 13, "move": "other"}, {"paragraph_index": 14, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence"], "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": "results"}, {"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", "alternative_measure", "subsample"]}, "tables": {"convention_notes": "Notes state the sample period, the estimation method (Fama-MacBeth, ordered logit, second-pass OLS), the weighting scheme (RW/VW), units (monthly returns in percent) and the panel structure. P-values appear in parentheses for some 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": ["joint hypothesis caveat", "'unlikely explanation' phrasing", "'consistent with' language", "explicit disclaimer of not taking a stance on whether characteristics are priced", "'mixed evidence' labeling"]}, "claim_evidence": [{"claim": "Market-to-value drives the entire value strategy return; value-to-book has insignificant return spread", "evidence_type": "table", "evidence_ref": "Table II, Table III, Figure 1"}, {"claim": "IST shock exposure of value strategy stems mainly from value-to-book", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Assets-in-place/operating leverage differences arise from value-to-book only", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Cash flow duration differences derive from unpriced value-to-book", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Analyst risk rating association with market-to-book driven by value-to-book", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Cash flow risk evidence is mixed depending on approach", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Ultimate consumption risk in returns is linked to market-to-value; Parker-Julliard model prices test assets", "evidence_type": "table", "evidence_ref": "Table V, Table X, Figure 2"}, {"claim": "Earnings announcement surprises reverse around portfolio formation for market-to-value, consistent with overextrapolation", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Market-to-value premium concentrates where limits to arbitrage bind and when arbitrage capital is low", "evidence_type": "table", "evidence_ref": "Tables XII-XIV"}], "structural_summary": "The paper motivates with the value premium and the unresolved debate over its source. It applies the RRV decomposition to split market-to-book into a priced component and an unpriced component. It then systematically tests whether characteristics and risk exposures proposed by prominent theories load on the priced component. Results proceed in three blocks: return predictability, aggregate risk and production-based explanations, and mispricing/limits to arbitrage. The intro previews each theory's test in turn, and the paper closes with a joint-hypothesis caveat.", "writing_profile_keywords": ["decomposition", "theory horse-race", "value premium", "portfolio sorts", "replication of prior tests", "limits to arbitrage", "joint hypothesis caveat"], "confidence": 0.85, "notes": "Standard errors are not described in the visible table notes. Section IV is labeled 'theory' by the role rule, but it consists of empirical tests of theories, so it is annotated as results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.33, "usage": {"input": 9727, "output": 2317, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086042, "request_id": "req_011CggqbFPiUXWGbNJTiJhvW", "cached": false, "ts": "2026-11-03T23:10:33+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/514cb7818a090ac577eed65bbe3edbf8bbcddfa50ff5bfb771f90a4145a8ab3b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-country conditional associations between firm-level corporate control or ownership concentration and legal origin, development, and institutional characteristics.\",\"strategy\":\"Firm-level probit, ordered probit, OLS and hierarchical models on a large hand-extended ORBIS sample. Legal origin dummies and institutional proxies are used, with controls for continent, industry, and development. Multiple proxies are used to address measurement error.\",\"key_assumptions\":[\"Legal origin is treated as a reduced-form correlate; no causal identification is claimed.\",\"Control proxies (20% cutoff, Shapley-Shubik) capture actual control.\"],\"threats_addressed\":[\"Measurement error in institutional proxies, addressed with multiple proxies.\",\"Firm size and age composition (Holderness critique), addressed via heterogeneity analysis.\",\"Confounding by continent, industry, and development level.\"],\"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\":7}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":18,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":30,\"move\":\"preview_of_approach\"},{\"paragraph_index\":30,\"move\":\"identification_defense\"},{\"paragraph_index\":31,\"move\":\"preview_of_results\"},{\"paragraph_index\":56,\"move\":\"contribution_statement\"},{\"paragraph_index\":56,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":57,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":56,\"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\":\"results\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"results\"},{\"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\":[\"alternative_measure\",\"subsample\",\"alternative_specification\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Notes name the estimator (probit ML average marginal effects, ordered probit, OLS) and define dependent variables and their categories in detail. Firm-level estimates are paired with country-level institutional proxies. Standard error treatment is not visible in the excerpts.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\",\"figures_note\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"correlates/associated with rather than causes\",\"consistent with / in line with theory\",\"explicit disclaimer that cross-country associations cannot establish causality\",\"suggests / may prevent\",\"first step framing\"]},\"claim_evidence\":[{\"claim\":\"Share of controlled firms is highest in French civil-law and lowest in common-law countries, net of continent, industry, and development.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-IV; Figure 6\"},{\"claim\":\"Equity blocks exist in over 80% of noncontrolled firms across regions.\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I; Section II\"},{\"claim\":\"The legal origin–control link holds for large, medium, and small firms and for young and old firms.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The negative income–control correlation appears only among large firms and is zero for small and medium firms.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Shareholder protection correlates with dispersed ownership, while creditor rights do not.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Legal formalism and entry regulation are only weakly related to control.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII-VIII\"},{\"claim\":\"Labor regulation and union power correlate strongly with corporate control.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; Figure 8\"}],\"structural_summary\":\"A large descriptive and correlational cross-country study. The introduction opens by noting the lack of consensus on basic correlations. It then previews a new hand-extended dataset and enumerates numbered findings in two blocks: legal origin with heterogeneity, and institutional correlates. Each finding is tied to supporting or challenged prior work, and the opening block closes with a synthesis favoring both legal-origin and political theories, followed by a roadmap. The body moves from data construction and control measures to world-wide descriptive patterns. It then runs legal-origin regressions with sensitivity checks, firm size and age heterogeneity, and institutional correlates. The conclusion restates the three-step findings and lists future research directions.\",\"writing_profile_keywords\":[\"law and finance\",\"descriptive anatomy\",\"new dataset\",\"enumerated findings\",\"cross-country correlations\",\"heterogeneity by firm size\",\"associational language\",\"theory adjudication\"],\"confidence\":0.8,\"notes\":\"The introduction paragraph list is fragmented by line breaks, so paragraph indices are approximate. The first contribution (the dataset) is mostly cut from the intro excerpt but is stated in the conclusion. Sections 4–5 are table and figure panel artifacts; section 5 contains the sensitivity analysis. The heterogeneity section is rule-labeled robustness but functions as results.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "international_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-country conditional associations between firm-level corporate control or ownership concentration and legal origin, development, and institutional characteristics.", "strategy": "Firm-level probit, ordered probit, OLS and hierarchical models on a large hand-extended ORBIS sample. Legal origin dummies and institutional proxies are used, with controls for continent, industry, and development. Multiple proxies are used to address measurement error.", "key_assumptions": ["Legal origin is treated as a reduced-form correlate; no causal identification is claimed.", "Control proxies (20% cutoff, Shapley-Shubik) capture actual control."], "threats_addressed": ["Measurement error in institutional proxies, addressed with multiple proxies.", "Firm size and age composition (Holderness critique), addressed via heterogeneity analysis.", "Confounding by continent, industry, and development level."], "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": 7}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 18, "move": "related_literature_positioning"}, {"paragraph_index": 30, "move": "preview_of_approach"}, {"paragraph_index": 30, "move": "identification_defense"}, {"paragraph_index": 31, "move": "preview_of_results"}, {"paragraph_index": 56, "move": "contribution_statement"}, {"paragraph_index": 56, "move": "policy_or_broader_implication"}, {"paragraph_index": 57, "move": "roadmap"}], "contribution_paragraph_index": 56, "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": "results"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "results"}, {"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": ["alternative_measure", "subsample", "alternative_specification", "alternative_sample_period"]}, "tables": {"convention_notes": "Notes name the estimator (probit ML average marginal effects, ordered probit, OLS) and define dependent variables and their categories in detail. Firm-level estimates are paired with country-level institutional proxies. Standard error treatment is not visible in the excerpts.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear", "figures_note": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["correlates/associated with rather than causes", "consistent with / in line with theory", "explicit disclaimer that cross-country associations cannot establish causality", "suggests / may prevent", "first step framing"]}, "claim_evidence": [{"claim": "Share of controlled firms is highest in French civil-law and lowest in common-law countries, net of continent, industry, and development.", "evidence_type": "table", "evidence_ref": "Tables II-IV; Figure 6"}, {"claim": "Equity blocks exist in over 80% of noncontrolled firms across regions.", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I; Section II"}, {"claim": "The legal origin–control link holds for large, medium, and small firms and for young and old firms.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "The negative income–control correlation appears only among large firms and is zero for small and medium firms.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Shareholder protection correlates with dispersed ownership, while creditor rights do not.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Legal formalism and entry regulation are only weakly related to control.", "evidence_type": "table", "evidence_ref": "Tables VII-VIII"}, {"claim": "Labor regulation and union power correlate strongly with corporate control.", "evidence_type": "table", "evidence_ref": "Table IX; Figure 8"}], "structural_summary": "A large descriptive and correlational cross-country study. The introduction opens by noting the lack of consensus on basic correlations. It then previews a new hand-extended dataset and enumerates numbered findings in two blocks: legal origin with heterogeneity, and institutional correlates. Each finding is tied to supporting or challenged prior work, and the opening block closes with a synthesis favoring both legal-origin and political theories, followed by a roadmap. The body moves from data construction and control measures to world-wide descriptive patterns. It then runs legal-origin regressions with sensitivity checks, firm size and age heterogeneity, and institutional correlates. The conclusion restates the three-step findings and lists future research directions.", "writing_profile_keywords": ["law and finance", "descriptive anatomy", "new dataset", "enumerated findings", "cross-country correlations", "heterogeneity by firm size", "associational language", "theory adjudication"], "confidence": 0.8, "notes": "The introduction paragraph list is fragmented by line breaks, so paragraph indices are approximate. The first contribution (the dataset) is mostly cut from the intro excerpt but is stated in the conclusion. Sections 4–5 are table and figure panel artifacts; section 5 contains the sensitivity analysis. The heterogeneity section is rule-labeled robustness but functions as results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 23.05, "usage": {"input": 9492, "output": 2281, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084382, "request_id": "req_011CggqmMSdP6J2v1avvVgLV", "cached": false, "ts": "2026-11-03T23:12:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/52851d7bd0d9124739586217bd63b0bddc7421cc0768630797fc4b76e1bdba14.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; analytical characterization of equilibrium in a continuous-time model\",\"key_assumptions\":[\"fixed exogenous riskless rate\",\"hedgers' asset demand independent of wealth\",\"arbitrageurs have CRRA utility\",\"no asymmetric information or leverage constraints\"],\"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\":2},\"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\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"}],\"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\":\"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\":\"No tables; results are delivered as propositions and numerical illustrations in figures with parameter values stated in the captions.\",\"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\":[\"modal 'can' for possibility results\",\"conditioning on parameter ranges (e.g., CRRA below one)\",\"explicit statement of simplifying assumptions and their scope\"]},\"claim_evidence\":[{\"claim\":\"Arbitrageur risk aversion equals static CRRA plus a forward-looking intertemporal hedging component\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.C.2\"},{\"claim\":\"More risk-averse hedgers can make arbitrageurs with CRRA below one endogenously more risk-averse\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Section II.C.2\"},{\"claim\":\"Risk premia can rise with arbitrageur wealth at low wealth even as Sharpe ratios fall, due to endogenous risk\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2; Section III.B\"},{\"claim\":\"Arbitrageur wealth is the single priced factor, and aggregate illiquidity captures it\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"Zero-supply assumption is without loss of generality with long-lived hedgers\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 12; Section V\"},{\"claim\":\"Aggregate liquidity is empirically time-varying and carries a positive premium\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction paragraph 3, footnote 2\"}],\"structural_summary\":\"A purely theoretical asset-pricing paper. The introduction motivates the work with the role of intermediary capital in liquidity provision. It states a minimal-friction modeling choice and previews analytical results on dynamic risk aversion, endogenous risk, and the priced arbitrageur-wealth factor proxied by aggregate illiquidity. It closes by positioning against margin-constraint and log-utility arbitrage models. The body moves from model setup to equilibrium with short-maturity assets, then long-maturity assets, liquidity risk, and a positive-supply extension. The conclusion restates the contributions and proposes stochastic hedger supply as an extension.\",\"writing_profile_keywords\":[\"continuous-time model\",\"analytical propositions\",\"endogenous risk\",\"intermediary asset pricing\",\"numerical illustrations\",\"literature differentiation\"],\"confidence\":0.85,\"notes\":\"No empirical analysis. Figures are numerical illustrations of the model, not data. Literature positioning occupies the latter introduction paragraphs and reads like a related-literature subsection, but the excerpts do not show it as separate.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "financial_intermediation_banking", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; analytical characterization of equilibrium in a continuous-time model", "key_assumptions": ["fixed exogenous riskless rate", "hedgers' asset demand independent of wealth", "arbitrageurs have CRRA utility", "no asymmetric information or leverage constraints"], "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": 2}, "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": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}], "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": "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": "No tables; results are delivered as propositions and numerical illustrations in figures with parameter values stated in the captions.", "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": ["modal 'can' for possibility results", "conditioning on parameter ranges (e.g., CRRA below one)", "explicit statement of simplifying assumptions and their scope"]}, "claim_evidence": [{"claim": "Arbitrageur risk aversion equals static CRRA plus a forward-looking intertemporal hedging component", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.C.2"}, {"claim": "More risk-averse hedgers can make arbitrageurs with CRRA below one endogenously more risk-averse", "evidence_type": "figure", "evidence_ref": "Figure 1; Section II.C.2"}, {"claim": "Risk premia can rise with arbitrageur wealth at low wealth even as Sharpe ratios fall, due to endogenous risk", "evidence_type": "figure", "evidence_ref": "Figure 2; Section III.B"}, {"claim": "Arbitrageur wealth is the single priced factor, and aggregate illiquidity captures it", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV"}, {"claim": "Zero-supply assumption is without loss of generality with long-lived hedgers", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 12; Section V"}, {"claim": "Aggregate liquidity is empirically time-varying and carries a positive premium", "evidence_type": "prior_literature", "evidence_ref": "Introduction paragraph 3, footnote 2"}], "structural_summary": "A purely theoretical asset-pricing paper. The introduction motivates the work with the role of intermediary capital in liquidity provision. It states a minimal-friction modeling choice and previews analytical results on dynamic risk aversion, endogenous risk, and the priced arbitrageur-wealth factor proxied by aggregate illiquidity. It closes by positioning against margin-constraint and log-utility arbitrage models. The body moves from model setup to equilibrium with short-maturity assets, then long-maturity assets, liquidity risk, and a positive-supply extension. The conclusion restates the contributions and proposes stochastic hedger supply as an extension.", "writing_profile_keywords": ["continuous-time model", "analytical propositions", "endogenous risk", "intermediary asset pricing", "numerical illustrations", "literature differentiation"], "confidence": 0.85, "notes": "No empirical analysis. Figures are numerical illustrations of the model, not data. Literature positioning occupies the latter introduction paragraphs and reads like a related-literature subsection, but the excerpts do not show it as separate."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.45, "usage": {"input": 6182, "output": 1880, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.063122, "request_id": "req_011CggqKoAxwiQXKFBm5JRfa", "cached": false, "ts": "2026-11-03T23:07:01+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/56d424900734b5ad8813e1bf225b7b19a92e228583614c06de6ade8b8e6b49f0.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); the analytical object is the CRA's ex ante effect on economic efficiency relative to a no-CRA benchmark\",\"strategy\":\"Equilibrium analysis of a bond financing game with feedback loop, compared against a no-CRA benchmark; real effects decomposed into informational and strategic components; comparative statics\",\"key_assumptions\":[\"rational creditors with dispersed private signals\",\"CRA cannot commit to rating rule\",\"partial verifiability constraint on ratings\",\"endogenous firm investment (moral hazard)\"],\"threats_addressed\":[],\"causal_language_strength\":3,\"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\":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\":\"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\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"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\":\"theory\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"No tables; analysis is carried out through propositions and lemmas, with proofs collected in Appendix A and additional results in an Internet Appendix.\",\"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\":[\"modal verbs (might, may, could)\",\"conditioning results on parameter regions\",\"explicitly stating effects can be positive or negative\"]},\"claim_evidence\":[{\"claim\":\"High ratings, though inflated, truncate creditor beliefs by ruling out very low-quality firms, lowering financing costs and altering investment\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Appendix A\"},{\"claim\":\"CRA's ex ante real effect is more likely negative when the upside of the risky inefficient project is high\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Section IV\"},{\"claim\":\"The informational effect always raises efficiency; negative effects arise purely from the strategic effect\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV.A; Figure 4\"},{\"claim\":\"With exogenous credit quality, rating information always increases efficiency\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Internet Appendix\"},{\"claim\":\"CRA effects become negligible when creditors have precise homogeneous information\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section VI\"},{\"claim\":\"Empirical work documents real effects of ratings on firm investment\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Sufi (2009); Bannier et al. (2012); Almeida et al. (2017)\"}],\"structural_summary\":\"The paper opens with a policy-driven puzzle: how can informative but inflated ratings reduce efficiency when creditors are rational? It sketches the feedback-loop mechanism and previews the main results, including the informational/strategic decomposition and the role of dispersed beliefs. It then positions itself against feedback and Bayesian persuasion models. The model is built sequentially: setup, a no-CRA benchmark, equilibrium rating strategies, and real effects relative to the benchmark. Comparative statics yield empirical predictions, followed by a section on belief dispersion. The conclusion restates the contributions, extends the mechanism to other settings and suggests further research. Proofs are in the appendix.\",\"writing_profile_keywords\":[\"pure theory\",\"feedback loop\",\"puzzle-driven motivation\",\"benchmark comparison\",\"effect decomposition\",\"comparative statics predictions\",\"proofs in appendix\",\"policy relevance\"],\"confidence\":0.85,\"notes\":\"Theoretical paper with no tables or data. Figures illustrate the model's timing and its effects using numerical parameter values. Robustness takes the form of model extensions (Section VI and the Internet Appendix) rather than empirical checks.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model); the analytical object is the CRA's ex ante effect on economic efficiency relative to a no-CRA benchmark", "strategy": "Equilibrium analysis of a bond financing game with feedback loop, compared against a no-CRA benchmark; real effects decomposed into informational and strategic components; comparative statics", "key_assumptions": ["rational creditors with dispersed private signals", "CRA cannot commit to rating rule", "partial verifiability constraint on ratings", "endogenous firm investment (moral hazard)"], "threats_addressed": [], "causal_language_strength": 3, "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": 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": "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": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 5, "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": "theory"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["other"]}, "tables": {"convention_notes": "No tables; analysis is carried out through propositions and lemmas, with proofs collected in Appendix A and additional results in an Internet Appendix.", "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": ["modal verbs (might, may, could)", "conditioning results on parameter regions", "explicitly stating effects can be positive or negative"]}, "claim_evidence": [{"claim": "High ratings, though inflated, truncate creditor beliefs by ruling out very low-quality firms, lowering financing costs and altering investment", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Appendix A"}, {"claim": "CRA's ex ante real effect is more likely negative when the upside of the risky inefficient project is high", "evidence_type": "figure", "evidence_ref": "Figure 3; Section IV"}, {"claim": "The informational effect always raises efficiency; negative effects arise purely from the strategic effect", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV.A; Figure 4"}, {"claim": "With exogenous credit quality, rating information always increases efficiency", "evidence_type": "equation_or_proof", "evidence_ref": "Internet Appendix"}, {"claim": "CRA effects become negligible when creditors have precise homogeneous information", "evidence_type": "equation_or_proof", "evidence_ref": "Section VI"}, {"claim": "Empirical work documents real effects of ratings on firm investment", "evidence_type": "prior_literature", "evidence_ref": "Sufi (2009); Bannier et al. (2012); Almeida et al. (2017)"}], "structural_summary": "The paper opens with a policy-driven puzzle: how can informative but inflated ratings reduce efficiency when creditors are rational? It sketches the feedback-loop mechanism and previews the main results, including the informational/strategic decomposition and the role of dispersed beliefs. It then positions itself against feedback and Bayesian persuasion models. The model is built sequentially: setup, a no-CRA benchmark, equilibrium rating strategies, and real effects relative to the benchmark. Comparative statics yield empirical predictions, followed by a section on belief dispersion. The conclusion restates the contributions, extends the mechanism to other settings and suggests further research. Proofs are in the appendix.", "writing_profile_keywords": ["pure theory", "feedback loop", "puzzle-driven motivation", "benchmark comparison", "effect decomposition", "comparative statics predictions", "proofs in appendix", "policy relevance"], "confidence": 0.85, "notes": "Theoretical paper with no tables or data. Figures illustrate the model's timing and its effects using numerical parameter values. Robustness takes the form of model extensions (Section VI and the Internet Appendix) rather than empirical checks."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.04, "usage": {"input": 7151, "output": 1982, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069038, "request_id": "req_011Cggr16XqKTQZDK4J6XPHg", "cached": false, "ts": "2026-11-03T23:15:54+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/5892cabec8710f58cffba0a4ab288b1da3a5a296465b0bda819175195e25657c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"mutual funds / taxation\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between a fund's prior three-year tax burden and its next-year before-tax, tax burden, and after-tax returns and alphas.\",\"strategy\":\"Annual predictive panel regressions of future fund performance on lagged average tax burden, with lagged controls and time and style fixed effects. These are complemented by a matched comparison of self-designated tax-managed funds against same-family, similar-style funds.\",\"key_assumptions\":[\"Lagged tax burden is not mechanically tied to future returns once controls and fixed effects are included.\",\"Matched funds share investment style and management teams, so they isolate deliberate tax management.\"],\"threats_addressed\":[\"Tax efficiency arising as a byproduct of investment style rather than deliberate management (addressed with self-designated fund matching and style controls).\",\"Risk or style adjustment (addressed with CAPM and multifactor alphas and style scores).\",\"Horizon and subperiod dependence.\",\"DC pension clienteles.\"],\"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\":false,\"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\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"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\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":5,\"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\":\"robustness\"},{\"idx\":6,\"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\",\"alternative_sample_period\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Regression tables are annual panel regressions on lagged tax burden with time and style fixed effects. Standard errors are fund-clustered and shown in parentheses, with stars marking 10/5/1% significance. Some notes state units (% per year), and returns in the matched-fund table are measured in excess of the market.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'is associated with'\",\"'indicate that'\",\"'seem to'\",\"'do not appear to'\",\"'consistent with'\"]},\"claim_evidence\":[{\"claim\":\"Equilibrium fund performance depends on the relative size of taxable and tax-exempt clienteles under decreasing returns to scale.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Figures 1-2\"},{\"claim\":\"A lower prior three-year tax burden predicts higher before-tax and after-tax returns over the following year.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table IV\"},{\"claim\":\"The before-tax outperformance of tax-efficient funds reflects lower trading costs, favorable style exposures and better selectivity.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table VI\"},{\"claim\":\"Self-designated tax-managed funds beat matched same-family funds after taxes without significant before-tax underperformance.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Tax burden is determined by investment style, investor trading behavior and market conditions.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"}],\"structural_summary\":\"The paper opens by explaining how fund distributions are taxed and framing the trade-off in tax management: constraints that may lower returns versus manager skill. A stylized Berk–Green-type model with heterogeneous tax clienteles follows. The data section defines the tax burden, and a short section examines its determinants. The core section runs predictive regressions of future before- and after-tax performance on prior tax burden, then extends them with risk adjustment, performance decomposition, style controls, alternative horizons, subperiods, distribution types and DC ratios. A matched analysis of self-designated tax-managed funds addresses the byproduct concern. The conclusion compares the importance of taxes to fund expenses and interprets the persistence of outperformance as investor inattention.\",\"writing_profile_keywords\":[\"theory-then-evidence\",\"predictive panel regressions\",\"performance decomposition\",\"puzzle framing\",\"matched-sample check\",\"mutual fund taxation\"],\"confidence\":0.82,\"notes\":\"The abstract truncates into the opening of the introduction. The Determinants and Self-Designated Fund sections are labeled 'other' by the role rule; I assigned them results and robustness/identification roles. The introduction excerpt is truncated, so a roadmap may exist but is not visible.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "mutual funds / taxation", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between a fund's prior three-year tax burden and its next-year before-tax, tax burden, and after-tax returns and alphas.", "strategy": "Annual predictive panel regressions of future fund performance on lagged average tax burden, with lagged controls and time and style fixed effects. These are complemented by a matched comparison of self-designated tax-managed funds against same-family, similar-style funds.", "key_assumptions": ["Lagged tax burden is not mechanically tied to future returns once controls and fixed effects are included.", "Matched funds share investment style and management teams, so they isolate deliberate tax management."], "threats_addressed": ["Tax efficiency arising as a byproduct of investment style rather than deliberate management (addressed with self-designated fund matching and style controls).", "Risk or style adjustment (addressed with CAPM and multifactor alphas and style scores).", "Horizon and subperiod dependence.", "DC pension clienteles."], "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": false, "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": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"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": "identification_defense"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}], "contribution_paragraph_index": 5, "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": "robustness"}, {"idx": 6, "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", "alternative_sample_period", "subsample", "other"]}, "tables": {"convention_notes": "Regression tables are annual panel regressions on lagged tax burden with time and style fixed effects. Standard errors are fund-clustered and shown in parentheses, with stars marking 10/5/1% significance. Some notes state units (% per year), and returns in the matched-fund table are measured in excess of the market.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'is associated with'", "'indicate that'", "'seem to'", "'do not appear to'", "'consistent with'"]}, "claim_evidence": [{"claim": "Equilibrium fund performance depends on the relative size of taxable and tax-exempt clienteles under decreasing returns to scale.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Figures 1-2"}, {"claim": "A lower prior three-year tax burden predicts higher before-tax and after-tax returns over the following year.", "evidence_type": "table", "evidence_ref": "Table III, Table IV"}, {"claim": "The before-tax outperformance of tax-efficient funds reflects lower trading costs, favorable style exposures and better selectivity.", "evidence_type": "table", "evidence_ref": "Table V, Table VI"}, {"claim": "Self-designated tax-managed funds beat matched same-family funds after taxes without significant before-tax underperformance.", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Tax burden is determined by investment style, investor trading behavior and market conditions.", "evidence_type": "table", "evidence_ref": "Table II"}], "structural_summary": "The paper opens by explaining how fund distributions are taxed and framing the trade-off in tax management: constraints that may lower returns versus manager skill. A stylized Berk–Green-type model with heterogeneous tax clienteles follows. The data section defines the tax burden, and a short section examines its determinants. The core section runs predictive regressions of future before- and after-tax performance on prior tax burden, then extends them with risk adjustment, performance decomposition, style controls, alternative horizons, subperiods, distribution types and DC ratios. A matched analysis of self-designated tax-managed funds addresses the byproduct concern. The conclusion compares the importance of taxes to fund expenses and interprets the persistence of outperformance as investor inattention.", "writing_profile_keywords": ["theory-then-evidence", "predictive panel regressions", "performance decomposition", "puzzle framing", "matched-sample check", "mutual fund taxation"], "confidence": 0.82, "notes": "The abstract truncates into the opening of the introduction. The Determinants and Self-Designated Fund sections are labeled 'other' by the role rule; I assigned them results and robustness/identification roles. 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": 18.47, "usage": {"input": 7129, "output": 2161, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07253, "request_id": "req_011CggqiToH3haSNfn47w1Sx", "cached": false, "ts": "2026-11-03T23:12:11+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/5c42355212cf0c12a943d37aac5c10f8ad205a4951f77c186999f9d1cb0b2e69.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of faster government payment (Quickpay) on employment growth of treated small-business contractors, and the net and crowding-out effects on local labor markets\",\"strategy\":\"Difference-in-differences in first differences, comparing establishments, county-by-sector cells and commuting zones by their exposure to accelerated federal contracts before and after the 2011 reform; dynamics and falsification tests; heterogeneity by labor market tightness\",\"key_assumptions\":[\"Parallel trends between treated and nontreated establishments or areas absent the reform\",\"Treatment defined by eligible government contracts in the two pre-reform years is not correlated with other shocks\",\"Reform timing is exogenous to firm-level employment prospects\"],\"threats_addressed\":[\"Pre-existing trends, addressed with dynamics tests\",\"Spurious effects, addressed with falsification tests\",\"Equilibrium spillovers to nontreated firms, addressed with aggregation to the commuting-zone level\"],\"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\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"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\":\"theory\"},{\"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\":[\"falsification\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes begin with a standardized description of a first-differenced difference-in-differences OLS design and state the level of observation and the treatment definition. One model-calibration table maps employment growth to the implied cost of finance.\",\"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\",\"can lead to\",\"potentially effective\",\"depending on labor market conditions\"]},\"claim_evidence\":[{\"claim\":\"Quickpay increased employment growth at treated establishments\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III\"},{\"claim\":\"Placebo/falsification tests do not show spurious effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Implied cost of external finance is about 40%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Aggregate employment rises only in slack labor markets; crowding out occurs in tight ones\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Table X\"},{\"claim\":\"Workers flow from low- to high-treatment sectors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Payment terms for small contractors fell after the reform\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"The paper motivates the question with post-crisis financing constraints and the government's role as a customer. It sketches a theory of working-capital financing of labor and describes the reform. It then estimates establishment-level difference-in-differences effects, with dynamics and falsification tests. Model-based magnitudes translate the employment response into implied financing frictions. The analysis then moves to county-by-sector and commuting-zone aggregation to reveal crowding out that depends on labor market tightness. The conclusion restates the findings and draws policy implications.\",\"writing_profile_keywords\":[\"policy reform\",\"difference-in-differences\",\"general equilibrium spillovers\",\"trade credit\",\"working capital\",\"labor markets\",\"small firms\"],\"confidence\":0.8,\"notes\":\"The roadmap places the identification strategy in Section II together with the data description. Standard error treatment is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "macro_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of faster government payment (Quickpay) on employment growth of treated small-business contractors, and the net and crowding-out effects on local labor markets", "strategy": "Difference-in-differences in first differences, comparing establishments, county-by-sector cells and commuting zones by their exposure to accelerated federal contracts before and after the 2011 reform; dynamics and falsification tests; heterogeneity by labor market tightness", "key_assumptions": ["Parallel trends between treated and nontreated establishments or areas absent the reform", "Treatment defined by eligible government contracts in the two pre-reform years is not correlated with other shocks", "Reform timing is exogenous to firm-level employment prospects"], "threats_addressed": ["Pre-existing trends, addressed with dynamics tests", "Spurious effects, addressed with falsification tests", "Equilibrium spillovers to nontreated firms, addressed with aggregation to the commuting-zone level"], "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": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "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": "theory"}, {"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": ["falsification", "alternative_specification"]}, "tables": {"convention_notes": "Table notes begin with a standardized description of a first-differenced difference-in-differences OLS design and state the level of observation and the treatment definition. One model-calibration table maps employment growth to the implied cost of finance.", "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", "can lead to", "potentially effective", "depending on labor market conditions"]}, "claim_evidence": [{"claim": "Quickpay increased employment growth at treated establishments", "evidence_type": "table", "evidence_ref": "Table II, Table III"}, {"claim": "Placebo/falsification tests do not show spurious effects", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Implied cost of external finance is about 40%", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Aggregate employment rises only in slack labor markets; crowding out occurs in tight ones", "evidence_type": "table", "evidence_ref": "Table VII, Table X"}, {"claim": "Workers flow from low- to high-treatment sectors", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Payment terms for small contractors fell after the reform", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "The paper motivates the question with post-crisis financing constraints and the government's role as a customer. It sketches a theory of working-capital financing of labor and describes the reform. It then estimates establishment-level difference-in-differences effects, with dynamics and falsification tests. Model-based magnitudes translate the employment response into implied financing frictions. The analysis then moves to county-by-sector and commuting-zone aggregation to reveal crowding out that depends on labor market tightness. The conclusion restates the findings and draws policy implications.", "writing_profile_keywords": ["policy reform", "difference-in-differences", "general equilibrium spillovers", "trade credit", "working capital", "labor markets", "small firms"], "confidence": 0.8, "notes": "The roadmap places the identification strategy in Section II together with the data description. Standard error treatment is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.26, "usage": {"input": 6013, "output": 1794, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.060726, "request_id": "req_011Cggr2mbMXEvNQCvi428w1", "cached": false, "ts": "2026-11-03T23:16:15+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/5cbc8b8f878c546bc89f3b0a76e77f1a599afbfc02ff42993933bff677640195.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Predictive relation between model-implied income hedging demand for HML and future HML returns, at the national and state level\",\"strategy\":\"Predictive time-series and within-state panel regressions of future HML returns on rolling-window covariance-based IHD estimates, with controls. Cross-sectional heterogeneity tests split by local ownership, occupation and income quantile.\",\"key_assumptions\":[\"Covariance-based IHD captures actual hedging-induced demand shifts\",\"Hedging demand shifts are systematic enough to move prices\",\"Controls absorb macroeconomic shocks\"],\"threats_addressed\":[\"macroeconomic shocks\",\"estimation biases\",\"known factor predictors\",\"validity of the demand measure (checked against brokerage holdings and fund flows)\"],\"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\":4},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"robustness_mention\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"identification_defense\"},{\"paragraph_index\":11,\"move\":\"identification_defense\"},{\"paragraph_index\":12,\"move\":\"contribution_statement\"},{\"paragraph_index\":12,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":13,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":12,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"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\":\"empirical_design\"},{\"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\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Long descriptive notes define each dependent variable, IHD construction (40-quarter rolling covariances) and control sets, and refer back to earlier tables. Tables move from summary statistics to validation, national predictive regressions, state panels, portfolio performance and factor-model alphas, and heterogeneity tests.\",\"units_stated_in_notes\":false,\"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\":[\"'potential' and 'could' when stating the conjecture\",\"'suggest' when interpreting results\",\"'at least part of' as a partial-attribution qualifier\",\"'reasonably well' when describing validation fit\"]},\"claim_evidence\":[{\"claim\":\"The model-implied IHD measure correlates positively with retail investors' relative value-growth portfolio weights\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"IHD is positively related to value fund flows and negatively related to growth fund flows\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"A one-SD increase in national IHD is associated with a 0.59% lower HML return the next month and 0.96% lower the next quarter\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"State-level IHD predicts state-level HML returns, especially among stocks with high local ownership\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"A long-short state portfolio based on IHD earns roughly 6% annualized risk-adjusted return\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII-VIII\"},{\"claim\":\"IHD for electrical engineers predicts HML returns while IHD for auto mechanics does not\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"IHD built from top-earner income growth has the most predictive power\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"IHD estimates vary substantially across states and over time\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"The paper builds a two-investor model that yields a covariance-based measure of income hedging demand. It motivates focusing on HML, constructs the measure from aggregate and state income data, and validates it against brokerage holdings and mutual fund flows. It then tests return predictability nationally and across states, forms a long-short portfolio, and adds heterogeneity tests by occupation, income tier and labor rigidity as support for the mechanism. The conclusion restates findings and suggests extensions. The proof sits in an appendix.\",\"writing_profile_keywords\":[\"model-motivated measure\",\"validation tests\",\"predictive regressions\",\"state-level variation\",\"mechanism heterogeneity tests\",\"long-short portfolio\",\"value factor\"],\"confidence\":0.8,\"notes\":[\"The excerpted roadmap misnumbers the aggregate predictability section as V; from the section list it appears to be IV.\",\"Section idx 5 is a heading fragment from parsing and continues Section IV's results.\",\"How standard errors are computed is not visible in the excerpts.\"]}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "household_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Predictive relation between model-implied income hedging demand for HML and future HML returns, at the national and state level", "strategy": "Predictive time-series and within-state panel regressions of future HML returns on rolling-window covariance-based IHD estimates, with controls. Cross-sectional heterogeneity tests split by local ownership, occupation and income quantile.", "key_assumptions": ["Covariance-based IHD captures actual hedging-induced demand shifts", "Hedging demand shifts are systematic enough to move prices", "Controls absorb macroeconomic shocks"], "threats_addressed": ["macroeconomic shocks", "estimation biases", "known factor predictors", "validity of the demand measure (checked against brokerage holdings and fund flows)"], "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": 4}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "robustness_mention"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "identification_defense"}, {"paragraph_index": 11, "move": "identification_defense"}, {"paragraph_index": 12, "move": "contribution_statement"}, {"paragraph_index": 12, "move": "related_literature_positioning"}, {"paragraph_index": 13, "move": "roadmap"}], "contribution_paragraph_index": 12, "contribution_types": ["new_evidence", "new_data_or_measure", "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": "empirical_design"}, {"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"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "alternative_measure", "falsification", "other"]}, "tables": {"convention_notes": "Long descriptive notes define each dependent variable, IHD construction (40-quarter rolling covariances) and control sets, and refer back to earlier tables. Tables move from summary statistics to validation, national predictive regressions, state panels, portfolio performance and factor-model alphas, and heterogeneity tests.", "units_stated_in_notes": false, "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": ["'potential' and 'could' when stating the conjecture", "'suggest' when interpreting results", "'at least part of' as a partial-attribution qualifier", "'reasonably well' when describing validation fit"]}, "claim_evidence": [{"claim": "The model-implied IHD measure correlates positively with retail investors' relative value-growth portfolio weights", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "IHD is positively related to value fund flows and negatively related to growth fund flows", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "A one-SD increase in national IHD is associated with a 0.59% lower HML return the next month and 0.96% lower the next quarter", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "State-level IHD predicts state-level HML returns, especially among stocks with high local ownership", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "A long-short state portfolio based on IHD earns roughly 6% annualized risk-adjusted return", "evidence_type": "table", "evidence_ref": "Tables VII-VIII"}, {"claim": "IHD for electrical engineers predicts HML returns while IHD for auto mechanics does not", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "IHD built from top-earner income growth has the most predictive power", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "IHD estimates vary substantially across states and over time", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "The paper builds a two-investor model that yields a covariance-based measure of income hedging demand. It motivates focusing on HML, constructs the measure from aggregate and state income data, and validates it against brokerage holdings and mutual fund flows. It then tests return predictability nationally and across states, forms a long-short portfolio, and adds heterogeneity tests by occupation, income tier and labor rigidity as support for the mechanism. The conclusion restates findings and suggests extensions. The proof sits in an appendix.", "writing_profile_keywords": ["model-motivated measure", "validation tests", "predictive regressions", "state-level variation", "mechanism heterogeneity tests", "long-short portfolio", "value factor"], "confidence": 0.8, "notes": ["The excerpted roadmap misnumbers the aggregate predictability section as V; from the section list it appears to be IV.", "Section idx 5 is a heading fragment from parsing and continues Section IV's results.", "How standard errors are computed is not visible in the excerpts."]}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.41, "usage": {"input": 8844, "output": 2355, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08327, "request_id": "req_011CggqSpDuxLiQRKv6SLfC9", "cached": false, "ts": "2026-11-03T23:08:39+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/5e3d09497c89adf5c4da84d06f2af2e028694369e816704e5c50be96b71218ba.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"law and finance / corporate credit\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of a lender's accumulated past bad loans on the probability that the same borrower delays repayment to that lender, and how this effect varies with local judicial (collateral-recovery) efficiency\",\"strategy\":\"Linear probability model on bank-firm-quarter Credit Register data. Firm-quarter fixed effects absorb borrower fundamentals, so variation comes from differences across a multi-bank firm's lenders. Bad loans are interacted with court-level duration of property execution proceedings. An IV for bank weakness combines 2007 sector-province portfolio shares with leave-out aggregate loss rates.\",\"key_assumptions\":[\"Firm-quarter effects fully absorb borrower-level fundamentals and demand\",\"The shift-share instrument (2007 portfolio weights times leave-out sector-province loss rates) affects delays only through bank weakness\",\"Court-level enforcement duration is not proxying for other regional traits such as culture or social capital\",\"Borrower-lender matching is not driven by unobservables correlated with delays\"],\"threats_addressed\":[\"reverse causality from borrower delays to bank health\",\"omitted variables\",\"borrower financial distress versus strategic delay\",\"endogenous borrower-lender matching\",\"law versus culture/social capital\",\"bank governance (mutual vs private)\",\"loan terms\",\"borrower and lender in different jurisdictions\"],\"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\":6}},\"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\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"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\":\"conclusion\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"},{\"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\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Descriptive tables come first: loan quality, court durations, and bank and firm characteristics, split by above- or below-median judicial inefficiency. Regression tables follow, with OLS and IV columns side by side. Notes define variables by reference to Appendix A. Standard errors are in parentheses, clustered at the bank level, with significance stars.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"results suggest\",\"it seems likely\",\"might impair\",\"qualitatively similar\"]},\"claim_evidence\":[{\"claim\":\"Lenders' past bad loans raise the probability that borrowers delay repayment to them\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The effect of bad loans on delays grows with slower collateral enforcement and is insignificant where courts are efficient\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 3\"},{\"claim\":\"In top-quartile legal inefficiency areas, a 1-SD rise in past bad loans raises delays by about 50% of the mean\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Late repayments reduce lender profitability\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Even safe borrowers selectively delay payment to weak banks where enforcement is poor\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII; Figures 4-5\"},{\"claim\":\"Firm-level total delays rise with exposure to weak lenders, more so under weak enforcement\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; Figure 6\"},{\"claim\":\"Results are robust to first delays, loan terms, same-jurisdiction pairs, matching, culture and governance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X; Table B2\"},{\"claim\":\"Italian bad loans were large and concentrated in nonfinancial business lending\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Introduction, Bank of Italy (2016)\"}],\"structural_summary\":\"The paper opens with the aggregate nonperforming-loan problem in Italy, states the headline finding with a magnitude, and explains identification via firm-quarter effects in multi-bank relationships. It then previews the IV defense against reverse causality and frames a strategic-versus-selective delay question with policy implications. A separate literature section positions the work against theories of repayment unraveling. A data section combines four registers. A long empirical section covers baseline, judicial-interaction, profit, borrower-risk and firm-level analyses, with embedded robustness subsections. A short conclusion claims priority and draws implications for legal enforcement and transparency.\",\"writing_profile_keywords\":[\"credit register\",\"firm-time fixed effects\",\"multiple bank relationships\",\"judicial efficiency interaction\",\"shift-share instrument\",\"marginal effect plots\",\"strategic default\",\"law and finance\"],\"confidence\":0.82,\"notes\":\"Section 3 is labeled empirical design by rule but mostly contains results; it is classified as results here. No explicit roadmap is visible in the truncated introduction. Whether the marginal-effect figures show confidence bands cannot be determined from the truncated captions.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "law and finance / corporate credit", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of a lender's accumulated past bad loans on the probability that the same borrower delays repayment to that lender, and how this effect varies with local judicial (collateral-recovery) efficiency", "strategy": "Linear probability model on bank-firm-quarter Credit Register data. Firm-quarter fixed effects absorb borrower fundamentals, so variation comes from differences across a multi-bank firm's lenders. Bad loans are interacted with court-level duration of property execution proceedings. An IV for bank weakness combines 2007 sector-province portfolio shares with leave-out aggregate loss rates.", "key_assumptions": ["Firm-quarter effects fully absorb borrower-level fundamentals and demand", "The shift-share instrument (2007 portfolio weights times leave-out sector-province loss rates) affects delays only through bank weakness", "Court-level enforcement duration is not proxying for other regional traits such as culture or social capital", "Borrower-lender matching is not driven by unobservables correlated with delays"], "threats_addressed": ["reverse causality from borrower delays to bank health", "omitted variables", "borrower financial distress versus strategic delay", "endogenous borrower-lender matching", "law versus culture/social capital", "bank governance (mutual vs private)", "loan terms", "borrower and lender in different jurisdictions"], "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": 6}}, "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": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence"], "roadmap_present": false, "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": "conclusion"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}, {"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": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Descriptive tables come first: loan quality, court durations, and bank and firm characteristics, split by above- or below-median judicial inefficiency. Regression tables follow, with OLS and IV columns side by side. Notes define variables by reference to Appendix A. Standard errors are in parentheses, clustered at the bank level, with significance stars.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["results suggest", "it seems likely", "might impair", "qualitatively similar"]}, "claim_evidence": [{"claim": "Lenders' past bad loans raise the probability that borrowers delay repayment to them", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "The effect of bad loans on delays grows with slower collateral enforcement and is insignificant where courts are efficient", "evidence_type": "table", "evidence_ref": "Table VI; Figure 3"}, {"claim": "In top-quartile legal inefficiency areas, a 1-SD rise in past bad loans raises delays by about 50% of the mean", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Late repayments reduce lender profitability", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Even safe borrowers selectively delay payment to weak banks where enforcement is poor", "evidence_type": "table", "evidence_ref": "Table VIII; Figures 4-5"}, {"claim": "Firm-level total delays rise with exposure to weak lenders, more so under weak enforcement", "evidence_type": "table", "evidence_ref": "Table IX; Figure 6"}, {"claim": "Results are robust to first delays, loan terms, same-jurisdiction pairs, matching, culture and governance", "evidence_type": "table", "evidence_ref": "Table X; Table B2"}, {"claim": "Italian bad loans were large and concentrated in nonfinancial business lending", "evidence_type": "descriptive_statistic", "evidence_ref": "Introduction, Bank of Italy (2016)"}], "structural_summary": "The paper opens with the aggregate nonperforming-loan problem in Italy, states the headline finding with a magnitude, and explains identification via firm-quarter effects in multi-bank relationships. It then previews the IV defense against reverse causality and frames a strategic-versus-selective delay question with policy implications. A separate literature section positions the work against theories of repayment unraveling. A data section combines four registers. A long empirical section covers baseline, judicial-interaction, profit, borrower-risk and firm-level analyses, with embedded robustness subsections. A short conclusion claims priority and draws implications for legal enforcement and transparency.", "writing_profile_keywords": ["credit register", "firm-time fixed effects", "multiple bank relationships", "judicial efficiency interaction", "shift-share instrument", "marginal effect plots", "strategic default", "law and finance"], "confidence": 0.82, "notes": "Section 3 is labeled empirical design by rule but mostly contains results; it is classified as results here. No explicit roadmap is visible in the truncated introduction. Whether the marginal-effect figures show confidence bands cannot be determined from the truncated captions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 26.62, "usage": {"input": 8029, "output": 3106, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.09503, "request_id": "req_011CggqtLH7iAkwihjHXjd1Q", "cached": false, "ts": "2026-11-03T23:14:33+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/614fa4ac11b1090d97276848e17f5c5f9afd7155de42064cd6440669f74256bc.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"corporate governance; mergers and acquisitions\",\"primary_method\":\"regression_discontinuity\",\"identification\":{\"estimand\":\"Intent-to-treat effect of passing a shareholder proposal to remove an ATP on five-year takeover probability, unconditional premium, and (bounded) conditional premium\",\"strategy\":\"RDD at the majority vote threshold for shareholder proposals; CIA-validated matching (Angrist-Rokkanen) to extend estimates away from the threshold; Lee (2009) bounds for selection into the target sample\",\"key_assumptions\":[\"No manipulation or pre-existing differences around the majority threshold\",\"Conditional independence of outcomes from the running variable given covariates\",\"Monotonicity of the ATP effect on selection into being a target\"],\"threats_addressed\":[\"Endogenous adoption of governance provisions\",\"Selection of which firms become targets\",\"Pre-trends or covariate imbalance at the threshold\",\"Alternative vote-share definitions, including treatment of abstentions\"],\"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\":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_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"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\":\"other\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"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\":\"theory\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"results\"},{\"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\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Table notes give the sample, data sources (Compustat, SDC), and variable definitions, including Compustat mnemonics. They also specify the estimator, e.g., Lee bounds or matching type. Panels separate outcomes or samples, such as the full sample versus the ±10% vote interval.\",\"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\":[\"bounds language ('can be as high as', 'never negative')\",\"explicit external-validity caveats\",\"ITT versus treatment-on-treated distinction\",\"'suggesting', 'likely', 'potentially'\"]},\"claim_evidence\":[{\"claim\":\"Passing an ATP-removal proposal at the threshold raises five-year takeover probability by about 9 percent and the expected premium by about 4 percent\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 1\"},{\"claim\":\"Away from the threshold, effects on takeover probability and unconditional premium are smaller but still positive\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The effect of ATP removal on the conditional premium is bounded below near zero and may reach a sizable four-week premium\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Over half of the value gain comes from higher takeover probability, with selection contributing materially\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"The positive premium in less protected firms reflects better bidder-target matching and synergies\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII-VIII\"},{\"claim\":\"The conditional independence assumption holds around the threshold\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Figure 2\"}],\"structural_summary\":\"The introduction challenges the accepted price-versus-probability trade-off of ATPs, then previews the empirical approach (RDD, validated matching, Lee bounds) and quantified results, followed by an external-validity caveat and a roadmap. The body first builds a decomposition framework and lays out the identification strategies and data. It then estimates effects on unconditional premiums and probabilities, bounds the conditional premium, decomposes the value effect, and explores channels. The conclusion restates the value-destruction channels.\",\"writing_profile_keywords\":[\"regression discontinuity\",\"shareholder proposals\",\"Lee bounds\",\"selection correction\",\"decomposition framework\",\"anti-takeover provisions\",\"causal framing\"],\"confidence\":0.85,\"notes\":\"Standard-error type and significance notation are not visible in the excerpted table notes. Robustness evidence appears within the results and in the Internet Appendix.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "corporate governance; mergers and acquisitions", "primary_method": "regression_discontinuity", "identification": {"estimand": "Intent-to-treat effect of passing a shareholder proposal to remove an ATP on five-year takeover probability, unconditional premium, and (bounded) conditional premium", "strategy": "RDD at the majority vote threshold for shareholder proposals; CIA-validated matching (Angrist-Rokkanen) to extend estimates away from the threshold; Lee (2009) bounds for selection into the target sample", "key_assumptions": ["No manipulation or pre-existing differences around the majority threshold", "Conditional independence of outcomes from the running variable given covariates", "Monotonicity of the ATP effect on selection into being a target"], "threats_addressed": ["Endogenous adoption of governance provisions", "Selection of which firms become targets", "Pre-trends or covariate imbalance at the threshold", "Alternative vote-share definitions, including treatment of abstentions"], "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": 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_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"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": "other"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "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": "theory"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "results"}, {"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": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "alternative_specification", "subsample"]}, "tables": {"convention_notes": "Table notes give the sample, data sources (Compustat, SDC), and variable definitions, including Compustat mnemonics. They also specify the estimator, e.g., Lee bounds or matching type. Panels separate outcomes or samples, such as the full sample versus the ±10% vote interval.", "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": ["bounds language ('can be as high as', 'never negative')", "explicit external-validity caveats", "ITT versus treatment-on-treated distinction", "'suggesting', 'likely', 'potentially'"]}, "claim_evidence": [{"claim": "Passing an ATP-removal proposal at the threshold raises five-year takeover probability by about 9 percent and the expected premium by about 4 percent", "evidence_type": "table", "evidence_ref": "Table II, Figure 1"}, {"claim": "Away from the threshold, effects on takeover probability and unconditional premium are smaller but still positive", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The effect of ATP removal on the conditional premium is bounded below near zero and may reach a sizable four-week premium", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Over half of the value gain comes from higher takeover probability, with selection contributing materially", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "The positive premium in less protected firms reflects better bidder-target matching and synergies", "evidence_type": "table", "evidence_ref": "Tables VII-VIII"}, {"claim": "The conditional independence assumption holds around the threshold", "evidence_type": "table", "evidence_ref": "Table III, Figure 2"}], "structural_summary": "The introduction challenges the accepted price-versus-probability trade-off of ATPs, then previews the empirical approach (RDD, validated matching, Lee bounds) and quantified results, followed by an external-validity caveat and a roadmap. The body first builds a decomposition framework and lays out the identification strategies and data. It then estimates effects on unconditional premiums and probabilities, bounds the conditional premium, decomposes the value effect, and explores channels. The conclusion restates the value-destruction channels.", "writing_profile_keywords": ["regression discontinuity", "shareholder proposals", "Lee bounds", "selection correction", "decomposition framework", "anti-takeover provisions", "causal framing"], "confidence": 0.85, "notes": "Standard-error type and significance notation are not visible in the excerpted table notes. Robustness evidence appears within the results and in the Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.25, "usage": {"input": 7364, "output": 2093, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07211, "request_id": "req_011CggqxYeNUY2XBZ2cXfCcx", "cached": false, "ts": "2026-11-03T23:15:21+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/62ba81ca78173b2f46d062cc1d05de1fa006f42a4d130c2cf52b86eebfabf73c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of the average participation or early-selling choices of local coworkers (same firm-location-window) on an employee's own ESPP participation and sale-timing decisions\",\"strategy\":\"Within-firm, same-election-window comparisons with metro-area (CBSA) fixed effects. Peer means are measured in an excluded random half-sample. This is supplemented by an IV approach in the style of Duflo-Saez and Case-Katz, using office demographic composition as instruments for peer choices. Heterogeneity tests by the presence of informed peers are used to identify a learning mechanism.\",\"key_assumptions\":[\"Office-level demographic composition affects own ESPP choices only through peer choices, after own traits are controlled (exclusion restriction)\",\"Within-firm, within-window office shocks are absorbed by location and demographic controls\",\"No endogenous sorting of workers to offices that is correlated with ESPP preferences\"],\"threats_addressed\":[\"Common office-level shocks not captured by controls\",\"Endogenous sorting into offices (dropping offices with extreme characteristic distributions)\",\"Reflection problem (excluded half-sample peer means)\",\"Uninformed mimicry versus learning (informed-peer presence tests)\",\"Firm-CBSA-month fixed effects for within-office heterogeneity\",\"Office size and headquarters-state exclusions\"]},\"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\":0,\"move\":\"tension_or_puzzle\"},{\"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\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"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\":\"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\":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\":\"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\":\"within_results\",\"types\":[\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the sample (a named 'Analysis Sample' built from a random half of each firm-location-window), the dependent indicator variables, peer means taken from the excluded half-sample, and the controls carried over from earlier tables. Summary and descriptive tables come before the regression tables. The standard-error description is not visible in the excerpts.\",\"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\":[\"'suggests the potential'\",\"'consistent with rational responses'\",\"'is associated with'\",\"explicit admission that the exclusion restriction cannot be tested directly\",\"'some of this effect could come from uninformed mimicry'\"]},\"claim_evidence\":[{\"claim\":\"ESPP participation averages only 43% despite sizable forgone gains\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Table II\"},{\"claim\":\"A 10 pp rise in local peer participation is associated with a 1.5 pp higher probability that an employee participates\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Peer effects on participation hold in IV estimates using office demographic composition\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Peer effects are weaker for uninformed workers in offices without high-information colleagues\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX, X\"},{\"claim\":\"Coworkers' early-sale behavior predicts an employee's own decision to sell within two weeks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XI, XII\"}],\"structural_summary\":\"The introduction opens with retail-investor mistakes and workplace plans as a possible learning channel. It then previews the data, results and magnitudes, defends identification with IV and sorting checks, and sketches the information mechanism. Three paragraphs of literature positioning follow, and there is no roadmap. The body has a data and institutional section and one long results section, with subsections on baseline estimates, heterogeneity, the mechanism and sale timing. A short conclusion restates the findings.\",\"writing_profile_keywords\":[\"peer effects\",\"workplace networks\",\"ESPP\",\"fixed effects\",\"instrumental variables\",\"learning mechanism\",\"household finance\"],\"confidence\":0.8,\"notes\":\"Section I is tagged 'literature' by the role rule, but it is a data and institutional section. Robustness checks are embedded in the results subsections and the introduction. The abstract's final sentence fragment comes from the introduction's opening sentence. Causal language strength is set to 3 because the paper claims peer influence while acknowledging the untestable exclusion restriction.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of the average participation or early-selling choices of local coworkers (same firm-location-window) on an employee's own ESPP participation and sale-timing decisions", "strategy": "Within-firm, same-election-window comparisons with metro-area (CBSA) fixed effects. Peer means are measured in an excluded random half-sample. This is supplemented by an IV approach in the style of Duflo-Saez and Case-Katz, using office demographic composition as instruments for peer choices. Heterogeneity tests by the presence of informed peers are used to identify a learning mechanism.", "key_assumptions": ["Office-level demographic composition affects own ESPP choices only through peer choices, after own traits are controlled (exclusion restriction)", "Within-firm, within-window office shocks are absorbed by location and demographic controls", "No endogenous sorting of workers to offices that is correlated with ESPP preferences"], "threats_addressed": ["Common office-level shocks not captured by controls", "Endogenous sorting into offices (dropping offices with extreme characteristic distributions)", "Reflection problem (excluded half-sample peer means)", "Uninformed mimicry versus learning (informed-peer presence tests)", "Firm-CBSA-month fixed effects for within-office heterogeneity", "Office size and headquarters-state exclusions"]}, "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": 0, "move": "tension_or_puzzle"}, {"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": "identification_defense"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"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": "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": 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": "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": "within_results", "types": ["subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Table notes define the sample (a named 'Analysis Sample' built from a random half of each firm-location-window), the dependent indicator variables, peer means taken from the excluded half-sample, and the controls carried over from earlier tables. Summary and descriptive tables come before the regression tables. The standard-error description is not visible in the excerpts.", "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": ["'suggests the potential'", "'consistent with rational responses'", "'is associated with'", "explicit admission that the exclusion restriction cannot be tested directly", "'some of this effect could come from uninformed mimicry'"]}, "claim_evidence": [{"claim": "ESPP participation averages only 43% despite sizable forgone gains", "evidence_type": "table", "evidence_ref": "Table I; Table II"}, {"claim": "A 10 pp rise in local peer participation is associated with a 1.5 pp higher probability that an employee participates", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Peer effects on participation hold in IV estimates using office demographic composition", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Peer effects are weaker for uninformed workers in offices without high-information colleagues", "evidence_type": "table", "evidence_ref": "Tables VIII, IX, X"}, {"claim": "Coworkers' early-sale behavior predicts an employee's own decision to sell within two weeks", "evidence_type": "table", "evidence_ref": "Tables XI, XII"}], "structural_summary": "The introduction opens with retail-investor mistakes and workplace plans as a possible learning channel. It then previews the data, results and magnitudes, defends identification with IV and sorting checks, and sketches the information mechanism. Three paragraphs of literature positioning follow, and there is no roadmap. The body has a data and institutional section and one long results section, with subsections on baseline estimates, heterogeneity, the mechanism and sale timing. A short conclusion restates the findings.", "writing_profile_keywords": ["peer effects", "workplace networks", "ESPP", "fixed effects", "instrumental variables", "learning mechanism", "household finance"], "confidence": 0.8, "notes": "Section I is tagged 'literature' by the role rule, but it is a data and institutional section. Robustness checks are embedded in the results subsections and the introduction. The abstract's final sentence fragment comes from the introduction's opening sentence. Causal language strength is set to 3 because the paper claims peer influence while acknowledging the untestable exclusion restriction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.35, "usage": {"input": 6594, "output": 2167, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07051, "request_id": "req_011Cggqcb3gWzwYn9zemPWva", "cached": false, "ts": "2026-11-03T23:10:50+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/651a22bab9b94ab82ce0d816f19b3ba6f4d086d352c89bd663bd746a1f50bb4e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"information economics / feedback between financial markets and real economy\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; analytical general equilibrium model with numerical illustrations\",\"key_assumptions\":[\"mutual learning between firms and financial markets\",\"monopolistic competition with constant markup\",\"CARA speculators\",\"OLG dynamic structure\"],\"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\":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\"}],\"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\":\"theory\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"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\":\"Two tables: one lists the parameter values used for the numerical illustration, with descriptions and sources; the other compares good versus bad steady-state equilibrium outcomes (GDP, investment, TFP, information precisions). No inferential statistics.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\",\"figures_note_placeholder_removed\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'possibly' for multiplicity of steady states\",\"acknowledging the model is too stylized to calibrate\",\"noting modeling assumptions such as constant markup\"]},\"claim_evidence\":[{\"claim\":\"Uncertainty in financial and real sectors spiked jointly in the Great Recession, motivating the question\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"VIX; Jurado, Ludvigson, and Ng (2015); Bloom et al. (2012)\"},{\"claim\":\"Mutual learning creates strategic complementarity in information production between firms and speculators\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Sections I–II; Appendix B\"},{\"claim\":\"General equilibrium model yields amplification, co-movement of uncertainties, contagion and cross-sectional implications\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C; Figures 2–4\"},{\"claim\":\"Dynamic OLG model can have two steady states (uncertainty traps) and a two-stage crisis\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5, 6, 8; Table II\"}],\"structural_summary\":\"The paper opens with stylized facts on joint uncertainty spikes and poses them as research questions. It then builds the theory incrementally:\\n- a one-firm baseline with exogenous information establishing two-way learning;\\n- endogenous information acquisition generating strategic complementarity;\\n- a macro GE extension that delivers four implications;\\n- a dynamic OLG extension that delivers uncertainty traps and transitional crisis dynamics.\\nNumerical illustrations with assigned parameters follow, then a conclusion restating the mechanism and noting extensions. Proofs are in the appendix.\",\"writing_profile_keywords\":[\"theory-building\",\"incremental model extension\",\"general equilibrium\",\"multiple equilibria\",\"stylized facts motivation\",\"numerical illustration\",\"proofs in appendix\"],\"confidence\":0.8,\"notes\":\"The introduction is captured as a single paragraph, so moves beyond the motivation and approach preview are not visible. The numerical section explicitly disclaims calibration.\"}", "json": {"genre": "theoretical", "subfield": "macro_finance", "subfield_secondary": "information economics / feedback between financial markets and real economy", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; analytical general equilibrium model with numerical illustrations", "key_assumptions": ["mutual learning between firms and financial markets", "monopolistic competition with constant markup", "CARA speculators", "OLG dynamic structure"], "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": 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"}], "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": "theory"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "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": "Two tables: one lists the parameter values used for the numerical illustration, with descriptions and sources; the other compares good versus bad steady-state equilibrium outcomes (GDP, investment, TFP, information precisions). No inferential statistics.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no", "figures_note_placeholder_removed": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'possibly' for multiplicity of steady states", "acknowledging the model is too stylized to calibrate", "noting modeling assumptions such as constant markup"]}, "claim_evidence": [{"claim": "Uncertainty in financial and real sectors spiked jointly in the Great Recession, motivating the question", "evidence_type": "prior_literature", "evidence_ref": "VIX; Jurado, Ludvigson, and Ng (2015); Bloom et al. (2012)"}, {"claim": "Mutual learning creates strategic complementarity in information production between firms and speculators", "evidence_type": "equation_or_proof", "evidence_ref": "Sections I–II; Appendix B"}, {"claim": "General equilibrium model yields amplification, co-movement of uncertainties, contagion and cross-sectional implications", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C; Figures 2–4"}, {"claim": "Dynamic OLG model can have two steady states (uncertainty traps) and a two-stage crisis", "evidence_type": "figure", "evidence_ref": "Figures 5, 6, 8; Table II"}], "structural_summary": "The paper opens with stylized facts on joint uncertainty spikes and poses them as research questions. It then builds the theory incrementally:\n- a one-firm baseline with exogenous information establishing two-way learning;\n- endogenous information acquisition generating strategic complementarity;\n- a macro GE extension that delivers four implications;\n- a dynamic OLG extension that delivers uncertainty traps and transitional crisis dynamics.\nNumerical illustrations with assigned parameters follow, then a conclusion restating the mechanism and noting extensions. Proofs are in the appendix.", "writing_profile_keywords": ["theory-building", "incremental model extension", "general equilibrium", "multiple equilibria", "stylized facts motivation", "numerical illustration", "proofs in appendix"], "confidence": 0.8, "notes": "The introduction is captured as a single paragraph, so moves beyond the motivation and approach preview are not visible. The numerical section explicitly disclaims calibration."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.08, "usage": {"input": 4806, "output": 1656, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.053138, "request_id": "req_011CggqNW87nM8jy5NKVkRHS", "cached": false, "ts": "2026-11-03T23:07:36+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/652a1a515d481af8b9a42a3f97d9bdf9a217bfcc143a4f32e81a1cd3e19e1f10.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"asset_pricing; macro_finance (dynamic contracting, limited commitment, risk management)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; the paper derives optimal policies from a continuous-time model with inalienable human capital and illustrates them numerically under a calibration\",\"key_assumptions\":[\"entrepreneur cannot irrevocably commit her human capital and can leave to recontract elsewhere\",\"complete, competitive financial markets apart from the inalienability friction\",\"risk-averse entrepreneur and diversified risk-averse investors\",\"quadratic capital adjustment costs (q theory)\",\"output observable and verifiable; no private saving\"],\"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\":4},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":2}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"other\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"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\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":13,\"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\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"theory\"},{\"idx\":7,\"role\":\"theory\"},{\"idx\":8,\"role\":\"theory\"},{\"idx\":9,\"role\":\"conclusion\"},{\"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\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Few tables. Two tables set out the correspondence between the primal optimization and dual contracting problems. A third lists calibration parameters with symbols and annualized values. There are no regression tables.\",\"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\":[\"'in general' qualifiers on main results\",\"'helps explain' / 'consistent with' when linking to empirical findings\",\"explicit acknowledgment of strong assumptions in conclusion\",\"'may' in describing alternative mechanisms\"],\"claim_evidence\":[{\"claim\":\"Relative to first-best, entrepreneur's net worth is overexposed to idiosyncratic risk and underexposed to systematic risk\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2 and 4; Section V.B–C\"},{\"claim\":\"Distortions grow as the firm approaches its endogenous debt capacity\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1–6\"},{\"claim\":\"Under inalienability the firm always underinvests and investment rises with liquidity\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Entrepreneur underconsumes; consumption is increasing and concave in liquidity\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"},{\"claim\":\"Liquidity/risk management problem is equivalent to an optimal long-term contract\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV, Tables I–II, Appendix B\"},{\"claim\":\"Optimality of debt financing from Hart and Moore is not robust in the general framework\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section VII, Figure 11\"},{\"claim\":\"Autarky as outside option yields very different predictions than recontracting\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 8; Section V.F\"},{\"claim\":\"Less constrained firms investing more in market portfolio matches empirical findings\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Duchin et al. (2017)\"}],\"structural_summary\":\"The introduction frames a frictionless benchmark, then introduces the inalienable-human-capital friction and claims a new theory of liquidity and risk management. It previews results on risk exposures and financial slack, and positions the paper at length against Hart–Moore, Ai–Li, Rampini–Viswanathan, Holmström–Tirole, the macro limited-commitment literature and the compensation literature, ending with a roadmap. The body builds the model, solves the first-best and the constrained case, and proves equivalence with a dual contracting problem. A calibrated numerical section illustrates the policies with figures. Extensions cover persistent shocks, the deterministic Hart–Moore case and two-sided commitment. The conclusion restates the contribution and names assumptions to relax. Proofs are in the appendices.\",\"writing_profile_keywords\":[\"dynamic contracting\",\"limited commitment\",\"continuous-time model\",\"primal-dual equivalence\",\"calibrated numerical illustration\",\"extensive literature positioning\",\"model extensions\",\"risk management theory\"],\"confidence\":0.85,\"notes\":\"The final abstract sentence is a parsing artifact (the start of the introduction). The quantitative section is classed as results even though it is numerical model illustration, not empirical estimation. Robustness here means model extensions and comparative statics, not empirical checks.\"}", "json": null, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 27.01, "usage": {"input": 11241, "output": 3417, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.114098, "request_id": "req_011CggqM8rvEuFUd3t8bW1JQ", "cached": false, "ts": "2026-11-03T23:07:30+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/654fb9e305590f84219c10a164a06a8e0e992b7b6f3b3592b19ae1b1b9f012ac.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"How much broker incentives (fees) and consumer search costs distort retail purchases toward dominated reverse convertible bonds, and the resulting losses to consumers\",\"strategy\":\"Within-set comparisons of simultaneously issued, otherwise-identical dominated and superior bonds using set fixed effects, plus maximum-likelihood estimation of a broker-intermediated search model with heterogeneous consumer sophistication and counterfactual search regimes\",\"key_assumptions\":[\"Reverse convertibles are fully described by a few observable features, so dominance is unambiguous\",\"Disclosed broker fees do not reflect differences in advice or bundled services\",\"Consumers use the same search strategies across markets (the reservation-utility distribution is constant)\",\"Consumers' beliefs about offered utilities match the empirical density\"],\"threats_addressed\":[\"Unobserved product characteristics and preference heterogeneity that could rationalize dispersion\",\"Endogeneity of fees with respect to unobserved product quality\"],\"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\":3},\"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\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":7,\"contribution_types\":[\"new_data_or_measure\",\"new_evidence\",\"new_theory_or_model\"],\"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\":\"other\"},{\"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\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Notes give the sample (SEC-registered one-year reverse convertibles, 2008-2012, largest 1% of issues dropped), the equation being estimated, and the fixed effects used (issuer, underlying equity, month, dominated/superior set). Stars mark 10/5/1% significance. Standard errors for the structural model come from the observed Fisher information matrix. Counterfactual tables report changes in expected returns.\",\"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\":[\"evidence consistent with\",\"the evidence suggests\",\"I argue\",\"likely not unique\",\"may not be acting\"]},\"claim_evidence\":[{\"claim\":\"Implied market values of reverse convertibles are widely dispersed (SD above 2.4% of notional).\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"When dominated and superior versions coexist, consumers buy more of the dominated one.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 4\"},{\"claim\":\"Products with higher broker fees have lower payoffs to consumers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Figure 5\"},{\"claim\":\"Brokers earn roughly twice the fee for selling dominated bonds.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Estimated search costs are a key friction, and brokers price discriminate by consumer sophistication.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V; Figure 7\"},{\"claim\":\"Consumer returns would change under random or optimally directed search compared with broker-directed search.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The introduction is motivated by price dispersion in financial products and policy concern over brokers not held to a fiduciary standard. It names data challenges in prior work and presents a novel reverse convertible dataset as a clean setting where dominance is unambiguous and broker fees are disclosed. It then previews three stylized facts and a broker-intermediated search model. Literature positioning and a roadmap close the introduction. The body moves from institutional background and data, to reduced-form evidence linking issue size and fees to product quality, to model setup and equilibrium, to maximum-likelihood estimation, and finally to search counterfactuals. The conclusion restates the findings and argues they likely generalize beyond this market.\",\"writing_profile_keywords\":[\"clean setting\",\"dominated products\",\"conflict of interest\",\"structural search model\",\"stylized facts\",\"counterfactuals\",\"novel data\",\"first-person singular\"],\"confidence\":0.8,\"notes\":\"The section labeled 'literature' by rule is actually institutional background and data. Robustness checks are not visible in the excerpts. Policy discussion appears in the introduction (the Obama fiduciary quote) but not explicitly in the conclusion excerpt.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "household_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "structural_estimation", "identification": {"estimand": "How much broker incentives (fees) and consumer search costs distort retail purchases toward dominated reverse convertible bonds, and the resulting losses to consumers", "strategy": "Within-set comparisons of simultaneously issued, otherwise-identical dominated and superior bonds using set fixed effects, plus maximum-likelihood estimation of a broker-intermediated search model with heterogeneous consumer sophistication and counterfactual search regimes", "key_assumptions": ["Reverse convertibles are fully described by a few observable features, so dominance is unambiguous", "Disclosed broker fees do not reflect differences in advice or bundled services", "Consumers use the same search strategies across markets (the reservation-utility distribution is constant)", "Consumers' beliefs about offered utilities match the empirical density"], "threats_addressed": ["Unobserved product characteristics and preference heterogeneity that could rationalize dispersion", "Endogeneity of fees with respect to unobserved product quality"], "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": 3}, "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": "policy_or_broader_implication"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 7, "contribution_types": ["new_data_or_measure", "new_evidence", "new_theory_or_model"], "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": "other"}, {"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": "none_seen", "types": []}, "tables": {"convention_notes": "Notes give the sample (SEC-registered one-year reverse convertibles, 2008-2012, largest 1% of issues dropped), the equation being estimated, and the fixed effects used (issuer, underlying equity, month, dominated/superior set). Stars mark 10/5/1% significance. Standard errors for the structural model come from the observed Fisher information matrix. Counterfactual tables report changes in expected returns.", "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": ["evidence consistent with", "the evidence suggests", "I argue", "likely not unique", "may not be acting"]}, "claim_evidence": [{"claim": "Implied market values of reverse convertibles are widely dispersed (SD above 2.4% of notional).", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "When dominated and superior versions coexist, consumers buy more of the dominated one.", "evidence_type": "table", "evidence_ref": "Table II, Figure 4"}, {"claim": "Products with higher broker fees have lower payoffs to consumers.", "evidence_type": "table", "evidence_ref": "Table III, Figure 5"}, {"claim": "Brokers earn roughly twice the fee for selling dominated bonds.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Estimated search costs are a key friction, and brokers price discriminate by consumer sophistication.", "evidence_type": "table", "evidence_ref": "Tables IV, V; Figure 7"}, {"claim": "Consumer returns would change under random or optimally directed search compared with broker-directed search.", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The introduction is motivated by price dispersion in financial products and policy concern over brokers not held to a fiduciary standard. It names data challenges in prior work and presents a novel reverse convertible dataset as a clean setting where dominance is unambiguous and broker fees are disclosed. It then previews three stylized facts and a broker-intermediated search model. Literature positioning and a roadmap close the introduction. The body moves from institutional background and data, to reduced-form evidence linking issue size and fees to product quality, to model setup and equilibrium, to maximum-likelihood estimation, and finally to search counterfactuals. The conclusion restates the findings and argues they likely generalize beyond this market.", "writing_profile_keywords": ["clean setting", "dominated products", "conflict of interest", "structural search model", "stylized facts", "counterfactuals", "novel data", "first-person singular"], "confidence": 0.8, "notes": "The section labeled 'literature' by rule is actually institutional background and data. Robustness checks are not visible in the excerpts. Policy discussion appears in the introduction (the Obama fiduciary quote) but not explicitly in the conclusion excerpt."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.59, "usage": {"input": 8663, "output": 2230, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080046, "request_id": "req_011CggqLYrHGW6qRXytiHu9C", "cached": false, "ts": "2026-11-03T23:07:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/6ef11269a772779a65ab49b17e05c3e9b2cb34a54f6378275574123a6c63062f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Effect of noise traders' reduced attention (distraction) on trading activity, liquidity, volatility and price reversals, especially in stocks predominantly owned by noise traders\",\"strategy\":\"Event study around days of sensational non-economic news, selected with a news pressure measure and treated as exogenous distraction shocks. Abnormal outcomes are measured against a 200-day window of days without economic news. Cross-sectional sample splits use proxies for retail ownership.\",\"key_assumptions\":[\"Sensational news events are exogenous to the stock market and carry no fundamental information\",\"These events distract noise traders more than informed speculators or market makers\",\"Estimation days without economic news provide a valid counterfactual\"],\"threats_addressed\":[\"confounding economic news\",\"reverse causality\",\"event list construction / keyword choice\",\"event clustering\",\"distraction of insiders or market makers\",\"investor sentiment\"],\"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\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"robustness_mention\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"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\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Event-study tables report abnormal values with t-statistics shown below each estimate. Notes state the sample period, the number of distraction events and the estimation window, and sort stocks into terciles. Variables are defined in the Appendix.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"we speculate\",\"may have\",\"suggest\",\"possible explanations\",\"alternative interpretation\"]},\"claim_evidence\":[{\"claim\":\"Distraction events reduce TV-watching investors' availability, validated by higher TV viewership on event days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Retail and institutional trading activity declines on distraction days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III, Figure 3\"},{\"claim\":\"Market-wide effects are weak overall\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"In small stocks, liquidity worsens 1-6%, volatility falls about 3%, and reversals fall about 4% on distraction days\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Events with larger trading declines show larger reductions in liquidity, volatility and reversals\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Distraction effects change across pre-Internet, Internet and algorithmic trading eras\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"In 1993-2006, effects are stronger among tech stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"In stocks with more algorithmic trading, effects are stronger for activity and volatility and attenuated for liquidity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The introduction opens with a gap: the attention literature focuses on informed agents and has neglected noise traders. It states two identification challenges and proposes sensational-news distraction days as the solution, then previews results with magnitudes and defends causality. A separate literature section follows, then data and methodology. Results move from investor-level trading to market outcomes, framed by a Kyle-model extension placed in the Internet Appendix. A dedicated robustness, endogeneity and alternatives section is followed by a technology and time-variation section. The conclusion restates findings, reconciles them with prior liquidity evidence, and offers speculative explanations plus a call for future work.\",\"writing_profile_keywords\":[\"natural experiment\",\"event study\",\"limited attention\",\"noise traders\",\"liquidity\",\"sample splits\",\"Kyle model\",\"technology and time trends\"],\"confidence\":0.82,\"notes\":\"The model is given only in the Internet Appendix and the excerpts contain no equations. Standard-error details are not visible: tables report t-statistics, and figures show confidence intervals (Figure 3 only). Nonparametric tests are mentioned in the introduction.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "behavioral_finance", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Effect of noise traders' reduced attention (distraction) on trading activity, liquidity, volatility and price reversals, especially in stocks predominantly owned by noise traders", "strategy": "Event study around days of sensational non-economic news, selected with a news pressure measure and treated as exogenous distraction shocks. Abnormal outcomes are measured against a 200-day window of days without economic news. Cross-sectional sample splits use proxies for retail ownership.", "key_assumptions": ["Sensational news events are exogenous to the stock market and carry no fundamental information", "These events distract noise traders more than informed speculators or market makers", "Estimation days without economic news provide a valid counterfactual"], "threats_addressed": ["confounding economic news", "reverse causality", "event list construction / keyword choice", "event clustering", "distraction of insiders or market makers", "investor sentiment"], "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": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "identification_defense"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "robustness_mention"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 0, "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": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_measure", "alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Event-study tables report abnormal values with t-statistics shown below each estimate. Notes state the sample period, the number of distraction events and the estimation window, and sort stocks into terciles. Variables are defined in the Appendix.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "we speculate", "may have", "suggest", "possible explanations", "alternative interpretation"]}, "claim_evidence": [{"claim": "Distraction events reduce TV-watching investors' availability, validated by higher TV viewership on event days", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Retail and institutional trading activity declines on distraction days", "evidence_type": "table", "evidence_ref": "Table II, Table III, Figure 3"}, {"claim": "Market-wide effects are weak overall", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "In small stocks, liquidity worsens 1-6%, volatility falls about 3%, and reversals fall about 4% on distraction days", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Events with larger trading declines show larger reductions in liquidity, volatility and reversals", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Distraction effects change across pre-Internet, Internet and algorithmic trading eras", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "In 1993-2006, effects are stronger among tech stocks", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "In stocks with more algorithmic trading, effects are stronger for activity and volatility and attenuated for liquidity", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The introduction opens with a gap: the attention literature focuses on informed agents and has neglected noise traders. It states two identification challenges and proposes sensational-news distraction days as the solution, then previews results with magnitudes and defends causality. A separate literature section follows, then data and methodology. Results move from investor-level trading to market outcomes, framed by a Kyle-model extension placed in the Internet Appendix. A dedicated robustness, endogeneity and alternatives section is followed by a technology and time-variation section. The conclusion restates findings, reconciles them with prior liquidity evidence, and offers speculative explanations plus a call for future work.", "writing_profile_keywords": ["natural experiment", "event study", "limited attention", "noise traders", "liquidity", "sample splits", "Kyle model", "technology and time trends"], "confidence": 0.82, "notes": "The model is given only in the Internet Appendix and the excerpts contain no equations. Standard-error details are not visible: tables report t-statistics, and figures show confidence intervals (Figure 3 only). Nonparametric tests are mentioned in the introduction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.83, "usage": {"input": 7211, "output": 2174, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073118, "request_id": "req_011Cggqh9GXt23tctuJLviJo", "cached": false, "ts": "2026-11-03T23:11:53+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/704f54e7e61f0d97dc788c26131eacde6051657ec2eb8c0e12bef9eea32f1fa3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How the inviter-invitee network is structured and how fast the idea spreads through it; how invested amounts (relative to income) relate to sponsor-investor differences in characteristics\",\"strategy\":\"Direct observation of word-of-mouth transmission through invitation-only membership, with power-law fitting and comparison to a simulated Poisson random network plus a calibrated Ponzi model; cross-sectional OLS with personal controls and inviter fixed effects\",\"key_assumptions\":[\"New investors learned of the scheme only from their inviter, since no public information existed\",\"Police interview data are free of the usual reporting and selection biases\",\"The simulated random network matches the real network on every dimension except degree distribution\"],\"threats_addressed\":[\"survey reporting and selection bias\",\"extreme tails of invested amounts (5% trimming)\",\"sponsor-level unobservables via inviter 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\":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\":2,\"move\":\"other\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":10,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":15,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":9,\"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\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Long descriptive notes define the panels and variables. The early tables are descriptive comparisons of sponsors versus nonsponsors and the actual versus simulated network; later tables report cross-sectional OLS of invested amount scaled by income. Currency (euros, thousands) and measurement dates are stated.\",\"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\",\"may be\",\"indicates\"]},\"claim_evidence\":[{\"claim\":\"The number of sponsored investors per sponsor approximately follows a power law, i.e., the network is scale-free\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4; KS tests\"},{\"claim\":\"The scale-free network spreads the idea far faster than a random network (15 vs 161 steps)\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"The cumulative number of investors follows an S-curve in network distance\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Investors invest more when their sponsor has relatively higher age, education, and income\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"The marginal effect of sponsor income peaks just above the investor's own income\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"Sponsors differ from nonsponsors in personal characteristics\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"}],\"structural_summary\":\"The paper opens with Shiller's social-epidemic view and introduces a unique Ponzi-scheme dataset from police investigation documents. Because entry was invitation-only, word-of-mouth transmission is directly observable, and the introduction defends the data quality. It previews network results (power law, simulation comparison, S-curve) and then investment-behavior regressions. Contributions are tied to the epidemic-diffusion, network-modeling, and peer-effects literatures before a roadmap. The body proceeds from background and data, to network analysis with simulation and calibration, to sponsor characteristics, to OLS regressions of invested amounts. The conclusion restates the finding that diffusion runs through a few hubs and notes welfare implications.\",\"writing_profile_keywords\":[\"unique data\",\"social network\",\"scale-free\",\"Ponzi scheme\",\"word-of-mouth\",\"simulation comparison\",\"descriptive-heavy\"],\"confidence\":0.8,\"notes\":\"Robustness checks are not visible in the excerpts, though an Internet Appendix exists. The regression analysis is associational. Section 1 is labeled literature by the rule but functions as background and data.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "household_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "How the inviter-invitee network is structured and how fast the idea spreads through it; how invested amounts (relative to income) relate to sponsor-investor differences in characteristics", "strategy": "Direct observation of word-of-mouth transmission through invitation-only membership, with power-law fitting and comparison to a simulated Poisson random network plus a calibrated Ponzi model; cross-sectional OLS with personal controls and inviter fixed effects", "key_assumptions": ["New investors learned of the scheme only from their inviter, since no public information existed", "Police interview data are free of the usual reporting and selection biases", "The simulated random network matches the real network on every dimension except degree distribution"], "threats_addressed": ["survey reporting and selection bias", "extreme tails of invested amounts (5% trimming)", "sponsor-level unobservables via inviter 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": 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": 2, "move": "other"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "literature_gap"}, {"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": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 10, "move": "mechanism_or_theory_sketch"}, {"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": "related_literature_positioning"}, {"paragraph_index": 15, "move": "roadmap"}], "contribution_paragraph_index": 9, "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": "none_seen", "types": []}, "tables": {"convention_notes": "Long descriptive notes define the panels and variables. The early tables are descriptive comparisons of sponsors versus nonsponsors and the actual versus simulated network; later tables report cross-sectional OLS of invested amount scaled by income. Currency (euros, thousands) and measurement dates are stated.", "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", "may be", "indicates"]}, "claim_evidence": [{"claim": "The number of sponsored investors per sponsor approximately follows a power law, i.e., the network is scale-free", "evidence_type": "figure", "evidence_ref": "Figure 4; KS tests"}, {"claim": "The scale-free network spreads the idea far faster than a random network (15 vs 161 steps)", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "The cumulative number of investors follows an S-curve in network distance", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Investors invest more when their sponsor has relatively higher age, education, and income", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "The marginal effect of sponsor income peaks just above the investor's own income", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "Sponsors differ from nonsponsors in personal characteristics", "evidence_type": "table", "evidence_ref": "Table V"}], "structural_summary": "The paper opens with Shiller's social-epidemic view and introduces a unique Ponzi-scheme dataset from police investigation documents. Because entry was invitation-only, word-of-mouth transmission is directly observable, and the introduction defends the data quality. It previews network results (power law, simulation comparison, S-curve) and then investment-behavior regressions. Contributions are tied to the epidemic-diffusion, network-modeling, and peer-effects literatures before a roadmap. The body proceeds from background and data, to network analysis with simulation and calibration, to sponsor characteristics, to OLS regressions of invested amounts. The conclusion restates the finding that diffusion runs through a few hubs and notes welfare implications.", "writing_profile_keywords": ["unique data", "social network", "scale-free", "Ponzi scheme", "word-of-mouth", "simulation comparison", "descriptive-heavy"], "confidence": 0.8, "notes": "Robustness checks are not visible in the excerpts, though an Internet Appendix exists. The regression analysis is associational. Section 1 is labeled literature by the rule but functions as background and data."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.31, "usage": {"input": 7671, "output": 2139, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074258, "request_id": "req_011CggqUmKbfb79TRYA2e7De", "cached": false, "ts": "2026-11-03T23:09:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/70a64c5b162459911bae3d62993fda236cf6548ef1487398101d4b48e4342e5f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"mutual funds / asset management industry organization; textual analysis\",\"primary_method\":\"machine_learning_textual\",\"identification\":{\"estimand\":\"Associations between text-based fund uniqueness and fund starts by family size, fees, investor flows, returns, and flow-performance sensitivity\",\"strategy\":\"Text-based uniqueness built from prospectus strategy descriptions, used in pooled OLS regressions with fund controls and category-by-month fixed effects, compared against holdings- and returns-based uniqueness; comparison of start rates by family size\",\"key_assumptions\":[\"Prospectus text captures product differentiation that holdings and returns characteristics miss\",\"Controls such as expenses (which include marketing) absorb confounding drivers of flows\"],\"threats_addressed\":[\"Measurement validity of the text measure, addressed with validation tests\",\"Overlap with holdings- and returns-based uniqueness\",\"Marketing expenses as an alternative channel for flows\",\"How fund starts are measured (counts vs. start rates scaled by family size)\"],\"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\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"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\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"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\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Table notes describe the sample period (1999–2015/16), the estimator (pooled OLS, monthly fund data), the dependent variable, and fixed effects such as category-by-month. Panels split samples or dependent variables (e.g., gross vs. net returns). The visible note excerpts 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\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'one possible explanation' versus 'an alternative explanation'\",\"'associated with' wording\",\"'weak negative' qualifiers\",\"'we try to fill this gap'\"]},\"claim_evidence\":[{\"claim\":\"Small and start-up fund families have higher start rates and launch more unique funds than large families\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 3\"},{\"claim\":\"New funds with above-median uniqueness attract about 50% (~$75M) more flows in their first 36 months\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4; Table VII\"},{\"claim\":\"Text-based uniqueness predicts new-fund flows beyond holdings/returns uniqueness and expense controls\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Text uniqueness has only a weak negative relation with returns of established funds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Unique funds show lower flow-performance sensitivity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"The text-based uniqueness measure is valid, i.e., it identifies distinctive fund types\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Smaller firms drive financial innovation, consistent with prior work\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Lerner (2006)\"}],\"structural_summary\":\"The introduction opens with industry growth statistics, flags unexplored questions, and motivates text-based uniqueness with an illustrative fund example. It then previews results for new and established funds, sketches an equilibrium intuition based on Berk and Green, and closes with contributions, literature positioning, and a roadmap. A hypothesis/framework section precedes the data and validation section, which constructs the measure. Two results sections follow: fund starts, fees, and flows; then equilibrium returns and flow-performance sensitivity. A short conclusion restates the contributions. Appendices give example texts, the word-filtering procedure, cutoff sensitivity, and readability/tonality analysis.\",\"writing_profile_keywords\":[\"textual analysis\",\"new measure\",\"mutual funds\",\"product differentiation\",\"financial innovation\",\"validation tests\",\"pooled OLS\",\"hypothesis framework\",\"equilibrium intuition\"],\"confidence\":0.8,\"notes\":\"Section IV is labeled 'theory' by the role rule but is substantively a results section (equilibrium empirical results). Standard error treatment is not visible in the excerpts. Robustness evidence visible in the excerpts is mainly the information-ratio cutoff appendix and the comparison of alternative uniqueness measures.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "mutual funds / asset management industry organization; textual analysis", "primary_method": "machine_learning_textual", "identification": {"estimand": "Associations between text-based fund uniqueness and fund starts by family size, fees, investor flows, returns, and flow-performance sensitivity", "strategy": "Text-based uniqueness built from prospectus strategy descriptions, used in pooled OLS regressions with fund controls and category-by-month fixed effects, compared against holdings- and returns-based uniqueness; comparison of start rates by family size", "key_assumptions": ["Prospectus text captures product differentiation that holdings and returns characteristics miss", "Controls such as expenses (which include marketing) absorb confounding drivers of flows"], "threats_addressed": ["Measurement validity of the text measure, addressed with validation tests", "Overlap with holdings- and returns-based uniqueness", "Marketing expenses as an alternative channel for flows", "How fund starts are measured (counts vs. start rates scaled by family size)"], "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": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"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": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "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": "conclusion"}, {"idx": 6, "role": "appendix"}, {"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", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Table notes describe the sample period (1999–2015/16), the estimator (pooled OLS, monthly fund data), the dependent variable, and fixed effects such as category-by-month. Panels split samples or dependent variables (e.g., gross vs. net returns). The visible note excerpts 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": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'one possible explanation' versus 'an alternative explanation'", "'associated with' wording", "'weak negative' qualifiers", "'we try to fill this gap'"]}, "claim_evidence": [{"claim": "Small and start-up fund families have higher start rates and launch more unique funds than large families", "evidence_type": "table", "evidence_ref": "Table IV; Figure 3"}, {"claim": "New funds with above-median uniqueness attract about 50% (~$75M) more flows in their first 36 months", "evidence_type": "figure", "evidence_ref": "Figure 4; Table VII"}, {"claim": "Text-based uniqueness predicts new-fund flows beyond holdings/returns uniqueness and expense controls", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Text uniqueness has only a weak negative relation with returns of established funds", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Unique funds show lower flow-performance sensitivity", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "The text-based uniqueness measure is valid, i.e., it identifies distinctive fund types", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Smaller firms drive financial innovation, consistent with prior work", "evidence_type": "prior_literature", "evidence_ref": "Lerner (2006)"}], "structural_summary": "The introduction opens with industry growth statistics, flags unexplored questions, and motivates text-based uniqueness with an illustrative fund example. It then previews results for new and established funds, sketches an equilibrium intuition based on Berk and Green, and closes with contributions, literature positioning, and a roadmap. A hypothesis/framework section precedes the data and validation section, which constructs the measure. Two results sections follow: fund starts, fees, and flows; then equilibrium returns and flow-performance sensitivity. A short conclusion restates the contributions. Appendices give example texts, the word-filtering procedure, cutoff sensitivity, and readability/tonality analysis.", "writing_profile_keywords": ["textual analysis", "new measure", "mutual funds", "product differentiation", "financial innovation", "validation tests", "pooled OLS", "hypothesis framework", "equilibrium intuition"], "confidence": 0.8, "notes": "Section IV is labeled 'theory' by the role rule but is substantively a results section (equilibrium empirical results). Standard error treatment is not visible in the excerpts. Robustness evidence visible in the excerpts is mainly the information-ratio cutoff appendix and the comparison of alternative uniqueness measures."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.81, "usage": {"input": 7811, "output": 2272, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077478, "request_id": "req_011Cggqiw8qWKTKSQ6Ny3F78", "cached": false, "ts": "2026-11-03T23:12:19+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/710ba5233f728cc9638a6fbdef7d7645a6886e83a33269ea49126e84e460b1a1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Effect of credit supply expansion on nontradable vs. tradable employment, nontradable prices, and later business-cycle amplitude across states and countries\",\"strategy\":\"Cross-state variation in the timing of 1980s US banking deregulation, used as an instrument or exposure score for loading on the 1982-1992 aggregate credit supply cycle. A first stage on leverage is shown. Long-difference and event-time regressions follow, with an international panel used for external validity.\",\"key_assumptions\":[\"Timing of deregulation is exogenous to current and anticipated business-cycle conditions\",\"Exclusion restriction: deregulation affects outcomes only through credit expansion\",\"Model-based mapping in which the household demand channel raises nontradable employment and prices relative to tradables\"],\"threats_addressed\":[\"Omitted variables correlated with deregulation timing\",\"Oil shocks\",\"Regulatory forbearance\",\"Industry trends\",\"China import penetration exposure\",\"Pre-existing higher cyclicality of early-deregulation states (placebo on earlier expansions)\",\"External validity of a single US episode\"],\"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\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\",\"new_method\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"literature\"},{\"idx\":7,\"role\":\"results\"},{\"idx\":8,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"placebo\",\"subsample\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Notes state the regression equation, define variables and data sources, and give SIC-based industry definitions. Heteroskedasticity-robust SEs appear in parentheses, with +, * and ** marking the 0.1, 0.05 and 0.01 levels. Some notes refer to Internet Appendix tables.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"robust\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"could be one reason\",\"may have\",\"suggest\",\"plausibly exogenous\",\"explained in part\"]},\"claim_evidence\":[{\"claim\":\"Earlier deregulation predicts a larger rise in household leverage during 1982-1989 (first stage)\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-IV, Figure 4\"},{\"claim\":\"Early-deregulation states saw relative nontradable employment growth with little change in tradable employment\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Figure 5\"},{\"claim\":\"Credit expansion raised consumer and nontradable prices in early-deregulation states\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 6\"},{\"claim\":\"Early-deregulation states showed no higher cyclicality in previous expansions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Early-deregulation states had an amplified boom and a more severe 1989-1992 recession\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII, Figure 7\"},{\"claim\":\"Household leverage growth in the boom is the strongest predictor of recession severity\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 8\"},{\"claim\":\"Nominal wages rose and did not reverse, consistent with downward nominal rigidity\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"},{\"claim\":\"Across 56 countries, household (not firm) credit growth is associated with rising nontradable-to-tradable employment, output and prices\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"In the 2000s US boom, household credit expansion is associated with similar patterns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"A household demand channel implies inflation and relative nontradable employment growth\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II, Figure 3\"}],\"structural_summary\":\"The introduction frames two channels of credit supply expansion and motivates distinguishing them. It then previews a test based on tradable/nontradable employment and prices, summarizes results on amplification and its mechanisms, defends identification via deregulation timing, and closes with a roadmap. The body proceeds through data, a simple equilibrium framework giving comparative statics, and the deregulation natural experiment with first stage and exclusion restriction. Main results on employment and prices follow, with placebo tests on earlier cycles, then business-cycle amplification and its explanations. A comparison with the deregulation literature and external-validity evidence from an international panel and the 2000s precede a short conclusion restating findings.\",\"writing_profile_keywords\":[\"natural experiment\",\"banking deregulation\",\"tradable vs nontradable\",\"model-guided test\",\"external validity panel\",\"placebo cycles\",\"scatterplot evidence\",\"event-time coefficients\"],\"confidence\":0.8,\"notes\":\"Section VII is labeled theory by rule but contains empirical results (international panel and 2000s boom), so it is coded as results. The framework section is a model used to derive testable predictions, not a standalone theory contribution. The abstract's final fragment is the start of the introduction and is ignored.\"}", "json": {"genre": "empirical", "subfield": "macro_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "instrumental_variables", "identification": {"estimand": "Effect of credit supply expansion on nontradable vs. tradable employment, nontradable prices, and later business-cycle amplitude across states and countries", "strategy": "Cross-state variation in the timing of 1980s US banking deregulation, used as an instrument or exposure score for loading on the 1982-1992 aggregate credit supply cycle. A first stage on leverage is shown. Long-difference and event-time regressions follow, with an international panel used for external validity.", "key_assumptions": ["Timing of deregulation is exogenous to current and anticipated business-cycle conditions", "Exclusion restriction: deregulation affects outcomes only through credit expansion", "Model-based mapping in which the household demand channel raises nontradable employment and prices relative to tradables"], "threats_addressed": ["Omitted variables correlated with deregulation timing", "Oil shocks", "Regulatory forbearance", "Industry trends", "China import penetration exposure", "Pre-existing higher cyclicality of early-deregulation states (placebo on earlier expansions)", "External validity of a single US episode"], "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": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"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": "related_literature_positioning"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "new_data_or_measure", "new_method"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "literature"}, {"idx": 7, "role": "results"}, {"idx": 8, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "placebo", "subsample", "alternative_measure"]}, "tables": {"convention_notes": "Notes state the regression equation, define variables and data sources, and give SIC-based industry definitions. Heteroskedasticity-robust SEs appear in parentheses, with +, * and ** marking the 0.1, 0.05 and 0.01 levels. Some notes refer to Internet Appendix tables.", "units_stated_in_notes": true, "standard_errors_description": "robust"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "could be one reason", "may have", "suggest", "plausibly exogenous", "explained in part"]}, "claim_evidence": [{"claim": "Earlier deregulation predicts a larger rise in household leverage during 1982-1989 (first stage)", "evidence_type": "table", "evidence_ref": "Tables III-IV, Figure 4"}, {"claim": "Early-deregulation states saw relative nontradable employment growth with little change in tradable employment", "evidence_type": "table", "evidence_ref": "Table V, Figure 5"}, {"claim": "Credit expansion raised consumer and nontradable prices in early-deregulation states", "evidence_type": "table", "evidence_ref": "Table VI, Figure 6"}, {"claim": "Early-deregulation states showed no higher cyclicality in previous expansions", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Early-deregulation states had an amplified boom and a more severe 1989-1992 recession", "evidence_type": "table", "evidence_ref": "Table VIII, Figure 7"}, {"claim": "Household leverage growth in the boom is the strongest predictor of recession severity", "evidence_type": "figure", "evidence_ref": "Figure 8"}, {"claim": "Nominal wages rose and did not reverse, consistent with downward nominal rigidity", "evidence_type": "figure", "evidence_ref": "Figure 9"}, {"claim": "Across 56 countries, household (not firm) credit growth is associated with rising nontradable-to-tradable employment, output and prices", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "In the 2000s US boom, household credit expansion is associated with similar patterns", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "A household demand channel implies inflation and relative nontradable employment growth", "evidence_type": "equation_or_proof", "evidence_ref": "Section II, Figure 3"}], "structural_summary": "The introduction frames two channels of credit supply expansion and motivates distinguishing them. It then previews a test based on tradable/nontradable employment and prices, summarizes results on amplification and its mechanisms, defends identification via deregulation timing, and closes with a roadmap. The body proceeds through data, a simple equilibrium framework giving comparative statics, and the deregulation natural experiment with first stage and exclusion restriction. Main results on employment and prices follow, with placebo tests on earlier cycles, then business-cycle amplification and its explanations. A comparison with the deregulation literature and external-validity evidence from an international panel and the 2000s precede a short conclusion restating findings.", "writing_profile_keywords": ["natural experiment", "banking deregulation", "tradable vs nontradable", "model-guided test", "external validity panel", "placebo cycles", "scatterplot evidence", "event-time coefficients"], "confidence": 0.8, "notes": "Section VII is labeled theory by rule but contains empirical results (international panel and 2000s boom), so it is coded as results. The framework section is a model used to derive testable predictions, not a standalone theory contribution. The abstract's final fragment is the start of the introduction and is ignored."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 29.09, "usage": {"input": 7959, "output": 3519, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.10301, "request_id": "req_011CggqkQ6MCF6cYDcAhF5pn", "cached": false, "ts": "2026-11-03T23:12:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/71b4085e8750e582f55ffcba0bae94d5e81c9f94d4ebcee643db7e29eedc5262.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"structural estimation of OTC search-and-bargaining model; insurance company investment\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"How network size relates to execution costs, plus structural primitives (search intensity, bargaining power, nontrade relationship value) used for welfare counterfactuals\",\"strategy\":\"Descriptive cross-sectional regressions of execution costs on network size with fixed effects, followed by minimum-distance estimation of a one-to-many search-and-bargaining model matched to network-size distribution and cost moments; counterfactual policy simulations\",\"key_assumptions\":[\"Model structure (winner-takes-all dealer races, Nash bargaining, repeat trading) captures insurer-dealer interaction\",\"Moments in Table VII are sensitive enough to separately identify parameters\",\"Merrill Lynch quote is a valid benchmark for execution costs\"],\"threats_addressed\":[\"parameter identification via sensitivity of moments\",\"alternative bargaining power specification (constant vs client-specific)\",\"heterogeneity across insurer, dealer, bond and day via clustering\"],\"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\":2},\"magnitude\":{\"present\":true,\"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\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":2,\"move\":\"literature_gap\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"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\":\"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\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Regression tables share a common caption convention, with later tables referring back to Table II. Dependent variables are log-transformed and execution costs are given in basis points relative to a quote benchmark. Significance is marked with stars. Structural estimates report sandwich standard errors in parentheses. Transition-matrix tables compare empirical and model-implied persistence.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"conditional welfare statements (unless nontrade value captured efficiently)\",\"'can decrease welfare'\",\"acknowledging model is stylized\",\"'suggest' for empirical patterns\"]},\"claim_evidence\":[{\"claim\":\"About one-third of insurers trade with a single dealer while some insurers have large dealer networks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table III\"},{\"claim\":\"Execution costs fall with network size up to roughly 20 dealers and rise beyond that\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Insurers' network size is persistent from year to year\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The estimated model quantitatively fits the network size distribution and the price–network relation\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6; Table VI\"},{\"claim\":\"Reducing repeat trading or unbundling nontrade services can raise costs or reduce welfare for some insurers\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7; Section IV.E\"},{\"claim\":\"Trading activity follows a power law\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"}],\"structural_summary\":\"The introduction motivates insurers' need for liquidity in OTC bond markets, then previews regulatory-data stylized facts. It sketches the one-to-many search model and identifies a gap: existing models predict either network size or prices, not both. It previews counterfactual policy results and closes with literature positioning and a roadmap. The body moves through data, three empirical stylized-fact subsections, model setup and predictions, and structural estimation with identification and fit, ending with policy counterfactuals. A short conclusion restates the facts and the policy relevance.\",\"writing_profile_keywords\":[\"stylized facts then structural model\",\"OTC search and bargaining\",\"regulatory transaction data\",\"counterfactual policy analysis\",\"power law distributions\",\"relationship trading\"],\"confidence\":0.8,\"notes\":\"Section IV combines estimation and policy counterfactuals; it is assigned the results role here even though the rule-based role was theory. Robustness is visible only through the parameter sensitivity table and alternative bargaining specifications.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "market_microstructure", "subfield_secondary": "structural estimation of OTC search-and-bargaining model; insurance company investment", "primary_method": "structural_estimation", "identification": {"estimand": "How network size relates to execution costs, plus structural primitives (search intensity, bargaining power, nontrade relationship value) used for welfare counterfactuals", "strategy": "Descriptive cross-sectional regressions of execution costs on network size with fixed effects, followed by minimum-distance estimation of a one-to-many search-and-bargaining model matched to network-size distribution and cost moments; counterfactual policy simulations", "key_assumptions": ["Model structure (winner-takes-all dealer races, Nash bargaining, repeat trading) captures insurer-dealer interaction", "Moments in Table VII are sensitive enough to separately identify parameters", "Merrill Lynch quote is a valid benchmark for execution costs"], "threats_addressed": ["parameter identification via sensitivity of moments", "alternative bargaining power specification (constant vs client-specific)", "heterogeneity across insurer, dealer, bond and day via clustering"], "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": 2}, "magnitude": {"present": true, "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": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 2, "move": "literature_gap"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 5, "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": "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": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Regression tables share a common caption convention, with later tables referring back to Table II. Dependent variables are log-transformed and execution costs are given in basis points relative to a quote benchmark. Significance is marked with stars. Structural estimates report sandwich standard errors in parentheses. Transition-matrix tables compare empirical and model-implied persistence.", "units_stated_in_notes": true, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["conditional welfare statements (unless nontrade value captured efficiently)", "'can decrease welfare'", "acknowledging model is stylized", "'suggest' for empirical patterns"]}, "claim_evidence": [{"claim": "About one-third of insurers trade with a single dealer while some insurers have large dealer networks", "evidence_type": "figure", "evidence_ref": "Figure 3; Table III"}, {"claim": "Execution costs fall with network size up to roughly 20 dealers and rise beyond that", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Insurers' network size is persistent from year to year", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The estimated model quantitatively fits the network size distribution and the price–network relation", "evidence_type": "figure", "evidence_ref": "Figures 5-6; Table VI"}, {"claim": "Reducing repeat trading or unbundling nontrade services can raise costs or reduce welfare for some insurers", "evidence_type": "figure", "evidence_ref": "Figure 7; Section IV.E"}, {"claim": "Trading activity follows a power law", "evidence_type": "figure", "evidence_ref": "Figure 2"}], "structural_summary": "The introduction motivates insurers' need for liquidity in OTC bond markets, then previews regulatory-data stylized facts. It sketches the one-to-many search model and identifies a gap: existing models predict either network size or prices, not both. It previews counterfactual policy results and closes with literature positioning and a roadmap. The body moves through data, three empirical stylized-fact subsections, model setup and predictions, and structural estimation with identification and fit, ending with policy counterfactuals. A short conclusion restates the facts and the policy relevance.", "writing_profile_keywords": ["stylized facts then structural model", "OTC search and bargaining", "regulatory transaction data", "counterfactual policy analysis", "power law distributions", "relationship trading"], "confidence": 0.8, "notes": "Section IV combines estimation and policy counterfactuals; it is assigned the results role here even though the rule-based role was theory. Robustness is visible only through the parameter sensitivity table and alternative bargaining specifications."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.26, "usage": {"input": 7923, "output": 2115, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074786, "request_id": "req_011CggqiREFHhNFu9KFJbT7t", "cached": false, "ts": "2026-11-03T23:12:09+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/740371daffcefcfdced488b88d3c6e4dc878eeca32376465d489829e5c0065de.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of the host-home interest rate differential (IOER rates) on a global bank's foreign-currency loan supply\",\"strategy\":\"Within-borrower and within-syndicate comparison across banks from different currency areas. Borrower-quarter and lender-quarter fixed effects isolate supply responses to IOER differentials. This is complemented by firm-level exposure analysis based on the past share of foreign banks.\",\"key_assumptions\":[\"Borrower-quarter fixed effects absorb time-varying loan demand\",\"Lender-quarter fixed effects absorb home-country shocks to banks\",\"Changes in IOER differentials are not driven by borrower-specific credit conditions\"],\"threats_addressed\":[\"Loan demand confounding\",\"Home-country shocks to bank health (e.g., Japanese banking crisis)\"],\"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\":3},\"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\":\"contribution_statement\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"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\":\"theory\"},{\"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 define dependent variables, units of observation (bank×currency×quarter, borrower-lender), and units (USD billions, percentage points, percent). The first table lists IOER rate changes by currency area.\",\"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\":\"low\",\"typical_devices\":[\"'tend to' phrasing\",\"'is associated with' for firm-level magnitudes\",\"model-conditional statements ('if capital constrained')\"]},\"claim_evidence\":[{\"claim\":\"With limited equity, lower domestic rates raise domestic lending and reduce foreign lending\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I model\"},{\"claim\":\"Introduction of IOER led to a large rise in foreign banks' Fed reserve holdings\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"A 25bp IOER differential lowers foreign-bank lending probability ~1.6% and volume ~3.6%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Firms with a larger past share of foreign banks get fewer and smaller loans after home-country easing\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Bank-level foreign currency lending share responds to the IOER difference\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The introduction motivates the paper with the scale of global bank activity, then claims a first framework linking monetary policy to loan supply across currencies via capital constraints. It previews bank-firm and firm-level results with magnitudes and defends identification through fixed effects. It positions the paper against the internal capital markets literature and closes with a roadmap. A simple analytical model derives testable hypotheses. Data from Call Reports, BIS and DealScan feed results that move from aggregate US and cross-country evidence to bank-, loan- and borrower-level regressions. The conclusion restates that spillovers run opposite to the internal markets view.\",\"writing_profile_keywords\":[\"model-then-test\",\"capital constraint mechanism\",\"syndicated loans\",\"fixed effects identification\",\"cross-border spillovers\",\"contrarian to existing view\"],\"confidence\":0.8,\"notes\":\"Standard error treatment and robustness checks are not visible in the excerpts. Significance conventions are unclear.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "international_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of the host-home interest rate differential (IOER rates) on a global bank's foreign-currency loan supply", "strategy": "Within-borrower and within-syndicate comparison across banks from different currency areas. Borrower-quarter and lender-quarter fixed effects isolate supply responses to IOER differentials. This is complemented by firm-level exposure analysis based on the past share of foreign banks.", "key_assumptions": ["Borrower-quarter fixed effects absorb time-varying loan demand", "Lender-quarter fixed effects absorb home-country shocks to banks", "Changes in IOER differentials are not driven by borrower-specific credit conditions"], "threats_addressed": ["Loan demand confounding", "Home-country shocks to bank health (e.g., Japanese banking crisis)"], "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": 3}, "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": "contribution_statement"}, {"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": "related_literature_positioning"}, {"paragraph_index": 3, "move": "literature_gap"}, {"paragraph_index": 4, "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": "theory"}, {"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 define dependent variables, units of observation (bank×currency×quarter, borrower-lender), and units (USD billions, percentage points, percent). The first table lists IOER rate changes by currency area.", "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": "low", "typical_devices": ["'tend to' phrasing", "'is associated with' for firm-level magnitudes", "model-conditional statements ('if capital constrained')"]}, "claim_evidence": [{"claim": "With limited equity, lower domestic rates raise domestic lending and reduce foreign lending", "evidence_type": "equation_or_proof", "evidence_ref": "Section I model"}, {"claim": "Introduction of IOER led to a large rise in foreign banks' Fed reserve holdings", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "A 25bp IOER differential lowers foreign-bank lending probability ~1.6% and volume ~3.6%", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Firms with a larger past share of foreign banks get fewer and smaller loans after home-country easing", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Bank-level foreign currency lending share responds to the IOER difference", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The introduction motivates the paper with the scale of global bank activity, then claims a first framework linking monetary policy to loan supply across currencies via capital constraints. It previews bank-firm and firm-level results with magnitudes and defends identification through fixed effects. It positions the paper against the internal capital markets literature and closes with a roadmap. A simple analytical model derives testable hypotheses. Data from Call Reports, BIS and DealScan feed results that move from aggregate US and cross-country evidence to bank-, loan- and borrower-level regressions. The conclusion restates that spillovers run opposite to the internal markets view.", "writing_profile_keywords": ["model-then-test", "capital constraint mechanism", "syndicated loans", "fixed effects identification", "cross-border spillovers", "contrarian to existing view"], "confidence": 0.8, "notes": "Standard error treatment and robustness checks are not visible in the excerpts. Significance conventions are unclear."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.14, "usage": {"input": 6314, "output": 1811, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06227, "request_id": "req_011Cggr23eUojpAyvc4DvqGv", "cached": false, "ts": "2026-11-03T23:16:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7471f998a4ad89e8a145407df6d2914f9b688eeb7e29d9a9aee52ad981ad0e88.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\":\"Predictive (time-series) association between lagged cyclical consumption and future h-quarter aggregate excess stock returns\",\"strategy\":\"In-sample and out-of-sample predictive regressions of future returns on detrended consumption (Hamilton projection). Small-sample inference uses IVX and bootstrap. Results are compared with calibrated Campbell–Cochrane habit model simulations.\",\"key_assumptions\":[\"Cyclical consumption measured via Hamilton (2018) projection captures deviations from trend\",\"Inference robust to persistent predictor (IVX/bootstrap validity)\",\"External habit model as interpretive framework\"],\"threats_addressed\":[\"Persistent regressor bias (Stambaugh)\",\"Look-ahead bias via recursive estimation out-of-sample\",\"Detrending method choice\",\"Temporal instability/subsample periods\",\"Subsumption by alternative predictors\"],\"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\":\"mechanism_or_theory_sketch\"},{\"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\":\"robustness_mention\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":4,\"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\":\"robustness\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes open by writing out the regression equation and defining the horizon, the return measure, and the predictor. Significance is marked with stars, and test statistics such as IVX-Wald and out-of-sample R2 are reported. Simulated-data tables state that values are in annualized percentages.\",\"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' framing\",\"'we conjecture'\",\"'suggest' for model-data link\",\"acknowledging fit with alternative model classes\"]},\"claim_evidence\":[{\"claim\":\"Low cyclical consumption predicts high future excess market returns across horizons\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Predictability survives persistence-robust IVX inference\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Predictive power is present in both good and bad times\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Out-of-sample forecasts beat constant expected return benchmark\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Cyclical consumption outperforms 19 alternative predictors out of sample\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Predictability extends to industry portfolios and international markets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, IX\"},{\"claim\":\"Simulated habit model generates an inverse cc-return relation similar to the data\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"}],\"structural_summary\":\"The introduction opens with a habit-formation mechanism and conjectures an inverse relation between detrended consumption and future returns. It then previews the empirical findings, the model simulation, the robustness and international extensions, and the horse race against existing predictors. The body proceeds in order: construction of the measure, in-sample and out-of-sample predictive regressions with state, detrending, stability and robustness splits, comparison with alternative predictors, and calibration of the external habit model. The conclusion restates the measure and its interpretation.\",\"writing_profile_keywords\":[\"return predictability\",\"consumption-based asset pricing\",\"habit formation\",\"out-of-sample tests\",\"horse race against predictors\",\"model calibration check\"],\"confidence\":0.85,\"notes\":\"No roadmap paragraph appears in the visible introduction. Section IV was auto-labeled data but functions as results comparing alternative predictors. Section I constructs the measure and is mapped to the data role. No specific magnitudes are visible in the intro excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Predictive (time-series) association between lagged cyclical consumption and future h-quarter aggregate excess stock returns", "strategy": "In-sample and out-of-sample predictive regressions of future returns on detrended consumption (Hamilton projection). Small-sample inference uses IVX and bootstrap. Results are compared with calibrated Campbell–Cochrane habit model simulations.", "key_assumptions": ["Cyclical consumption measured via Hamilton (2018) projection captures deviations from trend", "Inference robust to persistent predictor (IVX/bootstrap validity)", "External habit model as interpretive framework"], "threats_addressed": ["Persistent regressor bias (Stambaugh)", "Look-ahead bias via recursive estimation out-of-sample", "Detrending method choice", "Temporal instability/subsample periods", "Subsumption by alternative predictors"], "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": "mechanism_or_theory_sketch"}, {"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": "robustness_mention"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}], "contribution_paragraph_index": 4, "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": "robustness"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "subsample", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Table notes open by writing out the regression equation and defining the horizon, the return measure, and the predictor. Significance is marked with stars, and test statistics such as IVX-Wald and out-of-sample R2 are reported. Simulated-data tables state that values are in annualized percentages.", "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' framing", "'we conjecture'", "'suggest' for model-data link", "acknowledging fit with alternative model classes"]}, "claim_evidence": [{"claim": "Low cyclical consumption predicts high future excess market returns across horizons", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Predictability survives persistence-robust IVX inference", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Predictive power is present in both good and bad times", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Out-of-sample forecasts beat constant expected return benchmark", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Cyclical consumption outperforms 19 alternative predictors out of sample", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Predictability extends to industry portfolios and international markets", "evidence_type": "table", "evidence_ref": "Tables VII, IX"}, {"claim": "Simulated habit model generates an inverse cc-return relation similar to the data", "evidence_type": "table", "evidence_ref": "Table XII"}], "structural_summary": "The introduction opens with a habit-formation mechanism and conjectures an inverse relation between detrended consumption and future returns. It then previews the empirical findings, the model simulation, the robustness and international extensions, and the horse race against existing predictors. The body proceeds in order: construction of the measure, in-sample and out-of-sample predictive regressions with state, detrending, stability and robustness splits, comparison with alternative predictors, and calibration of the external habit model. The conclusion restates the measure and its interpretation.", "writing_profile_keywords": ["return predictability", "consumption-based asset pricing", "habit formation", "out-of-sample tests", "horse race against predictors", "model calibration check"], "confidence": 0.85, "notes": "No roadmap paragraph appears in the visible introduction. Section IV was auto-labeled data but functions as results comparing alternative predictors. Section I constructs the measure and is mapped to the data role. No specific magnitudes are visible in the intro excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.1, "usage": {"input": 6588, "output": 2000, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067146, "request_id": "req_011CggqpwP8pEVScFKF8GxuW", "cached": false, "ts": "2026-11-03T23:13:37+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/75e4885abcf57a6944f3b27ced9f3a61fac6cf3cf60d48bcb308c510a6d2d49b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance; credit risk pricing\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Loan premium: the gap between observed all-in-drawn loan spreads and the counterfactual spread the bond market would assign to the same loan's cash flows, accounting for seniority.\",\"strategy\":\"Pair each new loan with the same firm's outstanding unsecured bond quote on the loan start date. Calibrate a Merton-type structural model, and alternatively a reduced-form Duffie–Singleton model, to price the bond exactly, then compute the bond-implied loan spread.\",\"key_assumptions\":[\"Same-firm, same-date matching removes firm-time unobservables affecting credit risk pricing\",\"Credit models price seniority correctly\",\"Recovery specifications (bankruptcy costs or a calibrated beta distribution) are reasonable\",\"Distress, issuance and illiquidity costs are not implausibly large\"],\"threats_addressed\":[\"Firm-time unobservable factors\",\"Model misspecification in pricing seniority (tested with secured vs unsecured bonds)\",\"Risk premium on recoveries and tail risk\",\"Prepayment options and bond issuance costs\",\"Loan illiquidity\",\"External validity (in- vs out-of-sample issuers)\"],\"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\":3},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"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\":\"policy_or_broader_implication\"},{\"paragraph_index\":7,\"move\":\"other\"},{\"paragraph_index\":8,\"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\":\"data\"},{\"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_measure\",\"sensitivity_to_parameters\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes describe the sample source and period (DealScan/Compustat, Moody's URD, 1997–2017) and point to the Appendix for variable definitions. They mention winsorization at 1% and month fixed effects with within R2. Model tables split recovery specifications into panels.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\",\"convention_notes_placeholder\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggests' for inferences drawn from figures\",\"explicitly stating that the result does not imply bank borrowing costs more\",\"acknowledging difficulty pinning down the mechanism\",\"calling for future research on how the surplus is divided\"]},\"claim_evidence\":[{\"claim\":\"Loans recover far more than senior unsecured bonds in default (about 84% vs 35%).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Loan and bond spreads are similar for safe firms; the bond/loan spread ratio is well below the model-predicted four.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"The bond-implied loan spread is 140–170 bps below the observed spread, so roughly half of the spread is premium.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Implausibly high distress, issuance or illiquidity costs would be needed to justify observed loan pricing.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Secured vs unsecured bond spreads match the expected-loss ratio, so models price seniority adequately.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"The reduced-form default intensity approach yields similar premium estimates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V Panel C\"},{\"claim\":\"The premium varies with firm and loan characteristics.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"}],\"structural_summary\":\"The introduction opens with a gap on relative loan versus bond pricing. It then documents seniority differences and a figure-based puzzle (loan spreads are too high given loans' seniority), and previews a structural quantification of the premium plus checks on model validity. Data construction rests on a same-firm, same-date loan–bond matching design. The core section calibrates a Merton extension to bond prices to obtain counterfactual loan spreads, then runs comparative statics and robustness tests (reduced-form model, recovery risk premium, seniority pricing). A further section uses cross-sectional regressions and discussion to interpret the premium. The conclusion is short, bounds the claims and poses open questions about surplus sharing.\",\"writing_profile_keywords\":[\"novel matched sample\",\"structural credit model\",\"counterfactual pricing\",\"puzzle framing\",\"model validation\",\"single-author first person\",\"bounded interpretation\"],\"confidence\":0.8,\"notes\":\"Single-authored, uses first person. Statistical significance conventions are not visible in the excerpts. The causal question is about pricing rather than treatment effects, so causal language strength is low.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "corporate_finance; credit risk pricing", "primary_method": "structural_estimation", "identification": {"estimand": "Loan premium: the gap between observed all-in-drawn loan spreads and the counterfactual spread the bond market would assign to the same loan's cash flows, accounting for seniority.", "strategy": "Pair each new loan with the same firm's outstanding unsecured bond quote on the loan start date. Calibrate a Merton-type structural model, and alternatively a reduced-form Duffie–Singleton model, to price the bond exactly, then compute the bond-implied loan spread.", "key_assumptions": ["Same-firm, same-date matching removes firm-time unobservables affecting credit risk pricing", "Credit models price seniority correctly", "Recovery specifications (bankruptcy costs or a calibrated beta distribution) are reasonable", "Distress, issuance and illiquidity costs are not implausibly large"], "threats_addressed": ["Firm-time unobservable factors", "Model misspecification in pricing seniority (tested with secured vs unsecured bonds)", "Risk premium on recoveries and tail risk", "Prepayment options and bond issuance costs", "Loan illiquidity", "External validity (in- vs out-of-sample issuers)"], "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": 3}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"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": "policy_or_broader_implication"}, {"paragraph_index": 7, "move": "other"}, {"paragraph_index": 8, "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": "data"}, {"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_measure", "sensitivity_to_parameters", "falsification", "other"]}, "tables": {"convention_notes": "Notes describe the sample source and period (DealScan/Compustat, Moody's URD, 1997–2017) and point to the Appendix for variable definitions. They mention winsorization at 1% and month fixed effects with within R2. Model tables split recovery specifications into panels.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear", "convention_notes_placeholder": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggests' for inferences drawn from figures", "explicitly stating that the result does not imply bank borrowing costs more", "acknowledging difficulty pinning down the mechanism", "calling for future research on how the surplus is divided"]}, "claim_evidence": [{"claim": "Loans recover far more than senior unsecured bonds in default (about 84% vs 35%).", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Loan and bond spreads are similar for safe firms; the bond/loan spread ratio is well below the model-predicted four.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "The bond-implied loan spread is 140–170 bps below the observed spread, so roughly half of the spread is premium.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Implausibly high distress, issuance or illiquidity costs would be needed to justify observed loan pricing.", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Secured vs unsecured bond spreads match the expected-loss ratio, so models price seniority adequately.", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "The reduced-form default intensity approach yields similar premium estimates.", "evidence_type": "table", "evidence_ref": "Table V Panel C"}, {"claim": "The premium varies with firm and loan characteristics.", "evidence_type": "table", "evidence_ref": "Table VI"}], "structural_summary": "The introduction opens with a gap on relative loan versus bond pricing. It then documents seniority differences and a figure-based puzzle (loan spreads are too high given loans' seniority), and previews a structural quantification of the premium plus checks on model validity. Data construction rests on a same-firm, same-date loan–bond matching design. The core section calibrates a Merton extension to bond prices to obtain counterfactual loan spreads, then runs comparative statics and robustness tests (reduced-form model, recovery risk premium, seniority pricing). A further section uses cross-sectional regressions and discussion to interpret the premium. The conclusion is short, bounds the claims and poses open questions about surplus sharing.", "writing_profile_keywords": ["novel matched sample", "structural credit model", "counterfactual pricing", "puzzle framing", "model validation", "single-author first person", "bounded interpretation"], "confidence": 0.8, "notes": "Single-authored, uses first person. Statistical significance conventions are not visible in the excerpts. The causal question is about pricing rather than treatment effects, so causal language strength is low."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 25.91, "usage": {"input": 7033, "output": 2330, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075526, "request_id": "req_011Cggqjq8xDi5ZWghgazdr1", "cached": false, "ts": "2026-11-03T23:12:37+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/75f27005e9688ca4cfc2fa79e5eaa16c0a3c3c9df3ac2a5fad0ed2bacf1e2551.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"corporate_finance (structural credit risk)\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Firm-level asset variance risk premium and jump risk premia, plus their contribution to credit spreads\",\"strategy\":\"Firm-by-firm simulated maximum likelihood estimation of a structural SVJ credit model on CDS term structures and option-implied equity volatility, where two observables are assumed accurately priced and the rest observed with error; nested-model likelihood ratio tests; calibration to representative Baa and Aa firms\",\"key_assumptions\":[\"Structural first-passage model with exogenous default boundary in the baseline\",\"Variance risk premium is affine (proportional) in asset variance\",\"Two observables are measured without error and the others with error\",\"Heston-type asset variance dynamics\"],\"threats_addressed\":[\"Model misspecification via an extended affine risk premium alternative\",\"Short-maturity fit, addressed by adding jumps\",\"Endogenous default and optimal leverage, examined as extensions\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"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\":true,\"sentence_index\":3}},\"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\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"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\":\"robustness\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Calibration tables state the calibration targets (Sharpe ratio, default probability, recovery, rates) in their notes and compare risk-premium compositions. Estimation tables report cross-sectional quantiles of parameter estimates and RMSE pricing errors in basis points.\",\"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 of\",\"possible link\",\"economically reasonable\",\"to the best of our knowledge\"]},\"claim_evidence\":[{\"claim\":\"A reasonable asset variance risk premium lets the calibrated model match medium- and long-maturity spreads for Baa and Aa firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-II\"},{\"claim\":\"Jump risk premia help fit short-term spreads while the variance risk premium stays economically important\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Firm-level asset variance risk premia are negative and significant for most firms; mean implied equity variance risk premium is about -16.7%\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, IX\"},{\"claim\":\"The SVJ model dominates the nested SV and constant-volatility models in pricing performance\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table VIII\"},{\"claim\":\"The variance risk premium accounts for about 35% of five-year CDS spreads on average\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"Debt Sharpe ratios exceed equity Sharpe ratios for levered firms, linking the credit spread and distress puzzles\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"}],\"structural_summary\":\"The paper opens by framing the credit spread puzzle and building on the insight that risk premia matter more than default probabilities. It then extends the Leland model with priced stochastic asset volatility and derives semi-closed-form solutions. Calibrations to representative Baa and Aa firms show the puzzle is resolved at medium and long maturities. An alternative risk-premium specification serves as a check, and jumps are added to fit short maturities. The SVJ model is estimated firm by firm with simulated maximum likelihood on CDS and option data, and spreads are decomposed into loss and premium components. Extensions to endogenous default and optimal leverage follow, and the conclusion recaps the findings.\",\"writing_profile_keywords\":[\"structural credit model\",\"puzzle resolution\",\"calibration then estimation\",\"nested model comparison\",\"variance risk premium\",\"firm-level estimation\"],\"confidence\":0.8,\"notes\":\"How statistical significance is reported is not visible in the excerpts. The estimation section includes data and methodology subsections; it is labeled as results.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "corporate_finance (structural credit risk)", "primary_method": "structural_estimation", "identification": {"estimand": "Firm-level asset variance risk premium and jump risk premia, plus their contribution to credit spreads", "strategy": "Firm-by-firm simulated maximum likelihood estimation of a structural SVJ credit model on CDS term structures and option-implied equity volatility, where two observables are assumed accurately priced and the rest observed with error; nested-model likelihood ratio tests; calibration to representative Baa and Aa firms", "key_assumptions": ["Structural first-passage model with exogenous default boundary in the baseline", "Variance risk premium is affine (proportional) in asset variance", "Two observables are measured without error and the others with error", "Heston-type asset variance dynamics"], "threats_addressed": ["Model misspecification via an extended affine risk premium alternative", "Short-maturity fit, addressed by adding jumps", "Endogenous default and optimal leverage, examined as extensions"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "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": true, "sentence_index": 3}}, "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": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 1, "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": "robustness"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "Calibration tables state the calibration targets (Sharpe ratio, default probability, recovery, rates) in their notes and compare risk-premium compositions. Estimation tables report cross-sectional quantiles of parameter estimates and RMSE pricing errors in basis points.", "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 of", "possible link", "economically reasonable", "to the best of our knowledge"]}, "claim_evidence": [{"claim": "A reasonable asset variance risk premium lets the calibrated model match medium- and long-maturity spreads for Baa and Aa firms", "evidence_type": "table", "evidence_ref": "Tables I-II"}, {"claim": "Jump risk premia help fit short-term spreads while the variance risk premium stays economically important", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Firm-level asset variance risk premia are negative and significant for most firms; mean implied equity variance risk premium is about -16.7%", "evidence_type": "table", "evidence_ref": "Tables VII, IX"}, {"claim": "The SVJ model dominates the nested SV and constant-volatility models in pricing performance", "evidence_type": "figure", "evidence_ref": "Figure 3; Table VIII"}, {"claim": "The variance risk premium accounts for about 35% of five-year CDS spreads on average", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "Debt Sharpe ratios exceed equity Sharpe ratios for levered firms, linking the credit spread and distress puzzles", "evidence_type": "figure", "evidence_ref": "Figure 2"}], "structural_summary": "The paper opens by framing the credit spread puzzle and building on the insight that risk premia matter more than default probabilities. It then extends the Leland model with priced stochastic asset volatility and derives semi-closed-form solutions. Calibrations to representative Baa and Aa firms show the puzzle is resolved at medium and long maturities. An alternative risk-premium specification serves as a check, and jumps are added to fit short maturities. The SVJ model is estimated firm by firm with simulated maximum likelihood on CDS and option data, and spreads are decomposed into loss and premium components. Extensions to endogenous default and optimal leverage follow, and the conclusion recaps the findings.", "writing_profile_keywords": ["structural credit model", "puzzle resolution", "calibration then estimation", "nested model comparison", "variance risk premium", "firm-level estimation"], "confidence": 0.8, "notes": "How statistical significance is reported is not visible in the excerpts. The estimation section includes data and methodology subsections; it is labeled as results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.62, "usage": {"input": 7423, "output": 2087, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072226, "request_id": "req_011CggqRLX1KiMN2ygSHLMHR", "cached": false, "ts": "2026-11-03T23:08:17+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7666c87ce830de1edd907db1b083478629d59037a4b154b10c2c470673e53a41.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 protection sellers' CDS-portfolio capital shocks (dollar margin payments) on weekly log changes in a firm's CDS spread\",\"strategy\":\"Panel regressions of weekly spread changes on seller and buyer capital shocks that exclude positions on same-industry firms, with firm-level credit controls and macro controls. Supplemented by placebo tests, horizon dynamics, and a 2011 Japanese tsunami case study treated as an exogenous shock to seller capital.\",\"key_assumptions\":[\"Seller capital shocks built from positions on other-industry firms are not driven by the target firm's fundamentals\",\"Controls such as EDF, LGD and equity returns absorb changes in fundamental credit risk\",\"Sellers are not fully hedged, so mark-to-market losses change their risk appetite\"],\"threats_addressed\":[\"common fundamental shocks driving both seller losses and spreads\",\"hedging by sellers or buyers\",\"whether capital shocks are large enough to matter\",\"construction methodology of capital measures\",\"institution-wide versus desk-level constraints\"],\"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\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"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\":\"preview_of_results\"}],\"contribution_paragraph_index\":-1,\"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\":\"robustness\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"appendix\"}],\"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\",\"alternative_sample_period\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes write out the regression equation, list the control vectors, and define variables with their units (dollar changes in $ billion). Standard errors appear below the estimates, double-clustered, with stars marking p-value thresholds.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggest\",\"imply\",\"in part\",\"at least partially\",\"likely explanation\",\"potential unintended consequence\"]},\"claim_evidence\":[{\"claim\":\"A $1bn seller capital loss raises CDS spreads about 2.7% per week; seller shocks explain 12% of weekly spread variation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Desk-level seller shocks carry information beyond institution-wide capital constraints.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The spread effect of seller losses reverts fully after about nine weeks.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Incumbent sellers barely adjust positions after losses, and no new sellers enter.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Seller exposure to Japanese firms predicts U.S. CDS spread changes around the 2011 tsunami.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Asset managers have replaced dealers as the main net sellers of protection since 2010.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"}],\"structural_summary\":\"The introduction frames frictionless capital flow as a neoclassical benchmark, then sketches intermediary and slow-moving-capital theories and notes that causal identification is hard. It previews results paragraph by paragraph, giving magnitudes, desk-specific effects, horizon decay, mechanism, and dealer versus asset manager heterogeneity. The data section documents concentration in the market. The main section builds a simple pricing framework, runs a workhorse regression, and adds internal capital market, segmentation/placebo, horizon, and mechanism analyses. Robustness and additional analyses (hedging, shock size, tsunami case study, bond market) follow. The conclusion restates the segmentation interpretation and notes possible regulatory side effects.\",\"writing_profile_keywords\":[\"proprietary regulatory data\",\"slow-moving capital\",\"market segmentation\",\"first-person singular\",\"economic magnitude emphasis\",\"case-study corroboration\",\"policy implication\"],\"confidence\":0.8,\"notes\":\"The introduction excerpt is truncated, so contribution and roadmap paragraphs may exist but are not visible. Figure captions do not say whether confidence bands are shown.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of protection sellers' CDS-portfolio capital shocks (dollar margin payments) on weekly log changes in a firm's CDS spread", "strategy": "Panel regressions of weekly spread changes on seller and buyer capital shocks that exclude positions on same-industry firms, with firm-level credit controls and macro controls. Supplemented by placebo tests, horizon dynamics, and a 2011 Japanese tsunami case study treated as an exogenous shock to seller capital.", "key_assumptions": ["Seller capital shocks built from positions on other-industry firms are not driven by the target firm's fundamentals", "Controls such as EDF, LGD and equity returns absorb changes in fundamental credit risk", "Sellers are not fully hedged, so mark-to-market losses change their risk appetite"], "threats_addressed": ["common fundamental shocks driving both seller losses and spreads", "hedging by sellers or buyers", "whether capital shocks are large enough to matter", "construction methodology of capital measures", "institution-wide versus desk-level constraints"], "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": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"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": "preview_of_results"}], "contribution_paragraph_index": -1, "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": "robustness"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "appendix"}], "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", "alternative_sample_period", "other"]}, "tables": {"convention_notes": "Notes write out the regression equation, list the control vectors, and define variables with their units (dollar changes in $ billion). Standard errors appear below the estimates, double-clustered, with stars marking p-value thresholds.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggest", "imply", "in part", "at least partially", "likely explanation", "potential unintended consequence"]}, "claim_evidence": [{"claim": "A $1bn seller capital loss raises CDS spreads about 2.7% per week; seller shocks explain 12% of weekly spread variation.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Desk-level seller shocks carry information beyond institution-wide capital constraints.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The spread effect of seller losses reverts fully after about nine weeks.", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Incumbent sellers barely adjust positions after losses, and no new sellers enter.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Seller exposure to Japanese firms predicts U.S. CDS spread changes around the 2011 tsunami.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Asset managers have replaced dealers as the main net sellers of protection since 2010.", "evidence_type": "figure", "evidence_ref": "Figure 5"}], "structural_summary": "The introduction frames frictionless capital flow as a neoclassical benchmark, then sketches intermediary and slow-moving-capital theories and notes that causal identification is hard. It previews results paragraph by paragraph, giving magnitudes, desk-specific effects, horizon decay, mechanism, and dealer versus asset manager heterogeneity. The data section documents concentration in the market. The main section builds a simple pricing framework, runs a workhorse regression, and adds internal capital market, segmentation/placebo, horizon, and mechanism analyses. Robustness and additional analyses (hedging, shock size, tsunami case study, bond market) follow. The conclusion restates the segmentation interpretation and notes possible regulatory side effects.", "writing_profile_keywords": ["proprietary regulatory data", "slow-moving capital", "market segmentation", "first-person singular", "economic magnitude emphasis", "case-study corroboration", "policy implication"], "confidence": 0.8, "notes": "The introduction excerpt is truncated, so contribution and roadmap paragraphs may exist but are not visible. Figure captions do not say whether confidence bands are shown."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.31, "usage": {"input": 7618, "output": 2075, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072766, "request_id": "req_011CggqUBFjLBYv7wWYtZ7Ei", "cached": false, "ts": "2026-11-03T23:08:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/778309e1b77f4cb82c470d31ccb4d0e86d64c266336bb13287d4a6d00b44c169.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"corrigendum correcting an author affiliation for an article on brokers and order flow leakage\",\"primary_method\":\"none\",\"identification\":{\"estimand\":\"not applicable\",\"strategy\":\"not applicable\",\"key_assumptions\":[],\"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\":false,\"sentence_index\":-1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"other\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"replication_or_correction\"],\"roadmap_present\":false,\"literature_positioning_location\":\"none\"},\"section_roles\":[{\"idx\":0,\"role\":\"other\"}],\"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.\",\"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\":[]},\"claim_evidence\":[{\"claim\":\"A co-author's affiliation in the original acknowledgments footnote was wrong, and the corrected affiliation is supplied.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"paragraph 0\"}],\"structural_summary\":\"A brief administrative corrigendum. It identifies the page and footnote of the original article where an author's affiliation was misstated, then gives the correct affiliation. It contains no substantive argument, data, or results.\",\"writing_profile_keywords\":[\"corrigendum\",\"erratum\",\"affiliation correction\",\"administrative notice\",\"very short\"],\"confidence\":0.95,\"notes\":\"The entire document is a 25-word correction of an author affiliation. Most schema fields are not applicable. The subfield was inferred from the title of the original article. The second paragraph ('E1') appears to be an extraction artifact.\"}", "json": {"genre": "replication_or_correction", "subfield": "market_microstructure", "subfield_secondary": "corrigendum correcting an author affiliation for an article on brokers and order flow leakage", "primary_method": "none", "identification": {"estimand": "not applicable", "strategy": "not applicable", "key_assumptions": [], "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": false, "sentence_index": -1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "other"}], "contribution_paragraph_index": -1, "contribution_types": ["replication_or_correction"], "roadmap_present": false, "literature_positioning_location": "none"}, "section_roles": [{"idx": 0, "role": "other"}], "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.", "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": []}, "claim_evidence": [{"claim": "A co-author's affiliation in the original acknowledgments footnote was wrong, and the corrected affiliation is supplied.", "evidence_type": "none_stated", "evidence_ref": "paragraph 0"}], "structural_summary": "A brief administrative corrigendum. It identifies the page and footnote of the original article where an author's affiliation was misstated, then gives the correct affiliation. It contains no substantive argument, data, or results.", "writing_profile_keywords": ["corrigendum", "erratum", "affiliation correction", "administrative notice", "very short"], "confidence": 0.95, "notes": "The entire document is a 25-word correction of an author affiliation. Most schema fields are not applicable. The subfield was inferred from the title of the original article. The second paragraph ('E1') appears to be an extraction artifact."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 7.54, "usage": {"input": 512, "output": 863, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.020102, "request_id": "req_011Cggqgaq96smJZ3xDBorNf", "cached": false, "ts": "2026-11-03T23:11:35+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/77d2144cdcf32212b09787f35e32d51b2d88149df44c076c432fd56928105240.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; equilibrium analysis of a three-date model with incumbent pledgeability choice, debt, and auctions to experts and outsiders\",\"key_assumptions\":[\"Two periods with industry states G and B; state G for sure in period 2\",\"Experts are rational and never pay more than fundamental value\",\"Incumbent sets pledgeability, which is subject to moral hazard\",\"Debt contracts are fixed and pledgeability effects are hard to contract upon\"],\"threats_addressed\":[\"Robustness to dynamic effects\",\"Alternative assumptions on ability loss\",\"Long-term debt, public equity, and state-contingent contracts\"],\"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\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_theory_or_model\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"other\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"literature\"},{\"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\":[\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"No tables. Results are carried by propositions and lemmas, with proofs in the appendix. Figures depict states of nature, timing, equilibrium functions, and two descriptive time series used to discuss empirical implications.\",\"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\":[\"modal verbs (can, may, could)\",\"seems\",\"explicit acknowledgment that a dynamic extension is needed\",\"consistent-with appeals to prior evidence\"]},\"claim_evidence\":[{\"claim\":\"High anticipated liquidity raises the debt a firm can credibly repay, because wealthy experts bid full value at resale.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A\"},{\"claim\":\"High anticipated liquidity and the resulting high debt reduce the incumbent's incentive to raise pledgeability.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A.2\"},{\"claim\":\"In downturns after booms, firms may be sold to less productive outsiders, who raise pledgeability and later resell to experts.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Eisfeldt and Rampini (2006, 2008)\"},{\"claim\":\"Higher anticipated liquidity can reduce current access to finance.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.B.2, Figure 4\"},{\"claim\":\"Proxies for weak pledgeability, such as weak internal controls and covenant-lite loans, vary over the cycle.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 6 and 7\"}],\"structural_summary\":\"The paper opens with a macro puzzle about protracted downturns after booms and critiques debt-overhang explanations. It then sketches the mechanism by which liquidity, leverage, and pledgeability interact, positions itself against key prior work, and gives a roadmap. Next come the model framework and two motivating examples. Formal solutions are organized by industry case (after prosperity, normal times, after distress). These are followed by a robustness section on alternative contracts and assumptions, and a section deriving empirical implications with suggestive proxies and related literature. The conclusion recaps the cycle and outlines future work.\",\"writing_profile_keywords\":[\"applied theory\",\"financing cycles\",\"pledgeability\",\"verbal mechanism exposition\",\"motivating examples\",\"proofs in appendix\",\"empirical implications section\"],\"confidence\":0.85,\"notes\":\"Theory paper, so empirical-only fields are set to not applicable or none. Figures 6 and 7 are descriptive time series supporting the empirical implications, not tests. Only one equation is counted, probably reflecting parsing limitations.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "macro_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; equilibrium analysis of a three-date model with incumbent pledgeability choice, debt, and auctions to experts and outsiders", "key_assumptions": ["Two periods with industry states G and B; state G for sure in period 2", "Experts are rational and never pay more than fundamental value", "Incumbent sets pledgeability, which is subject to moral hazard", "Debt contracts are fixed and pledgeability effects are hard to contract upon"], "threats_addressed": ["Robustness to dynamic effects", "Alternative assumptions on ability loss", "Long-term debt, public equity, and state-contingent contracts"], "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": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"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": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_theory_or_model"], "roadmap_present": true, "literature_positioning_location": "both"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "other"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "literature"}, {"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": ["sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "No tables. Results are carried by propositions and lemmas, with proofs in the appendix. Figures depict states of nature, timing, equilibrium functions, and two descriptive time series used to discuss empirical implications.", "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": ["modal verbs (can, may, could)", "seems", "explicit acknowledgment that a dynamic extension is needed", "consistent-with appeals to prior evidence"]}, "claim_evidence": [{"claim": "High anticipated liquidity raises the debt a firm can credibly repay, because wealthy experts bid full value at resale.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A"}, {"claim": "High anticipated liquidity and the resulting high debt reduce the incumbent's incentive to raise pledgeability.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A.2"}, {"claim": "In downturns after booms, firms may be sold to less productive outsiders, who raise pledgeability and later resell to experts.", "evidence_type": "prior_literature", "evidence_ref": "Eisfeldt and Rampini (2006, 2008)"}, {"claim": "Higher anticipated liquidity can reduce current access to finance.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.B.2, Figure 4"}, {"claim": "Proxies for weak pledgeability, such as weak internal controls and covenant-lite loans, vary over the cycle.", "evidence_type": "figure", "evidence_ref": "Figures 6 and 7"}], "structural_summary": "The paper opens with a macro puzzle about protracted downturns after booms and critiques debt-overhang explanations. It then sketches the mechanism by which liquidity, leverage, and pledgeability interact, positions itself against key prior work, and gives a roadmap. Next come the model framework and two motivating examples. Formal solutions are organized by industry case (after prosperity, normal times, after distress). These are followed by a robustness section on alternative contracts and assumptions, and a section deriving empirical implications with suggestive proxies and related literature. The conclusion recaps the cycle and outlines future work.", "writing_profile_keywords": ["applied theory", "financing cycles", "pledgeability", "verbal mechanism exposition", "motivating examples", "proofs in appendix", "empirical implications section"], "confidence": 0.85, "notes": "Theory paper, so empirical-only fields are set to not applicable or none. Figures 6 and 7 are descriptive time series supporting the empirical implications, not tests. Only one equation is counted, probably reflecting parsing limitations."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.66, "usage": {"input": 6198, "output": 1997, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065526, "request_id": "req_011CggqeGKrAH8p6avJDdciw", "cached": false, "ts": "2026-11-03T23:11:12+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/783a341cf7ecd522f0cd48a9db3e1a500a117b42347e800549d0c85097ecfcac.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"career concerns / delegated investment; IPO market information frictions\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; the model compares a profit-motivated VC equilibrium with a reputation-motivated VC equilibrium\",\"key_assumptions\":[\"VCs care about reputation beyond profits because fixed fees depend on capital flows\",\"VC skill is the ability to identify good firms (being informed)\",\"Asymmetric information between VC-backed firms and IPO bidders\"],\"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\":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\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"other\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":2,\"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\":\"robustness\"},{\"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\":\"main_text_section\",\"types\":[\"alternative_specification\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"No tables. One figure shows the timeline of moves. Results are stated as lemmas, propositions and corollaries, with all 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 'can' for effects\",\"'arguably' when justifying assumptions\",\"conditional welfare statements (depends on abundance or profitability of good firms)\"]},\"claim_evidence\":[{\"claim\":\"Uninformed reputation-motivated VCs are biased against backing firms so as to avoid revealing their lack of information\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.C; Appendix A propositions\"},{\"claim\":\"Reputation motivation creates a certification effect that can prevent IPO market breakdowns\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A; Appendix A\"},{\"claim\":\"Reputation motivation lowers welfare when good firms are abundant or profitable\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.B; Appendix A corollaries\"},{\"claim\":\"The VC's equilibrium profits can exceed those of a purely profit-maximizing VC\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III\"},{\"claim\":\"VC reputation motivation stems from fixed fees on attracted capital, and picking good firms is central to VC skill\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction subsection A; Section I.D; Nelson et al. (2004), Gompers et al. (2006)\"},{\"claim\":\"Results survive extensions such as assets in place, IPO discount, nonlinear flow-performance and partially informed VCs\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"}],\"structural_summary\":\"Theory paper. The introduction contrasts the known harms of career concerns in secondary markets with the unexplored primary market, poses the question for VCs, and previews a counterintuitive result and its mechanism. It then discusses the realism of the assumptions and empirical proxies, and positions the paper against the career-concerns literature. The body sets up the model, characterizes the equilibria under profit and reputation motivation, compares their costs and benefits, and runs extensions as robustness checks. A short conclusion restates the main finding, and the proofs are collected in an appendix.\",\"writing_profile_keywords\":[\"career concerns\",\"certification effect\",\"counterintuitive result\",\"equilibrium comparison\",\"proofs in appendix\",\"extensions as robustness\",\"empirical proxies discussion\"],\"confidence\":0.85,\"notes\":\"The final abstract sentence fragment is truncated introduction text and was ignored. No roadmap paragraph is visible in the excerpts.\"}", "json": {"genre": "theoretical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "career concerns / delegated investment; IPO market information frictions", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; the model compares a profit-motivated VC equilibrium with a reputation-motivated VC equilibrium", "key_assumptions": ["VCs care about reputation beyond profits because fixed fees depend on capital flows", "VC skill is the ability to identify good firms (being informed)", "Asymmetric information between VC-backed firms and IPO bidders"], "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": 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": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "other"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "related_literature_positioning"}], "contribution_paragraph_index": 2, "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": "robustness"}, {"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": "main_text_section", "types": ["alternative_specification", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "No tables. One figure shows the timeline of moves. Results are stated as lemmas, propositions and corollaries, with all 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 'can' for effects", "'arguably' when justifying assumptions", "conditional welfare statements (depends on abundance or profitability of good firms)"]}, "claim_evidence": [{"claim": "Uninformed reputation-motivated VCs are biased against backing firms so as to avoid revealing their lack of information", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.C; Appendix A propositions"}, {"claim": "Reputation motivation creates a certification effect that can prevent IPO market breakdowns", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A; Appendix A"}, {"claim": "Reputation motivation lowers welfare when good firms are abundant or profitable", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.B; Appendix A corollaries"}, {"claim": "The VC's equilibrium profits can exceed those of a purely profit-maximizing VC", "evidence_type": "equation_or_proof", "evidence_ref": "Section III"}, {"claim": "VC reputation motivation stems from fixed fees on attracted capital, and picking good firms is central to VC skill", "evidence_type": "prior_literature", "evidence_ref": "Introduction subsection A; Section I.D; Nelson et al. (2004), Gompers et al. (2006)"}, {"claim": "Results survive extensions such as assets in place, IPO discount, nonlinear flow-performance and partially informed VCs", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV"}], "structural_summary": "Theory paper. The introduction contrasts the known harms of career concerns in secondary markets with the unexplored primary market, poses the question for VCs, and previews a counterintuitive result and its mechanism. It then discusses the realism of the assumptions and empirical proxies, and positions the paper against the career-concerns literature. The body sets up the model, characterizes the equilibria under profit and reputation motivation, compares their costs and benefits, and runs extensions as robustness checks. A short conclusion restates the main finding, and the proofs are collected in an appendix.", "writing_profile_keywords": ["career concerns", "certification effect", "counterintuitive result", "equilibrium comparison", "proofs in appendix", "extensions as robustness", "empirical proxies discussion"], "confidence": 0.85, "notes": "The final abstract sentence fragment is truncated introduction text and was ignored. No roadmap paragraph is visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.39, "usage": {"input": 4106, "output": 2370, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064618, "request_id": "req_011CggqLwaX1ACqpRRCHpaDU", "cached": false, "ts": "2026-11-03T23:07:20+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7a002050bd1e11dfc3aaa21fa1d2b4dcd0076eef8caa0ea7f936220ebad2aa63.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"survey_or_address\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"dynamic capital structure theory; collateral and commitment\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical address)\",\"strategy\":\"Stylized dynamic trade-off model comparing equilibrium leverage policies and shareholder value under full commitment, no commitment, and collateral-based commitment\",\"key_assumptions\":[\"firm cannot commit to future leverage choices\",\"unsecured creditors' claims are nonexclusive\",\"net funding cost advantage of debt (tax shields or pricing differentials)\",\"frequent, costless trade in securities\",\"equity value maximized at all times\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"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},\"notes\":\"no explicit method sentence\"},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"motivation\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":20,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_theory_or_model\",\"synthesis\"],\"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\":\"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 shown through equations and numerical model figures whose captions state parameter assumptions.\",\"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\":[\"explicit disclaimer that the goal is reframing, not a definitive theory\",\"modal verbs (may, likely, could)\",\"acknowledging simplified/stylized setting\",\"'of course' concessions about alternative mechanisms\"]},\"claim_evidence\":[{\"claim\":\"Without commitment and with frequent trade, leverage policy is indeterminate and does not raise shareholder value, an MM-like irrelevance result\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Figures 4-5\"},{\"claim\":\"Under full commitment the firm would rely almost entirely on debt, which counterfactually fits the data\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.A; Figure 2\"},{\"claim\":\"Secured debt against collateral lets the firm capture the funding cost advantage of leverage\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Figure 8\"},{\"claim\":\"Three qualitatively different leverage policies are all equilibrium-optimal and give the same initial equity value\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Collateral reduces debt overhang and expands the range of shareholder-beneficial investment\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"A regulatory debt cap can induce firms to borrow up to the cap because it acts as external commitment\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9; Section III.B\"},{\"claim\":\"Collateral's role in shielding creditors from dilution builds on prior work on collateral and nonexclusivity\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Donaldson, Gromb, and Piacentino (2017, 2018)\"}],\"structural_summary\":\"Presidential address that recasts dynamic capital structure as a commitment problem. The introduction opens with the unresolved empirical puzzle of leverage dynamics and sets out the thesis. It positions the paper against the collateral literature, lists empirical challenges, and previews key results through a single illustrative figure and bulleted findings, ending with a roadmap. Section I builds a stylized model of cash-flow-based borrowing, moving from deterministic to stochastic cases, and shows that leverage is irrelevant without commitment. Section II adds investment and collateral and shows that secured debt restores the benefits of leverage. Section III discusses alternative commitment mechanisms and regulatory caps, then sets out implications for theory, empirics and policy.\",\"writing_profile_keywords\":[\"presidential address\",\"first-person voice\",\"conceptual reframing\",\"stylized model\",\"numerical illustration figures\",\"no empirical tests\",\"bulleted result summaries\",\"policy implications\",\"research agenda\"],\"confidence\":0.85,\"notes\":\"Genre set to survey_or_address because the article is a presidential address. Its content is a theoretical model that relies on related work for general results. Introduction paragraph indices after 8 are fragmented bullet lines, and the roadmap sits in paragraph 20.\"}", "json": {"genre": "survey_or_address", "subfield": "corporate_finance", "subfield_secondary": "dynamic capital structure theory; collateral and commitment", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical address)", "strategy": "Stylized dynamic trade-off model comparing equilibrium leverage policies and shareholder value under full commitment, no commitment, and collateral-based commitment", "key_assumptions": ["firm cannot commit to future leverage choices", "unsecured creditors' claims are nonexclusive", "net funding cost advantage of debt (tax shields or pricing differentials)", "frequent, costless trade in securities", "equity value maximized at all times"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "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}, "notes": "no explicit method sentence"}, "introduction": {"moves": [{"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"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": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "motivation"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 20, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_theory_or_model", "synthesis"], "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": "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 shown through equations and numerical model figures whose captions state parameter assumptions.", "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": ["explicit disclaimer that the goal is reframing, not a definitive theory", "modal verbs (may, likely, could)", "acknowledging simplified/stylized setting", "'of course' concessions about alternative mechanisms"]}, "claim_evidence": [{"claim": "Without commitment and with frequent trade, leverage policy is indeterminate and does not raise shareholder value, an MM-like irrelevance result", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Figures 4-5"}, {"claim": "Under full commitment the firm would rely almost entirely on debt, which counterfactually fits the data", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.A; Figure 2"}, {"claim": "Secured debt against collateral lets the firm capture the funding cost advantage of leverage", "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Figure 8"}, {"claim": "Three qualitatively different leverage policies are all equilibrium-optimal and give the same initial equity value", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Collateral reduces debt overhang and expands the range of shareholder-beneficial investment", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "A regulatory debt cap can induce firms to borrow up to the cap because it acts as external commitment", "evidence_type": "figure", "evidence_ref": "Figure 9; Section III.B"}, {"claim": "Collateral's role in shielding creditors from dilution builds on prior work on collateral and nonexclusivity", "evidence_type": "prior_literature", "evidence_ref": "Donaldson, Gromb, and Piacentino (2017, 2018)"}], "structural_summary": "Presidential address that recasts dynamic capital structure as a commitment problem. The introduction opens with the unresolved empirical puzzle of leverage dynamics and sets out the thesis. It positions the paper against the collateral literature, lists empirical challenges, and previews key results through a single illustrative figure and bulleted findings, ending with a roadmap. Section I builds a stylized model of cash-flow-based borrowing, moving from deterministic to stochastic cases, and shows that leverage is irrelevant without commitment. Section II adds investment and collateral and shows that secured debt restores the benefits of leverage. Section III discusses alternative commitment mechanisms and regulatory caps, then sets out implications for theory, empirics and policy.", "writing_profile_keywords": ["presidential address", "first-person voice", "conceptual reframing", "stylized model", "numerical illustration figures", "no empirical tests", "bulleted result summaries", "policy implications", "research agenda"], "confidence": 0.85, "notes": "Genre set to survey_or_address because the article is a presidential address. Its content is a theoretical model that relies on related work for general results. Introduction paragraph indices after 8 are fragmented bullet lines, and the roadmap sits in paragraph 20."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.85, "usage": {"input": 8085, "output": 2145, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076034, "request_id": "req_011CggqPEHuN4Fnve9xYkmT2", "cached": false, "ts": "2026-11-03T23:07:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7a21a8e4b0bcc69196024989ad795d1bd784a26421878fe1f1971404ab839fa3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"not applicable (theory paper; comparative statics of equilibrium prices, returns, welfare with respect to proportional trading fees)\",\"strategy\":\"not applicable; dynamic multiagent general-equilibrium model with incomplete markets solved numerically via a novel synchronizing algorithm, with illustrative calibrated examples\",\"key_assumptions\":[\"finite-horizon discrete-time exchange economy on a tree/lattice\",\"incomplete markets with unhedgeable endowments giving an every-period trading motive\",\"proportional trading fees on the stock, liquid bond\",\"time-additive utility in illustrations\"],\"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\":3},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"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\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_theory_or_model\",\"new_method\"],\"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\":\"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\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Few tables; notes describe model quantities normalized by aggregate output, simulation path counts and data sources for calibration moments; no standard errors or significance markers.\",\"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\":\"moderate\",\"typical_devices\":[\"framing numerical results as illustrative rather than calibrated to real data\",\"'in calibrated examples' qualifiers\",\"'so far' framing when comparing competing models\",\"explicit statement of model limitations in appendix\"]},\"claim_evidence\":[{\"claim\":\"Trading fees raise both bond and stock prices via a precautionary-savings effect from higher individual consumption volatility\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 4 and 5\"},{\"claim\":\"Interest rate falls while equity premium, return volatility and Sharpe ratio rise with fees\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"Expected liquidity and liquidity risk generate deviations from the classic CCAPM, with time variation driven mainly by expected liquidity\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 8\"},{\"claim\":\"Adding a third trader does not shrink deviations from frictionless equilibrium; many trades are bilateral\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9; Table II\"},{\"claim\":\"Slow-moving capital under fees generates slow price reversal, similar to inattention models\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 10\"},{\"claim\":\"Analytical bounds and comparisons of prices with and without fees\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Appendix C (Proposition 2)\"}],\"structural_summary\":\"Theory-driven paper. The introduction motivates trading fees as a financial-sector service and poses three questions: mechanisms, pricing of liquidity risk, and slow price reversal. It previews the model and its results paragraph by paragraph, claims a methodological contribution (a Walrasian equilibrium concept and a synchronizing algorithm), positions the work against earlier transaction-cost equilibrium papers and dual methods, and ends with a roadmap. The body builds a general model, solves a frictionless benchmark, adds fees with analytical price bounds and the algorithm, then reports numerical illustrations across fee levels. Three applications follow (a CCAPM decomposition, three traders, impulse responses). The conclusion restates the contributions, while appendices hold the setup, FOCs, proof and algorithm details.\",\"writing_profile_keywords\":[\"general equilibrium\",\"trading fees\",\"numerical algorithm\",\"illustrative calibration\",\"comparative statics across fee levels\",\"figure-heavy results\",\"liquidity premia\",\"methodological contribution\"],\"confidence\":0.8,\"notes\":\"Classified as theoretical with calibration/simulation as the primary method; the empirical-identification fields are not applicable. Robustness consists mainly of varying fee levels and a three-trader extension. The 'question' move is implicit rather than stated in the abstract.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure", "primary_method": "calibration_simulation", "identification": {"estimand": "not applicable (theory paper; comparative statics of equilibrium prices, returns, welfare with respect to proportional trading fees)", "strategy": "not applicable; dynamic multiagent general-equilibrium model with incomplete markets solved numerically via a novel synchronizing algorithm, with illustrative calibrated examples", "key_assumptions": ["finite-horizon discrete-time exchange economy on a tree/lattice", "incomplete markets with unhedgeable endowments giving an every-period trading motive", "proportional trading fees on the stock, liquid bond", "time-additive utility in illustrations"], "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": 3}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"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": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_theory_or_model", "new_method"], "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": "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"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "Few tables; notes describe model quantities normalized by aggregate output, simulation path counts and data sources for calibration moments; no standard errors or significance markers.", "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": "moderate", "typical_devices": ["framing numerical results as illustrative rather than calibrated to real data", "'in calibrated examples' qualifiers", "'so far' framing when comparing competing models", "explicit statement of model limitations in appendix"]}, "claim_evidence": [{"claim": "Trading fees raise both bond and stock prices via a precautionary-savings effect from higher individual consumption volatility", "evidence_type": "figure", "evidence_ref": "Figures 4 and 5"}, {"claim": "Interest rate falls while equity premium, return volatility and Sharpe ratio rise with fees", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "Expected liquidity and liquidity risk generate deviations from the classic CCAPM, with time variation driven mainly by expected liquidity", "evidence_type": "figure", "evidence_ref": "Figure 8"}, {"claim": "Adding a third trader does not shrink deviations from frictionless equilibrium; many trades are bilateral", "evidence_type": "figure", "evidence_ref": "Figure 9; Table II"}, {"claim": "Slow-moving capital under fees generates slow price reversal, similar to inattention models", "evidence_type": "figure", "evidence_ref": "Figure 10"}, {"claim": "Analytical bounds and comparisons of prices with and without fees", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Appendix C (Proposition 2)"}], "structural_summary": "Theory-driven paper. The introduction motivates trading fees as a financial-sector service and poses three questions: mechanisms, pricing of liquidity risk, and slow price reversal. It previews the model and its results paragraph by paragraph, claims a methodological contribution (a Walrasian equilibrium concept and a synchronizing algorithm), positions the work against earlier transaction-cost equilibrium papers and dual methods, and ends with a roadmap. The body builds a general model, solves a frictionless benchmark, adds fees with analytical price bounds and the algorithm, then reports numerical illustrations across fee levels. Three applications follow (a CCAPM decomposition, three traders, impulse responses). The conclusion restates the contributions, while appendices hold the setup, FOCs, proof and algorithm details.", "writing_profile_keywords": ["general equilibrium", "trading fees", "numerical algorithm", "illustrative calibration", "comparative statics across fee levels", "figure-heavy results", "liquidity premia", "methodological contribution"], "confidence": 0.8, "notes": "Classified as theoretical with calibration/simulation as the primary method; the empirical-identification fields are not applicable. Robustness consists mainly of varying fee levels and a three-trader extension. The 'question' move is implicit rather than stated in the abstract."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 23.6, "usage": {"input": 10054, "output": 2796, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.09693, "request_id": "req_011CggqHRA69YfTqZ2qWpPsC", "cached": false, "ts": "2026-11-03T23:06:36+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7c19348481ee2de19f010bd4d19e7e63916952916535329b2c3409b0c6b13f18.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"methodological_econometric\",\"subfield\":\"financial_econometrics\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"machine_learning_textual\",\"identification\":{\"estimand\":\"SDF loading of a candidate factor, conditional on a selected subset of a high-dimensional set of existing control factors\",\"strategy\":\"Double-selection LASSO (two selection stages over control factors) followed by a two-pass cross-sectional regression and asymptotic inference from a derived central limit result\",\"key_assumptions\":[\"true asset pricing model is approximately low-dimensional (sparsity)\",\"linear SDF specification\",\"large cross-section and time-series asymptotics\"],\"threats_addressed\":[\"omitted variable bias from model selection mistakes\",\"regularization bias from LASSO\",\"tuning parameter sensitivity\",\"choice of test assets\"],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"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\":\"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_approach\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"paragraph_index\":8,\"move\":\"other\"},{\"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\":\"contribution_statement\"},{\"paragraph_index\":11,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_method\",\"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\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"sensitivity_to_parameters\",\"alternative_measure\",\"alternative_specification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Table notes spell out the control set (135 factors through 2011), the test assets (750 bivariate-sorted portfolios), and the sample period. Recursive tests report test-asset counts year by year. The robustness table repeats baseline column one for comparison.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\",\"figures\":null},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"describing test as conservative\",\"acknowledging inevitable LASSO selection mistakes\",\"approximate sparsity assumption\",\"robustness to tuning parameters\"]},\"claim_evidence\":[{\"claim\":\"Several factors introduced 2012-2016, notably profitability, add significant pricing power beyond earlier factors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Recursive application would have deemed only a few factors significant at introduction\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"SDF loading estimates are robust to tuning parameters even though the selected models vary\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-3\"},{\"claim\":\"Results are robust to alternative portfolios and to dimension-reduction methods such as Elastic Net and PCA\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Estimator performs well in finite samples and outperforms alternatives\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Internet Appendix simulations\"},{\"claim\":\"Valid asymptotic inference on the SDF loading despite selection mistakes\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.F; Internet Appendix\"}],\"structural_summary\":\"The paper motivates itself with the factor zoo problem and poses an open question. It then introduces a double-selection LASSO two-pass regression method that targets SDF loadings. The introduction previews four empirical exercises and emphasizes robustness to tuning parameters. The method is positioned against alpha-based tests and the LASSO post-selection literature. A methodology section derives inference, and an empirical section applies it to recent and recursively tested factors, with robustness subsections. A short conclusion restates the contribution and findings.\",\"writing_profile_keywords\":[\"methodological\",\"high-dimensional\",\"model selection\",\"LASSO\",\"SDF loadings\",\"factor zoo\",\"conservative test\",\"robustness to tuning\"],\"confidence\":0.8,\"notes\":\"Primary method has no exact category; machine_learning_textual was chosen to reflect LASSO regularization, though the paper is econometric and involves no text analysis. Figures 2-3 report t-statistic heat maps, so uncertainty is shown only partially. Monte Carlo evidence and proofs are in the Internet Appendix.\"}", "json": {"genre": "methodological_econometric", "subfield": "financial_econometrics", "subfield_secondary": "asset_pricing", "primary_method": "machine_learning_textual", "identification": {"estimand": "SDF loading of a candidate factor, conditional on a selected subset of a high-dimensional set of existing control factors", "strategy": "Double-selection LASSO (two selection stages over control factors) followed by a two-pass cross-sectional regression and asymptotic inference from a derived central limit result", "key_assumptions": ["true asset pricing model is approximately low-dimensional (sparsity)", "linear SDF specification", "large cross-section and time-series asymptotics"], "threats_addressed": ["omitted variable bias from model selection mistakes", "regularization bias from LASSO", "tuning parameter sensitivity", "choice of test assets"], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "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": "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_approach"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"paragraph_index": 8, "move": "other"}, {"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": "contribution_statement"}, {"paragraph_index": 11, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_method", "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": "conclusion"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["sensitivity_to_parameters", "alternative_measure", "alternative_specification", "subsample"]}, "tables": {"convention_notes": "Table notes spell out the control set (135 factors through 2011), the test assets (750 bivariate-sorted portfolios), and the sample period. Recursive tests report test-asset counts year by year. The robustness table repeats baseline column one for comparison.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed", "figures": null}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["describing test as conservative", "acknowledging inevitable LASSO selection mistakes", "approximate sparsity assumption", "robustness to tuning parameters"]}, "claim_evidence": [{"claim": "Several factors introduced 2012-2016, notably profitability, add significant pricing power beyond earlier factors", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Recursive application would have deemed only a few factors significant at introduction", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "SDF loading estimates are robust to tuning parameters even though the selected models vary", "evidence_type": "figure", "evidence_ref": "Figures 2-3"}, {"claim": "Results are robust to alternative portfolios and to dimension-reduction methods such as Elastic Net and PCA", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Estimator performs well in finite samples and outperforms alternatives", "evidence_type": "none_stated", "evidence_ref": "Internet Appendix simulations"}, {"claim": "Valid asymptotic inference on the SDF loading despite selection mistakes", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.F; Internet Appendix"}], "structural_summary": "The paper motivates itself with the factor zoo problem and poses an open question. It then introduces a double-selection LASSO two-pass regression method that targets SDF loadings. The introduction previews four empirical exercises and emphasizes robustness to tuning parameters. The method is positioned against alpha-based tests and the LASSO post-selection literature. A methodology section derives inference, and an empirical section applies it to recent and recursively tested factors, with robustness subsections. A short conclusion restates the contribution and findings.", "writing_profile_keywords": ["methodological", "high-dimensional", "model selection", "LASSO", "SDF loadings", "factor zoo", "conservative test", "robustness to tuning"], "confidence": 0.8, "notes": "Primary method has no exact category; machine_learning_textual was chosen to reflect LASSO regularization, though the paper is econometric and involves no text analysis. Figures 2-3 report t-statistic heat maps, so uncertainty is shown only partially. Monte Carlo evidence and proofs are in the Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.36, "usage": {"input": 6854, "output": 2079, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06979, "request_id": "req_011CggqnYk3fjFve8PEGzYkU", "cached": false, "ts": "2026-11-03T23:13:05+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/7e4b15d719f74f91511de4987b5567dcf374804b0cb684d38fe14449db6c6be4.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Incremental share of variation in corporate bond yield spread changes, and in their common unexplained component, accounted for by systematic OTC friction proxies\",\"strategy\":\"Bond-by-bond time-series regressions extending the CDGM structural-variable model with marketwide proxies for inventory, search and bargaining frictions; principal component analysis of residuals to assess the common factor\",\"key_assumptions\":[\"Constructed proxies capture the intensity of systematic inventory, search and bargaining frictions\",\"CDGM structural variables adequately control for fundamental drivers\"],\"threats_addressed\":[\"stale dealer quotes, handled by using transaction prices\",\"crisis-period dominance, handled by excluding the crisis\",\"other potential drivers of yield spread changes\",\"alternative definitions of CDGM variables\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":0},\"method\":{\"present\":true,\"sentence_index\":1},\"results\":{\"present\":true,\"sentence_index\":1},\"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\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"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\":\"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\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"data\"},{\"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\":[\"alternative_sample_period\",\"alternative_specification\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Notes spell out the regression equation and its variable vectors, the sample filter (bonds with at least 25 monthly observations) and the data source and period. Results come from bond-level regressions summarized across bonds.\",\"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' theory\",\"'suggests'/'indicates'\",\"'associated with'\",\"'support the implications of'\"]},\"claim_evidence\":[{\"claim\":\"Systematic inventory, search and bargaining frictions explain 23.4% of the variation in the common component of yield spread changes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"The CDGM model on transaction prices leaves a large common unexplained factor\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III\"},{\"claim\":\"Higher interdealer network connectivity and longer intermediation chains are associated with narrower yield spreads\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Search frictions capture 6.3% of the common component\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Results hold when the crisis period is excluded\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"The paper opens with the puzzle of a large unexplained common factor in yield spread changes and introduces OTC friction theory as a candidate explanation. It then previews the friction proxies and results, theory by theory, and positions itself against the liquidity and market-making literatures. The body describes the TRACE data, replicates the CDGM benchmark, and constructs systematic friction measures. It then tests each friction group separately and jointly, followed by robustness checks. A brief conclusion restates the key magnitudes.\",\"writing_profile_keywords\":[\"puzzle-driven\",\"benchmark replication\",\"theory-guided proxies\",\"explained variance\",\"OTC frictions\",\"network measures\"],\"confidence\":0.8,\"notes\":\"Sections III and IV are labelled 'other' by the role rule; Section III is mainly measure construction, so it is coded as data. Standard error conventions are not visible in the table notes.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure", "primary_method": "reduced_form_regression", "identification": {"estimand": "Incremental share of variation in corporate bond yield spread changes, and in their common unexplained component, accounted for by systematic OTC friction proxies", "strategy": "Bond-by-bond time-series regressions extending the CDGM structural-variable model with marketwide proxies for inventory, search and bargaining frictions; principal component analysis of residuals to assess the common factor", "key_assumptions": ["Constructed proxies capture the intensity of systematic inventory, search and bargaining frictions", "CDGM structural variables adequately control for fundamental drivers"], "threats_addressed": ["stale dealer quotes, handled by using transaction prices", "crisis-period dominance, handled by excluding the crisis", "other potential drivers of yield spread changes", "alternative definitions of CDGM variables"], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 0}, "method": {"present": true, "sentence_index": 1}, "results": {"present": true, "sentence_index": 1}, "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": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"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": "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": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "data"}, {"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": ["alternative_sample_period", "alternative_specification", "alternative_measure"]}, "tables": {"convention_notes": "Notes spell out the regression equation and its variable vectors, the sample filter (bonds with at least 25 monthly observations) and the data source and period. Results come from bond-level regressions summarized across bonds.", "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' theory", "'suggests'/'indicates'", "'associated with'", "'support the implications of'"]}, "claim_evidence": [{"claim": "Systematic inventory, search and bargaining frictions explain 23.4% of the variation in the common component of yield spread changes", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "The CDGM model on transaction prices leaves a large common unexplained factor", "evidence_type": "table", "evidence_ref": "Tables II, III"}, {"claim": "Higher interdealer network connectivity and longer intermediation chains are associated with narrower yield spreads", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Search frictions capture 6.3% of the common component", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Results hold when the crisis period is excluded", "evidence_type": "table", "evidence_ref": "Table X"}], "structural_summary": "The paper opens with the puzzle of a large unexplained common factor in yield spread changes and introduces OTC friction theory as a candidate explanation. It then previews the friction proxies and results, theory by theory, and positions itself against the liquidity and market-making literatures. The body describes the TRACE data, replicates the CDGM benchmark, and constructs systematic friction measures. It then tests each friction group separately and jointly, followed by robustness checks. A brief conclusion restates the key magnitudes.", "writing_profile_keywords": ["puzzle-driven", "benchmark replication", "theory-guided proxies", "explained variance", "OTC frictions", "network measures"], "confidence": 0.8, "notes": "Sections III and IV are labelled 'other' by the role rule; Section III is mainly measure construction, so it is coded as data. Standard error conventions are not visible in the table notes."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.15, "usage": {"input": 7427, "output": 1849, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067482, "request_id": "req_011CggqbdrpRSgxehFaNq2LC", "cached": false, "ts": "2026-11-03T23:10:35+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/812721c9d8310b09695447caf1d5f05c5f3fc031019d8a22d66a57c7e8f334d5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"economic history / derivatives market structure\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Causal effect of introducing central clearing (CLAM, 1882) on the share of coffee trade flows passing through Le Havre, relative to other harbors and control commodities\",\"strategy\":\"Triple difference-in-differences across harbor, commodity (coffee vs. non-cleared colonial commodities) and time around the 1882 CCP creation, extended to Europe-level import shares. Mechanism is inspected with trader-level data and narrative evidence.\",\"key_assumptions\":[\"Parallel trends in trade shares between Le Havre coffee and control harbors/commodities absent the CCP\",\"Timing of CLAM creation was not driven by anticipated trade-flow changes\",\"Control commodities are unaffected by central clearing\"],\"threats_addressed\":[\"Pre-trends in Le Havre trade flows\",\"Anticipation by traders, argued against through the contemporaneous debate\",\"Alternative mechanisms, discussed in a dedicated subsection\",\"Generalizability, checked with other commodities\"],\"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\":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\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":1,\"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\":\"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_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes cite the estimating equation number, the sample (e.g., 23 French harbors or country-level samples) and the dependent variable for each column. Descriptive share tables average over the pre-event window 1877–1882. The trader pool table uses two-sample pre/post t-tests.\",\"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\":[\"'arguably' for the exogeneity claim\",\"'consistent with' framing for mechanism evidence\",\"explicit caveats on external validity to modern mandatory clearing\"]},\"claim_evidence\":[{\"claim\":\"Central clearing shifted French trade flows in coffee toward Le Havre\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 1\"},{\"claim\":\"The effect extends to Europe-wide import shares from France\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Figure 2\"},{\"claim\":\"Futures volume rose sharply after CLAM creation while the low-quality trader share stayed low, indicating a missing market\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Figure 6, Table VII\"},{\"claim\":\"New trader share and entry rose and turnover fell after clearing, consistent with mitigated adverse selection\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5 and 7, Table VII\"},{\"claim\":\"Results also hold for other commodities\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"}],\"structural_summary\":\"The introduction frames contracting innovations as remedies for financial frictions and notes that identifying them is hard. It then presents the 1882 Le Havre CCP as a natural experiment and states two contributions: a causal effect on trade flows and a mechanism. A short theory section yields the hypothesis, followed by historical background and data. The main results use triple-DiD at the French and European levels, supplemented by narrative and consumption evidence. A long mechanism section separates missing-market and adverse-selection channels and alternatives. The conclusion caveats lessons for modern mandatory clearing.\",\"writing_profile_keywords\":[\"economic history\",\"natural experiment\",\"triple difference-in-differences\",\"hand-collected archival data\",\"mechanism decomposition\",\"derivatives clearing\",\"single-author first person\"],\"confidence\":0.8,\"notes\":\"The role_rule labels for sections 2 (literature) and 4 (robustness) were overridden: the excerpts show these sections are background/data and mechanism results. Standard error type and significance notation are not visible in the excerpts. Quantitative statements in the introduction include the 8.2% low-quality trader share and the 303% rise in futures volume.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "economic history / derivatives market structure", "primary_method": "difference_in_differences", "identification": {"estimand": "Causal effect of introducing central clearing (CLAM, 1882) on the share of coffee trade flows passing through Le Havre, relative to other harbors and control commodities", "strategy": "Triple difference-in-differences across harbor, commodity (coffee vs. non-cleared colonial commodities) and time around the 1882 CCP creation, extended to Europe-level import shares. Mechanism is inspected with trader-level data and narrative evidence.", "key_assumptions": ["Parallel trends in trade shares between Le Havre coffee and control harbors/commodities absent the CCP", "Timing of CLAM creation was not driven by anticipated trade-flow changes", "Control commodities are unaffected by central clearing"], "threats_addressed": ["Pre-trends in Le Havre trade flows", "Anticipation by traders, argued against through the contemporaneous debate", "Alternative mechanisms, discussed in a dedicated subsection", "Generalizability, checked with other commodities"], "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": 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": "preview_of_approach"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}], "contribution_paragraph_index": 1, "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": "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_measure", "other"]}, "tables": {"convention_notes": "Notes cite the estimating equation number, the sample (e.g., 23 French harbors or country-level samples) and the dependent variable for each column. Descriptive share tables average over the pre-event window 1877–1882. The trader pool table uses two-sample pre/post t-tests.", "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": ["'arguably' for the exogeneity claim", "'consistent with' framing for mechanism evidence", "explicit caveats on external validity to modern mandatory clearing"]}, "claim_evidence": [{"claim": "Central clearing shifted French trade flows in coffee toward Le Havre", "evidence_type": "table", "evidence_ref": "Table II, Figure 1"}, {"claim": "The effect extends to Europe-wide import shares from France", "evidence_type": "table", "evidence_ref": "Table V, Figure 2"}, {"claim": "Futures volume rose sharply after CLAM creation while the low-quality trader share stayed low, indicating a missing market", "evidence_type": "descriptive_statistic", "evidence_ref": "Figure 6, Table VII"}, {"claim": "New trader share and entry rose and turnover fell after clearing, consistent with mitigated adverse selection", "evidence_type": "figure", "evidence_ref": "Figures 5 and 7, Table VII"}, {"claim": "Results also hold for other commodities", "evidence_type": "table", "evidence_ref": "Table III"}], "structural_summary": "The introduction frames contracting innovations as remedies for financial frictions and notes that identifying them is hard. It then presents the 1882 Le Havre CCP as a natural experiment and states two contributions: a causal effect on trade flows and a mechanism. A short theory section yields the hypothesis, followed by historical background and data. The main results use triple-DiD at the French and European levels, supplemented by narrative and consumption evidence. A long mechanism section separates missing-market and adverse-selection channels and alternatives. The conclusion caveats lessons for modern mandatory clearing.", "writing_profile_keywords": ["economic history", "natural experiment", "triple difference-in-differences", "hand-collected archival data", "mechanism decomposition", "derivatives clearing", "single-author first person"], "confidence": 0.8, "notes": "The role_rule labels for sections 2 (literature) and 4 (robustness) were overridden: the excerpts show these sections are background/data and mechanism results. Standard error type and significance notation are not visible in the excerpts. Quantitative statements in the introduction include the 8.2% low-quality trader share and the 303% rise in futures volume."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.33, "usage": {"input": 6106, "output": 1992, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065058, "request_id": "req_011Cggqsme64xZsf5Pic2fBQ", "cached": false, "ts": "2026-11-03T23:14:15+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/82bf9ddad90a66cc44ee28e552d11876101b8ef17883a6a1a05903567309bf71.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Predictive relation between banks' aggregate income gap (net interest rate risk exposure) and future excess Treasury bond returns\",\"strategy\":\"Time-series predictive regressions of bond excess returns on the average bank income gap, with macro controls, real-time versions, and yield-based predictors (spanning test), complemented by structural estimation of a term structure model with banks\",\"key_assumptions\":[\"Banks are marginal investors in interest rate risk\",\"Income gap measures banks' net interest rate exposure\",\"Bank risk management links exposure to the price of risk in equilibrium\"],\"threats_addressed\":[\"Information already spanned by the yield curve\",\"Macroeconomic conditions driving both variables\",\"Look-ahead bias (real-time analysis)\",\"Asset or liability components alone driving results\",\"Liquidity risk as alternative explanation\",\"Demand/supply shocks to interest rate 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\":false,\"sentence_index\":-1},\"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\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":10,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":11,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":11,\"move\":\"preview_of_results\"},{\"paragraph_index\":12,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"contribution_statement\"},{\"paragraph_index\":13,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":13,\"move\":\"contribution_statement\"},{\"paragraph_index\":13,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":12,\"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\":\"empirical_design\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Long, largely repeated notes that restate the sample period (1986–2014) and the income gap definition from FR Y-9C filings. Model estimates are reported with bootstrapped 90% CIs. Standard error 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\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"interpret through the lens of\",\"natural interpretation\",\"explicit discussion of challenges to interpretation\"]},\"claim_evidence\":[{\"claim\":\"The average bank income gap strongly forecasts excess Treasury bond returns, more so at longer maturities\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 1\"},{\"claim\":\"Predictive power survives macro controls, real-time construction and yield-based predictors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV, V\"},{\"claim\":\"The model quantitatively fits the data and yields an estimate of banks' risk-taking willingness\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Only the net income gap predicts returns, not its asset or liability components alone\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Liquidity risk measures do not forecast bond premia\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"A one-sd income gap rise is associated with about a 231bp lower one-year five-year bond excess return\",\"evidence_type\":\"table\",\"evidence_ref\":\"Conclusion, Table II\"},{\"claim\":\"Banks' stock return exposure to rate risk, required by the theory, is elusive in the data\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The paper opens with the main predictive finding and its risk-management interpretation. It then builds the income gap measure and runs time-series forecasting regressions with several checks. Next comes a simple equilibrium model of banks bearing interest rate risk, followed by structural estimation of that model. A section of supporting tests is paired with an explicit discussion of challenges to the interpretation. The conclusion restates the magnitude and flags an unresolved puzzle for future work.\",\"writing_profile_keywords\":[\"intermediary asset pricing\",\"bond return predictability\",\"income gap\",\"Euler equation analogy\",\"self-critical interpretation\",\"structural estimation\"],\"confidence\":0.8,\"notes\":\"Introduction paragraphs 1–9 are figure axis fragments. Section IV's supporting tests were mapped to robustness, although they also function as further results. The statistical significance format is not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Predictive relation between banks' aggregate income gap (net interest rate risk exposure) and future excess Treasury bond returns", "strategy": "Time-series predictive regressions of bond excess returns on the average bank income gap, with macro controls, real-time versions, and yield-based predictors (spanning test), complemented by structural estimation of a term structure model with banks", "key_assumptions": ["Banks are marginal investors in interest rate risk", "Income gap measures banks' net interest rate exposure", "Bank risk management links exposure to the price of risk in equilibrium"], "threats_addressed": ["Information already spanned by the yield curve", "Macroeconomic conditions driving both variables", "Look-ahead bias (real-time analysis)", "Asset or liability components alone driving results", "Liquidity risk as alternative explanation", "Demand/supply shocks to interest rate 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": false, "sentence_index": -1}, "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": "preview_of_results"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 10, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 11, "move": "tension_or_puzzle"}, {"paragraph_index": 11, "move": "preview_of_results"}, {"paragraph_index": 12, "move": "related_literature_positioning"}, {"paragraph_index": 12, "move": "contribution_statement"}, {"paragraph_index": 13, "move": "related_literature_positioning"}, {"paragraph_index": 13, "move": "contribution_statement"}, {"paragraph_index": 13, "move": "roadmap"}], "contribution_paragraph_index": 12, "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": "empirical_design"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Long, largely repeated notes that restate the sample period (1986–2014) and the income gap definition from FR Y-9C filings. Model estimates are reported with bootstrapped 90% CIs. Standard error 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": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "interpret through the lens of", "natural interpretation", "explicit discussion of challenges to interpretation"]}, "claim_evidence": [{"claim": "The average bank income gap strongly forecasts excess Treasury bond returns, more so at longer maturities", "evidence_type": "table", "evidence_ref": "Table II, Figure 1"}, {"claim": "Predictive power survives macro controls, real-time construction and yield-based predictors", "evidence_type": "table", "evidence_ref": "Tables III, IV, V"}, {"claim": "The model quantitatively fits the data and yields an estimate of banks' risk-taking willingness", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Only the net income gap predicts returns, not its asset or liability components alone", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Liquidity risk measures do not forecast bond premia", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "A one-sd income gap rise is associated with about a 231bp lower one-year five-year bond excess return", "evidence_type": "table", "evidence_ref": "Conclusion, Table II"}, {"claim": "Banks' stock return exposure to rate risk, required by the theory, is elusive in the data", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The paper opens with the main predictive finding and its risk-management interpretation. It then builds the income gap measure and runs time-series forecasting regressions with several checks. Next comes a simple equilibrium model of banks bearing interest rate risk, followed by structural estimation of that model. A section of supporting tests is paired with an explicit discussion of challenges to the interpretation. The conclusion restates the magnitude and flags an unresolved puzzle for future work.", "writing_profile_keywords": ["intermediary asset pricing", "bond return predictability", "income gap", "Euler equation analogy", "self-critical interpretation", "structural estimation"], "confidence": 0.8, "notes": "Introduction paragraphs 1–9 are figure axis fragments. Section IV's supporting tests were mapped to robustness, although they also function as further results. The statistical significance format is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.08, "usage": {"input": 6785, "output": 2043, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068794, "request_id": "req_011CggqwQPFkSETN4MLwuy6e", "cached": false, "ts": "2026-11-03T23:15:05+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/8309841fa059de2eb2c85989720a04cef0b89ae8bf2db5b56f3a7083c4f042a5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"asset_pricing (option valuation); executive compensation\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Conditional daily rate at which employees voluntarily exercise vested in-the-money options, as a function of option, stock, firm and holder characteristics, and the implied option cost to the firm.\",\"strategy\":\"Reduced-form fractional-logistic exercise function (hazard-like) estimated by GMM that is robust to heteroskedasticity and within-employee correlation. The estimated function is then fed into simulation-based valuation.\",\"key_assumptions\":[\"The estimated exercise function describes future exercise behavior well enough to serve as a valuation input.\",\"Covariates capture the relevant drivers of exercise.\",\"Employee termination and cancellation hazards are modeled separately.\"],\"threats_addressed\":[\"Correlation across exercises by the same employee\",\"Heteroskedasticity\",\"Omitted-variable bias from using only standard exercise factors (shown via specifications 1-2)\"],\"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\":\"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\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_method\",\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"separate_section\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"results\"},{\"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\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable and specifications, and report coefficients with standard errors in parentheses. One table translates coefficients into percentage changes in the exercise rate for a one-SD change in each factor. Valuation tables state their base-case parameters (strike 100, 5% rate, 10% termination, SIC 3). Summary tables are grouped by one-digit SIC.\",\"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\":[\"ranges of magnitudes (e.g., 11% to 16%)\",\"'typically' qualifiers\",\"assumption-conditioned valuations\"]},\"claim_evidence\":[{\"claim\":\"Moving from annual or cliff vesting to monthly vesting lowers option value by roughly 11-16%.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Men exercise faster, lowering option value by 2-4%.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, VIII\"},{\"claim\":\"Top-ranked employees exercise more slowly, raising option value by 2-7%.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, VI, VIII\"},{\"claim\":\"Black-Scholes approximations used in practice produce valuation errors, and the new analytic approximation is more accurate.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Omitting portfolio and additional factors distorts exercise estimates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV specs (1)-(2)\"},{\"claim\":\"ESOs are a major share of CEO pay and their valuation is controversial in practice.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Murphy (2013); Meridian (2018); ISS/IBM example\"}],\"structural_summary\":\"The introduction motivates the topic with the economic importance of ESOs and a practical valuation controversy. It argues that standard theory and Black-Scholes modifications are inadequate, then previews a GMM fractional-logistic exercise model applied to proprietary data. A separate literature section contrasts utility-based theoretical models with empirical work. Next come the methodology, a long data section, estimation results with rank interactions, and a valuation section that quantifies cost effects, critiques practitioner approximations and proposes a new analytic approximation. The conclusion restates the method and findings.\",\"writing_profile_keywords\":[\"proprietary data\",\"reduced-form exercise function\",\"option valuation\",\"fractional logit GMM\",\"economic magnitudes in valuation terms\",\"practitioner relevance\",\"no figures\"],\"confidence\":0.75,\"notes\":\"The primary method is borderline between structural_estimation and reduced_form_regression: the exercise model is explicitly reduced-form, but it is used as an input to simulation-based valuation. Section IV's role_rule is 'empirical_design', but its content is estimation results, so it is labeled results. The excerpts do not show how standard errors are clustered beyond 'robust to correlation'. Robustness checks are not visible.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "asset_pricing (option valuation); executive compensation", "primary_method": "structural_estimation", "identification": {"estimand": "Conditional daily rate at which employees voluntarily exercise vested in-the-money options, as a function of option, stock, firm and holder characteristics, and the implied option cost to the firm.", "strategy": "Reduced-form fractional-logistic exercise function (hazard-like) estimated by GMM that is robust to heteroskedasticity and within-employee correlation. The estimated function is then fed into simulation-based valuation.", "key_assumptions": ["The estimated exercise function describes future exercise behavior well enough to serve as a valuation input.", "Covariates capture the relevant drivers of exercise.", "Employee termination and cancellation hazards are modeled separately."], "threats_addressed": ["Correlation across exercises by the same employee", "Heteroskedasticity", "Omitted-variable bias from using only standard exercise factors (shown via specifications 1-2)"], "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": "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": "roadmap"}], "contribution_paragraph_index": 0, "contribution_types": ["new_method", "new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "separate_section"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "results"}, {"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": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes define the dependent variable and specifications, and report coefficients with standard errors in parentheses. One table translates coefficients into percentage changes in the exercise rate for a one-SD change in each factor. Valuation tables state their base-case parameters (strike 100, 5% rate, 10% termination, SIC 3). Summary tables are grouped by one-digit SIC.", "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": ["ranges of magnitudes (e.g., 11% to 16%)", "'typically' qualifiers", "assumption-conditioned valuations"]}, "claim_evidence": [{"claim": "Moving from annual or cliff vesting to monthly vesting lowers option value by roughly 11-16%.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Men exercise faster, lowering option value by 2-4%.", "evidence_type": "table", "evidence_ref": "Tables IV, VIII"}, {"claim": "Top-ranked employees exercise more slowly, raising option value by 2-7%.", "evidence_type": "table", "evidence_ref": "Tables IV, VI, VIII"}, {"claim": "Black-Scholes approximations used in practice produce valuation errors, and the new analytic approximation is more accurate.", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Omitting portfolio and additional factors distorts exercise estimates.", "evidence_type": "table", "evidence_ref": "Table IV specs (1)-(2)"}, {"claim": "ESOs are a major share of CEO pay and their valuation is controversial in practice.", "evidence_type": "prior_literature", "evidence_ref": "Murphy (2013); Meridian (2018); ISS/IBM example"}], "structural_summary": "The introduction motivates the topic with the economic importance of ESOs and a practical valuation controversy. It argues that standard theory and Black-Scholes modifications are inadequate, then previews a GMM fractional-logistic exercise model applied to proprietary data. A separate literature section contrasts utility-based theoretical models with empirical work. Next come the methodology, a long data section, estimation results with rank interactions, and a valuation section that quantifies cost effects, critiques practitioner approximations and proposes a new analytic approximation. The conclusion restates the method and findings.", "writing_profile_keywords": ["proprietary data", "reduced-form exercise function", "option valuation", "fractional logit GMM", "economic magnitudes in valuation terms", "practitioner relevance", "no figures"], "confidence": 0.75, "notes": "The primary method is borderline between structural_estimation and reduced_form_regression: the exercise model is explicitly reduced-form, but it is used as an input to simulation-based valuation. Section IV's role_rule is 'empirical_design', but its content is estimation results, so it is labeled results. The excerpts do not show how standard errors are clustered beyond 'robust to correlation'. Robustness checks are not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.44, "usage": {"input": 5360, "output": 2038, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062994, "request_id": "req_011CggqKr2cigGf4YaUW4fBG", "cached": false, "ts": "2026-11-03T23:07:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/83a0bc298ba801da87d38c25cff633198201fc1bc6e1480e1d76a8a243b8d5f5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"mutual funds; reputation theory; rating coarseness\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; equilibrium characterization of a dynamic reputation game\",\"key_assumptions\":[\"fully rational investors who form beliefs from past performance\",\"the fund can acquire information, which then decays stochastically\",\"the fund decides on information acquisition after the type shock\"],\"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\":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\":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\":\"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\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"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\":\"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 paper has no tables. Figures are numerical or illustrative plots of equilibrium objects, and the parameter values are stated in captions.\",\"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\":[\"'may' for equilibrium possibilities\",\"'consistent with' empirical findings\",\"'our theory suggests'\"],\"typical_devices_note\":\"\"},\"claim_evidence\":[{\"claim\":\"Star upgrades are associated with large abnormal flows in prior empirical work, which motivates the puzzle.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Del Guercio and Tkac (2008); Reuter and Zitzewitz (2015)\"},{\"claim\":\"In equilibrium, reputation takes only finitely many values even though past performance is continuous (the star rating property).\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III, Proposition 2; Appendix A\"},{\"claim\":\"Two-star equilibria exist, may be multiple, and the set of equilibria varies with the information acquisition cost.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4, Figure 5\"},{\"claim\":\"An uninformed fund never strictly prefers to acquire information in equilibrium.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.C\"},{\"claim\":\"Higher-rated funds are more likely to be informed and perform better, consistent with the evidence on ratings' predictive power.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Blake and Morey (2000); Reuter and Zitzewitz (2015)\"},{\"claim\":\"The star rating property extends to equilibria with more than two groups.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section IV; Figures 8-9\"}],\"structural_summary\":\"The introduction opens with documented star-rating flow effects and poses a puzzle: why would rational investors respond to coarse ratings that add no new information? It then sketches a reputation model with endogenous information acquisition that decays stochastically. It previews two-star equilibria and the empirical implications they match, gives the intuition for why the star rating property is general, and positions the paper against the rating-coarseness and reputation literatures before a roadmap. The body sets up the model, then a preliminary analysis, two-star equilibria with their empirical implications, and a generalization to other equilibria. Proofs are in the appendix, and the conclusion restates the contributions.\",\"writing_profile_keywords\":[\"rational explanation of anomaly\",\"dynamic reputation game\",\"equilibrium characterization\",\"empirical consistency discussion\",\"proofs in appendix\",\"no empirical estimation\"],\"confidence\":0.85,\"notes\":\"This is a pure theory paper. Its empirical implications are linked to prior literature rather than tested.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "mutual funds; reputation theory; rating coarseness", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; equilibrium characterization of a dynamic reputation game", "key_assumptions": ["fully rational investors who form beliefs from past performance", "the fund can acquire information, which then decays stochastically", "the fund decides on information acquisition after the type shock"], "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": 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": 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": "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": 7, "move": "roadmap"}], "contribution_paragraph_index": 6, "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": "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 paper has no tables. Figures are numerical or illustrative plots of equilibrium objects, and the parameter values are stated in captions.", "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": ["'may' for equilibrium possibilities", "'consistent with' empirical findings", "'our theory suggests'"], "typical_devices_note": ""}, "claim_evidence": [{"claim": "Star upgrades are associated with large abnormal flows in prior empirical work, which motivates the puzzle.", "evidence_type": "prior_literature", "evidence_ref": "Del Guercio and Tkac (2008); Reuter and Zitzewitz (2015)"}, {"claim": "In equilibrium, reputation takes only finitely many values even though past performance is continuous (the star rating property).", "evidence_type": "equation_or_proof", "evidence_ref": "Section III, Proposition 2; Appendix A"}, {"claim": "Two-star equilibria exist, may be multiple, and the set of equilibria varies with the information acquisition cost.", "evidence_type": "figure", "evidence_ref": "Figure 4, Figure 5"}, {"claim": "An uninformed fund never strictly prefers to acquire information in equilibrium.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.C"}, {"claim": "Higher-rated funds are more likely to be informed and perform better, consistent with the evidence on ratings' predictive power.", "evidence_type": "prior_literature", "evidence_ref": "Blake and Morey (2000); Reuter and Zitzewitz (2015)"}, {"claim": "The star rating property extends to equilibria with more than two groups.", "evidence_type": "figure", "evidence_ref": "Section IV; Figures 8-9"}], "structural_summary": "The introduction opens with documented star-rating flow effects and poses a puzzle: why would rational investors respond to coarse ratings that add no new information? It then sketches a reputation model with endogenous information acquisition that decays stochastically. It previews two-star equilibria and the empirical implications they match, gives the intuition for why the star rating property is general, and positions the paper against the rating-coarseness and reputation literatures before a roadmap. The body sets up the model, then a preliminary analysis, two-star equilibria with their empirical implications, and a generalization to other equilibria. Proofs are in the appendix, and the conclusion restates the contributions.", "writing_profile_keywords": ["rational explanation of anomaly", "dynamic reputation game", "equilibrium characterization", "empirical consistency discussion", "proofs in appendix", "no empirical estimation"], "confidence": 0.85, "notes": "This is a pure theory paper. Its empirical implications are linked to prior literature rather than tested."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.82, "usage": {"input": 6115, "output": 1932, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.063894, "request_id": "req_011CggqqkTbFju7sfosbPTfb", "cached": false, "ts": "2026-11-03T23:13:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/8540515f7cc1e4b1e05f9bf0718b5db203a7e2254ac31ce665a7029addf00db0.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\":\"Cross-sectional relation between a country's trade network centrality and its unconditional interest rate differentials, currency risk premia, and consumption growth betas.\",\"strategy\":\"Panel regressions with year fixed effects on lagged standardized centrality, size and trade-to-GDP; currency portfolio sorts on centrality; time-series factor regressions of carry factors on the PMC factor; all guided by a general equilibrium network model.\",\"key_assumptions\":[\"Centrality is an unconditional, slowly changing country characteristic.\",\"Lagged centrality is predetermined relative to returns.\",\"The model's mechanism, common consumption exposure, maps to the empirical consumption betas.\"],\"threats_addressed\":[\"Confounding by country size (GDP share) and trade openness\",\"Structural changes in the trade network such as euro introduction and country entry and exit\",\"Pegged currencies\",\"CIP deviations after 2007\",\"Developed-country subset differences\"],\"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\":0}},\"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_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"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\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_evidence\",\"new_theory_or_model\",\"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\":\"conclusion\"},{\"idx\":4,\"role\":\"appendix\"},{\"idx\":5,\"role\":\"appendix\"},{\"idx\":6,\"role\":\"appendix\"},{\"idx\":7,\"role\":\"robustness\"},{\"idx\":8,\"role\":\"robustness\"},{\"idx\":9,\"role\":\"robustness\"},{\"idx\":10,\"role\":\"robustness\"},{\"idx\":11,\"role\":\"robustness\"},{\"idx\":12,\"role\":\"robustness\"},{\"idx\":13,\"role\":\"robustness\"},{\"idx\":14,\"role\":\"robustness\"},{\"idx\":15,\"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_sample_period\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes define variables, sorting procedures and fixed effects. Regressors are standardized and returns are annualized. Factor construction is described as long/short portfolios. Standard error method 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\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"consistent with the model's implications\",\"suggest\",\"helps explain\"]},\"claim_evidence\":[{\"claim\":\"Over 1988-2016, a peripheral-minus-central currency portfolio earns a Sharpe ratio similar to the unconditional carry trade.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table IV\"},{\"claim\":\"Interest rates and currency risk premia decline with trade network centrality, controlling for size and trade openness.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Figure 2\"},{\"claim\":\"Central countries' consumption growth covaries more with world consumption growth.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Figure 4\"},{\"claim\":\"The unconditional carry factor loads nearly one-for-one on PMC and leaves no unexplained excess return.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table VI\"},{\"claim\":\"In the model, central countries have higher exposure to global consumption shocks, hence lower interest rates and risk premia.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Propositions 1-2, Appendix A.4-A.5\"},{\"claim\":\"Results hold without the euro aggregate, without pegs, for developed countries only, and for a sample ending in 2007.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Appendix Tables B.V-B.IX\"}],\"structural_summary\":\"The introduction opens with the carry trade Sharpe ratio puzzle, then motivates trade network centrality with a network figure and a scatter of returns against centrality. It previews a multicountry network-production model and three empirical tests: regressions, portfolio sorts, and a factor comparison showing PMC subsumes the unconditional carry factor. Contributions to the networks, exchange rate, and risk-sharing literatures are positioned within the introduction. The model section derives two propositions; the data section builds the centrality measure and tests both propositions. A short conclusion restates the findings. Appendices hold the proofs and a battery of robustness portfolio sorts.\",\"writing_profile_keywords\":[\"model-then-test\",\"carry trade\",\"portfolio sorts\",\"network centrality\",\"first-person singular\",\"mechanism validation\"],\"confidence\":0.8,\"notes\":\"Single-author JF paper. The empirical section is labeled data but contains the main results. Standard error and significance conventions 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": "Cross-sectional relation between a country's trade network centrality and its unconditional interest rate differentials, currency risk premia, and consumption growth betas.", "strategy": "Panel regressions with year fixed effects on lagged standardized centrality, size and trade-to-GDP; currency portfolio sorts on centrality; time-series factor regressions of carry factors on the PMC factor; all guided by a general equilibrium network model.", "key_assumptions": ["Centrality is an unconditional, slowly changing country characteristic.", "Lagged centrality is predetermined relative to returns.", "The model's mechanism, common consumption exposure, maps to the empirical consumption betas."], "threats_addressed": ["Confounding by country size (GDP share) and trade openness", "Structural changes in the trade network such as euro introduction and country entry and exit", "Pegged currencies", "CIP deviations after 2007", "Developed-country subset differences"], "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": 0}}, "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_results"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"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": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_evidence", "new_theory_or_model", "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": "conclusion"}, {"idx": 4, "role": "appendix"}, {"idx": 5, "role": "appendix"}, {"idx": 6, "role": "appendix"}, {"idx": 7, "role": "robustness"}, {"idx": 8, "role": "robustness"}, {"idx": 9, "role": "robustness"}, {"idx": 10, "role": "robustness"}, {"idx": 11, "role": "robustness"}, {"idx": 12, "role": "robustness"}, {"idx": 13, "role": "robustness"}, {"idx": 14, "role": "robustness"}, {"idx": 15, "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_sample_period", "alternative_specification"]}, "tables": {"convention_notes": "Notes define variables, sorting procedures and fixed effects. Regressors are standardized and returns are annualized. Factor construction is described as long/short portfolios. Standard error method 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": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["consistent with the model's implications", "suggest", "helps explain"]}, "claim_evidence": [{"claim": "Over 1988-2016, a peripheral-minus-central currency portfolio earns a Sharpe ratio similar to the unconditional carry trade.", "evidence_type": "table", "evidence_ref": "Table III, Table IV"}, {"claim": "Interest rates and currency risk premia decline with trade network centrality, controlling for size and trade openness.", "evidence_type": "table", "evidence_ref": "Table I, Figure 2"}, {"claim": "Central countries' consumption growth covaries more with world consumption growth.", "evidence_type": "table", "evidence_ref": "Table II, Figure 4"}, {"claim": "The unconditional carry factor loads nearly one-for-one on PMC and leaves no unexplained excess return.", "evidence_type": "table", "evidence_ref": "Table V, Table VI"}, {"claim": "In the model, central countries have higher exposure to global consumption shocks, hence lower interest rates and risk premia.", "evidence_type": "equation_or_proof", "evidence_ref": "Propositions 1-2, Appendix A.4-A.5"}, {"claim": "Results hold without the euro aggregate, without pegs, for developed countries only, and for a sample ending in 2007.", "evidence_type": "table", "evidence_ref": "Appendix Tables B.V-B.IX"}], "structural_summary": "The introduction opens with the carry trade Sharpe ratio puzzle, then motivates trade network centrality with a network figure and a scatter of returns against centrality. It previews a multicountry network-production model and three empirical tests: regressions, portfolio sorts, and a factor comparison showing PMC subsumes the unconditional carry factor. Contributions to the networks, exchange rate, and risk-sharing literatures are positioned within the introduction. The model section derives two propositions; the data section builds the centrality measure and tests both propositions. A short conclusion restates the findings. Appendices hold the proofs and a battery of robustness portfolio sorts.", "writing_profile_keywords": ["model-then-test", "carry trade", "portfolio sorts", "network centrality", "first-person singular", "mechanism validation"], "confidence": 0.8, "notes": "Single-author JF paper. The empirical section is labeled data but contains the main results. Standard error and significance conventions are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.66, "usage": {"input": 8390, "output": 2283, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080014, "request_id": "req_011CggqM7zakNpsuj1KdsaEz", "cached": false, "ts": "2026-11-03T23:07:22+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/882a310eebfc9e638b5b603f29c2001b2f4dd484198e8e03f207c2f423429cb4.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"labor economics / consumer credit\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of removing a bankruptcy flag from credit reports on credit scores, credit card and mortgage borrowing, employment and earnings\",\"strategy\":\"Difference-in-differences comparing Chapter 13 filers, whose flags are removed at seven years, with Chapter 7 filers, whose flags remain until ten years, using staggered removal timing. A supplementary DiD exploits state bans on employer credit checks. Conditional correlations of 'hidden' flags with later outcomes are used for interpretation.\",\"key_assumptions\":[\"Absent flag removal, Chapter 13 and Chapter 7 filers would have followed parallel trends\",\"Flag removal timing is mechanical (seven years after filing) and not chosen by filers\",\"State-level credit-check bans affect labor demand but not lending markets\"],\"threats_addressed\":[\"Reverse causality between credit limits and balances, flagged as a caveat\",\"Offsetting labor supply and labor demand effects, addressed via price theory and the state-ban analysis\",\"Heterogeneity across demographics and the business cycle\",\"Limitations of the event-study counterfactual relative to the DiD design used here\"],\"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\":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\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":false,\"literature_positioning_location\":\"separate_section\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"literature\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Each DiD table reports the pre-removal mean and SD for Chapter 13 filers in column 1, followed by coefficients at successive post-removal horizons. Sample restrictions are spelled out in the notes.\",\"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\":[\"explicit caveats in the conclusion (three-year horizon, subgroup effects)\",\"warnings against interpreting estimates as a pure MPC\",\"'consistent with' phrasing\",\"'economically small' combined with tight confidence intervals\"]},\"claim_evidence\":[{\"claim\":\"Flag removal raises credit scores by about nine points immediately, lowering implied default probability.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 2\"},{\"claim\":\"Flag removal raises credit limits by about half and card balances by over 40% at a three-year horizon.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 3\"},{\"claim\":\"Mortgage holding rises by 2 pp within a year, concentrated in 2008-2011 removals.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 4\"},{\"claim\":\"Effects on employment and earnings are economically small and precisely estimated.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figure 5\"},{\"claim\":\"Small labor effects hold across demographic subgroups and the business cycle.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 6\"},{\"claim\":\"Hidden flags predict loan delinquency but not job retention or employment.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 7\"},{\"claim\":\"Offsetting supply and demand explanation rejected because wages and earnings do not change.\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section V.A price-theory argument\"}],\"structural_summary\":\"The introduction opens with policy debate over the use of credit reports, especially in hiring. It then describes the DiD design and the administrative data, and previews results in sequence with magnitudes: first-stage credit scores, credit card borrowing, mortgages, then null labor effects. Heterogeneity checks follow, ending with two candidate explanations and the hidden-flag predictive test. A separate literature section is followed by background and data, research design, results by outcome, and an interpretation section. The conclusion restates findings, draws policy implications and lists caveats.\",\"writing_profile_keywords\":[\"difference-in-differences\",\"administrative data\",\"precise null result\",\"contrasting outcomes\",\"credit reports\",\"labor market\",\"policy relevance\",\"magnitudes relative to SD\"],\"confidence\":0.85,\"notes\":\"The section labeled 'Background and Data' was rule-tagged as literature but functions as the data section. The paper emphasizes confidence intervals for the null labor result, but the standard-error method is not visible in the excerpts. A separate state-ban DiD is used as a supplementary test.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "labor economics / consumer credit", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of removing a bankruptcy flag from credit reports on credit scores, credit card and mortgage borrowing, employment and earnings", "strategy": "Difference-in-differences comparing Chapter 13 filers, whose flags are removed at seven years, with Chapter 7 filers, whose flags remain until ten years, using staggered removal timing. A supplementary DiD exploits state bans on employer credit checks. Conditional correlations of 'hidden' flags with later outcomes are used for interpretation.", "key_assumptions": ["Absent flag removal, Chapter 13 and Chapter 7 filers would have followed parallel trends", "Flag removal timing is mechanical (seven years after filing) and not chosen by filers", "State-level credit-check bans affect labor demand but not lending markets"], "threats_addressed": ["Reverse causality between credit limits and balances, flagged as a caveat", "Offsetting labor supply and labor demand effects, addressed via price theory and the state-ban analysis", "Heterogeneity across demographics and the business cycle", "Limitations of the event-study counterfactual relative to the DiD design used here"], "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": 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": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "move": "preview_of_results"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence"], "roadmap_present": false, "literature_positioning_location": "separate_section"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "literature"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Each DiD table reports the pre-removal mean and SD for Chapter 13 filers in column 1, followed by coefficients at successive post-removal horizons. Sample restrictions are spelled out in the notes.", "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": ["explicit caveats in the conclusion (three-year horizon, subgroup effects)", "warnings against interpreting estimates as a pure MPC", "'consistent with' phrasing", "'economically small' combined with tight confidence intervals"]}, "claim_evidence": [{"claim": "Flag removal raises credit scores by about nine points immediately, lowering implied default probability.", "evidence_type": "table", "evidence_ref": "Table II; Figure 2"}, {"claim": "Flag removal raises credit limits by about half and card balances by over 40% at a three-year horizon.", "evidence_type": "table", "evidence_ref": "Table III; Figure 3"}, {"claim": "Mortgage holding rises by 2 pp within a year, concentrated in 2008-2011 removals.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 4"}, {"claim": "Effects on employment and earnings are economically small and precisely estimated.", "evidence_type": "table", "evidence_ref": "Table V; Figure 5"}, {"claim": "Small labor effects hold across demographic subgroups and the business cycle.", "evidence_type": "table", "evidence_ref": "Table VI; Figure 6"}, {"claim": "Hidden flags predict loan delinquency but not job retention or employment.", "evidence_type": "table", "evidence_ref": "Table VII; Figure 7"}, {"claim": "Offsetting supply and demand explanation rejected because wages and earnings do not change.", "evidence_type": "none_stated", "evidence_ref": "Section V.A price-theory argument"}], "structural_summary": "The introduction opens with policy debate over the use of credit reports, especially in hiring. It then describes the DiD design and the administrative data, and previews results in sequence with magnitudes: first-stage credit scores, credit card borrowing, mortgages, then null labor effects. Heterogeneity checks follow, ending with two candidate explanations and the hidden-flag predictive test. A separate literature section is followed by background and data, research design, results by outcome, and an interpretation section. The conclusion restates findings, draws policy implications and lists caveats.", "writing_profile_keywords": ["difference-in-differences", "administrative data", "precise null result", "contrasting outcomes", "credit reports", "labor market", "policy relevance", "magnitudes relative to SD"], "confidence": 0.85, "notes": "The section labeled 'Background and Data' was rule-tagged as literature but functions as the data section. The paper emphasizes confidence intervals for the null labor result, but the standard-error method is not visible in the excerpts. A separate state-ban DiD is used as a supplementary test."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.83, "usage": {"input": 7585, "output": 2158, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074294, "request_id": "req_011CggqvJ3V9ZPSocBR3U532", "cached": false, "ts": "2026-11-03T23:14:52+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/885e16dd4123f8380aa168e299916677cc673f1c78a2a06d7c63aa3b61caf2f3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Marginal effect of investors' unrealized returns since purchase on announcement-window selling probability and abnormal volume/returns, conditional on earnings news\",\"strategy\":\"Cox proportional hazard models of fund-manager sales using Ancerno position-level data, controlling for earnings surprise; stock-level OLS regressions of abnormal volume and returns on capital gains overhang interacted with unexpected earnings, with portfolio sorts\",\"key_assumptions\":[\"Trades in short windows around earnings announcements are motivated by the announcements\",\"Proportional hazards assumption holds\",\"CGO proxies the representative investor's unrealized gain/loss position\",\"Controlling for earnings news isolates preference effects from belief shocks\"],\"threats_addressed\":[\"Belief shocks confounding preference-driven selling (controls for earnings surprise)\",\"Violation of proportional hazards assumption\",\"Tax-motivated trading and turn-of-year effects\",\"January-performance explanation for CGO return predictability (Novy-Marx)\",\"Alternative CGO measure (Frazzini)\"],\"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\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"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\":\"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\",\"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\":\"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\",\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes define the sample and estimator (OLS), refer to the Appendix for variable definitions, state that continuous regressors are ranked into quarterly deciles and scaled, and report significance thresholds for correlations.\",\"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\":[\"'consistent with' framing\",\"'suggest' verbs\",\"'generally consistent'\",\"ceteris paribus qualifiers\"]},\"claim_evidence\":[{\"claim\":\"Fund managers' marginal selling schedule approximates an inverse V-shape around zero unrealized returns, steeper in the loss domain\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Figure 2\"},{\"claim\":\"Gain-position holders sell more and loss-position holders sell less as earnings news improves\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Hazard-of-sale for 22–26% winners is about 1.24 times that of 22–26% losers\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Stockholders' CGO moderates abnormal seller-initiated volume around announcements\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"CGO is associated with short-window return underreaction to earnings news\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V, Table VI, Figure 4, Figure 5\"}],\"structural_summary\":\"The introduction motivates the work with a puzzle: preference-based accounts of the disposition effect are questioned by evidence of V-shaped selling. It then previews investor-level hazard models and stock-level volume and return tests. The results are presented with a magnitude example and an emotion-based interpretation, followed by contribution claims positioned against prior literature and a roadmap. The body moves from data, to investor-level tests (motivation, method, proportional-hazards checks), to two stock-level test sections. A combined robustness-and-conclusion section summarizes untabulated and Internet Appendix checks and suggests directions for future research.\",\"writing_profile_keywords\":[\"disposition effect\",\"prospect theory\",\"hazard model\",\"earnings announcements\",\"multi-level evidence\",\"first-person singular\",\"consistent-with hedging\"],\"confidence\":0.8,\"notes\":\"Single-author paper written in the first person. Robustness checks are partly untabulated and partly placed in the Internet Appendix. Standard-error treatment is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "asset_pricing", "primary_method": "reduced_form_regression", "identification": {"estimand": "Marginal effect of investors' unrealized returns since purchase on announcement-window selling probability and abnormal volume/returns, conditional on earnings news", "strategy": "Cox proportional hazard models of fund-manager sales using Ancerno position-level data, controlling for earnings surprise; stock-level OLS regressions of abnormal volume and returns on capital gains overhang interacted with unexpected earnings, with portfolio sorts", "key_assumptions": ["Trades in short windows around earnings announcements are motivated by the announcements", "Proportional hazards assumption holds", "CGO proxies the representative investor's unrealized gain/loss position", "Controlling for earnings news isolates preference effects from belief shocks"], "threats_addressed": ["Belief shocks confounding preference-driven selling (controls for earnings surprise)", "Violation of proportional hazards assumption", "Tax-motivated trading and turn-of-year effects", "January-performance explanation for CGO return predictability (Novy-Marx)", "Alternative CGO measure (Frazzini)"], "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": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "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": "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", "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": "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", "subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Notes define the sample and estimator (OLS), refer to the Appendix for variable definitions, state that continuous regressors are ranked into quarterly deciles and scaled, and report significance thresholds for correlations.", "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": ["'consistent with' framing", "'suggest' verbs", "'generally consistent'", "ceteris paribus qualifiers"]}, "claim_evidence": [{"claim": "Fund managers' marginal selling schedule approximates an inverse V-shape around zero unrealized returns, steeper in the loss domain", "evidence_type": "figure", "evidence_ref": "Figure 1, Figure 2"}, {"claim": "Gain-position holders sell more and loss-position holders sell less as earnings news improves", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Hazard-of-sale for 22–26% winners is about 1.24 times that of 22–26% losers", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Stockholders' CGO moderates abnormal seller-initiated volume around announcements", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "CGO is associated with short-window return underreaction to earnings news", "evidence_type": "table", "evidence_ref": "Table V, Table VI, Figure 4, Figure 5"}], "structural_summary": "The introduction motivates the work with a puzzle: preference-based accounts of the disposition effect are questioned by evidence of V-shaped selling. It then previews investor-level hazard models and stock-level volume and return tests. The results are presented with a magnitude example and an emotion-based interpretation, followed by contribution claims positioned against prior literature and a roadmap. The body moves from data, to investor-level tests (motivation, method, proportional-hazards checks), to two stock-level test sections. A combined robustness-and-conclusion section summarizes untabulated and Internet Appendix checks and suggests directions for future research.", "writing_profile_keywords": ["disposition effect", "prospect theory", "hazard model", "earnings announcements", "multi-level evidence", "first-person singular", "consistent-with hedging"], "confidence": 0.8, "notes": "Single-author paper written in the first person. Robustness checks are partly untabulated and partly placed in the Internet Appendix. Standard-error treatment is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.25, "usage": {"input": 7927, "output": 2100, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074502, "request_id": "req_011CggqJTkdqUhgQRaTD2TLv", "cached": false, "ts": "2026-11-03T23:06:44+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/88a18d057a327bcaf9a06e0c09495cb70dbc4a94345c617d8341761aff97cfe3.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theory): comparative-static sign of equilibrium bank lending with respect to the capital requirement\",\"strategy\":\"Analytical first-order approach in a bank model with government guarantees, supplemented by numerical examples and a calibration to a global bank in 2017\",\"key_assumptions\":[\"government guarantee creates an implicit subsidy\",\"heterogeneity in residual cash flows (e.g., legacy vs. new loans)\",\"perfect correlation and diminishing returns among new loans for tractability\",\"first-order condition uniquely pins down lending\"],\"threats_addressed\":[\"full bank specialization implying nil FSE\",\"sensitivity of the lending response to calibration parameters\",\"corporate income tax\",\"discrete loan choice\"],\"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\":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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"other\"},{\"paragraph_index\":9,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":10,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"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\":\"other\"},{\"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\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"The only table is a calibration table that lists each parameter with its value, definition, calculation and sources, aimed at a major international bank in 2017. There are no regression tables.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\",\"notes_placeholder_removed\":\"n/a\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"modal 'can' for possibility results\",\"'more conservative interpretation' fallback claim\",\"'in some circumstances'\",\"'likely' in calibration statements\"]},\"claim_evidence\":[{\"claim\":\"A higher capital requirement can increase lending when the forced safety effect dominates the liability composition effect\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II, Appendix B\"},{\"claim\":\"The guarantee overhang can lead banks to pass on positive NPV loans\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III\"},{\"claim\":\"Calibrated to a global bank in 2017, the FSE fully offsets negative forces, giving a slightly positive lending response\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Table I, Figure 8\"},{\"claim\":\"At pre-crisis capital requirement levels, lending responses were more likely negative; the response depends on parameters\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 9\"},{\"claim\":\"With discrete heterogeneous loans, the optimal portfolio can expand with the capital requirement\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5, Appendix C, Figure D1\"}],\"structural_summary\":\"The paper opens with the policy debate and the conventional wisdom that capital requirements reduce lending, then states a challenge to it. It builds intuition for the mechanism over several paragraphs: guarantee overhang, residual cash flows, the default boundary and the forced safety effect. Contributions are listed as numbered claims and positioned against debt-overhang and specialization models, followed by a roadmap. The body sets up the environment, derives the sign of the lending response through a first-order approach, and works through special cases of residual cash flow heterogeneity. It ends with a calibration and sensitivity section that assesses empirical relevance and links the model to predictions. The conclusion restates the contribution with a conservative fallback interpretation, and proofs and numerical details sit in the appendices.\",\"writing_profile_keywords\":[\"contrarian thesis\",\"challenge conventional wisdom\",\"named mechanism\",\"intuition-first introduction\",\"calibration for relevance\",\"policy-oriented\",\"proofs in appendix\"],\"confidence\":0.85,\"notes\":\"Theory paper with a calibration section, so empirical identification fields are marked not applicable. The figures object should contain only uncertainty_shown; the extra 'notes_placeholder_removed' key does not belong there.\"}", "json": {"genre": "theoretical", "subfield": "financial_intermediation_banking", "subfield_secondary": "macro_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theory): comparative-static sign of equilibrium bank lending with respect to the capital requirement", "strategy": "Analytical first-order approach in a bank model with government guarantees, supplemented by numerical examples and a calibration to a global bank in 2017", "key_assumptions": ["government guarantee creates an implicit subsidy", "heterogeneity in residual cash flows (e.g., legacy vs. new loans)", "perfect correlation and diminishing returns among new loans for tractability", "first-order condition uniquely pins down lending"], "threats_addressed": ["full bank specialization implying nil FSE", "sensitivity of the lending response to calibration parameters", "corporate income tax", "discrete loan choice"], "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": 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": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"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": "contribution_statement"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "other"}, {"paragraph_index": 9, "move": "policy_or_broader_implication"}, {"paragraph_index": 10, "move": "roadmap"}], "contribution_paragraph_index": 5, "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": "other"}, {"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": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters", "alternative_specification"]}, "tables": {"convention_notes": "The only table is a calibration table that lists each parameter with its value, definition, calculation and sources, aimed at a major international bank in 2017. There are no regression tables.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no", "notes_placeholder_removed": "n/a"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["modal 'can' for possibility results", "'more conservative interpretation' fallback claim", "'in some circumstances'", "'likely' in calibration statements"]}, "claim_evidence": [{"claim": "A higher capital requirement can increase lending when the forced safety effect dominates the liability composition effect", "evidence_type": "equation_or_proof", "evidence_ref": "Section II, Appendix B"}, {"claim": "The guarantee overhang can lead banks to pass on positive NPV loans", "evidence_type": "equation_or_proof", "evidence_ref": "Section III"}, {"claim": "Calibrated to a global bank in 2017, the FSE fully offsets negative forces, giving a slightly positive lending response", "evidence_type": "figure", "evidence_ref": "Table I, Figure 8"}, {"claim": "At pre-crisis capital requirement levels, lending responses were more likely negative; the response depends on parameters", "evidence_type": "figure", "evidence_ref": "Figure 9"}, {"claim": "With discrete heterogeneous loans, the optimal portfolio can expand with the capital requirement", "evidence_type": "figure", "evidence_ref": "Figure 5, Appendix C, Figure D1"}], "structural_summary": "The paper opens with the policy debate and the conventional wisdom that capital requirements reduce lending, then states a challenge to it. It builds intuition for the mechanism over several paragraphs: guarantee overhang, residual cash flows, the default boundary and the forced safety effect. Contributions are listed as numbered claims and positioned against debt-overhang and specialization models, followed by a roadmap. The body sets up the environment, derives the sign of the lending response through a first-order approach, and works through special cases of residual cash flow heterogeneity. It ends with a calibration and sensitivity section that assesses empirical relevance and links the model to predictions. The conclusion restates the contribution with a conservative fallback interpretation, and proofs and numerical details sit in the appendices.", "writing_profile_keywords": ["contrarian thesis", "challenge conventional wisdom", "named mechanism", "intuition-first introduction", "calibration for relevance", "policy-oriented", "proofs in appendix"], "confidence": 0.85, "notes": "Theory paper with a calibration section, so empirical identification fields are marked not applicable. The figures object should contain only uncertainty_shown; the extra 'notes_placeholder_removed' key does not belong there."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.34, "usage": {"input": 6725, "output": 2186, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071414, "request_id": "req_011Cggr1yBqizZ1bPf8Jo8V9", "cached": false, "ts": "2026-11-03T23:16:07+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/8bbc87c88e1496850ec660dad89c60398530d95d622de4a5858a90117f55f6a2.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Marginal propensity to consume out of unrealized capital gains and out of dividend income, estimated separately by wealth group\",\"strategy\":\"Simulated IV: actual changes in capital gains and dividends are instrumented with passive gains and dividends computed from lagged portfolio weights. Regressions are in first differences, run by wealth group, and exclude holdings in the household's own industry.\",\"key_assumptions\":[\"Lagged portfolio weights are not determined by future shocks that jointly drive returns and consumption\",\"Portfolios are sticky enough that passive returns strongly predict actual gains (relevance)\",\"Consumption imputed from the budget constraint is measured without systematic bias correlated with returns\"],\"threats_addressed\":[\"Endogenous portfolio rebalancing (e.g., liquidation for expenditure shocks)\",\"Time-invariant household heterogeneity\",\"Correlation between own-industry stock returns and labor income\",\"Measurement error in capital gains and imputed consumption\",\"Correlated changes in housing and other financial wealth\",\"Common income shocks (same-employer comparisons)\",\"Realized versus unrealized gains\"],\"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},\"interpretation_note_placeholder\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"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\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"both\"},\"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\":\"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\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes write out the regression equation and state the instruments (passive capital gains and passive dividends) and the controls. Results are reported separately by wealth group, with each cell often a separate regression. Monetary units (1,000 SEK) are given in the summary statistics note. The OLS table is placed before the IV table so the two can be compared.\",\"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' framing for mechanisms\",\"'suggests' when stating broader implications\",\"noting that high-MPC households own a small share of stockholdings\"]},\"claim_evidence\":[{\"claim\":\"MPC out of unrealized capital gains is about 3% for the top 30% of the wealth distribution and about 23% for the bottom half\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"MPC out of dividends exceeds MPC out of capital gains in all wealth groups\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Estimates are robust to controlling for realized capital gains\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Measurement error does not drive the MPC estimates\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VII\"},{\"claim\":\"MPC heterogeneity is driven by liquid wealth relative to income rather than wealth level per se\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Among unconstrained households, older households have a higher MPC out of capital gains\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Results are not driven by other wealth changes or common employer income shocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X-XII\"},{\"claim\":\"Prior estimates of MPC out of stock wealth range from 0% to 10%\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction paragraph 6; Table A.2\"}],\"structural_summary\":\"The introduction opens with the macro and policy relevance of stock wealth effects, then names the gap in causal estimates and the specific identification challenges. It lays out a layered defense (first differences, wealth-group splits, own-industry exclusion, simulated IV from lagged weights) before previewing headline MPC magnitudes and mechanism links to buffer-stock theory. Literature positioning (dividend demand, MPC, wealth effects) and a roadmap close the introduction. The body moves from data and design to OLS-versus-IV main results, then a long measurement-error section combining theory and auxiliary brokerage data. Heterogeneity tests provide mechanism evidence and robustness tables follow. The conclusion restates the two advantages and three findings, framed as consistent with theory.\",\"writing_profile_keywords\":[\"administrative data\",\"simulated instrument\",\"marginal propensity to consume\",\"wealth heterogeneity\",\"measurement error analysis\",\"buffer stock interpretation\",\"mental accounting of dividends\"],\"confidence\":0.8,\"notes\":\"The abstract_moves object contains an extra key ('interpretation_note_placeholder') that is not in the schema; with additionalProperties set to false this output will fail validation and the key should be removed. The abstract's final sentence is an interpretation and was not coded as a contribution. The standard-error type is not visible in the excerpts. Section idx 5 (Heterogeneity) is pre-labelled robustness but functions as mechanism results. The Measurement Error section is tagged as data by the rule but serves a robustness role. Appendix sections (idx 8–11) are table/appendix headings.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "macro_finance", "primary_method": "instrumental_variables", "identification": {"estimand": "Marginal propensity to consume out of unrealized capital gains and out of dividend income, estimated separately by wealth group", "strategy": "Simulated IV: actual changes in capital gains and dividends are instrumented with passive gains and dividends computed from lagged portfolio weights. Regressions are in first differences, run by wealth group, and exclude holdings in the household's own industry.", "key_assumptions": ["Lagged portfolio weights are not determined by future shocks that jointly drive returns and consumption", "Portfolios are sticky enough that passive returns strongly predict actual gains (relevance)", "Consumption imputed from the budget constraint is measured without systematic bias correlated with returns"], "threats_addressed": ["Endogenous portfolio rebalancing (e.g., liquidation for expenditure shocks)", "Time-invariant household heterogeneity", "Correlation between own-industry stock returns and labor income", "Measurement error in capital gains and imputed consumption", "Correlated changes in housing and other financial wealth", "Common income shocks (same-employer comparisons)", "Realized versus unrealized gains"], "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}, "interpretation_note_placeholder": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 0, "move": "literature_gap"}, {"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": 3, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_identification", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "both"}, "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": "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": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "subsample", "alternative_measure", "other"]}, "tables": {"convention_notes": "Table notes write out the regression equation and state the instruments (passive capital gains and passive dividends) and the controls. Results are reported separately by wealth group, with each cell often a separate regression. Monetary units (1,000 SEK) are given in the summary statistics note. The OLS table is placed before the IV table so the two can be compared.", "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' framing for mechanisms", "'suggests' when stating broader implications", "noting that high-MPC households own a small share of stockholdings"]}, "claim_evidence": [{"claim": "MPC out of unrealized capital gains is about 3% for the top 30% of the wealth distribution and about 23% for the bottom half", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "MPC out of dividends exceeds MPC out of capital gains in all wealth groups", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Estimates are robust to controlling for realized capital gains", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Measurement error does not drive the MPC estimates", "evidence_type": "table", "evidence_ref": "Tables VI-VII"}, {"claim": "MPC heterogeneity is driven by liquid wealth relative to income rather than wealth level per se", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Among unconstrained households, older households have a higher MPC out of capital gains", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Results are not driven by other wealth changes or common employer income shocks", "evidence_type": "table", "evidence_ref": "Tables X-XII"}, {"claim": "Prior estimates of MPC out of stock wealth range from 0% to 10%", "evidence_type": "prior_literature", "evidence_ref": "Introduction paragraph 6; Table A.2"}], "structural_summary": "The introduction opens with the macro and policy relevance of stock wealth effects, then names the gap in causal estimates and the specific identification challenges. It lays out a layered defense (first differences, wealth-group splits, own-industry exclusion, simulated IV from lagged weights) before previewing headline MPC magnitudes and mechanism links to buffer-stock theory. Literature positioning (dividend demand, MPC, wealth effects) and a roadmap close the introduction. The body moves from data and design to OLS-versus-IV main results, then a long measurement-error section combining theory and auxiliary brokerage data. Heterogeneity tests provide mechanism evidence and robustness tables follow. The conclusion restates the two advantages and three findings, framed as consistent with theory.", "writing_profile_keywords": ["administrative data", "simulated instrument", "marginal propensity to consume", "wealth heterogeneity", "measurement error analysis", "buffer stock interpretation", "mental accounting of dividends"], "confidence": 0.8, "notes": "The abstract_moves object contains an extra key ('interpretation_note_placeholder') that is not in the schema; with additionalProperties set to false this output will fail validation and the key should be removed. The abstract's final sentence is an interpretation and was not coded as a contribution. The standard-error type is not visible in the excerpts. Section idx 5 (Heterogeneity) is pre-labelled robustness but functions as mechanism results. The Measurement Error section is tagged as data by the rule but serves a robustness role. Appendix sections (idx 8–11) are table/appendix headings."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 21.86, "usage": {"input": 8375, "output": 2579, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.085874, "request_id": "req_011Cggr3AcwbXweLSSq79Z7b", "cached": false, "ts": "2026-11-03T23:16:28+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9240787a12c8f80016a47c7be0c41b1f821a66854185f4b05e28ca0675c868a7.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"household_finance (mortgage contract design; contract theory under asymmetric information)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theory paper; target is existence conditions for a noncontingent contracts equilibrium)\",\"strategy\":\"not applicable; analytical model of competitive lending with lender private information about index quality, with equilibrium refinements\",\"key_assumptions\":[\"Lenders privately observe the quality of the index as a measure of the true external state\",\"Lenders are risk-averse with respect to promises (value larger promises more in bad external states)\",\"Competitive lenders; standard belief refinements govern off-equilibrium beliefs\",\"Index is observable and verifiable, while the true state is revealed only after renegotiation is no longer possible\"],\"threats_addressed\":[],\"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\":5},\"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\":\"literature_gap\"},{\"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\":\"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\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":8,\"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\":\"robustness\"},{\"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\":[\"other\"]},\"tables\":{\"convention_notes\":\"A single appendix table lists functional forms and parameter values for the numerical costly-state-verification example. It contains no regression output.\",\"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\":[\"conditional framing ('under the conditions described in our model')\",\"sufficient-conditions phrasing rather than necessity\",\"modal verbs ('might', 'can arise', 'may be concerned')\",\"'arguably', 'loosely speaking'\"],\"typical_devices_note\":\"none\"} ,\"claim_evidence\":[{\"claim\":\"Contracts, notably mortgages, widely fail to condition repayments on available indices despite proposals favoring indexation\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Intro paras 0 and 7 (e.g., Caplin et al. 2007; Mian and Sufi 2015)\"},{\"claim\":\"A noncontingent equilibrium exists in a two-state numerical example because a contingent offer carries a premium that low-quality-index lenders gain from\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Intro paras 1-2; Section I\"},{\"claim\":\"Sufficient conditions guarantee a noncontingent equilibrium even though it is ex ante Pareto-inferior to full risk-sharing\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section V; Proposition 3; Appendix B\"},{\"claim\":\"The conditions of the general theorem hold in a stylized mortgage model with default\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Proposition 1 proof in Appendix B\"},{\"claim\":\"A costly-state-verification model satisfies the key assumptions under specific functional forms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Appendix A; Table A.I\"},{\"claim\":\"Under monopoly lending the noncontingent equilibrium persists while the full-information equilibrium does not\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section VI.B\"}],\"structural_summary\":\"The paper opens with the puzzle that contracts rarely index to public measures such as house-price indices. It rejects incomplete-contract and transaction-cost explanations and proposes an asymmetric-information mechanism. A numerical example in the introduction sketches two equilibria and the 'lemon insurance' intuition. Extended positioning against the incomplete-contracts, insurance and mortgage-design literatures follows, then a roadmap. The body escalates in generality: a simple example, a mortgage application, the general model and equilibrium definition, preliminary zero-profit analysis, the main sufficient-conditions theorem, and extensions (profits, monopoly, alternative forms of adverse selection). A short conclusion restates the theory. Appendices give a costly-state-verification numerical example and the proofs.\",\"writing_profile_keywords\":[\"pure theory\",\"asymmetric information\",\"contract theory\",\"example-first exposition\",\"sufficient conditions\",\"equilibrium refinement\",\"mortgage indexation\",\"literature contrast\"],\"confidence\":0.85,\"notes\":\"No empirical analysis; empirical-only fields are set to none or unclear. Section IV ('Preliminary Analysis') is tagged theory. Section VI (extensions) is mapped to robustness with type 'other' because it checks how the model's results change under alternative assumptions.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "household_finance (mortgage contract design; contract theory under asymmetric information)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theory paper; target is existence conditions for a noncontingent contracts equilibrium)", "strategy": "not applicable; analytical model of competitive lending with lender private information about index quality, with equilibrium refinements", "key_assumptions": ["Lenders privately observe the quality of the index as a measure of the true external state", "Lenders are risk-averse with respect to promises (value larger promises more in bad external states)", "Competitive lenders; standard belief refinements govern off-equilibrium beliefs", "Index is observable and verifiable, while the true state is revealed only after renegotiation is no longer possible"], "threats_addressed": [], "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": 5}, "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": "literature_gap"}, {"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": "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": "related_literature_positioning"}, {"paragraph_index": 7, "move": "policy_or_broader_implication"}, {"paragraph_index": 8, "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": "robustness"}, {"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": ["other"]}, "tables": {"convention_notes": "A single appendix table lists functional forms and parameter values for the numerical costly-state-verification example. It contains no regression output.", "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": ["conditional framing ('under the conditions described in our model')", "sufficient-conditions phrasing rather than necessity", "modal verbs ('might', 'can arise', 'may be concerned')", "'arguably', 'loosely speaking'"], "typical_devices_note": "none"}, "claim_evidence": [{"claim": "Contracts, notably mortgages, widely fail to condition repayments on available indices despite proposals favoring indexation", "evidence_type": "prior_literature", "evidence_ref": "Intro paras 0 and 7 (e.g., Caplin et al. 2007; Mian and Sufi 2015)"}, {"claim": "A noncontingent equilibrium exists in a two-state numerical example because a contingent offer carries a premium that low-quality-index lenders gain from", "evidence_type": "equation_or_proof", "evidence_ref": "Intro paras 1-2; Section I"}, {"claim": "Sufficient conditions guarantee a noncontingent equilibrium even though it is ex ante Pareto-inferior to full risk-sharing", "evidence_type": "equation_or_proof", "evidence_ref": "Section V; Proposition 3; Appendix B"}, {"claim": "The conditions of the general theorem hold in a stylized mortgage model with default", "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Proposition 1 proof in Appendix B"}, {"claim": "A costly-state-verification model satisfies the key assumptions under specific functional forms", "evidence_type": "table", "evidence_ref": "Appendix A; Table A.I"}, {"claim": "Under monopoly lending the noncontingent equilibrium persists while the full-information equilibrium does not", "evidence_type": "equation_or_proof", "evidence_ref": "Section VI.B"}], "structural_summary": "The paper opens with the puzzle that contracts rarely index to public measures such as house-price indices. It rejects incomplete-contract and transaction-cost explanations and proposes an asymmetric-information mechanism. A numerical example in the introduction sketches two equilibria and the 'lemon insurance' intuition. Extended positioning against the incomplete-contracts, insurance and mortgage-design literatures follows, then a roadmap. The body escalates in generality: a simple example, a mortgage application, the general model and equilibrium definition, preliminary zero-profit analysis, the main sufficient-conditions theorem, and extensions (profits, monopoly, alternative forms of adverse selection). A short conclusion restates the theory. Appendices give a costly-state-verification numerical example and the proofs.", "writing_profile_keywords": ["pure theory", "asymmetric information", "contract theory", "example-first exposition", "sufficient conditions", "equilibrium refinement", "mortgage indexation", "literature contrast"], "confidence": 0.85, "notes": "No empirical analysis; empirical-only fields are set to none or unclear. Section IV ('Preliminary Analysis') is tagged theory. Section VI (extensions) is mapped to robustness with type 'other' because it checks how the model's results change under alternative assumptions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.45, "usage": {"input": 6680, "output": 2327, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074054, "request_id": "req_011Cggqf9Fj9HKLBS3fthWkz", "cached": false, "ts": "2026-11-03T23:11:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9571185376b3300c161d181c60541157ca0595a30f12534a6ea63dd002cc01f8.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Average treatment effect on subjective return expectations of eliciting forecasts in returns rather than prices (task) and of showing return bar charts rather than price line charts (stimulus).\",\"strategy\":\"Randomized 2x2 between-subject experiment across three studies (lab students, lab with incentives, finance professionals with real-world information); treatment dummies in regressions with demographic controls.\",\"key_assumptions\":[\"Random assignment to the format treatments\",\"Expectations stated in price and return formats can be converted into comparable units\",\"Elicited forecasts reflect subjects' true beliefs\"],\"threats_addressed\":[\"Confounding of chart format and question format, separated by the 2x2 design\",\"Lack of incentives or expertise, varied across studies\",\"Artificial information environment, tested with real-world information in Study 3\",\"Alternative explanations such as dividend yield, risk premium and simple heuristics\"],\"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\":2},\"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\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"robustness_mention\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":3,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\"],\"roadmap_present\":true,\"literature_positioning_location\":\"unclear\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"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\":\"robustness\"},{\"idx\":11,\"role\":\"appendix\"},{\"idx\":12,\"role\":\"robustness\"},{\"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\":[\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes name the dependent variable, the treatment dummies and the demographic controls, and state the estimator (fixed or random effects panel, probit marginal effects). The demographics table reports Mann-Whitney z-statistics. The past-returns table states subject-level clustering.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'can be explained in part'\",\"'indicates that'\",\"'potential explanations'\",\"'could thus'\"]},\"claim_evidence\":[{\"claim\":\"Asking subjects to forecast returns rather than prices raises expectations by about 1.1-2.4 pp per month.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 6\"},{\"claim\":\"Showing return bar charts instead of price line charts lowers expectations.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 6\"},{\"claim\":\"Format effects are larger for subjects relying on intuition (low cognitive reflection).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Return formats affect the probability of forecasting a negative return, consistent with reluctance toward negative numbers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Table VI\"},{\"claim\":\"Price charts induce extrapolation from recent returns only, while return formats draw attention to the full history.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Use of price vs return formats in the field is large and largely arbitrary, and was affected by KIID regulation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I; Figure 1\"},{\"claim\":\"Chart and question formats affect forecast accuracy.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; Figure 8\"}],\"structural_summary\":\"The paper motivates with the normative invariance principle and the question of whether investors process prices and returns equivalently. A field overview documents arbitrary format use. It then describes a three-study 2x2 between-subject experimental design. Main treatment effects are presented next, followed by mechanism sections on intuition and number processing, a section on which past returns matter, and one on forecast accuracy. The conclusion restates the findings and stresses regulatory and robo-advisor implications.\",\"writing_profile_keywords\":[\"experimental finance\",\"framing effects\",\"2x2 between-subject design\",\"expectation formation\",\"cognitive reflection\",\"policy relevance\",\"mechanism exploration\"],\"confidence\":0.8,\"notes\":\"literature_positioning_location is set to 'unclear', which is not in the allowed list; no related-literature move is visible in the truncated intro excerpts, and the closest allowed value is 'none'. Robustness checks appear in appendix headings and are also mentioned within the intro. Figures 6 and 8 show 95% confidence intervals while the design figures do not, hence 'mixed'. Statistical significance conventions are not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "household_finance", "primary_method": "survey_or_experiment", "identification": {"estimand": "Average treatment effect on subjective return expectations of eliciting forecasts in returns rather than prices (task) and of showing return bar charts rather than price line charts (stimulus).", "strategy": "Randomized 2x2 between-subject experiment across three studies (lab students, lab with incentives, finance professionals with real-world information); treatment dummies in regressions with demographic controls.", "key_assumptions": ["Random assignment to the format treatments", "Expectations stated in price and return formats can be converted into comparable units", "Elicited forecasts reflect subjects' true beliefs"], "threats_addressed": ["Confounding of chart format and question format, separated by the 2x2 design", "Lack of incentives or expertise, varied across studies", "Artificial information environment, tested with real-world information in Study 3", "Alternative explanations such as dividend yield, risk premium and simple heuristics"], "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": 2}, "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": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "robustness_mention"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "policy_or_broader_implication"}, {"paragraph_index": 3, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence"], "roadmap_present": true, "literature_positioning_location": "unclear"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "other"}, {"idx": 2, "role": "empirical_design"}, {"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": "robustness"}, {"idx": 11, "role": "appendix"}, {"idx": 12, "role": "robustness"}, {"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": ["alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Table notes name the dependent variable, the treatment dummies and the demographic controls, and state the estimator (fixed or random effects panel, probit marginal effects). The demographics table reports Mann-Whitney z-statistics. The past-returns table states subject-level clustering.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'can be explained in part'", "'indicates that'", "'potential explanations'", "'could thus'"]}, "claim_evidence": [{"claim": "Asking subjects to forecast returns rather than prices raises expectations by about 1.1-2.4 pp per month.", "evidence_type": "table", "evidence_ref": "Table III; Figure 6"}, {"claim": "Showing return bar charts instead of price line charts lowers expectations.", "evidence_type": "table", "evidence_ref": "Table III; Figure 6"}, {"claim": "Format effects are larger for subjects relying on intuition (low cognitive reflection).", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Return formats affect the probability of forecasting a negative return, consistent with reluctance toward negative numbers.", "evidence_type": "table", "evidence_ref": "Table V; Table VI"}, {"claim": "Price charts induce extrapolation from recent returns only, while return formats draw attention to the full history.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Use of price vs return formats in the field is large and largely arbitrary, and was affected by KIID regulation.", "evidence_type": "table", "evidence_ref": "Table I; Figure 1"}, {"claim": "Chart and question formats affect forecast accuracy.", "evidence_type": "table", "evidence_ref": "Table IX; Figure 8"}], "structural_summary": "The paper motivates with the normative invariance principle and the question of whether investors process prices and returns equivalently. A field overview documents arbitrary format use. It then describes a three-study 2x2 between-subject experimental design. Main treatment effects are presented next, followed by mechanism sections on intuition and number processing, a section on which past returns matter, and one on forecast accuracy. The conclusion restates the findings and stresses regulatory and robo-advisor implications.", "writing_profile_keywords": ["experimental finance", "framing effects", "2x2 between-subject design", "expectation formation", "cognitive reflection", "policy relevance", "mechanism exploration"], "confidence": 0.8, "notes": "literature_positioning_location is set to 'unclear', which is not in the allowed list; no related-literature move is visible in the truncated intro excerpts, and the closest allowed value is 'none'. Robustness checks appear in appendix headings and are also mentioned within the intro. Figures 6 and 8 show 95% confidence intervals while the design figures do not, hence 'mixed'. Statistical significance conventions are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.11, "usage": {"input": 7270, "output": 2185, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073574, "request_id": "req_011CggqZwGmiz6AggyVuWXij", "cached": false, "ts": "2026-11-03T23:10:15+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/97cb73be2b9f18ccf2e44e1c250633c83d69d28ec48db0e4cca3867dfe56c7b5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance; labor and technology adoption\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-sectional association between a firm's within-industry routine-task labor share (RShare) and its systematic risk, expected returns, and the response of investment and employment to aggregate shocks.\",\"strategy\":\"Within-industry portfolio sorts on RShare (single and double sorts), unconditional and conditional CAPM betas, cash-flow and GDP betas, and panel regressions with controls. Mechanism tests interact RShare with aggregate shocks (GDP growth, recessions) for machine investment and routine employment.\",\"key_assumptions\":[\"RShare proxies for the presence of labor-technology switching options\",\"Within-industry comparisons absorb industry-level confounds\",\"Betas and future returns proxy for systematic risk and expected returns\"],\"threats_addressed\":[\"operating leverage as an alternative explanation\",\"known firm characteristics that predict risk (via double sorts and panel controls)\",\"asymmetry between recessions and expansions (negative versus positive shocks)\"],\"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\":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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_theory_or_model\",\"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\":\"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\":\"within_results\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Notes define sorting procedures (end-of-June, within Fama-French 17 industries, value-weighted quintiles), define variables, and state sample periods. Return units are given in some notes (e.g., monthly excess returns, annual returns in percentages). 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\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'it seems that'\",\"'suggesting that'\",\"'do seem to be able to'\",\"'to the best of my knowledge'\",\"'likely to interrupt'\"]},\"claim_evidence\":[{\"claim\":\"Betas of high-RShare portfolios are over 20% lower than those of low-RShare portfolios, with a high-minus-low return spread of about -3.9% per year.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV\"},{\"claim\":\"High-RShare firms have lower exposure to cash-flow news and to GDP growth.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"RShare predicts lower conditional betas and returns after controlling for firm characteristics.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"In downturns, high-RShare firms invest more in machines and cut more routine-task labor than industry peers.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX; Figure 2\"},{\"claim\":\"High-RShare firms cut costs more and lose less value during recessions, but not during expansions.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"The switching option is exercised when productivity falls below a threshold, which lowers firm risk.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I; Appendix A\"},{\"claim\":\"Routine-task job loss is concentrated in recessions.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Jaimovich and Siu (2014); Figure 1\"}],\"structural_summary\":\"The paper opens with labor-economics evidence that routine jobs are automated mainly in recessions. It frames labor-technology substitution as a state-contingent investment option distinct from growth options. A simple real-options switching model then predicts lower risk for firms with routine-task labor. Next, an establishment-level measure of routine-labor share is built from occupational data. The empirical section proceeds in three steps: characteristics, then asset-pricing tests via sorts, betas and panel regressions, then mechanism tests on investment, employment and performance around aggregate shocks. It also separates the mechanism from an operating-leverage explanation. The conclusion restates the insight, and proofs and data details are placed in appendices.\",\"writing_profile_keywords\":[\"model-guided empirics\",\"real options\",\"portfolio sorts\",\"mechanism tests\",\"new measure\",\"first-person singular\",\"cross-sectional asset pricing\"],\"confidence\":0.8,\"notes\":\"Single-author paper. The significance reporting format and standard-error treatment are not visible in the excerpts. The causal claims concern the theoretical mechanism; the empirical tests are framed as consistent with the model.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance; labor and technology adoption", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-sectional association between a firm's within-industry routine-task labor share (RShare) and its systematic risk, expected returns, and the response of investment and employment to aggregate shocks.", "strategy": "Within-industry portfolio sorts on RShare (single and double sorts), unconditional and conditional CAPM betas, cash-flow and GDP betas, and panel regressions with controls. Mechanism tests interact RShare with aggregate shocks (GDP growth, recessions) for machine investment and routine employment.", "key_assumptions": ["RShare proxies for the presence of labor-technology switching options", "Within-industry comparisons absorb industry-level confounds", "Betas and future returns proxy for systematic risk and expected returns"], "threats_addressed": ["operating leverage as an alternative explanation", "known firm characteristics that predict risk (via double sorts and panel controls)", "asymmetry between recessions and expansions (negative versus positive shocks)"], "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": 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": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 3, "contribution_types": ["new_theory_or_model", "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": "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": "within_results", "types": ["alternative_measure", "alternative_specification", "subsample"]}, "tables": {"convention_notes": "Notes define sorting procedures (end-of-June, within Fama-French 17 industries, value-weighted quintiles), define variables, and state sample periods. Return units are given in some notes (e.g., monthly excess returns, annual returns in percentages). 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": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'it seems that'", "'suggesting that'", "'do seem to be able to'", "'to the best of my knowledge'", "'likely to interrupt'"]}, "claim_evidence": [{"claim": "Betas of high-RShare portfolios are over 20% lower than those of low-RShare portfolios, with a high-minus-low return spread of about -3.9% per year.", "evidence_type": "table", "evidence_ref": "Tables III, IV"}, {"claim": "High-RShare firms have lower exposure to cash-flow news and to GDP growth.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "RShare predicts lower conditional betas and returns after controlling for firm characteristics.", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "In downturns, high-RShare firms invest more in machines and cut more routine-task labor than industry peers.", "evidence_type": "table", "evidence_ref": "Tables VIII, IX; Figure 2"}, {"claim": "High-RShare firms cut costs more and lose less value during recessions, but not during expansions.", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "The switching option is exercised when productivity falls below a threshold, which lowers firm risk.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I; Appendix A"}, {"claim": "Routine-task job loss is concentrated in recessions.", "evidence_type": "prior_literature", "evidence_ref": "Jaimovich and Siu (2014); Figure 1"}], "structural_summary": "The paper opens with labor-economics evidence that routine jobs are automated mainly in recessions. It frames labor-technology substitution as a state-contingent investment option distinct from growth options. A simple real-options switching model then predicts lower risk for firms with routine-task labor. Next, an establishment-level measure of routine-labor share is built from occupational data. The empirical section proceeds in three steps: characteristics, then asset-pricing tests via sorts, betas and panel regressions, then mechanism tests on investment, employment and performance around aggregate shocks. It also separates the mechanism from an operating-leverage explanation. The conclusion restates the insight, and proofs and data details are placed in appendices.", "writing_profile_keywords": ["model-guided empirics", "real options", "portfolio sorts", "mechanism tests", "new measure", "first-person singular", "cross-sectional asset pricing"], "confidence": 0.8, "notes": "Single-author paper. The significance reporting format and standard-error treatment are not visible in the excerpts. The causal claims concern the theoretical mechanism; the empirical tests are framed as consistent with the model."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.86, "usage": {"input": 6410, "output": 2189, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070214, "request_id": "req_011CggqQxUx2n31r4CQYqMnf", "cached": false, "ts": "2026-11-03T23:08:13+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9a853263dbbd0967af274478400265b59fb2c8e4e093dbac44749d816aba5edf.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"high-frequency trading and price discovery\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Contribution of market vs. limit orders by HFTs and non-HFTs to efficient price variance (price discovery), and how this varies with volatility\",\"strategy\":\"Event-time VAR (Hasbrouck-style, extended by Fleming-Mizrach-Nguyen) on regulatory message data with participant identifiers; impulse responses, variance decompositions, information shares; pooled OLS of daily shares on volatility\",\"key_assumptions\":[\"Causal ordering in VAR given by event time sequence\",\"HFT classification of participant IDs is accurate\",\"Small cross-exchange reporting time differences do not drive results\"],\"threats_addressed\":[\"Cross-exchange timestamp differences (one-second calendar-time information shares)\",\"Public information explaining HFT price impact\",\"Front-running of non-HFT orders\"],\"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\":false,\"sentence_index\":-1},\"preview\":null},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"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\":\"policy_or_broader_implication\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"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\":\"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\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes repeat sample (15 non-cross-listed TSX 60 stocks, Oct 2012–Jun 2013), report stock-day averages of VAR IRFs, variance decompositions and OLS coefficients, and define variables. Price impacts are given in basis points and market cap in billions of CAD.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"consistent with\",\"suggests\",\"lend some support\",\"does not imply\",\"seemingly at odds\"]},\"claim_evidence\":[{\"claim\":\"HFT limit orders contribute over twice as much to price discovery as HFT market orders; for non-HFTs, market orders contribute more than limit orders\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, VII\"},{\"claim\":\"More aggressive orders have higher individual price impact, but limit orders dominate price discovery because they are far more numerous\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV, VI\"},{\"claim\":\"HFT limit orders account for about 30% of price discovery vs about 15% for non-HFT limit orders\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Limit order submissions and their price discovery share fall with volatility, driven by HFTs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"HFTs dominate price discovery via limit orders on each exchange, even at one-second sampling\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X, XI\"},{\"claim\":\"HFTs react more than non-HFTs to events on other exchanges, consistent with cross-market information integration\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XII, XIII\"},{\"claim\":\"Positive limit order price impact is reconciled with prior negative impact of HFT liquidity-supplying trades through nonexecution\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Brogaard, Hendershott, and Riordan (2014)\"}],\"structural_summary\":\"The introduction contrasts the traditional trades-versus-quotes view with modern limit order books and proposes a VAR decomposition of price discovery by order type and trader type. It previews the headline results and reconciles them with prior trade-based evidence. It then links the findings to GPR and Hoffmann models, using volatility to discriminate between them, and discusses stability, adverse selection and front-running concerns. A data section is followed by two results sections, market-wide and then within/across exchanges, all presented through tables. The conclusion restates the findings, discusses regulatory trade-offs, and flags future research.\",\"writing_profile_keywords\":[\"market microstructure\",\"high-frequency traders\",\"limit orders\",\"VAR price discovery\",\"regulatory message data\",\"theory-consistent interpretation\",\"table-heavy\"],\"confidence\":0.8,\"notes\":\"The primary method is a Hasbrouck-style VAR, coded as 'other'. Standard error and significance conventions are not visible in the excerpts. The excerpts contain no figures and no equations.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "high-frequency trading and price discovery", "primary_method": "other", "identification": {"estimand": "Contribution of market vs. limit orders by HFTs and non-HFTs to efficient price variance (price discovery), and how this varies with volatility", "strategy": "Event-time VAR (Hasbrouck-style, extended by Fleming-Mizrach-Nguyen) on regulatory message data with participant identifiers; impulse responses, variance decompositions, information shares; pooled OLS of daily shares on volatility", "key_assumptions": ["Causal ordering in VAR given by event time sequence", "HFT classification of participant IDs is accurate", "Small cross-exchange reporting time differences do not drive results"], "threats_addressed": ["Cross-exchange timestamp differences (one-second calendar-time information shares)", "Public information explaining HFT price impact", "Front-running of non-HFT orders"], "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": false, "sentence_index": -1}, "preview": null}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"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": "policy_or_broader_implication"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "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": "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": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure"]}, "tables": {"convention_notes": "Table notes repeat sample (15 non-cross-listed TSX 60 stocks, Oct 2012–Jun 2013), report stock-day averages of VAR IRFs, variance decompositions and OLS coefficients, and define variables. Price impacts are given in basis points and market cap in billions of CAD.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["consistent with", "suggests", "lend some support", "does not imply", "seemingly at odds"]}, "claim_evidence": [{"claim": "HFT limit orders contribute over twice as much to price discovery as HFT market orders; for non-HFTs, market orders contribute more than limit orders", "evidence_type": "table", "evidence_ref": "Tables IV, VII"}, {"claim": "More aggressive orders have higher individual price impact, but limit orders dominate price discovery because they are far more numerous", "evidence_type": "table", "evidence_ref": "Tables III, IV, VI"}, {"claim": "HFT limit orders account for about 30% of price discovery vs about 15% for non-HFT limit orders", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Limit order submissions and their price discovery share fall with volatility, driven by HFTs", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "HFTs dominate price discovery via limit orders on each exchange, even at one-second sampling", "evidence_type": "table", "evidence_ref": "Tables X, XI"}, {"claim": "HFTs react more than non-HFTs to events on other exchanges, consistent with cross-market information integration", "evidence_type": "table", "evidence_ref": "Tables XII, XIII"}, {"claim": "Positive limit order price impact is reconciled with prior negative impact of HFT liquidity-supplying trades through nonexecution", "evidence_type": "prior_literature", "evidence_ref": "Brogaard, Hendershott, and Riordan (2014)"}], "structural_summary": "The introduction contrasts the traditional trades-versus-quotes view with modern limit order books and proposes a VAR decomposition of price discovery by order type and trader type. It previews the headline results and reconciles them with prior trade-based evidence. It then links the findings to GPR and Hoffmann models, using volatility to discriminate between them, and discusses stability, adverse selection and front-running concerns. A data section is followed by two results sections, market-wide and then within/across exchanges, all presented through tables. The conclusion restates the findings, discusses regulatory trade-offs, and flags future research.", "writing_profile_keywords": ["market microstructure", "high-frequency traders", "limit orders", "VAR price discovery", "regulatory message data", "theory-consistent interpretation", "table-heavy"], "confidence": 0.8, "notes": "The primary method is a Hasbrouck-style VAR, coded as 'other'. Standard error and significance conventions are not visible in the excerpts. The excerpts contain no figures and no equations."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.05, "usage": {"input": 8092, "output": 2179, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076742, "request_id": "req_011CggqPYdiPkQ99kXXDFk76", "cached": false, "ts": "2026-11-03T23:07:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9b6cfe2f36badfe4b83636ced2568bfb69fd6ea40fd9df65b5a10a31b940a0f0.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"international_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Cross-country association between the cyclicality of LC bond returns (bond-stock beta) and the LC share of sovereign debt, plus a model-implied mechanism.\",\"strategy\":\"Cross-country regressions and correlations with controls, followed by a two-period model with and without commitment, calibrated separately to emerging and developed markets.\",\"key_assumptions\":[\"Bond-stock beta proxies for the hedging value of LC debt to the borrower\",\"Emerging markets lack commitment and developed markets have commitment\",\"Lender SDF is correlated with domestic output\",\"Stock returns are proportional to output gap surprises\"],\"threats_addressed\":[\"Controls for local-US stock beta, exchange rate regime, and commodity export share\",\"Alternative debt-share measures (total vs external, TIC, other sources)\",\"LC minus FC beta to separate currency effects\",\"Risk-neutral lender counterfactual within the model\"],\"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\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"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\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"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\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_measure\",\"alternative_specification\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Notes define each variable and its units (percent, shares between 0 and 1, annualized natural units). They list controls and data sources, and link model moments to equations in the internet appendix.\",\"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\":[\"'seemingly' qualifiers\",\"framing the model as a simple framework\",\"conditional statements (e.g., provided risk aversion exceeds zero)\",\"'to the best of our knowledge'\"],\"typical_devices\":[\"'seemingly' qualifiers\",\"framing the model as a simple framework\",\"conditional statements (e.g., provided risk aversion exceeds zero)\",\"'to the best of our knowledge'\"]},\"claim_evidence\":[{\"claim\":\"Countries with lower LC debt shares have more procyclical LC bond returns, i.e., better hedging value for the borrower.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table III, Figure 1\"},{\"claim\":\"LC bond-stock betas correlate with inflation cyclicality and with a text-based monetary credibility measure.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Bonds with higher local betas also have higher US stock betas and higher risk premia.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Without commitment, governments choose more countercyclical inflation and tilt borrowing toward FC debt.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.B-C\"},{\"claim\":\"The calibrated model matches EM/DM moments only when lenders are risk-averse.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 5\"}],\"structural_summary\":\"The paper opens with standard optimal-debt theory and states a puzzle: countries whose LC debt hedges best borrow least in LC. It documents this pattern with cross-country regressions, links it to US-stock betas, risk premia and inflation credibility, then builds a two-period commitment/no-commitment model with risk-averse lenders. The model is solved analytically in a special case and calibrated to EM versus DM moments. The conclusion restates the mechanism and its implication for debt management.\",\"writing_profile_keywords\":[\"stylized fact then model\",\"puzzle resolution\",\"commitment vs discretion\",\"cross-country evidence\",\"calibration\",\"international asset pricing\"],\"confidence\":0.8,\"notes\":\"Standard-error type and significance conventions are not visible in the table excerpts. The empirical evidence is cross-sectional and associational. The 'hedging' object contains a duplicated 'typical_devices' key with identical content; most JSON parsers will keep the last instance.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "international_finance", "subfield_secondary": "macro_finance", "primary_method": "calibration_simulation", "identification": {"estimand": "Cross-country association between the cyclicality of LC bond returns (bond-stock beta) and the LC share of sovereign debt, plus a model-implied mechanism.", "strategy": "Cross-country regressions and correlations with controls, followed by a two-period model with and without commitment, calibrated separately to emerging and developed markets.", "key_assumptions": ["Bond-stock beta proxies for the hedging value of LC debt to the borrower", "Emerging markets lack commitment and developed markets have commitment", "Lender SDF is correlated with domestic output", "Stock returns are proportional to output gap surprises"], "threats_addressed": ["Controls for local-US stock beta, exchange rate regime, and commodity export share", "Alternative debt-share measures (total vs external, TIC, other sources)", "LC minus FC beta to separate currency effects", "Risk-neutral lender counterfactual within the model"], "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": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"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": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": 5, "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": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_measure", "alternative_specification", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Notes define each variable and its units (percent, shares between 0 and 1, annualized natural units). They list controls and data sources, and link model moments to equations in the internet appendix.", "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": ["'seemingly' qualifiers", "framing the model as a simple framework", "conditional statements (e.g., provided risk aversion exceeds zero)", "'to the best of our knowledge'"]}, "claim_evidence": [{"claim": "Countries with lower LC debt shares have more procyclical LC bond returns, i.e., better hedging value for the borrower.", "evidence_type": "table", "evidence_ref": "Table II, Table III, Figure 1"}, {"claim": "LC bond-stock betas correlate with inflation cyclicality and with a text-based monetary credibility measure.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Bonds with higher local betas also have higher US stock betas and higher risk premia.", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Without commitment, governments choose more countercyclical inflation and tilt borrowing toward FC debt.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.B-C"}, {"claim": "The calibrated model matches EM/DM moments only when lenders are risk-averse.", "evidence_type": "table", "evidence_ref": "Table VI, Figure 5"}], "structural_summary": "The paper opens with standard optimal-debt theory and states a puzzle: countries whose LC debt hedges best borrow least in LC. It documents this pattern with cross-country regressions, links it to US-stock betas, risk premia and inflation credibility, then builds a two-period commitment/no-commitment model with risk-averse lenders. The model is solved analytically in a special case and calibrated to EM versus DM moments. The conclusion restates the mechanism and its implication for debt management.", "writing_profile_keywords": ["stylized fact then model", "puzzle resolution", "commitment vs discretion", "cross-country evidence", "calibration", "international asset pricing"], "confidence": 0.8, "notes": "Standard-error type and significance conventions are not visible in the table excerpts. The empirical evidence is cross-sectional and associational. The 'hedging' object contains a duplicated 'typical_devices' key with identical content; most JSON parsers will keep the last instance."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.84, "usage": {"input": 6826, "output": 2058, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069258, "request_id": "req_011Cggr2HTvcF7VStyWU41Q9", "cached": false, "ts": "2026-11-03T23:16:11+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/9d04601f4376306a0fd449122bcade4d09ac773b3da832824526c716ab2b21fd.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Average excess stock returns by week of the FOMC cycle (even versus odd weeks), and how much of the equity premium is attributable to Fed monetary policy news.\",\"strategy\":\"Regress daily excess returns on FOMC-cycle dummies. Rule out non-Fed explanations with controls for macro releases, earnings and reserve periods. Link the pattern to the Fed through the timing of intermeeting target changes, Fed funds futures changes, Board meeting interactions, Fed-put predictability regressions and changes in equity premium bounds.\",\"key_assumptions\":[\"The irregular FOMC calendar is unrelated to other news calendars.\",\"Board meetings and intermeeting changes proxy for the timing of Fed news.\",\"The Martin (2017) bounds measure the equity premium.\"],\"threats_addressed\":[\"Macroeconomic data release timing\",\"Corporate earnings announcements\",\"Reserve maintenance periods\",\"Data mining / small sample, via out-of-sample period 2014–2016\",\"Formal Fed releases and speeches as an alternative 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\":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\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"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\":\"policy_or_broader_implication\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"}],\"contribution_paragraph_index\":0,\"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\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_sample_period\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes define FOMC-cycle week dummies, give units in percent (0.1 = 10 bps per day), report heteroskedasticity-robust t-statistics in parentheses with stars, state data sources, and include a narrative table of leak examples.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"robust\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'evidence suggests'\",\"'make it likely'\",\"'appears'\",\"'we argue'\",\"'suggests that'\"]},\"claim_evidence\":[{\"claim\":\"Since 1994 the equity premium is earned entirely in even weeks of FOMC cycle time.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Table I\"},{\"claim\":\"The pattern is also present in international stock indices.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Non-Fed news (macro releases, earnings) does not explain the even-week pattern.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Intermeeting target changes and Fed funds futures declines concentrate in even weeks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V\"},{\"claim\":\"The even-week effect is driven disproportionately by days following Board of Governors meetings.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Even-week returns mean-revert after market declines and predict target changes (Fed put).\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII–IX; Figure 4\"},{\"claim\":\"Even-week reductions in the equity premium account for most of the high even-week returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X; Figure 5\"},{\"claim\":\"Official Fed releases and speeches do not account for the even-week returns; informal communication does.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XI, XII; Figure 6\"}],\"structural_summary\":\"Opens with questions about the Fed's impact on stocks and notes that prior work focuses on announcement days. Section I presents the new fact (an even-week pattern), with statistical significance, out-of-sample checks, trading strategies, international evidence and cross-sectional evidence. Section II rules out non-Fed explanations, then ties the pattern to the Fed through several pieces of evidence. Section III argues the mechanism runs through reductions in the equity premium (a downside-risk channel). Section IV examines formal versus informal communication channels and Fed motives. The paper closes with a policy discussion and a conclusion.\",\"writing_profile_keywords\":[\"new stylized fact\",\"calendar-time anomaly\",\"monetary policy and stocks\",\"narrative evidence\",\"Fed put\",\"informal communication\",\"causal linking via multiple evidence pieces\"],\"confidence\":0.8,\"notes\":\"Causal claims rest on a triangulation of several pieces of evidence rather than a single quasi-experiment. Section IV's rule label of 'robustness' is overridden here as results, since it reports channel analysis rather than robustness checks.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Average excess stock returns by week of the FOMC cycle (even versus odd weeks), and how much of the equity premium is attributable to Fed monetary policy news.", "strategy": "Regress daily excess returns on FOMC-cycle dummies. Rule out non-Fed explanations with controls for macro releases, earnings and reserve periods. Link the pattern to the Fed through the timing of intermeeting target changes, Fed funds futures changes, Board meeting interactions, Fed-put predictability regressions and changes in equity premium bounds.", "key_assumptions": ["The irregular FOMC calendar is unrelated to other news calendars.", "Board meetings and intermeeting changes proxy for the timing of Fed news.", "The Martin (2017) bounds measure the equity premium."], "threats_addressed": ["Macroeconomic data release timing", "Corporate earnings announcements", "Reserve maintenance periods", "Data mining / small sample, via out-of-sample period 2014–2016", "Formal Fed releases and speeches as an alternative 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": 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": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"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": "policy_or_broader_implication"}, {"paragraph_index": 4, "move": "preview_of_results"}], "contribution_paragraph_index": 0, "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": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "within_results", "types": ["alternative_sample_period", "subsample", "other"]}, "tables": {"convention_notes": "Notes define FOMC-cycle week dummies, give units in percent (0.1 = 10 bps per day), report heteroskedasticity-robust t-statistics in parentheses with stars, state data sources, and include a narrative table of leak examples.", "units_stated_in_notes": true, "standard_errors_description": "robust"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'evidence suggests'", "'make it likely'", "'appears'", "'we argue'", "'suggests that'"]}, "claim_evidence": [{"claim": "Since 1994 the equity premium is earned entirely in even weeks of FOMC cycle time.", "evidence_type": "figure", "evidence_ref": "Figure 1; Table I"}, {"claim": "The pattern is also present in international stock indices.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Non-Fed news (macro releases, earnings) does not explain the even-week pattern.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Intermeeting target changes and Fed funds futures declines concentrate in even weeks.", "evidence_type": "table", "evidence_ref": "Tables IV, V"}, {"claim": "The even-week effect is driven disproportionately by days following Board of Governors meetings.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Even-week returns mean-revert after market declines and predict target changes (Fed put).", "evidence_type": "table", "evidence_ref": "Tables VII–IX; Figure 4"}, {"claim": "Even-week reductions in the equity premium account for most of the high even-week returns.", "evidence_type": "table", "evidence_ref": "Table X; Figure 5"}, {"claim": "Official Fed releases and speeches do not account for the even-week returns; informal communication does.", "evidence_type": "table", "evidence_ref": "Tables XI, XII; Figure 6"}], "structural_summary": "Opens with questions about the Fed's impact on stocks and notes that prior work focuses on announcement days. Section I presents the new fact (an even-week pattern), with statistical significance, out-of-sample checks, trading strategies, international evidence and cross-sectional evidence. Section II rules out non-Fed explanations, then ties the pattern to the Fed through several pieces of evidence. Section III argues the mechanism runs through reductions in the equity premium (a downside-risk channel). Section IV examines formal versus informal communication channels and Fed motives. The paper closes with a policy discussion and a conclusion.", "writing_profile_keywords": ["new stylized fact", "calendar-time anomaly", "monetary policy and stocks", "narrative evidence", "Fed put", "informal communication", "causal linking via multiple evidence pieces"], "confidence": 0.8, "notes": "Causal claims rest on a triangulation of several pieces of evidence rather than a single quasi-experiment. Section IV's rule label of 'robustness' is overridden here as results, since it reports channel analysis rather than robustness checks."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.26, "usage": {"input": 8132, "output": 2184, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077002, "request_id": "req_011CggqThnVdAkgQomuh7jLL", "cached": false, "ts": "2026-11-03T23:08:50+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a220807b8df84f59afa957ecaae7b06207a80a128d0517ce79478aacdfcb4b53.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"corporate_finance (risk management and hedging)\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of financial institutions' net worth on the intensity of interest rate and foreign exchange hedging, controlling for risk exposures.\",\"strategy\":\"Difference-in-differences with 2009 as the treatment year. Treated institutions have below-median mortgage-weighted local house price changes, or below-median housing supply elasticity. A triple-differences design adds real estate exposure terciles. Supporting analyses use propensity score matching, IV with house price changes as the instrument, placebo treatment years, and institution and year fixed effects.\",\"key_assumptions\":[\"Parallel trends between treated and control institutions absent house price shocks\",\"House-price-driven loan losses affect hedging only through net worth, not through risk exposures or lending opportunities\",\"Real estate exposure isolates the net worth channel in the triple differences\"],\"threats_addressed\":[\"changes in lending opportunities\",\"changes in risk exposures\",\"reduction in assets alone\",\"sophistication and fixed costs of hedging\",\"risk shifting\",\"regulatory capital concerns\",\"pre-trends (placebo years)\",\"external validity (IV)\"],\"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\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"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\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"resolution_of_puzzle\"===\"x\"?\"other\":\"tension_or_puzzle\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"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\":\"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\":\"mixed\",\"types\":[\"placebo\",\"alternative_measure\",\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state the specification family (pooled OLS, Tobit, fixed effects), the fixed effects included, panel structure, data level (BHC), and sample restrictions. The appendix defines variables with source codes and winsorization rules.\",\"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\":[\"'not likely due to'\",\"'suggest'\",\"'consistent with the theory'\",\"'arguably'\",\"'unlikely to be driving'\"]},\"claim_evidence\":[{\"claim\":\"Institutions with higher net worth hedge more, both across institutions and within institutions over time.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Treated institutions facing house-price net worth shocks cut hedging by up to one-half relative to controls.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Figure 2\"},{\"claim\":\"Treatment reduces net worth but leaves risk exposures unchanged.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Placebo treatment years show no comparable effect.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Results hold when treatment is defined by housing supply elasticity.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Trading rises with net worth, which argues against a risk-shifting explanation.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"IV estimates using house price changes support external validity.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"}],\"structural_summary\":\"The introduction frames bank risk management as poorly understood and policy-relevant, then previews a stylized fact (hedging rises with net worth) and a novel house-price-shock identification strategy with magnitudes. It defends against alternative explanations point by point and positions the findings against the size–hedging puzzle before giving a roadmap. A short theory section derives the hypothesis from collateral-constrained models. The data section details measurement. The results section moves from stylized facts to DiD, triple differences, matching, mechanism checks, placebo tests and an alternative treatment definition. A final section covers IV external validity and alternative hypotheses (risk shifting, regulatory capital). The conclusion restates results, rebuts alternatives, and draws monetary and macroprudential implications.\",\"writing_profile_keywords\":[\"theory-guided empirics\",\"stylized fact then identification\",\"difference-in-differences\",\"triple differences\",\"alternative hypotheses rebuttal\",\"bank holding companies\",\"hedging\"],\"confidence\":0.8,\"notes\":\"The article is marked as retracted. Section IV is labeled theory by the rule-based role, but it actually contains IV and alternative-hypothesis checks, so it was annotated as robustness. Standard error type is not visible in the excerpts.\"}", "json": null, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.75, "usage": {"input": 7119, "output": 2249, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07425, "request_id": "req_011CggqhcsTSD3PztSsA4YEH", "cached": false, "ts": "2026-11-03T23:11:59+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a31d50d30927029b05a279d9ade02c179b1f4e5dea2a256df7c35af9a7b6babe.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Predictive relations: whether a latent persistent dividend growth component forecasts future dividend, GDP and consumption growth and stock returns, and how dividend news components relate to same-day returns, volatility and jumps.\",\"strategy\":\"A latent components model (persistent mean, stochastic-volatility transitory shocks, and jumps whose intensity depends on the number of announcing firms) is fitted to daily firm-matched dividend growth. Its filtered persistent component is then used in predictive regressions and in a log-linearized present value model with cross-equation restrictions, plus out-of-sample forecast comparisons.\",\"key_assumptions\":[\"The components model correctly separates persistent, transitory and jump dynamics in daily dividend growth\",\"The Campbell-Shiller log-linear present value approximation holds\",\"Year-over-year firm-matched dividend changes remove seasonality and composition effects\"],\"threats_addressed\":[\"Lumpiness and composition shifts from varying numbers of announcing firms\",\"Seasonality in firm cash flows\",\"Comparison against alternative dividend growth predictors and standard return predictors\",\"Sensitivity to data frequency and weighting scheme\",\"Firm heterogeneity\"],\"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\":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\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":13,\"move\":\"roadmap\"},{\"paragraph_index\":13,\"move\":\"robustness_mention\"}],\"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\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"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\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes restate the model equations and the cross-equation restrictions being tested. They define panels by frequency or predictor set and report t-statistics in square brackets. Out-of-sample tables benchmark against the 14 Goyal-Welch predictors.\",\"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\":[\"results suggest\",\"consistent with model constraints\",\"at least to some extent\",\"generally / likely\"]},\"claim_evidence\":[{\"claim\":\"The persistent dividend growth component significantly predicts future aggregate dividend growth at quarterly and annual horizons.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The model-based persistent component compares favorably with alternative dividend growth predictors from prior work.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II Panel B; Figure 8\"},{\"claim\":\"The persistent component positively predicts future GDP and consumption growth.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The persistent component predicts daily stock returns, and the present-value cross-equation restrictions are not rejected.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"At monthly horizon the measure beats all but one of 14 standard predictors out of sample.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Jump probability in dividend growth falls as more firms announce dividends on a given day.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6; Table I\"},{\"claim\":\"Dividend news components help explain concurrent returns, volatility and return jumps.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The introduction motivates the problem of irregular daily cash flow news and notes that prior work relies on low-frequency data. It then previews a three-component decomposition and the forecasting results. A second strand frames the scarcity of high-frequency cash flow data as a limitation for asset pricing tests and previews present-value-based return predictability, ending with a roadmap. The body builds a daily bottom-up dividend series and estimates the latent components model. It tests dividend and macro predictability against benchmarks and across frequencies. It then embeds the measure in a predictive present value model with restriction tests and out-of-sample comparisons, and closes with contemporaneous return dynamics. The conclusion restates the method and the main findings.\",\"writing_profile_keywords\":[\"high-frequency cash flows\",\"latent components model\",\"present value restrictions\",\"predictability\",\"benchmark comparison\",\"out-of-sample\",\"new measure\"],\"confidence\":0.75,\"notes\":\"Primary method set to 'other': it is a latent state-space/jump components time-series model combined with predictive regressions, with no causal design. Section II is tagged 'theory' by rule but functions as econometric design and estimation. Significance language: t-statistics are visible but stars are not explicitly seen. The Internet Appendix holds additional robustness checks. Introduction paragraphs 5-12 are a fragmented roadmap sentence.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "other", "identification": {"estimand": "Predictive relations: whether a latent persistent dividend growth component forecasts future dividend, GDP and consumption growth and stock returns, and how dividend news components relate to same-day returns, volatility and jumps.", "strategy": "A latent components model (persistent mean, stochastic-volatility transitory shocks, and jumps whose intensity depends on the number of announcing firms) is fitted to daily firm-matched dividend growth. Its filtered persistent component is then used in predictive regressions and in a log-linearized present value model with cross-equation restrictions, plus out-of-sample forecast comparisons.", "key_assumptions": ["The components model correctly separates persistent, transitory and jump dynamics in daily dividend growth", "The Campbell-Shiller log-linear present value approximation holds", "Year-over-year firm-matched dividend changes remove seasonality and composition effects"], "threats_addressed": ["Lumpiness and composition shifts from varying numbers of announcing firms", "Seasonality in firm cash flows", "Comparison against alternative dividend growth predictors and standard return predictors", "Sensitivity to data frequency and weighting scheme", "Firm heterogeneity"], "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": 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": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 13, "move": "roadmap"}, {"paragraph_index": 13, "move": "robustness_mention"}], "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": "data"}, {"idx": 2, "role": "empirical_design"}, {"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": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "subsample", "other"]}, "tables": {"convention_notes": "Table notes restate the model equations and the cross-equation restrictions being tested. They define panels by frequency or predictor set and report t-statistics in square brackets. Out-of-sample tables benchmark against the 14 Goyal-Welch predictors.", "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": ["results suggest", "consistent with model constraints", "at least to some extent", "generally / likely"]}, "claim_evidence": [{"claim": "The persistent dividend growth component significantly predicts future aggregate dividend growth at quarterly and annual horizons.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The model-based persistent component compares favorably with alternative dividend growth predictors from prior work.", "evidence_type": "table", "evidence_ref": "Table II Panel B; Figure 8"}, {"claim": "The persistent component positively predicts future GDP and consumption growth.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The persistent component predicts daily stock returns, and the present-value cross-equation restrictions are not rejected.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "At monthly horizon the measure beats all but one of 14 standard predictors out of sample.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Jump probability in dividend growth falls as more firms announce dividends on a given day.", "evidence_type": "figure", "evidence_ref": "Figures 5-6; Table I"}, {"claim": "Dividend news components help explain concurrent returns, volatility and return jumps.", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The introduction motivates the problem of irregular daily cash flow news and notes that prior work relies on low-frequency data. It then previews a three-component decomposition and the forecasting results. A second strand frames the scarcity of high-frequency cash flow data as a limitation for asset pricing tests and previews present-value-based return predictability, ending with a roadmap. The body builds a daily bottom-up dividend series and estimates the latent components model. It tests dividend and macro predictability against benchmarks and across frequencies. It then embeds the measure in a predictive present value model with restriction tests and out-of-sample comparisons, and closes with contemporaneous return dynamics. The conclusion restates the method and the main findings.", "writing_profile_keywords": ["high-frequency cash flows", "latent components model", "present value restrictions", "predictability", "benchmark comparison", "out-of-sample", "new measure"], "confidence": 0.75, "notes": "Primary method set to 'other': it is a latent state-space/jump components time-series model combined with predictive regressions, with no causal design. Section II is tagged 'theory' by rule but functions as econometric design and estimation. Significance language: t-statistics are visible but stars are not explicitly seen. The Internet Appendix holds additional robustness checks. Introduction paragraphs 5-12 are a fragmented roadmap sentence."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 30.8, "usage": {"input": 7832, "output": 3546, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.103042, "request_id": "req_011CggqudMorJJbKh1iaEnRz", "cached": false, "ts": "2026-11-03T23:14:55+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a35d5711921df19327d84ec38609827824c189a4e1a81cabf03a260cb943bf06.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-sectional risk premium (portfolio alpha) of firms in low versus high shipping-cost industries, and the sign of the price of foreign productivity shock risk\",\"strategy\":\"Portfolio sorts on industry shipping costs and weight-to-value, with five-factor Fama-French alphas; conditional sorts on size and profitability; betas on Chinese import growth; checks of earnings-announcement returns and analyst forecast errors; a calibrated Melitz-type GE model\",\"key_assumptions\":[\"Shipping costs capture structural, slow-moving barriers to trade that determine exposure to globalization\",\"Five-factor alphas reflect compensation for priced risk rather than mispricing\",\"Chinese import growth proxies for foreign productivity shocks\"],\"threats_addressed\":[\"loadings on known risk factors\",\"mispricing or a wedge between ex post realized and ex ante expected returns (tested with earnings announcements and analyst forecasts)\",\"sample specificity (European stocks, subperiods)\",\"weighting scheme (value-weighted portfolios)\"],\"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\":true,\"sentence_index\":5}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"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\":\"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\":\"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_sample_period\",\"alternative_specification\",\"sensitivity_to_parameters\"]},\"tables\":{\"convention_notes\":\"Table notes define how shipping costs and weight-to-value are constructed, give the sample (exchanges, SIC range, period), and state that monthly alphas are multiplied by 12 to annualize. Panels split the shipping-cost and weight-to-value sorts. The calibration tables give parameter sources and report simulated moments with Newey-West standard errors.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"mixed\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggesting that\",\"can be consistent with\",\"we hope to stimulate\",\"we leave to future research\",\"may help rationalize\"]},\"claim_evidence\":[{\"claim\":\"Shipping costs vary substantially across industries, persist over time, and correlate negatively with trade flows\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-III\"},{\"claim\":\"The long high-SC, short low-SC portfolio earns about -7% per year, and five-factor alphas are large\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV-V\"},{\"claim\":\"The premium also holds in European stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"No evidence of mispricing from earnings-announcement returns or analyst forecast errors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII-VIII\"},{\"claim\":\"The premium is concentrated among small, less profitable firms, which load negatively on Chinese import growth\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX-X\"},{\"claim\":\"The premium is stronger in industries with high trade elasticity and strong business-stealing effects, as the model predicts\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XI-XII\"},{\"claim\":\"With perfect risk-sharing the model fails; with limited risk-sharing consumption falls after foreign shocks\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-3, Table XIV\"}],\"structural_summary\":\"The introduction moves from the mixed domestic effects of globalization to an asset-pricing lens. It previews the shipping-cost measure, the portfolio results and robustness, and the displacement interpretation, which it frames as a puzzle. It then sketches the model, previews the calibration outcome, positions the paper in four literatures, and closes with a roadmap. Section I builds the measure and estimates the premium, with robustness and mispricing checks folded into the results. Section II develops and calibrates a Melitz-type model and tests its cross-sectional predictions. A short conclusion restates the findings and leaves alternative frictions to future work.\",\"writing_profile_keywords\":[\"portfolio sorts\",\"risk premium\",\"puzzle framing\",\"trade model with asset prices\",\"calibration\",\"limited risk-sharing\",\"shipping costs proxy\"],\"confidence\":0.8,\"notes\":\"The empirical part relies on portfolio sorts with factor alphas; reduced_form_regression is the closest method category. The excerpts do not show how significance is reported in the main tables.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "international_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-sectional risk premium (portfolio alpha) of firms in low versus high shipping-cost industries, and the sign of the price of foreign productivity shock risk", "strategy": "Portfolio sorts on industry shipping costs and weight-to-value, with five-factor Fama-French alphas; conditional sorts on size and profitability; betas on Chinese import growth; checks of earnings-announcement returns and analyst forecast errors; a calibrated Melitz-type GE model", "key_assumptions": ["Shipping costs capture structural, slow-moving barriers to trade that determine exposure to globalization", "Five-factor alphas reflect compensation for priced risk rather than mispricing", "Chinese import growth proxies for foreign productivity shocks"], "threats_addressed": ["loadings on known risk factors", "mispricing or a wedge between ex post realized and ex ante expected returns (tested with earnings announcements and analyst forecasts)", "sample specificity (European stocks, subperiods)", "weighting scheme (value-weighted portfolios)"], "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": true, "sentence_index": 5}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"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": "preview_of_results"}, {"paragraph_index": 4, "move": "tension_or_puzzle"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"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": "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": "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_sample_period", "alternative_specification", "sensitivity_to_parameters"]}, "tables": {"convention_notes": "Table notes define how shipping costs and weight-to-value are constructed, give the sample (exchanges, SIC range, period), and state that monthly alphas are multiplied by 12 to annualize. Panels split the shipping-cost and weight-to-value sorts. The calibration tables give parameter sources and report simulated moments with Newey-West standard errors.", "units_stated_in_notes": true, "standard_errors_description": "mixed"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggesting that", "can be consistent with", "we hope to stimulate", "we leave to future research", "may help rationalize"]}, "claim_evidence": [{"claim": "Shipping costs vary substantially across industries, persist over time, and correlate negatively with trade flows", "evidence_type": "table", "evidence_ref": "Tables I-III"}, {"claim": "The long high-SC, short low-SC portfolio earns about -7% per year, and five-factor alphas are large", "evidence_type": "table", "evidence_ref": "Tables IV-V"}, {"claim": "The premium also holds in European stocks", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "No evidence of mispricing from earnings-announcement returns or analyst forecast errors", "evidence_type": "table", "evidence_ref": "Tables VII-VIII"}, {"claim": "The premium is concentrated among small, less profitable firms, which load negatively on Chinese import growth", "evidence_type": "table", "evidence_ref": "Tables IX-X"}, {"claim": "The premium is stronger in industries with high trade elasticity and strong business-stealing effects, as the model predicts", "evidence_type": "table", "evidence_ref": "Tables XI-XII"}, {"claim": "With perfect risk-sharing the model fails; with limited risk-sharing consumption falls after foreign shocks", "evidence_type": "figure", "evidence_ref": "Figures 1-3, Table XIV"}], "structural_summary": "The introduction moves from the mixed domestic effects of globalization to an asset-pricing lens. It previews the shipping-cost measure, the portfolio results and robustness, and the displacement interpretation, which it frames as a puzzle. It then sketches the model, previews the calibration outcome, positions the paper in four literatures, and closes with a roadmap. Section I builds the measure and estimates the premium, with robustness and mispricing checks folded into the results. Section II develops and calibrates a Melitz-type model and tests its cross-sectional predictions. A short conclusion restates the findings and leaves alternative frictions to future work.", "writing_profile_keywords": ["portfolio sorts", "risk premium", "puzzle framing", "trade model with asset prices", "calibration", "limited risk-sharing", "shipping costs proxy"], "confidence": 0.8, "notes": "The empirical part relies on portfolio sorts with factor alphas; reduced_form_regression is the closest method category. The excerpts do not show how significance is reported in the main tables."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.31, "usage": {"input": 8773, "output": 2352, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082926, "request_id": "req_011CggqUyzLMpPtf6uLRAVKr", "cached": false, "ts": "2026-11-03T23:09:09+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a65cfd8c4ab55eb8cc233c8b12d96619ae6be0dadf2ba58f495cbd9498c0a83b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance (heterogeneous beliefs, speculative bubbles, short selling)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; equilibrium characterization via HJB-type PDE with comparative statics and limiting cases\",\"key_assumptions\":[\"risk-neutral agents with fluctuating heterogeneous beliefs about a Markov state\",\"quadratic costs-of-carry, possibly asymmetric for long and short positions\",\"finite horizon, continuous time, constant interest rate\",\"exogenous asset supply\"],\"threats_addressed\":[],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"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\":3},\"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\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"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\":\"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\":[\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"No tables; results are stated as theorems and comparative statics, with proofs collected in an appendix.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"modal verbs (may, could)\",\"the model suggests\",\"counterfactual framing with 'probably'\",\"noting the setup approximates risk aversion rather than modeling it directly\"]},\"claim_evidence\":[{\"claim\":\"There is a unique equilibrium price function, characterized by a PDE of Hamilton-Jacobi-Bellman type\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Theorem 1, Section I; Appendix F\"},{\"claim\":\"Higher supply lowers the price; cheaper long positions or costlier short positions raise it\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.A; Appendix F.B\"},{\"claim\":\"Increasing costs-of-carry create a delay option that can push the price below the buy-and-hold price even without shorting\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.C; Appendix E\"},{\"claim\":\"The equilibrium coincides with the solution of a limited-instrument planner's problem that maximizes the initial price\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV; Appendix F.D\"},{\"claim\":\"An unexpected drop in shorting costs may deflate a bubble, consistent with the subprime collapse\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Conclusion; Cordell, Huang, and Williams (2011)\"}],\"structural_summary\":\"A theory paper. The introduction opens with historical motivation about bubbles and resale speculation, then contrasts the classic Harrison-Kreps assumptions (no shorting, constant carry costs) with the paper's model, which uses asymmetric quadratic carry costs. It sketches the HJB characterization and previews comparative-statics and planner results. Two paragraphs position the paper against heterogeneous-belief and search-friction literatures, followed by a roadmap. The body proceeds from existence and the PDE, to comparative statics and limiting cases, to resale and delay options illustrated with examples, to a planner interpretation. The conclusion restates the contribution and applies it to synthetic CDOs and the subprime bubble. Extensions and proofs are placed in appendices.\",\"writing_profile_keywords\":[\"continuous-time model\",\"heterogeneous beliefs\",\"short selling\",\"HJB equation\",\"resale option\",\"delay option\",\"comparative statics\",\"planner's problem\",\"proofs in appendix\",\"historical motivation\"],\"confidence\":0.85,\"notes\":\"Purely theoretical paper with no tables or figures. The robustness-like extensions (linear costs, heterogeneous costs, value-based costs) appear in the appendices. The policy-flavored subprime discussion in the conclusion is hedged.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance (heterogeneous beliefs, speculative bubbles, short selling)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; equilibrium characterization via HJB-type PDE with comparative statics and limiting cases", "key_assumptions": ["risk-neutral agents with fluctuating heterogeneous beliefs about a Markov state", "quadratic costs-of-carry, possibly asymmetric for long and short positions", "finite horizon, continuous time, constant interest rate", "exogenous asset supply"], "threats_addressed": [], "causal_language_strength": 2, "uses_causal_claim": false}, "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": 3}, "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": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "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": "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": ["alternative_specification", "other"]}, "tables": {"convention_notes": "No tables; results are stated as theorems and comparative statics, with proofs collected in an appendix.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["modal verbs (may, could)", "the model suggests", "counterfactual framing with 'probably'", "noting the setup approximates risk aversion rather than modeling it directly"]}, "claim_evidence": [{"claim": "There is a unique equilibrium price function, characterized by a PDE of Hamilton-Jacobi-Bellman type", "evidence_type": "equation_or_proof", "evidence_ref": "Theorem 1, Section I; Appendix F"}, {"claim": "Higher supply lowers the price; cheaper long positions or costlier short positions raise it", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.A; Appendix F.B"}, {"claim": "Increasing costs-of-carry create a delay option that can push the price below the buy-and-hold price even without shorting", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.C; Appendix E"}, {"claim": "The equilibrium coincides with the solution of a limited-instrument planner's problem that maximizes the initial price", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV; Appendix F.D"}, {"claim": "An unexpected drop in shorting costs may deflate a bubble, consistent with the subprime collapse", "evidence_type": "prior_literature", "evidence_ref": "Conclusion; Cordell, Huang, and Williams (2011)"}], "structural_summary": "A theory paper. The introduction opens with historical motivation about bubbles and resale speculation, then contrasts the classic Harrison-Kreps assumptions (no shorting, constant carry costs) with the paper's model, which uses asymmetric quadratic carry costs. It sketches the HJB characterization and previews comparative-statics and planner results. Two paragraphs position the paper against heterogeneous-belief and search-friction literatures, followed by a roadmap. The body proceeds from existence and the PDE, to comparative statics and limiting cases, to resale and delay options illustrated with examples, to a planner interpretation. The conclusion restates the contribution and applies it to synthetic CDOs and the subprime bubble. Extensions and proofs are placed in appendices.", "writing_profile_keywords": ["continuous-time model", "heterogeneous beliefs", "short selling", "HJB equation", "resale option", "delay option", "comparative statics", "planner's problem", "proofs in appendix", "historical motivation"], "confidence": 0.85, "notes": "Purely theoretical paper with no tables or figures. The robustness-like extensions (linear costs, heterogeneous costs, value-based costs) appear in the appendices. The policy-flavored subprime discussion in the conclusion is hedged."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.22, "usage": {"input": 6742, "output": 2065, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069062, "request_id": "req_011CggqkupG6grN4bUB7DvWi", "cached": false, "ts": "2026-11-03T23:12:42+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/a7ad10f0ce1a66b55d0ca5ddc23ed62f9a2ac9b011fb8bf1cc6a66042463c698.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of issuer-paid rating optimism (S&P rating above investor-paid EJR rating) on secondary-market bond spreads and on the probability of choosing private over public debt\",\"strategy\":\"Validation of the overrating measure through default and incentive-conflict regressions, hazard models and a DiD on broker mergers/closures. Symmetric one-notch split-rating comparisons of bond spreads. 2SLS using analyst coverage shocks. Becker-Ivashina conditional-issuance approach to debt choice. Spread changes around private debt issuances.\",\"key_assumptions\":[\"Without inflation, S&P and EJR ratings are equally informative about default\",\"Brokerage mergers/closures exogenously raise S&P's leeway to inflate ratings\",\"Conditioning on new debt issuance holds the firm's demand for debt fixed\",\"Rating differences reflect incentive conflicts rather than structural methodological differences\"],\"threats_addressed\":[\"Structural differences between agencies' rating methods\",\"Selection in primary-market pricing, addressed by using secondary markets\",\"Endogeneity of ratings inflation, addressed by the exogenous analyst-coverage shock\",\"Confounding of demand and supply in debt choice\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"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\":1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3},\"other\":null},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":9,\"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\":\"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_measure\"]},\"tables\":{\"convention_notes\":\"Table notes state the model type (linear probability, OLS, Cox hazard, 2SLS second stage), the dependent variable definition, the sample panel and period (1999-2012), and event windows. Some coefficients are rescaled by 100 for exposition. Table I and Table AI tabulate model outcomes.\",\"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\":[\"results suggest\",\"consistent with\",\"evidence is more consistent with\",\"may lead\"]},\"claim_evidence\":[{\"claim\":\"S&P overrating is more frequent for issuers contributing more rating revenue and with stronger rating-shopping incentives\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"S&P converges toward EJR faster when it was initially more optimistic, indicating lower timeliness\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V-VI\"},{\"claim\":\"Loss of analyst coverage from broker mergers or closures raises the likelihood of S&P overrating\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Bonds of firms S&P overrates carry higher spreads than symmetric split-rated counterparts\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-IX\"},{\"claim\":\"Overrated firms are more likely to issue private debt than public bonds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"The overrating spread premium shrinks after private debt issuance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XI-XII\"},{\"claim\":\"Inflation in issuer-paid ratings pools risks and worsens information asymmetry even when investors price it\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I, Appendix A\"}],\"structural_summary\":\"Opens with the role of rating agencies and prior evidence on issuer-pay bias. A simple model (proofs in the appendix) motivates the tests. A long validation section establishes S&P-minus-EJR overrating as a quality measure, using default regressions, conflict proxies, convergence dynamics and a broker-merger DiD. The paper then tests bond-spread pricing via symmetric split ratings and 2SLS, followed by debt-choice substitution and spread changes after private debt issuance. A short conclusion restates the findings.\",\"writing_profile_keywords\":[\"model-motivated empirics\",\"measure validation\",\"split ratings\",\"quasi-natural experiment\",\"debt choice\",\"lemons discount\"],\"confidence\":0.8,\"notes\":\"Standard error type and significance notation are not visible in the excerpts. Robustness checks with alternative quality measures are mentioned only briefly in the introduction.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of issuer-paid rating optimism (S&P rating above investor-paid EJR rating) on secondary-market bond spreads and on the probability of choosing private over public debt", "strategy": "Validation of the overrating measure through default and incentive-conflict regressions, hazard models and a DiD on broker mergers/closures. Symmetric one-notch split-rating comparisons of bond spreads. 2SLS using analyst coverage shocks. Becker-Ivashina conditional-issuance approach to debt choice. Spread changes around private debt issuances.", "key_assumptions": ["Without inflation, S&P and EJR ratings are equally informative about default", "Brokerage mergers/closures exogenously raise S&P's leeway to inflate ratings", "Conditioning on new debt issuance holds the firm's demand for debt fixed", "Rating differences reflect incentive conflicts rather than structural methodological differences"], "threats_addressed": ["Structural differences between agencies' rating methods", "Selection in primary-market pricing, addressed by using secondary markets", "Endogeneity of ratings inflation, addressed by the exogenous analyst-coverage shock", "Confounding of demand and supply in debt choice"], "causal_language_strength": 3, "uses_causal_claim": true}, "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": 1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}, "other": null}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"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": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "related_literature_positioning"}], "contribution_paragraph_index": 9, "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": "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_measure"]}, "tables": {"convention_notes": "Table notes state the model type (linear probability, OLS, Cox hazard, 2SLS second stage), the dependent variable definition, the sample panel and period (1999-2012), and event windows. Some coefficients are rescaled by 100 for exposition. Table I and Table AI tabulate model outcomes.", "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": ["results suggest", "consistent with", "evidence is more consistent with", "may lead"]}, "claim_evidence": [{"claim": "S&P overrating is more frequent for issuers contributing more rating revenue and with stronger rating-shopping incentives", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "S&P converges toward EJR faster when it was initially more optimistic, indicating lower timeliness", "evidence_type": "table", "evidence_ref": "Tables V-VI"}, {"claim": "Loss of analyst coverage from broker mergers or closures raises the likelihood of S&P overrating", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Bonds of firms S&P overrates carry higher spreads than symmetric split-rated counterparts", "evidence_type": "table", "evidence_ref": "Tables VIII-IX"}, {"claim": "Overrated firms are more likely to issue private debt than public bonds", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "The overrating spread premium shrinks after private debt issuance", "evidence_type": "table", "evidence_ref": "Tables XI-XII"}, {"claim": "Inflation in issuer-paid ratings pools risks and worsens information asymmetry even when investors price it", "evidence_type": "equation_or_proof", "evidence_ref": "Section I, Appendix A"}], "structural_summary": "Opens with the role of rating agencies and prior evidence on issuer-pay bias. A simple model (proofs in the appendix) motivates the tests. A long validation section establishes S&P-minus-EJR overrating as a quality measure, using default regressions, conflict proxies, convergence dynamics and a broker-merger DiD. The paper then tests bond-spread pricing via symmetric split ratings and 2SLS, followed by debt-choice substitution and spread changes after private debt issuance. A short conclusion restates the findings.", "writing_profile_keywords": ["model-motivated empirics", "measure validation", "split ratings", "quasi-natural experiment", "debt choice", "lemons discount"], "confidence": 0.8, "notes": "Standard error type and significance notation are not visible in the excerpts. Robustness checks with alternative quality measures are mentioned only briefly in the introduction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.76, "usage": {"input": 9214, "output": 2253, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08271, "request_id": "req_011CggqWjkPiLmRJbaqij4mP", "cached": false, "ts": "2026-11-03T23:09:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/abfd93bd40dbe8e9ee87d0e9fe106d2fe250a52d78a9c7f5e114ecdfd4365a0a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model); comparative statics of intermediary size, risk and firm leverage with respect to safe asset demand and QE\",\"strategy\":\"general equilibrium model; social planner allocation shown to be replicated by a competitive equilibrium; comparative statics\",\"key_assumptions\":[\"agency cost of intermediary risk bearing\",\"household utility from riskless assets\",\"exogenous Lucas-tree output with aggregate and idiosyncratic risk\",\"nominal rigidities and zero lower bound for the QE-at-ZLB analysis\"],\"threats_addressed\":[],\"causal_language_strength\":3,\"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\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"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\"],\"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\":\"conclusion\"},{\"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\":\"No tables. Results are conveyed through propositions, equations and four illustrative figures, one of which shows empirical balance sheets.\",\"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\":[\"'under some conditions'\",\"'can decrease'\",\"'broadly consistent with'\",\"'may provide additional insights'\"]},\"claim_evidence\":[{\"claim\":\"Intermediaries hold diversified risky debt while households hold riskless assets and equities, broadly matching US bank and household balance sheets\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Higher safe asset demand leads to larger, riskier intermediaries and more levered nonfinancial firms\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4; Section II.A\"},{\"claim\":\"QE reduces intermediary riskiness, consistent with empirical evidence\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Chodorow-Reich (2014)\"},{\"claim\":\"Debt and equity are priced by different kernels, with intermediaries charging a higher price of systematic risk\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 2; Figure 3\"},{\"claim\":\"The competitive equilibrium replicates the social planner allocation\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.C; Appendix A\"}],\"structural_summary\":\"Opens by motivating the topic with empirical evidence on the safe-asset premium. The introduction then sketches the model mechanism: intermediaries face an agency cost of bearing risk and back deposits with diversified firm debt. It previews comparative statics on safe asset demand and QE, linking each to prior empirical findings, and closes with a roadmap. The body presents a baseline two-period model, solves the planner problem and shows decentralization. Policy applications follow (the subprime boom, QE in equities versus debt, QE at the zero lower bound), then a short section on asset pricing predictions checked against existing evidence. The conclusion summarizes and lists future research directions. Proofs are in the appendix.\",\"writing_profile_keywords\":[\"general equilibrium\",\"safe assets\",\"intermediary agency cost\",\"market segmentation\",\"quantitative easing\",\"planner decentralization\",\"policy application\"],\"confidence\":0.85,\"notes\":\"Pure theory paper. Model predictions are tied to empirical facts through citations and one descriptive figure; no new estimation is performed.\"}", "json": {"genre": "theoretical", "subfield": "financial_intermediation_banking", "subfield_secondary": "macro_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model); comparative statics of intermediary size, risk and firm leverage with respect to safe asset demand and QE", "strategy": "general equilibrium model; social planner allocation shown to be replicated by a competitive equilibrium; comparative statics", "key_assumptions": ["agency cost of intermediary risk bearing", "household utility from riskless assets", "exogenous Lucas-tree output with aggregate and idiosyncratic risk", "nominal rigidities and zero lower bound for the QE-at-ZLB analysis"], "threats_addressed": [], "causal_language_strength": 3, "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": "preview_of_approach"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"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": "mechanism_or_theory_sketch"}, {"paragraph_index": 4, "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"], "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": "conclusion"}, {"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": "No tables. Results are conveyed through propositions, equations and four illustrative figures, one of which shows empirical balance sheets.", "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": ["'under some conditions'", "'can decrease'", "'broadly consistent with'", "'may provide additional insights'"]}, "claim_evidence": [{"claim": "Intermediaries hold diversified risky debt while households hold riskless assets and equities, broadly matching US bank and household balance sheets", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Higher safe asset demand leads to larger, riskier intermediaries and more levered nonfinancial firms", "evidence_type": "figure", "evidence_ref": "Figure 4; Section II.A"}, {"claim": "QE reduces intermediary riskiness, consistent with empirical evidence", "evidence_type": "prior_literature", "evidence_ref": "Chodorow-Reich (2014)"}, {"claim": "Debt and equity are priced by different kernels, with intermediaries charging a higher price of systematic risk", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 2; Figure 3"}, {"claim": "The competitive equilibrium replicates the social planner allocation", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.C; Appendix A"}], "structural_summary": "Opens by motivating the topic with empirical evidence on the safe-asset premium. The introduction then sketches the model mechanism: intermediaries face an agency cost of bearing risk and back deposits with diversified firm debt. It previews comparative statics on safe asset demand and QE, linking each to prior empirical findings, and closes with a roadmap. The body presents a baseline two-period model, solves the planner problem and shows decentralization. Policy applications follow (the subprime boom, QE in equities versus debt, QE at the zero lower bound), then a short section on asset pricing predictions checked against existing evidence. The conclusion summarizes and lists future research directions. Proofs are in the appendix.", "writing_profile_keywords": ["general equilibrium", "safe assets", "intermediary agency cost", "market segmentation", "quantitative easing", "planner decentralization", "policy application"], "confidence": 0.85, "notes": "Pure theory paper. Model predictions are tied to empirical facts through citations and one descriptive figure; no new estimation is performed."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.66, "usage": {"input": 5307, "output": 1818, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.058382, "request_id": "req_011Cggr1gWVWEqzHKu6HntmR", "cached": false, "ts": "2026-11-03T23:16:01+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ad2b078317f8b54857863ed2aa97503df085fec159a82397761a9c97eb2644e2.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-sectional risk prices (lambda) on exposures (betas) to multi-quarter capital share growth across portfolio cross-sections\",\"strategy\":\"Two-pass Fama-MacBeth expected return-beta regressions using long-horizon (H-quarter) capital share betas, with horse races against Fama-French and intermediary factors; household-level income-share proxy as an alternative SDF\",\"key_assumptions\":[\"Linearized SDF in capital share growth (and consumption) is an adequate representation\",\"Long-horizon betas capture the relevant risk exposure\",\"Mimicking income-share factor based on capital share proxies top-wealth consumption growth\"],\"threats_addressed\":[\"Explanatory power could reflect aggregate consumption risk rather than capital share risk\",\"Overlap with existing pricing factors (SMB, HML, intermediary factors)\",\"Estimation uncertainty in two-pass regressions, handled via bootstrapped confidence intervals\"],\"causal_language_strength\":2,\"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\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"other\"},{\"paragraph_index\":6,\"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\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"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\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Risk-price tables share boilerplate notes: estimates scaled by 100, bootstrapped 95% confidence intervals in square brackets, sample period stated, and the cross-sectional regression equation reproduced. Fit is reported via adjusted R2, RMSE/RMSR and BIC. The preliminary income-share regressions instead use OLS t-statistics with significance stars.\",\"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\":[\"'consistent with' / 'commensurate with' framing for theory links\",\"explicit acknowledgment of portfolios the factor fails to price\",\"'our results suggest' phrasing\"]},\"claim_evidence\":[{\"claim\":\"Stock wealth is highly concentrated, with the top 5% holding most of the market value\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Top stock-wealth groups' income shares co-move with the capital share\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Capital share betas explain average returns across equity characteristic portfolios with high fit\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Table III\"},{\"claim\":\"Capital share exposure subsumes SMB and HML in pricing equity portfolios\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Capital share risk competes favorably with intermediary factors for equities and other asset classes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V-VI\"},{\"claim\":\"A top-10% consumption proxy from household data has similar explanatory power\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"The factor fails to price industry, FX, commodity and momentum portfolios\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Introduction paragraph 5\"}],\"structural_summary\":\"The introduction opens with the search for an empirically relevant SDF and immediately states the headline finding. It then challenges representative-agent pricing using wealth-concentration facts and sketches the redistributive-shock mechanism. Results are previewed via a figure, the paper is positioned against the limited-participation and redistributive-risk literatures, and the authors openly state the factor's failures before giving a roadmap. The body proceeds through data with preliminary distributional evidence, the econometric SDF/beta setup, and a results section of sequential horse races, beta spreads and a household-data SDF. A short conclusion restates the findings and their theoretical consistency.\",\"writing_profile_keywords\":[\"single-factor asset pricing\",\"capital share\",\"Fama-MacBeth\",\"bootstrapped confidence intervals\",\"horse race against existing factors\",\"heterogeneous-agent motivation\",\"candid limitations\"],\"confidence\":0.8,\"notes\":\"The Econometric Model section is assigned the empirical_design role rather than theory because it sets up the estimation framework. Figure uncertainty is marked unclear because the captions do not mention error bands. The 'other' move in paragraph 5 marks a discussion of the factor's limitations.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-sectional risk prices (lambda) on exposures (betas) to multi-quarter capital share growth across portfolio cross-sections", "strategy": "Two-pass Fama-MacBeth expected return-beta regressions using long-horizon (H-quarter) capital share betas, with horse races against Fama-French and intermediary factors; household-level income-share proxy as an alternative SDF", "key_assumptions": ["Linearized SDF in capital share growth (and consumption) is an adequate representation", "Long-horizon betas capture the relevant risk exposure", "Mimicking income-share factor based on capital share proxies top-wealth consumption growth"], "threats_addressed": ["Explanatory power could reflect aggregate consumption risk rather than capital share risk", "Overlap with existing pricing factors (SMB, HML, intermediary factors)", "Estimation uncertainty in two-pass regressions, handled via bootstrapped confidence intervals"], "causal_language_strength": 2, "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": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"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": "related_literature_positioning"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "other"}, {"paragraph_index": 6, "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": "data"}, {"idx": 2, "role": "empirical_design"}, {"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": "within_results", "types": ["alternative_specification", "alternative_measure", "subsample"]}, "tables": {"convention_notes": "Risk-price tables share boilerplate notes: estimates scaled by 100, bootstrapped 95% confidence intervals in square brackets, sample period stated, and the cross-sectional regression equation reproduced. Fit is reported via adjusted R2, RMSE/RMSR and BIC. The preliminary income-share regressions instead use OLS t-statistics with significance stars.", "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": ["'consistent with' / 'commensurate with' framing for theory links", "explicit acknowledgment of portfolios the factor fails to price", "'our results suggest' phrasing"]}, "claim_evidence": [{"claim": "Stock wealth is highly concentrated, with the top 5% holding most of the market value", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Top stock-wealth groups' income shares co-move with the capital share", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Capital share betas explain average returns across equity characteristic portfolios with high fit", "evidence_type": "figure", "evidence_ref": "Figure 1; Table III"}, {"claim": "Capital share exposure subsumes SMB and HML in pricing equity portfolios", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Capital share risk competes favorably with intermediary factors for equities and other asset classes", "evidence_type": "table", "evidence_ref": "Tables V-VI"}, {"claim": "A top-10% consumption proxy from household data has similar explanatory power", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "The factor fails to price industry, FX, commodity and momentum portfolios", "evidence_type": "none_stated", "evidence_ref": "Introduction paragraph 5"}], "structural_summary": "The introduction opens with the search for an empirically relevant SDF and immediately states the headline finding. It then challenges representative-agent pricing using wealth-concentration facts and sketches the redistributive-shock mechanism. Results are previewed via a figure, the paper is positioned against the limited-participation and redistributive-risk literatures, and the authors openly state the factor's failures before giving a roadmap. The body proceeds through data with preliminary distributional evidence, the econometric SDF/beta setup, and a results section of sequential horse races, beta spreads and a household-data SDF. A short conclusion restates the findings and their theoretical consistency.", "writing_profile_keywords": ["single-factor asset pricing", "capital share", "Fama-MacBeth", "bootstrapped confidence intervals", "horse race against existing factors", "heterogeneous-agent motivation", "candid limitations"], "confidence": 0.8, "notes": "The Econometric Model section is assigned the empirical_design role rather than theory because it sets up the estimation framework. Figure uncertainty is marked unclear because the captions do not mention error bands. The 'other' move in paragraph 5 marks a discussion of the factor's limitations."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.68, "usage": {"input": 6192, "output": 2135, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068262, "request_id": "req_011CggqQTyCQqugpGngRYxWr", "cached": false, "ts": "2026-11-03T23:08:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b08414a5d5c5f032f4057e9357372be0dee529718bf1344baecaa1b956922ead.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance (misallocation, real effects of mispricing, lumpy investment)\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Aggregate real efficiency (value) losses from investment distortions attributable to cross-sectional alpha wedges, measured as a counterfactual gain from eliminating alphas\",\"strategy\":\"Estimate a dynamic lumpy-investment model by GMM on cross-sectional moments, with subjective-belief wedges disciplined by empirical alphas, then compare it with a counterfactual economy without alphas\",\"key_assumptions\":[\"Alphas from a postulated benchmark asset pricing model measure informational inefficiencies (joint-hypothesis caveat)\",\"Firms maximize market value under prevailing distorted subjective beliefs\",\"Model structure (stochastic lumpy adjustment, alpha process dynamics) is correctly specified\",\"Macroeconomic parameters calibrated from the literature\"],\"threats_addressed\":[\"joint-hypothesis problem / alternative view that alphas reflect omitted risk factors\",\"general equilibrium effects\",\"independence and persistence of the alpha process\",\"role of adjustment frictions\",\"cash-raising channel for the asset growth–alpha relation\"],\"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\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"motivation\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"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\":\"results\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"empirical_design\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"other\"},{\"idx\":8,\"role\":\"robustness\"},{\"idx\":9,\"role\":\"other\"},{\"idx\":10,\"role\":\"conclusion\"},{\"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\"},{\"idx\":21,\"role\":\"appendix\"},{\"idx\":22,\"role\":\"appendix\"},{\"idx\":23,\"role\":\"appendix\"},{\"idx\":24,\"role\":\"appendix\"},{\"idx\":25,\"role\":\"appendix\"},{\"idx\":26,\"role\":\"appendix\"},{\"idx\":27,\"role\":\"appendix\"},{\"idx\":28,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"confidence_intervals\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Only two tables are shown. The first reports decile portfolio returns and CAPM alphas for the value and investment anomalies. The second lists calibrated macro parameters and GMM-estimated firm parameters, with normalization notes. Most quantitative results are delivered through figures.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'can'/'may' modal verbs\",\"'raises the possibility'\",\"conditional framing ('if alphas reflect inefficiencies', 'conditional on the view')\",\"'potential' gains\",\"'plausibly'\",\"explicit disclaimer of being a first step rather than the final word\"]},\"claim_evidence\":[{\"claim\":\"Value and investment anomalies replicate over 1975–2014 consistent with prior literature\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Estimated model matches targeted moments on size, Tobin's q, value premium and investment\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Model replicates untargeted investment-alpha relation\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Aggregate efficiency gains from eliminating alphas can be substantial and rise with the value spread\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Gains are concentrated among firms with particular Tobin's q levels\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"Alpha persistence strongly affects aggregate gains, so alpha magnitude alone is a poor indicator\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6\"},{\"claim\":\"Cash-raising channel does not explain the asset growth–alpha relation\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Internet Appendix (referenced in intro)\"},{\"claim\":\"Alphas alone are uninformative about real distortions\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I one-period example\"}],\"structural_summary\":\"The paper opens with the anomalies literature and asks how large the real losses are if alphas reflect inefficiency. It positions itself against misallocation work in macro and acknowledges the joint-hypothesis issue. A one-period example motivates the argument, followed by reduced-form replication of the value and investment anomalies. The core is a continuous-time lumpy-investment model with closed-form distributions, estimated by GMM. Counterfactuals then remove the alpha wedges to quantify efficiency gains. Later sections give an efficient-markets reinterpretation, robustness discussions of modeling assumptions, and implications for ranking anomalies. The conclusion is hedged and lists extensions.\",\"writing_profile_keywords\":[\"structural model\",\"counterfactual\",\"GMM\",\"lumpy investment\",\"joint-hypothesis caveat\",\"misallocation\",\"figure-heavy results\",\"hedged quantification\"],\"confidence\":0.8,\"notes\":\"Section idx 6 ('prd') is a parsing artifact that holds the continuation of the counterfactual analysis. Several appendix headings are equation fragments. The roadmap refers to 'Section X' for the conclusion, but the conclusion is numbered IX. Model-fit figures show 95% bounds, while the counterfactual figures do not appear to show uncertainty.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance (misallocation, real effects of mispricing, lumpy investment)", "primary_method": "structural_estimation", "identification": {"estimand": "Aggregate real efficiency (value) losses from investment distortions attributable to cross-sectional alpha wedges, measured as a counterfactual gain from eliminating alphas", "strategy": "Estimate a dynamic lumpy-investment model by GMM on cross-sectional moments, with subjective-belief wedges disciplined by empirical alphas, then compare it with a counterfactual economy without alphas", "key_assumptions": ["Alphas from a postulated benchmark asset pricing model measure informational inefficiencies (joint-hypothesis caveat)", "Firms maximize market value under prevailing distorted subjective beliefs", "Model structure (stochastic lumpy adjustment, alpha process dynamics) is correctly specified", "Macroeconomic parameters calibrated from the literature"], "threats_addressed": ["joint-hypothesis problem / alternative view that alphas reflect omitted risk factors", "general equilibrium effects", "independence and persistence of the alpha process", "role of adjustment frictions", "cash-raising channel for the asset growth–alpha relation"], "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": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "motivation"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "roadmap"}], "contribution_paragraph_index": 4, "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": "results"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "empirical_design"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "other"}, {"idx": 8, "role": "robustness"}, {"idx": 9, "role": "other"}, {"idx": 10, "role": "conclusion"}, {"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"}, {"idx": 21, "role": "appendix"}, {"idx": 22, "role": "appendix"}, {"idx": 23, "role": "appendix"}, {"idx": 24, "role": "appendix"}, {"idx": 25, "role": "appendix"}, {"idx": 26, "role": "appendix"}, {"idx": 27, "role": "appendix"}, {"idx": 28, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "confidence_intervals", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters", "alternative_specification", "other"]}, "tables": {"convention_notes": "Only two tables are shown. The first reports decile portfolio returns and CAPM alphas for the value and investment anomalies. The second lists calibrated macro parameters and GMM-estimated firm parameters, with normalization notes. Most quantitative results are delivered through figures.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'can'/'may' modal verbs", "'raises the possibility'", "conditional framing ('if alphas reflect inefficiencies', 'conditional on the view')", "'potential' gains", "'plausibly'", "explicit disclaimer of being a first step rather than the final word"]}, "claim_evidence": [{"claim": "Value and investment anomalies replicate over 1975–2014 consistent with prior literature", "evidence_type": "table", "evidence_ref": "Table I"}, {"claim": "Estimated model matches targeted moments on size, Tobin's q, value premium and investment", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Model replicates untargeted investment-alpha relation", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Aggregate efficiency gains from eliminating alphas can be substantial and rise with the value spread", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Gains are concentrated among firms with particular Tobin's q levels", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "Alpha persistence strongly affects aggregate gains, so alpha magnitude alone is a poor indicator", "evidence_type": "figure", "evidence_ref": "Figure 6"}, {"claim": "Cash-raising channel does not explain the asset growth–alpha relation", "evidence_type": "none_stated", "evidence_ref": "Internet Appendix (referenced in intro)"}, {"claim": "Alphas alone are uninformative about real distortions", "evidence_type": "equation_or_proof", "evidence_ref": "Section I one-period example"}], "structural_summary": "The paper opens with the anomalies literature and asks how large the real losses are if alphas reflect inefficiency. It positions itself against misallocation work in macro and acknowledges the joint-hypothesis issue. A one-period example motivates the argument, followed by reduced-form replication of the value and investment anomalies. The core is a continuous-time lumpy-investment model with closed-form distributions, estimated by GMM. Counterfactuals then remove the alpha wedges to quantify efficiency gains. Later sections give an efficient-markets reinterpretation, robustness discussions of modeling assumptions, and implications for ranking anomalies. The conclusion is hedged and lists extensions.", "writing_profile_keywords": ["structural model", "counterfactual", "GMM", "lumpy investment", "joint-hypothesis caveat", "misallocation", "figure-heavy results", "hedged quantification"], "confidence": 0.8, "notes": "Section idx 6 ('prd') is a parsing artifact that holds the continuation of the counterfactual analysis. Several appendix headings are equation fragments. The roadmap refers to 'Section X' for the conclusion, but the conclusion is numbered IX. Model-fit figures show 95% bounds, while the counterfactual figures do not appear to show uncertainty."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 21.76, "usage": {"input": 10062, "output": 2815, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.097342, "request_id": "req_011CggqPjU63RJeaRiqxi2n4", "cached": false, "ts": "2026-11-03T23:08:00+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b52d80ce5109b2e900d2f54a5b72d3df9b08820aa9447c8b98e3b39b43da8e1c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"asset_pricing (mutual fund flows, sustainable investing)\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Causal effect of being assigned a given Morningstar sustainability globe rating on subsequent mutual fund net flows, and on liquidation.\",\"strategy\":\"Quasi-exogenous shock from the March 2016 publication of Morningstar globe ratings. Post-publication flows are compared across globe categories with time fixed effects. This is supplemented by regression discontinuity at globe breakpoints, nearest-neighbor matching on pre-period characteristics, and a placebo test, plus a lab/MTurk experiment on mechanisms.\",\"key_assumptions\":[\"Ratings repackage public information and do not change fund fundamentals.\",\"Funds in different globe categories were not systematically different in pre-publication flows.\",\"Within-category construction makes ratings largely uncorrelated with investment style.\",\"Continuity of fund characteristics around the percentile breakpoints for the RD.\"],\"threats_addressed\":[\"Pre-existing differences across globe categories (pre-period plot, matching, controlling for preperiod)\",\"Investors responding to continuous sustainability information rather than the discrete rating (percentile and score controls, RD)\",\"Placebo timing\",\"Institutional constraints as an alternative explanation\",\"Rational return expectations (performance tests)\"],\"causal_language_strength\":5,\"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\":0},\"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\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"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\":\"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\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"placebo\",\"alternative_specification\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Notes open by stating what the table shows. They then specify the dependent variable and regressors (globe dummies, percentile, score), panel structure (value- vs equal-weighted; MBA vs MTurk), and estimators (conventional vs bias-corrected RD; nearest-neighbor matching). Standard-error details are truncated in the excerpts.\",\"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\":\"moderate\",\"typical_devices\":[\"'consistent with' interpretations\",\"'suggests' for mechanism evidence\",\"explicit statement that mechanism hypotheses cannot be definitively distinguished\",\"'one interpretation is'\"],\"typical_devices_note\":\"\"},\"claim_evidence\":[{\"claim\":\"Investors marketwide value sustainability: low-globe funds lost over $12B in flows, high-globe funds gained over $24B.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Section III.E\"},{\"claim\":\"Funds did not differ systematically in flows before publication of the ratings.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1; Table V\"},{\"claim\":\"Investors respond to the discrete globe categories, mainly the extremes, rather than to percentile ranks or raw scores.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table IV, Figure 4\"},{\"claim\":\"High-sustainability funds do not outperform low-sustainability funds.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Experimental subjects expect higher returns and lower risk from higher globe ratings, consistent with affect.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6; Table IX\"},{\"claim\":\"Subjects considering environmental or social factors allocate more to five-globe funds than their expectations explain.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The introduction frames whether the average investor values sustainability and immediately states the causal answer. It then defends the Morningstar rating launch as a quasi-exogenous, information-free shock and previews three sets of results: flow results, discrete/extreme-category salience results, and experimental mechanism results. Literature positioning sits inside the introduction, followed by a roadmap. The body describes the rating system and the proprietary data. A large section establishes the flow effects through attention evidence, base regressions, within-globe RD, pre-period matching and economic magnitudes. A second section tests institutional, rational-return and nonpecuniary/affect explanations, using share-class splits, performance regressions and an MBA/MTurk experiment. A one-paragraph conclusion restates the findings.\",\"writing_profile_keywords\":[\"natural experiment\",\"salience\",\"fund flows\",\"ESG ratings\",\"causal framing\",\"mechanism decomposition\",\"experiment complement\",\"dollar magnitudes\"],\"confidence\":0.85,\"notes\":\"The format of statistical-significance reporting is not visible in the truncated table notes. Robustness checks (matching, placebo, controlling for the preperiod) are embedded as subsections of the results section. Hedging is uneven: the main causal result is stated strongly, while mechanism claims are explicitly hedged.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "asset_pricing (mutual fund flows, sustainable investing)", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Causal effect of being assigned a given Morningstar sustainability globe rating on subsequent mutual fund net flows, and on liquidation.", "strategy": "Quasi-exogenous shock from the March 2016 publication of Morningstar globe ratings. Post-publication flows are compared across globe categories with time fixed effects. This is supplemented by regression discontinuity at globe breakpoints, nearest-neighbor matching on pre-period characteristics, and a placebo test, plus a lab/MTurk experiment on mechanisms.", "key_assumptions": ["Ratings repackage public information and do not change fund fundamentals.", "Funds in different globe categories were not systematically different in pre-publication flows.", "Within-category construction makes ratings largely uncorrelated with investment style.", "Continuity of fund characteristics around the percentile breakpoints for the RD."], "threats_addressed": ["Pre-existing differences across globe categories (pre-period plot, matching, controlling for preperiod)", "Investors responding to continuous sustainability information rather than the discrete rating (percentile and score controls, RD)", "Placebo timing", "Institutional constraints as an alternative explanation", "Rational return expectations (performance tests)"], "causal_language_strength": 5, "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": 0}, "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": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "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": "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": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["placebo", "alternative_specification", "alternative_measure"]}, "tables": {"convention_notes": "Notes open by stating what the table shows. They then specify the dependent variable and regressors (globe dummies, percentile, score), panel structure (value- vs equal-weighted; MBA vs MTurk), and estimators (conventional vs bias-corrected RD; nearest-neighbor matching). Standard-error details are truncated in the excerpts.", "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": "moderate", "typical_devices": ["'consistent with' interpretations", "'suggests' for mechanism evidence", "explicit statement that mechanism hypotheses cannot be definitively distinguished", "'one interpretation is'"], "typical_devices_note": ""}, "claim_evidence": [{"claim": "Investors marketwide value sustainability: low-globe funds lost over $12B in flows, high-globe funds gained over $24B.", "evidence_type": "table", "evidence_ref": "Table III; Section III.E"}, {"claim": "Funds did not differ systematically in flows before publication of the ratings.", "evidence_type": "figure", "evidence_ref": "Figure 1; Table V"}, {"claim": "Investors respond to the discrete globe categories, mainly the extremes, rather than to percentile ranks or raw scores.", "evidence_type": "table", "evidence_ref": "Table III, Table IV, Figure 4"}, {"claim": "High-sustainability funds do not outperform low-sustainability funds.", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Experimental subjects expect higher returns and lower risk from higher globe ratings, consistent with affect.", "evidence_type": "figure", "evidence_ref": "Figure 6; Table IX"}, {"claim": "Subjects considering environmental or social factors allocate more to five-globe funds than their expectations explain.", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The introduction frames whether the average investor values sustainability and immediately states the causal answer. It then defends the Morningstar rating launch as a quasi-exogenous, information-free shock and previews three sets of results: flow results, discrete/extreme-category salience results, and experimental mechanism results. Literature positioning sits inside the introduction, followed by a roadmap. The body describes the rating system and the proprietary data. A large section establishes the flow effects through attention evidence, base regressions, within-globe RD, pre-period matching and economic magnitudes. A second section tests institutional, rational-return and nonpecuniary/affect explanations, using share-class splits, performance regressions and an MBA/MTurk experiment. A one-paragraph conclusion restates the findings.", "writing_profile_keywords": ["natural experiment", "salience", "fund flows", "ESG ratings", "causal framing", "mechanism decomposition", "experiment complement", "dollar magnitudes"], "confidence": 0.85, "notes": "The format of statistical-significance reporting is not visible in the truncated table notes. Robustness checks (matching, placebo, controlling for the preperiod) are embedded as subsections of the results section. Hedging is uneven: the main causal result is stated strongly, while mechanism claims are explicitly hedged."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 23.65, "usage": {"input": 7356, "output": 2462, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079458, "request_id": "req_011CggqYL4ASoYxTq9jTgLev", "cached": false, "ts": "2026-11-03T23:09:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b5f17c509d784f17eefaa6904dccef60456c165983dad4d3d6c2731402242574.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"corporate_finance (optimal contracting, firm dynamics)\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Association between industry-level idiosyncratic volatility and the average firm-level capital-income-to-sales ratio; model-implied effect of volatility and rents on aggregate versus average capital share.\",\"strategy\":\"Stylized facts from Compustat/CRSP, a calibrated industry equilibrium model with entry and exit and worker insurance, comparative statics across stationary equilibria, and within-industry regressions of capital share on idiosyncratic volatility.\",\"key_assumptions\":[\"Shareholders are diversified while workers are risk-averse, so optimal contracts insure workers.\",\"Comparisons are between stationary equilibria rather than transition dynamics.\",\"Compustat public firms are representative for the mechanism.\"],\"threats_addressed\":[\"Pay-for-performance exposure in compensation (Appendix D).\",\"Public versus private firm differences (Appendix B).\",\"Timing mismatch between stationary comparisons and observed data dynamics.\"],\"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\":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\":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\":\"contribution_statement\"},{\"paragraph_index\":6,\"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\":\"data\"},{\"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\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"appendix\",\"types\":[\"alternative_specification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Few tables. A calibration table pairs data targets with model moments, a table compares data changes with model-predicted changes across periods, and an industry regression table follows. Notes describe sample filters and variable construction.\",\"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\":[\"consistent with\",\"largely\",\"is associated with\",\"caveat that calibration compares stationary equilibria\"]},\"claim_evidence\":[{\"claim\":\"Aggregate capital share of public firms rose from about 40% to 60% since 1980.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Aggregate factor shares are driven by mega-firms, while the average firm's capital share fell.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2, 3, 9\"},{\"claim\":\"Doubling rents and raising volatility from 20% to 40% replicates the rise in aggregate and the fall in average capital share.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Within-industry increases in idiosyncratic volatility are associated with lower average capital-income-to-sales ratios.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Dispersion effect is strongest in health and technology industries, which saw large volatility increases.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 10\"},{\"claim\":\"Higher volatility lowers the exit threshold and fattens the right tail of the stationary productivity distribution.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III, Figure 7, Appendix C\"}],\"structural_summary\":\"The paper opens with a divergence puzzle: aggregate capital share rose while firm-level shares fell. A stylized-facts section documents this divergence. A dynamic industry equilibrium model with worker insurance and endogenous exit follows, along with its equilibrium analysis. The model is then calibrated and run through comparative statics across stationary equilibria. Supporting empirical evidence comes from size-quantile, industry and regression analyses. Appendices cover data, private firms, proofs, pay-for-performance and valuation.\",\"writing_profile_keywords\":[\"stylized facts first\",\"model as laboratory\",\"selection mechanism\",\"calibration\",\"national income accounting\",\"supporting evidence\",\"differentiation from Autor et al.\"],\"confidence\":0.8,\"notes\":\"Empirical evidence is framed as consistent with the model rather than causal. The type of standard errors is not visible. Uncertainty bands in figures cannot be assessed from the captions.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "macro_finance", "subfield_secondary": "corporate_finance (optimal contracting, firm dynamics)", "primary_method": "calibration_simulation", "identification": {"estimand": "Association between industry-level idiosyncratic volatility and the average firm-level capital-income-to-sales ratio; model-implied effect of volatility and rents on aggregate versus average capital share.", "strategy": "Stylized facts from Compustat/CRSP, a calibrated industry equilibrium model with entry and exit and worker insurance, comparative statics across stationary equilibria, and within-industry regressions of capital share on idiosyncratic volatility.", "key_assumptions": ["Shareholders are diversified while workers are risk-averse, so optimal contracts insure workers.", "Comparisons are between stationary equilibria rather than transition dynamics.", "Compustat public firms are representative for the mechanism."], "threats_addressed": ["Pay-for-performance exposure in compensation (Appendix D).", "Public versus private firm differences (Appendix B).", "Timing mismatch between stationary comparisons and observed data dynamics."], "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": 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": 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": "contribution_statement"}, {"paragraph_index": 6, "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": "data"}, {"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"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "appendix", "types": ["alternative_specification", "subsample", "other"]}, "tables": {"convention_notes": "Few tables. A calibration table pairs data targets with model moments, a table compares data changes with model-predicted changes across periods, and an industry regression table follows. Notes describe sample filters and variable construction.", "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": ["consistent with", "largely", "is associated with", "caveat that calibration compares stationary equilibria"]}, "claim_evidence": [{"claim": "Aggregate capital share of public firms rose from about 40% to 60% since 1980.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "Aggregate factor shares are driven by mega-firms, while the average firm's capital share fell.", "evidence_type": "figure", "evidence_ref": "Figures 2, 3, 9"}, {"claim": "Doubling rents and raising volatility from 20% to 40% replicates the rise in aggregate and the fall in average capital share.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Within-industry increases in idiosyncratic volatility are associated with lower average capital-income-to-sales ratios.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Dispersion effect is strongest in health and technology industries, which saw large volatility increases.", "evidence_type": "figure", "evidence_ref": "Figure 10"}, {"claim": "Higher volatility lowers the exit threshold and fattens the right tail of the stationary productivity distribution.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III, Figure 7, Appendix C"}], "structural_summary": "The paper opens with a divergence puzzle: aggregate capital share rose while firm-level shares fell. A stylized-facts section documents this divergence. A dynamic industry equilibrium model with worker insurance and endogenous exit follows, along with its equilibrium analysis. The model is then calibrated and run through comparative statics across stationary equilibria. Supporting empirical evidence comes from size-quantile, industry and regression analyses. Appendices cover data, private firms, proofs, pay-for-performance and valuation.", "writing_profile_keywords": ["stylized facts first", "model as laboratory", "selection mechanism", "calibration", "national income accounting", "supporting evidence", "differentiation from Autor et al."], "confidence": 0.8, "notes": "Empirical evidence is framed as consistent with the model rather than causal. The type of standard errors is not visible. Uncertainty bands in figures cannot be assessed from the captions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.24, "usage": {"input": 8847, "output": 2097, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078122, "request_id": "req_011CggqQGkkh38ZFtxi7NPsr", "cached": false, "ts": "2026-11-03T23:08:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b6704e407393e5c0d02f04f9f1a4c690eaf7b6f0a44923d39a06e6ff1c88536b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"structural estimation of asymmetric information in credit markets / macro-finance (credit crunch)\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"Magnitude of the information gap (incumbent lenders' private information) and its contribution to the decline in lending during the 2007-2009 crisis\",\"strategy\":\"Two-stage discrete-choice model of renewal and new relationship formation. Bank lending-capacity shocks separate private from common information. Renewers serve as a benchmark, and the second stage uses non-renewing firms while correcting for selection on common information.\",\"key_assumptions\":[\"Bank shocks to lending ability are unrelated to unobservable borrower characteristics\",\"Distribution of borrower unobservables and the lending rule are common across lenders up to a matching error orthogonal to the previous lender's shock\"],\"threats_addressed\":[\"Confounding of private information with common information unobservable to the econometrician\",\"Mechanical selection of non-renewing firms on common information\",\"Matching errors (Monte Carlo)\"],\"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\":3},\"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\":\"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\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"new_method\",\"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\":\"data\"},{\"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\":\"confidence_intervals\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"subsample\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Probit tables report marginal effects ×100. Second-stage estimates use nonlinear least squares on non-renewers with bootstrapped 95% CIs. Notes define renewal via lead-lender overlap, the crisis-exposure construction, and sample restrictions. Counterfactual lending is computed with the information gap set to zero.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"bootstrap\",\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'appears to be' qualifiers on the main finding\",\"'unlikely to explain'\",\"explicit statement of maintained assumptions\",\"'a likely friction' for alternative mechanisms\",\"confining the conclusion to the syndicated loan market\"]},\"claim_evidence\":[{\"claim\":\"Lending relationships in the syndicated market are very sticky; few firms form new relationships relative to renewals\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Table I\"},{\"claim\":\"Lender crisis exposure lowers renewal probability with the existing lender\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Estimated information gap is small\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Information gap explains little of the aggregate post-crisis lending decline\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Naive models ignoring common information greatly overstate the friction's effect\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Reduced-form patterns consistent with the inference hypothesis are mostly driven by common information\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table AIII; Figure 6\"},{\"claim\":\"Bank shocks matter for firm borrowing in this market\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Chodorow-Reich (2014)\"}]},\"structural_summary\":\"The introduction motivates the work with relationship stickiness and its aggregate fragility, then states the main contribution: a first direct estimate of the information gap. It frames the key challenge of separating private from common information and sketches the comparison with renewers and a three-layer-information discrete-choice model. Identifying assumptions are stated, followed by a negative quantitative result contrasted with the prevailing view, and an alternative covenant channel is proposed. Related literature is discussed inside the introduction. The body moves from data and bank-shock measurement to the model and identification, then to two-stage estimation, aggregate counterfactuals, bias from naive models, policy, and generalizability. A short conclusion restates the method and finding.\",\"writing_profile_keywords\":[\"structural discrete choice\",\"adverse selection\",\"identification of private vs common information\",\"negative result\",\"bank shocks\",\"first-person singular\",\"quantification\",\"syndicated loans\"],\"confidence\":0.8,\"notes\":\"Single-authored. The main finding is a null/small-magnitude result. Robustness via subsamples (Table AIV), alternative bank shocks (Fig A2) and Monte Carlo (Fig A3) sits in the appendix. Section 2 heading appears to be a parsing fragment of Section I.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "structural estimation of asymmetric information in credit markets / macro-finance (credit crunch)", "primary_method": "structural_estimation", "identification": {"estimand": "Magnitude of the information gap (incumbent lenders' private information) and its contribution to the decline in lending during the 2007-2009 crisis", "strategy": "Two-stage discrete-choice model of renewal and new relationship formation. Bank lending-capacity shocks separate private from common information. Renewers serve as a benchmark, and the second stage uses non-renewing firms while correcting for selection on common information.", "key_assumptions": ["Bank shocks to lending ability are unrelated to unobservable borrower characteristics", "Distribution of borrower unobservables and the lending rule are common across lenders up to a matching error orthogonal to the previous lender's shock"], "threats_addressed": ["Confounding of private information with common information unobservable to the econometrician", "Mechanical selection of non-renewing firms on common information", "Matching errors (Monte Carlo)"], "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": 3}, "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": "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": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "tension_or_puzzle"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_identification", "new_method", "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": "data"}, {"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": "confidence_intervals", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["subsample", "alternative_measure", "other"]}, "tables": {"convention_notes": "Probit tables report marginal effects ×100. Second-stage estimates use nonlinear least squares on non-renewers with bootstrapped 95% CIs. Notes define renewal via lead-lender overlap, the crisis-exposure construction, and sample restrictions. Counterfactual lending is computed with the information gap set to zero.", "units_stated_in_notes": true, "standard_errors_description": "bootstrap", "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'appears to be' qualifiers on the main finding", "'unlikely to explain'", "explicit statement of maintained assumptions", "'a likely friction' for alternative mechanisms", "confining the conclusion to the syndicated loan market"]}, "claim_evidence": [{"claim": "Lending relationships in the syndicated market are very sticky; few firms form new relationships relative to renewals", "evidence_type": "descriptive_statistic", "evidence_ref": "Table I"}, {"claim": "Lender crisis exposure lowers renewal probability with the existing lender", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Estimated information gap is small", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Information gap explains little of the aggregate post-crisis lending decline", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Naive models ignoring common information greatly overstate the friction's effect", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Reduced-form patterns consistent with the inference hypothesis are mostly driven by common information", "evidence_type": "table", "evidence_ref": "Table AIII; Figure 6"}, {"claim": "Bank shocks matter for firm borrowing in this market", "evidence_type": "prior_literature", "evidence_ref": "Chodorow-Reich (2014)"}]}, "structural_summary": "The introduction motivates the work with relationship stickiness and its aggregate fragility, then states the main contribution: a first direct estimate of the information gap. It frames the key challenge of separating private from common information and sketches the comparison with renewers and a three-layer-information discrete-choice model. Identifying assumptions are stated, followed by a negative quantitative result contrasted with the prevailing view, and an alternative covenant channel is proposed. Related literature is discussed inside the introduction. The body moves from data and bank-shock measurement to the model and identification, then to two-stage estimation, aggregate counterfactuals, bias from naive models, policy, and generalizability. A short conclusion restates the method and finding.", "writing_profile_keywords": ["structural discrete choice", "adverse selection", "identification of private vs common information", "negative result", "bank shocks", "first-person singular", "quantification", "syndicated loans"], "confidence": 0.8, "notes": "Single-authored. The main finding is a null/small-magnitude result. Robustness via subsamples (Table AIV), alternative bank shocks (Fig A2) and Monte Carlo (Fig A3) sits in the appendix. Section 2 heading appears to be a parsing fragment of Section I."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.23, "usage": {"input": 7173, "output": 2263, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074746, "request_id": "req_011CggqtCrPoWtU55iEzQ87w", "cached": false, "ts": "2026-11-03T23:14:24+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/b8a092d7b206323d782b0ced46b1623f59824a9473877ed7489a87971ee3fae9.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"corporate governance / managerial economics\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of a reduction in firms' exposure to their CEO (days of CEO hospitalization) on firm performance (OROA, investment, other outcomes), holding the firm-CEO match fixed\",\"strategy\":\"Panel regressions of performance on CEO hospitalization days with firm-CEO fixed effects. Supplemented by matched event/nonevent firm comparisons, an IV using the national share of lengthy stays per diagnosis, non-stress-related and deadliest-disease subsamples, and placebos using hospitalizations of other executives.\",\"key_assumptions\":[\"Hospitalizations reduce CEO productive capacity\",\"Hospitalization length is not driven by firm performance (reverse causality), conditional on instrument\",\"Timing and incidence of hospitalization are not caused by poor performance\",\"Medical costs do not directly affect firm performance (public financing of hospitals)\"],\"threats_addressed\":[\"reverse causality from performance to length of stay\",\"stress-related illness triggered by poor performance\",\"strategic timing of treatment\",\"firm-CEO matching endogeneity / demand-side fixed effects\",\"effects driven by CEO turnover firms\"],\"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\":5}},\"introduction\":{\"moves\":[{\"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\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"other\"},{\"paragraph_index\":10,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":7,\"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\":\"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\",\"placebo\",\"falsification\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variables (OROA, net income to assets, distress, log sales) and the hospitalization-day regressors, and state sample filters (nonfinancial, nonutility, non-government-owned Danish limited liability firms, 1996-2012). Panels separate means, mean differences and medians, and address distinct identification challenges. Standard-error treatment is not visible in the note excerpts.\",\"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\":[\"'we posit' premise framing\",\"'consistent with' wording\",\"explicit external-validity caveat for public firms\",\"alternative interpretation acknowledged for turnover result\",\"'to our knowledge' priority claim\"],\"claim_evidence\":[{\"claim\":\"A 10-day CEO hospital stay lowers operating profitability by about 5.8% of its mean\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Results survive IV, non-stress-related diagnoses, and deadliest-disease subsamples\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Hospitalizations of non-CEO senior executives do not produce similar performance effects\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"CEO effects are larger for younger CEOs, growing and family firms, and human-capital-intensive industries\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"CEO hospitalization raises the probability of CEO turnover\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Performance effects hold in firms without CEO turnover\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"},{\"claim\":\"Event firms underperform matched firms around hospitalization\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-2\"}],\"structural_summary\":\"The paper opens directly with its identification idea: variation in CEO exposure from hospitalization, holding the firm-CEO match fixed. It then justifies Denmark as the setting and previews the headline magnitude and the turnover finding. The introduction goes on to defend identification against reverse causality and coincident timing, followed by enumerated contributions, an external-validity caveat and a roadmap. The body moves from data, to baseline univariate, matched and multivariate results with causality and magnitude subsections, to a uniqueness test using other executives, heterogeneity splits and turnover. A short conclusion restates the contribution.\",\"writing_profile_keywords\":[\"administrative registry data\",\"firm-CEO fixed effects\",\"shock-based identification\",\"instrumental variable robustness\",\"placebo on other executives\",\"enumerated contributions\",\"Denmark private firms\"],\"confidence\":0.8,\"notes\":\"Heterogeneity section labeled robustness by rule but is substantively a results section. Figure 2 shows 95% CIs; Figure 1 apparently does not. The intro's contribution list is truncated (first, second, then fifth visible). The abstract's 'CEO contingency plans are valuable' is a mild implication, but the conclusion excerpt is truncated.\"}}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "corporate governance / managerial economics", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of a reduction in firms' exposure to their CEO (days of CEO hospitalization) on firm performance (OROA, investment, other outcomes), holding the firm-CEO match fixed", "strategy": "Panel regressions of performance on CEO hospitalization days with firm-CEO fixed effects. Supplemented by matched event/nonevent firm comparisons, an IV using the national share of lengthy stays per diagnosis, non-stress-related and deadliest-disease subsamples, and placebos using hospitalizations of other executives.", "key_assumptions": ["Hospitalizations reduce CEO productive capacity", "Hospitalization length is not driven by firm performance (reverse causality), conditional on instrument", "Timing and incidence of hospitalization are not caused by poor performance", "Medical costs do not directly affect firm performance (public financing of hospitals)"], "threats_addressed": ["reverse causality from performance to length of stay", "stress-related illness triggered by poor performance", "strategic timing of treatment", "firm-CEO matching endogeneity / demand-side fixed effects", "effects driven by CEO turnover firms"], "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": 5}}, "introduction": {"moves": [{"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": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "identification_defense"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "other"}, {"paragraph_index": 10, "move": "roadmap"}], "contribution_paragraph_index": 7, "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": "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", "placebo", "falsification", "alternative_measure"]}, "tables": {"convention_notes": "Table notes define the dependent variables (OROA, net income to assets, distress, log sales) and the hospitalization-day regressors, and state sample filters (nonfinancial, nonutility, non-government-owned Danish limited liability firms, 1996-2012). Panels separate means, mean differences and medians, and address distinct identification challenges. Standard-error treatment is not visible in the note excerpts.", "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": ["'we posit' premise framing", "'consistent with' wording", "explicit external-validity caveat for public firms", "alternative interpretation acknowledged for turnover result", "'to our knowledge' priority claim"], "claim_evidence": [{"claim": "A 10-day CEO hospital stay lowers operating profitability by about 5.8% of its mean", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Results survive IV, non-stress-related diagnoses, and deadliest-disease subsamples", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Hospitalizations of non-CEO senior executives do not produce similar performance effects", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "CEO effects are larger for younger CEOs, growing and family firms, and human-capital-intensive industries", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "CEO hospitalization raises the probability of CEO turnover", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Performance effects hold in firms without CEO turnover", "evidence_type": "table", "evidence_ref": "Table XII"}, {"claim": "Event firms underperform matched firms around hospitalization", "evidence_type": "figure", "evidence_ref": "Figures 1-2"}], "structural_summary": "The paper opens directly with its identification idea: variation in CEO exposure from hospitalization, holding the firm-CEO match fixed. It then justifies Denmark as the setting and previews the headline magnitude and the turnover finding. The introduction goes on to defend identification against reverse causality and coincident timing, followed by enumerated contributions, an external-validity caveat and a roadmap. The body moves from data, to baseline univariate, matched and multivariate results with causality and magnitude subsections, to a uniqueness test using other executives, heterogeneity splits and turnover. A short conclusion restates the contribution.", "writing_profile_keywords": ["administrative registry data", "firm-CEO fixed effects", "shock-based identification", "instrumental variable robustness", "placebo on other executives", "enumerated contributions", "Denmark private firms"], "confidence": 0.8, "notes": "Heterogeneity section labeled robustness by rule but is substantively a results section. Figure 2 shows 95% CIs; Figure 1 apparently does not. The intro's contribution list is truncated (first, second, then fifth visible). The abstract's 'CEO contingency plans are valuable' is a mild implication, but the conclusion excerpt is truncated."}}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.94, "usage": {"input": 7426, "output": 2295, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076398, "request_id": "req_011CggqrHqDv5zh1vDTiNiU2", "cached": false, "ts": "2026-11-03T23:13:59+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/bafb10791c70c11ecc952851fe195a2769a1fac2933a717ed11f196b11416a1b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"corporate_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Determinants of post-IPO VC financing, plus investor and firm returns associated with these financings; the value of having pre-IPO backing from VCs prone to post-IPO funding\",\"strategy\":\"Competing-hypothesis tests. Firm-level linear probability models compare four financing types. VC×firm-level choice regressions follow. Returns are measured as buy-and-hold abnormal returns to investors, announcement CARs and long-run returns. Calendar-time long-short portfolios and delisting logits are sorted on the pre-IPO VC's propensity to fund after the IPO.\",\"key_assumptions\":[\"Proxies for information asymmetry and VC agency costs capture the intended constructs\",\"A VC's historical propensity to fund post-IPO serves as a proxy for a firm's access to post-IPO VC financing\",\"Comparing with non-VC PIPEs isolates the VC comparative advantage\"],\"threats_addressed\":[\"Alternative agency explanation tested against the information asymmetry explanation\",\"Comparison with other financing sources, including PIPEs without VC participation\",\"Heterogeneity by VC rank\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"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\":0},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"motivation\"},{\"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\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"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\":\"data\"},{\"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\"}],\"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 open by defining the sample, cross-referencing Table I. They then describe the model (linear probability, logit, calendar-time portfolio regressions, CARs against size-decile benchmarks) and the panel structure.\",\"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\":[\"consistent with\",\"suggest\",\"suggestive evidence\",\"we find no evidence\"]},\"claim_evidence\":[{\"claim\":\"About 15% of VC-backed IPO firms obtain additional VC funding within five years of the IPO\",\"evidence_type\":\"descriptive_statistic\",\"evidence_ref\":\"Figure 1; Table I\"},{\"claim\":\"Only post-IPO VC financings are linked to both high firm information asymmetry and a VC informational advantage\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"VCs with prior investment or board seats are more likely to fund post-IPO; agency proxies show no association\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"VCs earn roughly 31 percentage points of abnormal return over 12 months, unlike non-VC PIPE investors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IV, V\"},{\"claim\":\"Market reactions around post-IPO VC investments are positive\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"A long-short portfolio sorted on the pre-IPO VC's post-IPO funding propensity earns positive returns, and such firms delist less\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"}],\"structural_summary\":\"The paper opens by challenging conventional wisdom with a descriptive fact: VCs frequently fund firms after their IPO. It frames two competing explanations, information asymmetry versus agency. Data and descriptives come first. Successive results sections then test the hypotheses through determinants of financing type, VC-level choice, returns to VC investors, market reactions and long-run returns, and a portfolio and delisting analysis based on pre-IPO VC propensity. The conclusion restates the finding favoring information asymmetry and draws a practical implication for private firms choosing VCs.\",\"writing_profile_keywords\":[\"stylized fact\",\"competing hypotheses\",\"multi-angle evidence\",\"venture capital\",\"post-IPO financing\",\"consistent-with language\"],\"confidence\":0.8,\"notes\":\"No explicit robustness section appears in the excerpts; an Internet Appendix is mentioned but its contents are not shown. The excerpts conflict on sample start year (1985 vs 1995).\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "corporate_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Determinants of post-IPO VC financing, plus investor and firm returns associated with these financings; the value of having pre-IPO backing from VCs prone to post-IPO funding", "strategy": "Competing-hypothesis tests. Firm-level linear probability models compare four financing types. VC×firm-level choice regressions follow. Returns are measured as buy-and-hold abnormal returns to investors, announcement CARs and long-run returns. Calendar-time long-short portfolios and delisting logits are sorted on the pre-IPO VC's propensity to fund after the IPO.", "key_assumptions": ["Proxies for information asymmetry and VC agency costs capture the intended constructs", "A VC's historical propensity to fund post-IPO serves as a proxy for a firm's access to post-IPO VC financing", "Comparing with non-VC PIPEs isolates the VC comparative advantage"], "threats_addressed": ["Alternative agency explanation tested against the information asymmetry explanation", "Comparison with other financing sources, including PIPEs without VC participation", "Heterogeneity by VC rank"], "causal_language_strength": 2, "uses_causal_claim": false}, "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": 0}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "motivation"}, {"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": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"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": "data"}, {"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"}], "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 open by defining the sample, cross-referencing Table I. They then describe the model (linear probability, logit, calendar-time portfolio regressions, CARs against size-decile benchmarks) and the panel structure.", "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": ["consistent with", "suggest", "suggestive evidence", "we find no evidence"]}, "claim_evidence": [{"claim": "About 15% of VC-backed IPO firms obtain additional VC funding within five years of the IPO", "evidence_type": "descriptive_statistic", "evidence_ref": "Figure 1; Table I"}, {"claim": "Only post-IPO VC financings are linked to both high firm information asymmetry and a VC informational advantage", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "VCs with prior investment or board seats are more likely to fund post-IPO; agency proxies show no association", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "VCs earn roughly 31 percentage points of abnormal return over 12 months, unlike non-VC PIPE investors", "evidence_type": "table", "evidence_ref": "Tables IV, V"}, {"claim": "Market reactions around post-IPO VC investments are positive", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "A long-short portfolio sorted on the pre-IPO VC's post-IPO funding propensity earns positive returns, and such firms delist less", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}], "structural_summary": "The paper opens by challenging conventional wisdom with a descriptive fact: VCs frequently fund firms after their IPO. It frames two competing explanations, information asymmetry versus agency. Data and descriptives come first. Successive results sections then test the hypotheses through determinants of financing type, VC-level choice, returns to VC investors, market reactions and long-run returns, and a portfolio and delisting analysis based on pre-IPO VC propensity. The conclusion restates the finding favoring information asymmetry and draws a practical implication for private firms choosing VCs.", "writing_profile_keywords": ["stylized fact", "competing hypotheses", "multi-angle evidence", "venture capital", "post-IPO financing", "consistent-with language"], "confidence": 0.8, "notes": "No explicit robustness section appears in the excerpts; an Internet Appendix is mentioned but its contents are not shown. The excerpts conflict on sample start year (1985 vs 1995)."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.7, "usage": {"input": 8562, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077842, "request_id": "req_011CggqpSUa1DkNYCkcNFXGw", "cached": false, "ts": "2026-11-03T23:13:31+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/bb6e92f688613c5d57e8f5b9df1f3ed7c76ae3a84f8b244baf08fdf603bfebf1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"methodological_econometric\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"The wedge (bias) between empirically observed responses to exogenous shocks and theory-implied causal effects defined as comparative statics, in a dynamic linear-quadratic stock-accumulation model.\",\"strategy\":\"The model serves as a laboratory. Shocks are exogenous Markov chain transitions, so endogeneity is absent by construction. Shock responses are derived analytically and compared with known comparative statics, yielding bias, correction and extrapolation formulas. A numerical tax example uses an estimated tax-rate Markov chain.\",\"key_assumptions\":[\"Linear profits/benefit flows and linear-quadratic stock accumulation (adjustment) costs\",\"Rational expectations over a known shock-generating (Markov) process\",\"Shocks are exogenous (random assignment holds)\",\"For equality results: martingale benefit flow, or infinitesimal shock probabilities\"],\"threats_addressed\":[\"Endogeneity/selection deliberately ruled out so that dynamic-uncertainty bias is isolated\",\"Sensitivity to technology assumptions (alternative technologies section)\",\"External validity across data-generating processes\"],\"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\":2},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_theory_or_model\",\"new_method\",\"new_identification\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"theory\"},{\"idx\":2,\"role\":\"other\"},{\"idx\":3,\"role\":\"theory\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"theory\"},{\"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\"},{\"idx\":13,\"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 report model-implied shock responses, transition probabilities and response-to-causal-effect ratios. Notes state the assumed parameters (shock probabilities, risk-free rate, depreciation rate) and use panels for alternative state-space sizes. No inferential statistics are reported.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"modal verbs (can, may)\",\"conditioning on 'plausible' or baseline parameterizations\",\"'in some settings/circumstances' qualifiers\",\"'in general' framing of results\"]},\"claim_evidence\":[{\"claim\":\"With binary shocks, attenuation bias exceeds half the causal effect once the shock probability is above about 5%.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II propositions (binary-shock correction formula)\"},{\"claim\":\"Under permanent shocks, responses can overshoot or carry the wrong sign, depending on realized versus expected changes.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II, Proposition 7\"},{\"claim\":\"Martingale profitability is sufficient for shock responses to equal causal effects.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Appendix proofs\"},{\"claim\":\"With an estimated tax-rate process, shock responses are roughly half the theory-implied causal effects.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-IV\"},{\"claim\":\"Shock responses vary with the shock-generating process, e.g., decreasing versus mean-reverting tax rates.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I-II, Figure 1\"},{\"claim\":\"Shock-based identification designs have grown sharply in corporate finance and governance research.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Bowen, Fresard, and Taillard (2016); Atanasov and Black (2016)\"}],\"structural_summary\":\"The introduction opens with authoritative quotations equating comparative statics with causal effects. It then critiques the reduced-form toolkit for ignoring dynamic uncertainty. Next it frames an analytical model as a laboratory in which endogeneity is absent, and previews striking bias results (wrong signs, overshooting, attenuation above 50%) with an applied tax illustration and welfare implications. A constructive contribution follows: correction formulas, extrapolation methods and sufficient identifying conditions. The introduction closes with extended literature positioning against Lucas, tax-investment papers and Keane-Wolpin. The body develops the model, then the shock-response propositions, identifying assumptions, external validity/extrapolation, a calibrated tax example and alternative technologies. The conclusion restates the theory-evidence gap and the constructive prescription; proofs are in the appendix.\",\"writing_profile_keywords\":[\"methodological critique\",\"analytical propositions\",\"comparative statics versus shock responses\",\"model as laboratory\",\"bias correction formulas\",\"external validity\",\"dynamic uncertainty\",\"numerical illustration\",\"quotation-driven motivation\"],\"confidence\":0.8,\"notes\":\"Genre is borderline between theoretical and methodological; classified as methodological because the purpose is inference guidance for empiricists. The numerical section uses an estimated tax Markov chain but no regressions. A roadmap was not visible in the excerpts. The abstract's final fragment is spillover from the introduction.\"}", "json": {"genre": "methodological_econometric", "subfield": "corporate_finance", "subfield_secondary": "financial_econometrics", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "The wedge (bias) between empirically observed responses to exogenous shocks and theory-implied causal effects defined as comparative statics, in a dynamic linear-quadratic stock-accumulation model.", "strategy": "The model serves as a laboratory. Shocks are exogenous Markov chain transitions, so endogeneity is absent by construction. Shock responses are derived analytically and compared with known comparative statics, yielding bias, correction and extrapolation formulas. A numerical tax example uses an estimated tax-rate Markov chain.", "key_assumptions": ["Linear profits/benefit flows and linear-quadratic stock accumulation (adjustment) costs", "Rational expectations over a known shock-generating (Markov) process", "Shocks are exogenous (random assignment holds)", "For equality results: martingale benefit flow, or infinitesimal shock probabilities"], "threats_addressed": ["Endogeneity/selection deliberately ruled out so that dynamic-uncertainty bias is isolated", "Sensitivity to technology assumptions (alternative technologies section)", "External validity across data-generating processes"], "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": 2}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"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": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "related_literature_positioning"}], "contribution_paragraph_index": 6, "contribution_types": ["new_theory_or_model", "new_method", "new_identification"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "theory"}, {"idx": 2, "role": "other"}, {"idx": 3, "role": "theory"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "theory"}, {"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"}, {"idx": 13, "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 report model-implied shock responses, transition probabilities and response-to-causal-effect ratios. Notes state the assumed parameters (shock probabilities, risk-free rate, depreciation rate) and use panels for alternative state-space sizes. No inferential statistics are reported.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["modal verbs (can, may)", "conditioning on 'plausible' or baseline parameterizations", "'in some settings/circumstances' qualifiers", "'in general' framing of results"]}, "claim_evidence": [{"claim": "With binary shocks, attenuation bias exceeds half the causal effect once the shock probability is above about 5%.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II propositions (binary-shock correction formula)"}, {"claim": "Under permanent shocks, responses can overshoot or carry the wrong sign, depending on realized versus expected changes.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II, Proposition 7"}, {"claim": "Martingale profitability is sufficient for shock responses to equal causal effects.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Appendix proofs"}, {"claim": "With an estimated tax-rate process, shock responses are roughly half the theory-implied causal effects.", "evidence_type": "table", "evidence_ref": "Tables III-IV"}, {"claim": "Shock responses vary with the shock-generating process, e.g., decreasing versus mean-reverting tax rates.", "evidence_type": "table", "evidence_ref": "Tables I-II, Figure 1"}, {"claim": "Shock-based identification designs have grown sharply in corporate finance and governance research.", "evidence_type": "prior_literature", "evidence_ref": "Bowen, Fresard, and Taillard (2016); Atanasov and Black (2016)"}], "structural_summary": "The introduction opens with authoritative quotations equating comparative statics with causal effects. It then critiques the reduced-form toolkit for ignoring dynamic uncertainty. Next it frames an analytical model as a laboratory in which endogeneity is absent, and previews striking bias results (wrong signs, overshooting, attenuation above 50%) with an applied tax illustration and welfare implications. A constructive contribution follows: correction formulas, extrapolation methods and sufficient identifying conditions. The introduction closes with extended literature positioning against Lucas, tax-investment papers and Keane-Wolpin. The body develops the model, then the shock-response propositions, identifying assumptions, external validity/extrapolation, a calibrated tax example and alternative technologies. The conclusion restates the theory-evidence gap and the constructive prescription; proofs are in the appendix.", "writing_profile_keywords": ["methodological critique", "analytical propositions", "comparative statics versus shock responses", "model as laboratory", "bias correction formulas", "external validity", "dynamic uncertainty", "numerical illustration", "quotation-driven motivation"], "confidence": 0.8, "notes": "Genre is borderline between theoretical and methodological; classified as methodological because the purpose is inference guidance for empiricists. The numerical section uses an estimated tax Markov chain but no regressions. A roadmap was not visible in the excerpts. The abstract's final fragment is spillover from the introduction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 28.67, "usage": {"input": 8616, "output": 3392, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.103098, "request_id": "req_011CggqjEbZWiv8dnnTe1ysK", "cached": false, "ts": "2026-11-03T23:12:32+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c0167b441b0060e75bf38951919a4929908409eb84b379ec0ea10d157381fd63.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Shape of a common nonlinear function of lagged VIX that forecasts excess stock and bond returns, plus asset-specific loadings on that function\",\"strategy\":\"Sieve reduced-rank (SRR) panel predictive regressions. Each asset's excess return is regressed on sieve expansions of lagged VIX, and the coefficient matrix is reduced to rank one. Inference uses Hodrick-style reverse regressions; there is no causal design.\",\"key_assumptions\":[\"A rank-one common nonlinear factor drives expected returns across assets up to affine transformations\",\"Reduced-rank estimation is optimal under conditionally Gaussian errors with a fixed number of regressors\",\"Asymptotic validity of the proposed test statistics under overlapping observations\"],\"threats_addressed\":[\"Omitted standard predictors (dividend yield, default spread, term spread, VRP)\",\"Influence of the 2008-09 crisis tail events, via a 1990-2007 subsample\",\"Treasury-specific volatility (MOVE) as an alternative predictor\",\"Overlapping-horizon inference\",\"Out-of-sample forecasting performance\"],\"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\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"robustness_mention\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":10,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_evidence\",\"new_method\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"results\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"theory\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"mixed\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_sample_period\",\"subsample\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes state the regression equation, forecasting horizons (6/12/18 months) and sample period. Tables report Hodrick (1992) t-statistics, LM test statistics with p-values in parentheses, and significance stars in the fund flow table.\",\"units_stated_in_notes\":false,\"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\":[\"consistent with\",\"suggests\",\"points toward\",\"in line with\",\"tend to\"]},\"claim_evidence\":[{\"claim\":\"Lagged VIX forecasts stock and bond returns nonlinearly, with mirror-image shapes across the two asset classes\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Figure 4\"},{\"claim\":\"Linear VIX predictability is insignificant, but polynomial specifications are significant across horizons\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Figures 2-3\"},{\"claim\":\"SRR estimates show a highly significant common nonlinear function in the full sample and in the pre-crisis sample\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-III\"},{\"claim\":\"Tests do not reject that stocks and bonds load on the same nonlinear function, interpreted as flight-to-safety\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The SRR model improves out-of-sample forecasts relative to a mean-only benchmark\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Mutual fund flows depend on VIX with a shape similar to the return-forecasting function, supporting an asset-manager mechanism\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI, Figure 7\"},{\"claim\":\"Forecasting slopes line up with risk factor loadings in a dynamic asset pricing model\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"}],\"structural_summary\":\"The paper opens with flight-to-safety theories and the difficulty of documenting nonlinear pricing empirically. It previews a mirror-image nonlinear VIX-return relation and introduces a new panel sieve reduced-rank estimator with accompanying inference tests. Motivating univariate polynomial and nonparametric evidence comes first, followed by derivation of the estimator, main results and robustness checks. An economic-interpretation section links the results to dynamic asset pricing, mutual fund flows and intermediary VaR theories. A short conclusion restates the findings.\",\"writing_profile_keywords\":[\"return predictability\",\"nonparametric panel estimator\",\"flight-to-safety\",\"VIX\",\"mirror image\",\"method plus evidence\",\"theory-consistent interpretation\"],\"confidence\":0.8,\"notes\":\"Primary method is set to 'other' because the paper's core tool is a new sieve reduced-rank estimator rather than standard reduced-form regression. Section idx 3 is a parsing artifact (its heading is an equation fragment) that continues Section II, covering SRR estimation and robustness. The Internet Appendix holds additional results. Figure 7's caption mentions panels showing predicted fund flows, but whether figures display confidence bands is not visible.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "other", "identification": {"estimand": "Shape of a common nonlinear function of lagged VIX that forecasts excess stock and bond returns, plus asset-specific loadings on that function", "strategy": "Sieve reduced-rank (SRR) panel predictive regressions. Each asset's excess return is regressed on sieve expansions of lagged VIX, and the coefficient matrix is reduced to rank one. Inference uses Hodrick-style reverse regressions; there is no causal design.", "key_assumptions": ["A rank-one common nonlinear factor drives expected returns across assets up to affine transformations", "Reduced-rank estimation is optimal under conditionally Gaussian errors with a fixed number of regressors", "Asymptotic validity of the proposed test statistics under overlapping observations"], "threats_addressed": ["Omitted standard predictors (dividend yield, default spread, term spread, VRP)", "Influence of the 2008-09 crisis tail events, via a 1990-2007 subsample", "Treasury-specific volatility (MOVE) as an alternative predictor", "Overlapping-horizon inference", "Out-of-sample forecasting performance"], "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": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 7, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "robustness_mention"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 10, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 11, "move": "related_literature_positioning"}, {"paragraph_index": 12, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_evidence", "new_method"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "results"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "theory"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "mixed", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_sample_period", "subsample", "alternative_measure", "other"]}, "tables": {"convention_notes": "Notes state the regression equation, forecasting horizons (6/12/18 months) and sample period. Tables report Hodrick (1992) t-statistics, LM test statistics with p-values in parentheses, and significance stars in the fund flow table.", "units_stated_in_notes": false, "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": ["consistent with", "suggests", "points toward", "in line with", "tend to"]}, "claim_evidence": [{"claim": "Lagged VIX forecasts stock and bond returns nonlinearly, with mirror-image shapes across the two asset classes", "evidence_type": "figure", "evidence_ref": "Figure 1, Figure 4"}, {"claim": "Linear VIX predictability is insignificant, but polynomial specifications are significant across horizons", "evidence_type": "table", "evidence_ref": "Table I, Figures 2-3"}, {"claim": "SRR estimates show a highly significant common nonlinear function in the full sample and in the pre-crisis sample", "evidence_type": "table", "evidence_ref": "Tables II-III"}, {"claim": "Tests do not reject that stocks and bonds load on the same nonlinear function, interpreted as flight-to-safety", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The SRR model improves out-of-sample forecasts relative to a mean-only benchmark", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Mutual fund flows depend on VIX with a shape similar to the return-forecasting function, supporting an asset-manager mechanism", "evidence_type": "table", "evidence_ref": "Table VI, Figure 7"}, {"claim": "Forecasting slopes line up with risk factor loadings in a dynamic asset pricing model", "evidence_type": "figure", "evidence_ref": "Figure 5"}], "structural_summary": "The paper opens with flight-to-safety theories and the difficulty of documenting nonlinear pricing empirically. It previews a mirror-image nonlinear VIX-return relation and introduces a new panel sieve reduced-rank estimator with accompanying inference tests. Motivating univariate polynomial and nonparametric evidence comes first, followed by derivation of the estimator, main results and robustness checks. An economic-interpretation section links the results to dynamic asset pricing, mutual fund flows and intermediary VaR theories. A short conclusion restates the findings.", "writing_profile_keywords": ["return predictability", "nonparametric panel estimator", "flight-to-safety", "VIX", "mirror image", "method plus evidence", "theory-consistent interpretation"], "confidence": 0.8, "notes": "Primary method is set to 'other' because the paper's core tool is a new sieve reduced-rank estimator rather than standard reduced-form regression. Section idx 3 is a parsing artifact (its heading is an equation fragment) that continues Section II, covering SRR estimation and robustness. The Internet Appendix holds additional results. Figure 7's caption mentions panels showing predicted fund flows, but whether figures display confidence bands is not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.81, "usage": {"input": 8060, "output": 2276, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078554, "request_id": "req_011CggqRmRTbEKkYxM768Y6K", "cached": false, "ts": "2026-11-03T23:08:25+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c4d7af1ca4a3f704b3a1cf28e9fc7681276f874e10ab7a56f049f113aa3275dd.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Causal effect of a foreclosure on a resetting ARM on default and foreclosure probability among neighboring properties within 0.10 miles (foreclosure spillover)\",\"strategy\":\"IV/2SLS in which the within-month interest rate shock at ARM reset, driven by index choice (LIBOR vs Treasury) and lookback period, instruments for the foreclosure of the resetting loan; first stage, reduced form and IV estimates are contrasted with OLS\",\"key_assumptions\":[\"Within-month reset rate variation from index and lookback terms is unrelated to other determinants of loan performance\",\"Exclusion restriction: the resetting loan's rate shock affects neighbors' defaults only through its foreclosure\",\"Neighboring loans have different contract terms, so they are not directly hit by the same shock\"],\"threats_addressed\":[\"Common local shocks such as a plant closure, addressed by showing effects are concentrated at very small distances\",\"Pre-existing neighborhood distress, addressed with a placebo on prior foreclosures\",\"Borrower selection into index type\",\"Demand versus supply in refinancing, addressed with mortgage inquiries\"],\"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\":1},\"magnitude\":{\"present\":true,\"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\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_identification\",\"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\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"placebo\",\"subsample\",\"falsification\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes state the regression equations (first stage, reduced form, 2SLS) and describe the samples (Deeds vs BlackBox). Heterogeneity and placebo tables reuse the main IV specification. Summary statistics are split by index type.\",\"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\",\"suggest\",\"likely reflects\",\"may help to explain\",\"potentially point to\",\"highly unlikely ex ante\"]},\"claim_evidence\":[{\"claim\":\"A 100bp higher reset rate raises the 12-month foreclosure probability by about 2.5 points against an 8% baseline\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 6\"},{\"claim\":\"Each instrumented foreclosure raises neighbor default probability by about 2.1%, or 0.3-0.6 additional foreclosures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II, Table IV\"},{\"claim\":\"LIBOR-Treasury spread and lookback volatility produce sizable within-month payment differences\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1-4\"},{\"claim\":\"Neighboring sale prices fall about 1-3% after a resetting ARM forecloses\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"Neighbor refinancing falls by one-third while mortgage inquiries do not, indicating a bank supply response\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 8\"},{\"claim\":\"Spillovers are strongest in areas with few prior foreclosures and are concentrated at short distances\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 10\"},{\"claim\":\"Tract-level foreclosure activity raises mortgage denials, citing collateral reasons\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 12\"},{\"claim\":\"Placebo: the instrument does not predict prior neighbor foreclosures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The paper opens with crisis-scale motivation and lists candidate spillover channels. It then introduces two contract-detail instruments, previews first-stage, reduced-form and IV magnitudes, and walks through mechanism results (prices, refinancing supply, borrower learning) before aggregate HMDA evidence. Literature positioning sits inside the introduction, followed by a roadmap. The body covers data and strategy, main IV results, mechanisms, macro channels, and a short placebo robustness section. The conclusion restates the findings, favours an informational interpretation, and closes with bank-run analogies and policy implications.\",\"writing_profile_keywords\":[\"instrumental variables\",\"contract-detail instruments\",\"peer effects\",\"foreclosure spillovers\",\"mechanism decomposition\",\"single-author first person\",\"geocoded administrative data\"],\"confidence\":0.8,\"notes\":\"The roadmap labels Section V as the conclusion, but in the section list Section V is Robustness and VI is the Conclusion. The abstract is truncated and has no explicit gap or contribution sentence. Standard error treatment is not visible in the excerpts. Mechanism and macro sections were classified as results despite the rule-based robustness label.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "instrumental_variables", "identification": {"estimand": "Causal effect of a foreclosure on a resetting ARM on default and foreclosure probability among neighboring properties within 0.10 miles (foreclosure spillover)", "strategy": "IV/2SLS in which the within-month interest rate shock at ARM reset, driven by index choice (LIBOR vs Treasury) and lookback period, instruments for the foreclosure of the resetting loan; first stage, reduced form and IV estimates are contrasted with OLS", "key_assumptions": ["Within-month reset rate variation from index and lookback terms is unrelated to other determinants of loan performance", "Exclusion restriction: the resetting loan's rate shock affects neighbors' defaults only through its foreclosure", "Neighboring loans have different contract terms, so they are not directly hit by the same shock"], "threats_addressed": ["Common local shocks such as a plant closure, addressed by showing effects are concentrated at very small distances", "Pre-existing neighborhood distress, addressed with a placebo on prior foreclosures", "Borrower selection into index type", "Demand versus supply in refinancing, addressed with mortgage inquiries"], "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": 1}, "magnitude": {"present": true, "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": "literature_gap"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "identification_defense"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "identification_defense"}, {"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": "preview_of_results"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_identification", "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": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["placebo", "subsample", "falsification", "alternative_measure"]}, "tables": {"convention_notes": "Table notes state the regression equations (first stage, reduced form, 2SLS) and describe the samples (Deeds vs BlackBox). Heterogeneity and placebo tables reuse the main IV specification. Summary statistics are split by index type.", "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", "suggest", "likely reflects", "may help to explain", "potentially point to", "highly unlikely ex ante"]}, "claim_evidence": [{"claim": "A 100bp higher reset rate raises the 12-month foreclosure probability by about 2.5 points against an 8% baseline", "evidence_type": "table", "evidence_ref": "Table II; Figure 6"}, {"claim": "Each instrumented foreclosure raises neighbor default probability by about 2.1%, or 0.3-0.6 additional foreclosures", "evidence_type": "table", "evidence_ref": "Table II, Table IV"}, {"claim": "LIBOR-Treasury spread and lookback volatility produce sizable within-month payment differences", "evidence_type": "figure", "evidence_ref": "Figures 1-4"}, {"claim": "Neighboring sale prices fall about 1-3% after a resetting ARM forecloses", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "Neighbor refinancing falls by one-third while mortgage inquiries do not, indicating a bank supply response", "evidence_type": "table", "evidence_ref": "Table III; Figure 8"}, {"claim": "Spillovers are strongest in areas with few prior foreclosures and are concentrated at short distances", "evidence_type": "table", "evidence_ref": "Table VI; Figure 10"}, {"claim": "Tract-level foreclosure activity raises mortgage denials, citing collateral reasons", "evidence_type": "table", "evidence_ref": "Table VII; Figure 12"}, {"claim": "Placebo: the instrument does not predict prior neighbor foreclosures", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "The paper opens with crisis-scale motivation and lists candidate spillover channels. It then introduces two contract-detail instruments, previews first-stage, reduced-form and IV magnitudes, and walks through mechanism results (prices, refinancing supply, borrower learning) before aggregate HMDA evidence. Literature positioning sits inside the introduction, followed by a roadmap. The body covers data and strategy, main IV results, mechanisms, macro channels, and a short placebo robustness section. The conclusion restates the findings, favours an informational interpretation, and closes with bank-run analogies and policy implications.", "writing_profile_keywords": ["instrumental variables", "contract-detail instruments", "peer effects", "foreclosure spillovers", "mechanism decomposition", "single-author first person", "geocoded administrative data"], "confidence": 0.8, "notes": "The roadmap labels Section V as the conclusion, but in the section list Section V is Robustness and VI is the Conclusion. The abstract is truncated and has no explicit gap or contribution sentence. Standard error treatment is not visible in the excerpts. Mechanism and macro sections were classified as results despite the rule-based robustness label."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.4, "usage": {"input": 10593, "output": 2469, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.092546, "request_id": "req_011CggqU1ZKgq8Gy1WyCxH3S", "cached": false, "ts": "2026-11-03T23:08:56+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/c764aa4df1930139041f995ceca5719d1bdc8e60695c98b84bf8211f49e9fcd5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"survey_or_address\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical address)\",\"strategy\":\"not applicable; illustrative theoretical models (fables) with analytical propositions\",\"key_assumptions\":[\"agents fail to fully adjust for transmission bias in received signals\",\"random pairwise meetings among agents\"],\"threats_addressed\":[],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":0},\"gap\":{\"present\":true,\"sentence_index\":2},\"method\":{\"present\":true,\"sentence_index\":3},\"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\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"literature_gap\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":8,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":9,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":5,\"contribution_types\":[\"new_theory_or_model\",\"synthesis\"],\"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\":\"conclusion\"},{\"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. Figures are a schematic summary of themes, model timelines and simulated or expected price paths.\",\"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\":[\"modal verbs (may, can help explain)\",\"framing models as illustrative fables\",\"noting work-in-progress status\",\"suggest rather than show\"]},\"claim_evidence\":[{\"claim\":\"Social transmission bias compounds recursively, so small biases can produce large aggregate effects such as booms and bubbles\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV, Propositions 1-11, Appendix\"},{\"claim\":\"Self-enhancing transmission favors the spread of high-variance or high-skew active strategies without a direct psychological preference\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section II; Han, Hirshleifer, and Walden (2019a)\"},{\"claim\":\"Visibility bias in consumption leads to overconsumption\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section III; Han, Hirshleifer, and Walden (2019b)\"},{\"claim\":\"Biased information percolation generates bubble price paths with distinct phases\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Contagion with buzz in an asset market yields damped oscillating prices through overreaction and overcorrection\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Socially emergent behavior can mimic individual propensities, so inferences from behavioral tests may be overstated\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Introduction; Section VI; Figure 2\"}],\"structural_summary\":\"A presidential address that proposes a new paradigm and defines its core concept, then contrasts it with earlier paradigm shifts to identify a gap in the toolkit. It works through five model-based fables: three summarize prior coauthored work, one is a long new model with formal propositions and proofs in the appendix, and one sketches work-in-progress. It closes with a list of emergent themes that act as the conclusion, including suggested empirical proxies.\",\"writing_profile_keywords\":[\"presidential address\",\"paradigm proposal\",\"fables\",\"analytical models\",\"first-person voice\",\"conceptual synthesis\",\"propositions with appendix proofs\",\"emergent themes\"],\"confidence\":0.8,\"notes\":\"No conclusion paragraphs were provided; the Emergent Themes section is treated as the conclusion. Fable 4 is the paper's main new model and contains propositions; Fables 1-3 summarize other work.\"}", "json": {"genre": "survey_or_address", "subfield": "behavioral_finance", "subfield_secondary": "asset_pricing", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical address)", "strategy": "not applicable; illustrative theoretical models (fables) with analytical propositions", "key_assumptions": ["agents fail to fully adjust for transmission bias in received signals", "random pairwise meetings among agents"], "threats_addressed": [], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": true, "sentence_index": 0}, "gap": {"present": true, "sentence_index": 2}, "method": {"present": true, "sentence_index": 3}, "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": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "tension_or_puzzle"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "literature_gap"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 8, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 9, "move": "roadmap"}], "contribution_paragraph_index": 5, "contribution_types": ["new_theory_or_model", "synthesis"], "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": "conclusion"}, {"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. Figures are a schematic summary of themes, model timelines and simulated or expected price paths.", "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": ["modal verbs (may, can help explain)", "framing models as illustrative fables", "noting work-in-progress status", "suggest rather than show"]}, "claim_evidence": [{"claim": "Social transmission bias compounds recursively, so small biases can produce large aggregate effects such as booms and bubbles", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV, Propositions 1-11, Appendix"}, {"claim": "Self-enhancing transmission favors the spread of high-variance or high-skew active strategies without a direct psychological preference", "evidence_type": "prior_literature", "evidence_ref": "Section II; Han, Hirshleifer, and Walden (2019a)"}, {"claim": "Visibility bias in consumption leads to overconsumption", "evidence_type": "prior_literature", "evidence_ref": "Section III; Han, Hirshleifer, and Walden (2019b)"}, {"claim": "Biased information percolation generates bubble price paths with distinct phases", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Contagion with buzz in an asset market yields damped oscillating prices through overreaction and overcorrection", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Socially emergent behavior can mimic individual propensities, so inferences from behavioral tests may be overstated", "evidence_type": "none_stated", "evidence_ref": "Introduction; Section VI; Figure 2"}], "structural_summary": "A presidential address that proposes a new paradigm and defines its core concept, then contrasts it with earlier paradigm shifts to identify a gap in the toolkit. It works through five model-based fables: three summarize prior coauthored work, one is a long new model with formal propositions and proofs in the appendix, and one sketches work-in-progress. It closes with a list of emergent themes that act as the conclusion, including suggested empirical proxies.", "writing_profile_keywords": ["presidential address", "paradigm proposal", "fables", "analytical models", "first-person voice", "conceptual synthesis", "propositions with appendix proofs", "emergent themes"], "confidence": 0.8, "notes": "No conclusion paragraphs were provided; the Emergent Themes section is treated as the conclusion. Fable 4 is the paper's main new model and contains propositions; Fables 1-3 summarize other work."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.83, "usage": {"input": 4983, "output": 1849, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.057706, "request_id": "req_011CggqqmysEjCvXEvLQrPdC", "cached": false, "ts": "2026-11-03T23:13:47+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ca2731f10192511f085eff32d161d0394967411500523c7c89867d637097321b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"corporate governance / organizational economics (cheap talk, control rights)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; equilibrium analysis of a principal-agent cheap-talk model with costly intervention\",\"key_assumptions\":[\"incomplete contracts: actions and messages not contractible\",\"top-down information structure with privately informed principal\",\"cheap talk communication a la Crawford and Sobel\",\"agent biased toward overinvestment\",\"costly intervention by principal\"],\"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\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":1,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"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\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"conclusion\"},{\"idx\":5,\"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 stated as propositions, with proofs in Appendix A and additional material in an Internet Appendix. Figures plot numerical examples at stated parameter values.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"conditional statements tied to parameter regions (e.g., low intervention cost, severe agency problem)\",\"modal verbs such as 'can harm' and 'can reduce'\"]},\"claim_evidence\":[{\"claim\":\"A principal's power to intervene limits her ability to communicate private information effectively.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.C; Appendix A\"},{\"claim\":\"Intervention can harm the principal, especially when the intervention cost is low or the agency problem is severe.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I.D.1; Figure 2\"},{\"claim\":\"Welfare effects of intervention on the agent are characterized.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section I.D.2; Figure 3\"},{\"claim\":\"Expected intervention is lower in equilibrium with effective communication than without communication.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Internet Appendix Section I.D\"},{\"claim\":\"Pay-for-performance contracts alter the principal's welfare from intervention.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Section II.A; Figure 4\"}],\"structural_summary\":\"The paper opens with an epigraph, then motivates the topic with governance practice and notes that the interplay of communication and control is poorly understood. It states the main result and its implications up front, then describes the cheap-talk model. The main Analysis section covers setup, equilibrium, the effect of intervention on communication, and welfare. Extensions relax assumptions: incentive pay, an agent's intervention cost, and an informed agent. The paper then derives testable predictions and applies them to managerial leadership, boards, private equity, and activism. A brief conclusion restates the result, and formal proofs are in the appendix.\",\"writing_profile_keywords\":[\"cheap talk\",\"principal-agent\",\"control rights\",\"intervention\",\"corporate governance\",\"counterintuitive result\",\"propositions and proofs\",\"applications section\"],\"confidence\":0.85,\"notes\":\"Pure theory paper. Empirical predictions are offered in Section III but not tested. The introduction excerpt is truncated, so later intro moves (literature positioning, roadmap) are not visible; the roadmap is set to false based on what is visible.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "corporate governance / organizational economics (cheap talk, control rights)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; equilibrium analysis of a principal-agent cheap-talk model with costly intervention", "key_assumptions": ["incomplete contracts: actions and messages not contractible", "top-down information structure with privately informed principal", "cheap talk communication a la Crawford and Sobel", "agent biased toward overinvestment", "costly intervention by principal"], "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": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "other"}, {"paragraph_index": 1, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "policy_or_broader_implication"}, {"paragraph_index": 1, "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": "theory"}, {"idx": 2, "role": "theory"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "conclusion"}, {"idx": 5, "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 stated as propositions, with proofs in Appendix A and additional material in an Internet Appendix. Figures plot numerical examples at stated parameter values.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["conditional statements tied to parameter regions (e.g., low intervention cost, severe agency problem)", "modal verbs such as 'can harm' and 'can reduce'"]}, "claim_evidence": [{"claim": "A principal's power to intervene limits her ability to communicate private information effectively.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.C; Appendix A"}, {"claim": "Intervention can harm the principal, especially when the intervention cost is low or the agency problem is severe.", "evidence_type": "equation_or_proof", "evidence_ref": "Section I.D.1; Figure 2"}, {"claim": "Welfare effects of intervention on the agent are characterized.", "evidence_type": "figure", "evidence_ref": "Section I.D.2; Figure 3"}, {"claim": "Expected intervention is lower in equilibrium with effective communication than without communication.", "evidence_type": "equation_or_proof", "evidence_ref": "Internet Appendix Section I.D"}, {"claim": "Pay-for-performance contracts alter the principal's welfare from intervention.", "evidence_type": "figure", "evidence_ref": "Section II.A; Figure 4"}], "structural_summary": "The paper opens with an epigraph, then motivates the topic with governance practice and notes that the interplay of communication and control is poorly understood. It states the main result and its implications up front, then describes the cheap-talk model. The main Analysis section covers setup, equilibrium, the effect of intervention on communication, and welfare. Extensions relax assumptions: incentive pay, an agent's intervention cost, and an informed agent. The paper then derives testable predictions and applies them to managerial leadership, boards, private equity, and activism. A brief conclusion restates the result, and formal proofs are in the appendix.", "writing_profile_keywords": ["cheap talk", "principal-agent", "control rights", "intervention", "corporate governance", "counterintuitive result", "propositions and proofs", "applications section"], "confidence": 0.85, "notes": "Pure theory paper. Empirical predictions are offered in Section III but not tested. The introduction excerpt is truncated, so later intro moves (literature positioning, roadmap) are not visible; the roadmap is set to false based on what is visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.25, "usage": {"input": 3306, "output": 1692, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.047858, "request_id": "req_011CggqgU1dCjbFbp3AVf6A9", "cached": false, "ts": "2026-11-03T23:11:40+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/cb6cee47a4f2ab19e457837a07091a576eebea18113f0cd7a30b3a63b056c231.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"not applicable (theory paper); model-implied conditional and unconditional predictability of characteristic-sorted portfolio returns\",\"strategy\":\"Dynamic behavioral model with an economy-wide disruption factor and overconfident investors, calibrated to US data from 1964 to 2014 and analyzed through simulated 50-year histories\",\"key_assumptions\":[\"risk-neutral investors discounting at a constant rate\",\"representative agent with common beliefs and information\",\"frictionless financing with Modigliani-Miller holding\",\"investors overconfident about their ability to assess the disruption climate\",\"zero unconditional return predictability over long samples\"],\"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\":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\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":9,\"move\":\"preview_of_approach\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"other\"},{\"paragraph_index\":11,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":13,\"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\":\"theory\"},{\"idx\":2,\"role\":\"theory\"},{\"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\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Tables report calibrated parameters and statistics from simulated data: betas on the disruption factor, conditional Sharpe ratios, autocorrelations and Sharpe-ratio percentiles. Notes describe how each statistic is computed and point to the corresponding text sections.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"explicit acknowledgment of stylized assumptions\",\"statements that the model does not capture all features of the data\",\"'up to' qualifiers on simulated frequencies\",\"framing as testable implications rather than established facts\"]},\"claim_evidence\":[{\"claim\":\"Portfolios favoring newcomers load positively on disruption and those favoring incumbents load negatively.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Overestimating the disruption climate predicts low subsequent returns for newcomer-favoring strategies, and underestimation predicts the reverse.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Characteristic-sorted portfolio returns are positively autocorrelated, so momentum across portfolios is profitable.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Return persistence makes high 50-year Sharpe ratios (e.g., 0.40) occur in up to 20% of simulated histories.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 1\"},{\"claim\":\"The calibration broadly matches salient features of US firm data.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I and II\"}],\"structural_summary\":\"The paper opens by noting the documented characteristic anomalies and the scarcity of work on their conditional behavior. It then builds a continuous-time behavioral model in which a disruption factor shifts profits from incumbents to newcomers and overconfident investors misread that factor. The model is calibrated to US data, and simulated panels are used to show portfolio exposures to disruption, conditional predictability and return autocorrelation. A final analysis asks how often historical-length samples would show high Sharpe ratios. The paper closes by acknowledging its simplifications and stating testable implications.\",\"writing_profile_keywords\":[\"behavioral asset pricing\",\"dynamic model\",\"calibration\",\"simulation\",\"overconfidence\",\"value/profitability/asset growth anomalies\",\"testable implications\",\"stylized model caveats\"],\"confidence\":0.8,\"notes\":\"The introduction paragraph list contains broken fragments (indices 1-8 are single words), so moves are assigned to indices 0 and 9-13. The paper is a calibrated theoretical model with no empirical identification. Statistics come from simulated samples, and standard errors are not applicable.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "calibration_simulation", "identification": {"estimand": "not applicable (theory paper); model-implied conditional and unconditional predictability of characteristic-sorted portfolio returns", "strategy": "Dynamic behavioral model with an economy-wide disruption factor and overconfident investors, calibrated to US data from 1964 to 2014 and analyzed through simulated 50-year histories", "key_assumptions": ["risk-neutral investors discounting at a constant rate", "representative agent with common beliefs and information", "frictionless financing with Modigliani-Miller holding", "investors overconfident about their ability to assess the disruption climate", "zero unconditional return predictability over long samples"], "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": 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": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 9, "move": "preview_of_approach"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "other"}, {"paragraph_index": 11, "move": "preview_of_results"}, {"paragraph_index": 11, "move": "related_literature_positioning"}, {"paragraph_index": 12, "move": "related_literature_positioning"}, {"paragraph_index": 13, "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": "theory"}, {"idx": 2, "role": "theory"}, {"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": "none_seen", "types": []}, "tables": {"convention_notes": "Tables report calibrated parameters and statistics from simulated data: betas on the disruption factor, conditional Sharpe ratios, autocorrelations and Sharpe-ratio percentiles. Notes describe how each statistic is computed and point to the corresponding text sections.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["explicit acknowledgment of stylized assumptions", "statements that the model does not capture all features of the data", "'up to' qualifiers on simulated frequencies", "framing as testable implications rather than established facts"]}, "claim_evidence": [{"claim": "Portfolios favoring newcomers load positively on disruption and those favoring incumbents load negatively.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Overestimating the disruption climate predicts low subsequent returns for newcomer-favoring strategies, and underestimation predicts the reverse.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Characteristic-sorted portfolio returns are positively autocorrelated, so momentum across portfolios is profitable.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Return persistence makes high 50-year Sharpe ratios (e.g., 0.40) occur in up to 20% of simulated histories.", "evidence_type": "table", "evidence_ref": "Table VI; Figure 1"}, {"claim": "The calibration broadly matches salient features of US firm data.", "evidence_type": "table", "evidence_ref": "Tables I and II"}], "structural_summary": "The paper opens by noting the documented characteristic anomalies and the scarcity of work on their conditional behavior. It then builds a continuous-time behavioral model in which a disruption factor shifts profits from incumbents to newcomers and overconfident investors misread that factor. The model is calibrated to US data, and simulated panels are used to show portfolio exposures to disruption, conditional predictability and return autocorrelation. A final analysis asks how often historical-length samples would show high Sharpe ratios. The paper closes by acknowledging its simplifications and stating testable implications.", "writing_profile_keywords": ["behavioral asset pricing", "dynamic model", "calibration", "simulation", "overconfidence", "value/profitability/asset growth anomalies", "testable implications", "stylized model caveats"], "confidence": 0.8, "notes": "The introduction paragraph list contains broken fragments (indices 1-8 are single words), so moves are assigned to indices 0 and 9-13. The paper is a calibrated theoretical model with no empirical identification. Statistics come from simulated samples, and standard errors are not applicable."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 22.13, "usage": {"input": 5729, "output": 2576, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07523, "request_id": "req_011CggqbGrGUfGpf6dcpqk1x", "cached": false, "ts": "2026-11-03T23:10:37+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/cc9bc1fb238a779ecd51580db2319875f4d025ed3ad960de667ecd81a8404234.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"Diversification and performance gains, measured by correlations, Sharpe ratios, spanning-test statistics and certainty-equivalent returns, from adding trade-weighted developed-market portfolios that target emerging economies (EE indices) to a developed-market benchmark.\",\"strategy\":\"Build EE indices by weighting developed-market industry portfolios by their exports to each emerging country, then benchmark them against EM-mimicking portfolios using DCC correlations, factor alphas, mean-variance and step-down spanning tests, out-of-sample minimum-variance portfolios, certainty-equivalent rates, and firm-level FactSet Revere reconstructions.\",\"key_assumptions\":[\"Industry exports to an emerging country proxy for exposure to that country's economic activity.\",\"Past return moments are informative about diversification benefits.\",\"EM-mimicking portfolios are the appropriate benchmark.\"],\"threats_addressed\":[\"Industry-level data may be too coarse (addressed with firm-level data).\",\"Comtrade data mix public and private firms.\",\"Performance differences driven by alphas (addressed with demeaned spanning tests).\",\"In-sample bias (addressed with out-of-sample minimum-variance portfolios).\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"sentence_index\":-1},\"gap\":{\"present\":true,\"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\":4}},\"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\":2,\"move\":\"related_literature_positioning\"},{\"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\":\"policy_or_broader_implication\"},{\"paragraph_index\":7,\"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\":\"results\"},{\"idx\":3,\"role\":\"robustness\"},{\"idx\":4,\"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\"]},\"tables\":{\"convention_notes\":\"Notes open with 'This table reports...', define the test and benchmark assets, write out the regression models explicitly, and organize content into panels. Comparisons are made across EE, EM-mimicking and MSCI DM portfolios.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"no\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'suggest' for interpreting results\",\"'may offer' benefits\",\"'we expect' for extensions\",\"'potentially' in firm-level framing\"]},\"claim_evidence\":[{\"claim\":\"Listed firms represent a small share of emerging economies.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"EE portfolio returns load more strongly on emerging-economy activity proxies than EM-mimicking indices do.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"EE index correlations with developed markets remain low and stable over time.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2-3\"},{\"claim\":\"EE portfolios have higher Sharpe ratios than DM and EM-mimicking portfolios.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Figure 4\"},{\"claim\":\"EE indices significantly shift the mean-variance frontier.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V-VI\"},{\"claim\":\"Firm-level EE indices show low correlations and higher Sharpe ratios.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X, Figures 6-7\"}],\"structural_summary\":\"The paper opens by noting that equity indices understate diversification gains because emerging markets' listed sectors are small. It proposes trade-weighted developed-market portfolios (EE indices) as a new route to emerging-economy exposure and positions this against the EM-mimicking approach in prior work. It then validates that EE indices track economic activity. Results present correlations, return moments, alphas, spanning tests, out-of-sample portfolios and certainty equivalents. A firm-level section confirms the findings, and the conclusion restates the 'best of both worlds' message and points to applications to other nontraded assets.\",\"writing_profile_keywords\":[\"portfolio construction\",\"international diversification\",\"benchmark comparison\",\"spanning tests\",\"method generalization\",\"numbered findings\"],\"confidence\":0.8,\"notes\":\"The primary method is portfolio construction with performance evaluation, coded as 'other'. Statistical significance conventions are not visible in the excerpts. Robustness is spread across Section III and an Internet Appendix.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "international_finance", "primary_method": "other", "identification": {"estimand": "Diversification and performance gains, measured by correlations, Sharpe ratios, spanning-test statistics and certainty-equivalent returns, from adding trade-weighted developed-market portfolios that target emerging economies (EE indices) to a developed-market benchmark.", "strategy": "Build EE indices by weighting developed-market industry portfolios by their exports to each emerging country, then benchmark them against EM-mimicking portfolios using DCC correlations, factor alphas, mean-variance and step-down spanning tests, out-of-sample minimum-variance portfolios, certainty-equivalent rates, and firm-level FactSet Revere reconstructions.", "key_assumptions": ["Industry exports to an emerging country proxy for exposure to that country's economic activity.", "Past return moments are informative about diversification benefits.", "EM-mimicking portfolios are the appropriate benchmark."], "threats_addressed": ["Industry-level data may be too coarse (addressed with firm-level data).", "Comtrade data mix public and private firms.", "Performance differences driven by alphas (addressed with demeaned spanning tests).", "In-sample bias (addressed with out-of-sample minimum-variance portfolios)."], "causal_language_strength": 2, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "sentence_index": -1}, "gap": {"present": true, "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": 4}}, "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": 2, "move": "related_literature_positioning"}, {"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": "policy_or_broader_implication"}, {"paragraph_index": 7, "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": "results"}, {"idx": 3, "role": "robustness"}, {"idx": 4, "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"]}, "tables": {"convention_notes": "Notes open with 'This table reports...', define the test and benchmark assets, write out the regression models explicitly, and organize content into panels. Comparisons are made across EE, EM-mimicking and MSCI DM portfolios.", "units_stated_in_notes": false, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "no"}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'suggest' for interpreting results", "'may offer' benefits", "'we expect' for extensions", "'potentially' in firm-level framing"]}, "claim_evidence": [{"claim": "Listed firms represent a small share of emerging economies.", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "EE portfolio returns load more strongly on emerging-economy activity proxies than EM-mimicking indices do.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "EE index correlations with developed markets remain low and stable over time.", "evidence_type": "figure", "evidence_ref": "Figures 2-3"}, {"claim": "EE portfolios have higher Sharpe ratios than DM and EM-mimicking portfolios.", "evidence_type": "table", "evidence_ref": "Table III, Figure 4"}, {"claim": "EE indices significantly shift the mean-variance frontier.", "evidence_type": "table", "evidence_ref": "Tables V-VI"}, {"claim": "Firm-level EE indices show low correlations and higher Sharpe ratios.", "evidence_type": "table", "evidence_ref": "Table X, Figures 6-7"}], "structural_summary": "The paper opens by noting that equity indices understate diversification gains because emerging markets' listed sectors are small. It proposes trade-weighted developed-market portfolios (EE indices) as a new route to emerging-economy exposure and positions this against the EM-mimicking approach in prior work. It then validates that EE indices track economic activity. Results present correlations, return moments, alphas, spanning tests, out-of-sample portfolios and certainty equivalents. A firm-level section confirms the findings, and the conclusion restates the 'best of both worlds' message and points to applications to other nontraded assets.", "writing_profile_keywords": ["portfolio construction", "international diversification", "benchmark comparison", "spanning tests", "method generalization", "numbered findings"], "confidence": 0.8, "notes": "The primary method is portfolio construction with performance evaluation, coded as 'other'. Statistical significance conventions are not visible in the excerpts. Robustness is spread across Section III and an Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.75, "usage": {"input": 7734, "output": 2062, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07297, "request_id": "req_011CggqWQoNfKa6nntCsaErj", "cached": false, "ts": "2026-11-03T23:09:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ce6960dc55d960a84a6a597149fe7c6595619d259e9603c79e80ae16d343f9ed.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"international_finance\",\"primary_method\":\"other\",\"identification\":{\"estimand\":\"How strongly dealers' quote changes co-move after efficient-price shocks (network links), and how a dealer's network centrality relates to how informative its quotes are, how fast it absorbs shocks, and how wide its spreads are.\",\"strategy\":\"Hasbrouck-style VECM with a common efficient-price trend. Partial correlations of multivariate impulse responses form edges, and price inefficiencies give node positions. The authors run Granger causality tests between centrality and inefficiency, condition on public news releases, and use the surprise 2015 removal of the EUR/CHF peg as a natural experiment for how the network forms.\",\"key_assumptions\":[\"Dealer and EBS quotes are cointegrated with a single common stochastic trend that can be read as the efficient price.\",\"Quote synchronicity after shocks reflects information linkages rather than physical trading connections.\",\"The removal of the peg was an exogenous move from a no-learning regime to a learning regime.\"],\"threats_addressed\":[\"Confusing co-activity networks with networks based on physical trade links.\",\"Choice of impulse-response horizon for the network snapshot.\"],\"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\":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\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_method\",\"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\":\"empirical_design\"},{\"idx\":3,\"role\":\"empirical_design\"},{\"idx\":4,\"role\":\"data\"},{\"idx\":5,\"role\":\"results\"},{\"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\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes define each network or information measure (adjacency matrix as partial correlations, InfoShare lower bound, InfoSpeed, centrality thresholds 0.2/0.1) and the sample construction (five-minute subsamples, dealer markets DLR1-8 plus EBS). Titles are often phrased as hypothesis questions.\",\"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\":[\"'largely support theory'\",\"'in a sense'\",\"'to the best of our knowledge'\",\"'we believe'\",\"'hopefully inspire future researchers'\"]},\"claim_evidence\":[{\"claim\":\"The information-revelation network is nontrivial, stable over time, and has a core-periphery structure.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Table I\"},{\"claim\":\"Central dealers' quotes are more informative at the shock and reach the full impact faster.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4; Table II; Table IV\"},{\"claim\":\"Central dealers quote wider bid-ask spreads, consistent with recouping adverse-selection costs.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"When part of the information is released publicly via news, links weaken and revelation slows, consistent with reduced search.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"After the peg removal the market centralized on EBS, then gradually decentralized toward the steady-state network.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 6; Table AI; Figure A1\"}],\"structural_summary\":\"Opens with the fragmentation of trading as motivation and a gap: work on information aggregation in dealer networks is only theoretical. The paper then derives five hypotheses from that theory. Next it builds an econometric network map from a VECM, with impulse-response partial correlations as edges and price inefficiency as node positions, and details estimation and inference. It describes the hybrid FX market and the EUR/CHF quote data, and tests each hypothesis in its own results subsection. The 2015 peg-removal crash serves as a natural experiment on how the network forms. It closes by restating the findings, noting the method applies to other markets, and posing open questions about welfare and regulation.\",\"writing_profile_keywords\":[\"hypothesis-testing structure\",\"theory-motivated empirics\",\"network methodology\",\"cross-disciplinary analogy (neuroscience)\",\"natural experiment\",\"FX microstructure\",\"descriptive network maps\"],\"confidence\":0.75,\"notes\":\"Primary method is coded 'other' because the core tool is a VECM/impulse-response network construction, not standard regression. Subsection D (Hypothesis 4) is missing from the section list but is implied. Statistical inference details are in Section III.B but are not visible in the excerpts. No explicit robustness section is visible.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "international_finance", "primary_method": "other", "identification": {"estimand": "How strongly dealers' quote changes co-move after efficient-price shocks (network links), and how a dealer's network centrality relates to how informative its quotes are, how fast it absorbs shocks, and how wide its spreads are.", "strategy": "Hasbrouck-style VECM with a common efficient-price trend. Partial correlations of multivariate impulse responses form edges, and price inefficiencies give node positions. The authors run Granger causality tests between centrality and inefficiency, condition on public news releases, and use the surprise 2015 removal of the EUR/CHF peg as a natural experiment for how the network forms.", "key_assumptions": ["Dealer and EBS quotes are cointegrated with a single common stochastic trend that can be read as the efficient price.", "Quote synchronicity after shocks reflects information linkages rather than physical trading connections.", "The removal of the peg was an exogenous move from a no-learning regime to a learning regime."], "threats_addressed": ["Confusing co-activity networks with networks based on physical trade links.", "Choice of impulse-response horizon for the network snapshot."], "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": 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": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 4, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_method", "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": "empirical_design"}, {"idx": 3, "role": "empirical_design"}, {"idx": 4, "role": "data"}, {"idx": 5, "role": "results"}, {"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": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes define each network or information measure (adjacency matrix as partial correlations, InfoShare lower bound, InfoSpeed, centrality thresholds 0.2/0.1) and the sample construction (five-minute subsamples, dealer markets DLR1-8 plus EBS). Titles are often phrased as hypothesis questions.", "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": ["'largely support theory'", "'in a sense'", "'to the best of our knowledge'", "'we believe'", "'hopefully inspire future researchers'"]}, "claim_evidence": [{"claim": "The information-revelation network is nontrivial, stable over time, and has a core-periphery structure.", "evidence_type": "table", "evidence_ref": "Table III; Table I"}, {"claim": "Central dealers' quotes are more informative at the shock and reach the full impact faster.", "evidence_type": "figure", "evidence_ref": "Figure 4; Table II; Table IV"}, {"claim": "Central dealers quote wider bid-ask spreads, consistent with recouping adverse-selection costs.", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "When part of the information is released publicly via news, links weaken and revelation slows, consistent with reduced search.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "After the peg removal the market centralized on EBS, then gradually decentralized toward the steady-state network.", "evidence_type": "figure", "evidence_ref": "Figure 6; Table AI; Figure A1"}], "structural_summary": "Opens with the fragmentation of trading as motivation and a gap: work on information aggregation in dealer networks is only theoretical. The paper then derives five hypotheses from that theory. Next it builds an econometric network map from a VECM, with impulse-response partial correlations as edges and price inefficiency as node positions, and details estimation and inference. It describes the hybrid FX market and the EUR/CHF quote data, and tests each hypothesis in its own results subsection. The 2015 peg-removal crash serves as a natural experiment on how the network forms. It closes by restating the findings, noting the method applies to other markets, and posing open questions about welfare and regulation.", "writing_profile_keywords": ["hypothesis-testing structure", "theory-motivated empirics", "network methodology", "cross-disciplinary analogy (neuroscience)", "natural experiment", "FX microstructure", "descriptive network maps"], "confidence": 0.75, "notes": "Primary method is coded 'other' because the core tool is a VECM/impulse-response network construction, not standard regression. Subsection D (Hypothesis 4) is missing from the section list but is implied. Statistical inference details are in Section III.B but are not visible in the excerpts. No explicit robustness section is visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 26.51, "usage": {"input": 6824, "output": 3044, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08897, "request_id": "req_011CggqXyTSCGQhDCwQ9nDa7", "cached": false, "ts": "2026-11-03T23:09:57+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ced67225c925ba08dce1bfb2063b3fcd4f360f742fb69d037435b9b34eee76a1.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model)\",\"strategy\":\"not applicable; equilibrium characterization and comparative statics in a matching model with Bayesian persuasion\",\"key_assumptions\":[\"advisers compensated only upon customer investment (conflicted fee)\",\"rational customers heterogeneous in ex ante information quality\",\"frictionless matching market with pairwise stability\",\"advisers on short side of market (scarce) for fee irrelevance\"],\"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\":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\":\"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\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":4,\"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\":\"theory\"},{\"idx\":4,\"role\":\"robustness\"},{\"idx\":5,\"role\":\"other\"},{\"idx\":6,\"role\":\"conclusion\"},{\"idx\":7,\"role\":\"appendix\"},{\"idx\":8,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"No tables; results are stated as propositions, with proofs in the appendix and illustrative 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\":[\"conditioning results on model assumptions (e.g., 'in our model', 'when advisers are scarce')\",\"'suggest' when drawing empirical implications\"]},\"claim_evidence\":[{\"claim\":\"Competitive equilibrium features dispersed, partially disclosed information under conflicted fees\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Appendix A\"},{\"claim\":\"Banning or lowering conflicted fees changes information quality but not customer welfare when advisers are scarce\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.A\"},{\"claim\":\"Financial literacy education for least informed customers raises all customers' welfare via spillover\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III.B\"},{\"claim\":\"Advised customers earn lower returns due to selection, rationalizing empirical findings\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Section V; Bergstresser et al. (2009), Chalmers and Reuter (2010)\"},{\"claim\":\"Equilibrium information policy is a threshold policy in customer type\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3\"},{\"claim\":\"Sorting result holds under more general information structures and heterogeneity\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV.A\"}],\"structural_summary\":\"A pure theory paper. The introduction opens with the policy relevance of conflicted advice and fee regulation, then names an unexplained point: how advice markets work in equilibrium. It previews a matching-plus-persuasion model and gives intuitive results on information distortion, fee irrelevance, literacy spillovers and selection in returns. It closes by positioning against the Bayesian persuasion, matching and financial-advice literatures. The body sets out the model, characterizes optimal policies and matching, derives regulatory comparative statics, runs robustness extensions and translates the results into empirical implications. Proofs and an alternative game form sit in appendices.\",\"writing_profile_keywords\":[\"equilibrium matching model\",\"Bayesian persuasion\",\"policy-oriented theory\",\"counterintuitive irrelevance result\",\"intuition-driven exposition\",\"proofs in appendix\"],\"confidence\":0.85,\"notes\":\"No tables or data. The robustness section consists of model generalizations rather than empirical checks; Appendix B gives an alternative extensive form. Empirical implications are reconciled with prior evidence only qualitatively.\"}", "json": {"genre": "theoretical", "subfield": "financial_intermediation_banking", "subfield_secondary": "household_finance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model)", "strategy": "not applicable; equilibrium characterization and comparative statics in a matching model with Bayesian persuasion", "key_assumptions": ["advisers compensated only upon customer investment (conflicted fee)", "rational customers heterogeneous in ex ante information quality", "frictionless matching market with pairwise stability", "advisers on short side of market (scarce) for fee irrelevance"], "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": 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": "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": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}], "contribution_paragraph_index": 4, "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": "theory"}, {"idx": 4, "role": "robustness"}, {"idx": 5, "role": "other"}, {"idx": 6, "role": "conclusion"}, {"idx": 7, "role": "appendix"}, {"idx": 8, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "other"]}, "tables": {"convention_notes": "No tables; results are stated as propositions, with proofs in the appendix and illustrative 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": ["conditioning results on model assumptions (e.g., 'in our model', 'when advisers are scarce')", "'suggest' when drawing empirical implications"]}, "claim_evidence": [{"claim": "Competitive equilibrium features dispersed, partially disclosed information under conflicted fees", "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Appendix A"}, {"claim": "Banning or lowering conflicted fees changes information quality but not customer welfare when advisers are scarce", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.A"}, {"claim": "Financial literacy education for least informed customers raises all customers' welfare via spillover", "evidence_type": "equation_or_proof", "evidence_ref": "Section III.B"}, {"claim": "Advised customers earn lower returns due to selection, rationalizing empirical findings", "evidence_type": "prior_literature", "evidence_ref": "Section V; Bergstresser et al. (2009), Chalmers and Reuter (2010)"}, {"claim": "Equilibrium information policy is a threshold policy in customer type", "evidence_type": "figure", "evidence_ref": "Figure 3"}, {"claim": "Sorting result holds under more general information structures and heterogeneity", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV.A"}], "structural_summary": "A pure theory paper. The introduction opens with the policy relevance of conflicted advice and fee regulation, then names an unexplained point: how advice markets work in equilibrium. It previews a matching-plus-persuasion model and gives intuitive results on information distortion, fee irrelevance, literacy spillovers and selection in returns. It closes by positioning against the Bayesian persuasion, matching and financial-advice literatures. The body sets out the model, characterizes optimal policies and matching, derives regulatory comparative statics, runs robustness extensions and translates the results into empirical implications. Proofs and an alternative game form sit in appendices.", "writing_profile_keywords": ["equilibrium matching model", "Bayesian persuasion", "policy-oriented theory", "counterintuitive irrelevance result", "intuition-driven exposition", "proofs in appendix"], "confidence": 0.85, "notes": "No tables or data. The robustness section consists of model generalizations rather than empirical checks; Appendix B gives an alternative extensive form. Empirical implications are reconciled with prior evidence only qualitatively."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.32, "usage": {"input": 5809, "output": 1979, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06361, "request_id": "req_011CggqjhRcLvVKUTfkJ2JEH", "cached": false, "ts": "2026-11-03T23:12:26+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/cee6f622dfc9abc7ae159d8daed818a4276313752bdd31dc72d9510c8488df5b.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"high-frequency trading / speed differentials and liquidity\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Effect of removing speed differentials among traders on adverse selection, trading costs, volume and gains from trade\",\"strategy\":\"Precipitation episodes along the Chicago–New York path that disrupt microwave networks serve as exogenous intraday shocks to speed differentials; complemented by a difference-in-differences event study of a microwave provider opening its network to subscribers (Quincy launch)\",\"key_assumptions\":[\"Precipitation disrupts microwave transmission and slows the fastest traders\",\"Weather affects liquidity only through the speed channel (exclusion), not through mood or pessimism\",\"The Quincy launch removed speed differentials without coinciding confounding shocks\"],\"threats_addressed\":[\"Weather-induced investor mood or pessimism (Internet Appendix)\",\"Premise that precipitation slows information transmission (Figure 2)\",\"Controls for volume and volatility in regressions\",\"Short duration of episodes limiting volume effects, addressed with the long-term event\"],\"causal_language_strength\":3,\"uses_causal_claim\":true},\"abstract_moves\":{\"question\":{\"present\":true,\"sentence_index\":1},\"gap\":{\"present\":true,\"sentence_index\":1},\"method\":{\"present\":true,\"sentence_index\":2},\"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\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"other\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"},{\"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\",\"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\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_measure\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes describe the sample and data source (DTAQ, ITCH, CME) in each title and write out the panel regression equation. Panels contrast heavy- and low-precipitation intervals before showing regression coefficients. Data are aggregated into 15-minute intervals.\",\"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\":[\"'associated with' wording\",\"'consistent with the notion'\",\"'seems to suggest'\",\"acknowledging that 30-minute episodes may be too short\",\"'it is possible, however'\"]},\"claim_evidence\":[{\"claim\":\"Heavy precipitation slows futures-to-ETF information transmission, supporting the premise that microwave networks are disrupted\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Precipitation episodes are associated with lower trade price impacts (adverse selection)\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Spreads and trading costs decline during precipitation episodes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Limit order aggressiveness changes when speed differentials are removed\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Total volume falls in rain because arbitrage volume falls, while end-user volume is unchanged\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"Long-term removal of speed differentials via the Quincy launch lowers adverse selection and costs and raises volume\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Weather-induced mood does not drive the results\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Internet Appendix\"}],\"structural_summary\":\"The introduction sets up a theoretical tension: speed differentials may help or hurt liquidity. It then introduces weather disruptions of microwave links as an exogenous shock and previews the liquidity results. It concedes that the gains-from-trade evidence is inconclusive and introduces a second, long-horizon event to address that. Literature positioning follows, covering both theory and weather-as-shock studies, then an enumerated contribution list and a roadmap. The body gives institutional background on transmission physics, describes the data, and presents results in sequence: premise validation, adverse selection, spreads, order aggressiveness, volume/volatility, arbitrage vs end-user volume, and the event study. The conclusion restates the findings.\",\"writing_profile_keywords\":[\"natural experiment\",\"weather shock\",\"high-frequency data\",\"institutional detail\",\"theory-testing\",\"two complementary settings\",\"candid limitation\"],\"confidence\":0.8,\"notes\":\"Standard errors and significance conventions are not visible in the table note excerpts. Section I is labeled 'other' because it covers institutional background, not a literature review.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "high-frequency trading / speed differentials and liquidity", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Effect of removing speed differentials among traders on adverse selection, trading costs, volume and gains from trade", "strategy": "Precipitation episodes along the Chicago–New York path that disrupt microwave networks serve as exogenous intraday shocks to speed differentials; complemented by a difference-in-differences event study of a microwave provider opening its network to subscribers (Quincy launch)", "key_assumptions": ["Precipitation disrupts microwave transmission and slows the fastest traders", "Weather affects liquidity only through the speed channel (exclusion), not through mood or pessimism", "The Quincy launch removed speed differentials without coinciding confounding shocks"], "threats_addressed": ["Weather-induced investor mood or pessimism (Internet Appendix)", "Premise that precipitation slows information transmission (Figure 2)", "Controls for volume and volatility in regressions", "Short duration of episodes limiting volume effects, addressed with the long-term event"], "causal_language_strength": 3, "uses_causal_claim": true}, "abstract_moves": {"question": {"present": true, "sentence_index": 1}, "gap": {"present": true, "sentence_index": 1}, "method": {"present": true, "sentence_index": 2}, "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": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "other"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "contribution_statement"}, {"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", "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": true}, "robustness": {"location": "mixed", "types": ["alternative_measure", "falsification", "other"]}, "tables": {"convention_notes": "Notes describe the sample and data source (DTAQ, ITCH, CME) in each title and write out the panel regression equation. Panels contrast heavy- and low-precipitation intervals before showing regression coefficients. Data are aggregated into 15-minute intervals.", "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": ["'associated with' wording", "'consistent with the notion'", "'seems to suggest'", "acknowledging that 30-minute episodes may be too short", "'it is possible, however'"]}, "claim_evidence": [{"claim": "Heavy precipitation slows futures-to-ETF information transmission, supporting the premise that microwave networks are disrupted", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Precipitation episodes are associated with lower trade price impacts (adverse selection)", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Spreads and trading costs decline during precipitation episodes", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Limit order aggressiveness changes when speed differentials are removed", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Total volume falls in rain because arbitrage volume falls, while end-user volume is unchanged", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "Long-term removal of speed differentials via the Quincy launch lowers adverse selection and costs and raises volume", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Weather-induced mood does not drive the results", "evidence_type": "none_stated", "evidence_ref": "Internet Appendix"}], "structural_summary": "The introduction sets up a theoretical tension: speed differentials may help or hurt liquidity. It then introduces weather disruptions of microwave links as an exogenous shock and previews the liquidity results. It concedes that the gains-from-trade evidence is inconclusive and introduces a second, long-horizon event to address that. Literature positioning follows, covering both theory and weather-as-shock studies, then an enumerated contribution list and a roadmap. The body gives institutional background on transmission physics, describes the data, and presents results in sequence: premise validation, adverse selection, spreads, order aggressiveness, volume/volatility, arbitrage vs end-user volume, and the event study. The conclusion restates the findings.", "writing_profile_keywords": ["natural experiment", "weather shock", "high-frequency data", "institutional detail", "theory-testing", "two complementary settings", "candid limitation"], "confidence": 0.8, "notes": "Standard errors and significance conventions are not visible in the table note excerpts. Section I is labeled 'other' because it covers institutional background, not a literature review."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.1, "usage": {"input": 6230, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069154, "request_id": "req_011CggqzdcpsN7TKQCHG7Uoy", "cached": false, "ts": "2026-11-03T23:15:50+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d00a527123aa739d540a9c652d017e00d716308ddc7acac385ebb16e70230870.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"corporate_finance (capital structure / structural credit risk models)\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Cross-sectional pricing of unlevered market beta, size, BTM and volatility in leverage-adjusted (unlevered) equity returns, compared with levered stock returns\",\"strategy\":\"Unlever stock returns in two ways: data-driven scaling by (1 - leverage), and a parametric Merton or Leland-Toft structural model. Then run portfolio sorts and Fama-MacBeth regressions on the unlevered returns.\",\"key_assumptions\":[\"Linear asset pricing models allow levered risk to be decomposed into asset risk plus leverage-related risk\",\"The structural credit model (Merton/Leland-Toft) correctly maps equity returns to asset returns\",\"Book liabilities and assumed debt maturity proxy adequately for the debt structure\"],\"threats_addressed\":[\"Model dependence of the unlevering (checked with Leland-Toft and the scaling approach)\",\"Debt maturity and face-value measurement choices\",\"Portfolio weighting choices\",\"Nonlinearity of leverage that regression controls fail to capture (heteroskedasticity)\"],\"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\":\"tension_or_puzzle\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"robustness_mention\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"new_evidence\",\"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\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"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\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Panels report value-weighted returns for 25 double-sorted or 5 single-sorted portfolios, followed by Fama-MacBeth regression panels with t-statistics in parentheses. Notes define the leverage measure and its timing, and give the simulation parameters.\",\"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\":[\"'in part' qualifiers\",\"'suggest'\",\"'virtually disappear' / 'weakened'\",\"'consistent with theoretical predictions'\"]},\"claim_evidence\":[{\"claim\":\"Unlevered market beta is more strongly priced than levered beta, even after controlling for size and BTM\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables I, III, IV; Figure 2\"},{\"claim\":\"The value premium virtually disappears and the size effect weakens for unlevered returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV\"},{\"claim\":\"Leverage induces heteroskedasticity in stock returns that unlevering removes\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Figure 3\"},{\"claim\":\"Controlling for leverage in regressions does not adequately address its nonlinear effect\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The volatility puzzle holds for levered returns but not for unlevered returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Results are robust to debt maturity, face value of debt, the Leland-Toft model and weighting choices\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI, VII\"},{\"claim\":\"Derivative of expected levered returns with respect to levered volatility is negative\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Appendix, Equation (3)\"}],\"structural_summary\":\"The paper opens with a question and motivates it with well-known anomalies, then positions itself as the first to test hypotheses on unlevered returns. A theory section derives predictions for asset beta, size, BTM and volatility. After the data section, the empirical work unlevers returns twice: first by data-driven leverage scaling, then with the Merton model, and shows why leverage controls in regressions are inadequate. A dedicated robustness section varies the model and measurement choices, and a separate section treats volatility. The conclusion restates the contribution, and an appendix contains a proof.\",\"writing_profile_keywords\":[\"unlevered returns\",\"anomalies\",\"structural credit models\",\"Fama-MacBeth\",\"portfolio sorts\",\"theory-derived hypotheses\",\"parsimonious factor model\"],\"confidence\":0.8,\"notes\":\"The introduction excerpt is truncated, so a roadmap may exist but is not visible. The standard-error type is not stated in the visible excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "corporate_finance (capital structure / structural credit risk models)", "primary_method": "reduced_form_regression", "identification": {"estimand": "Cross-sectional pricing of unlevered market beta, size, BTM and volatility in leverage-adjusted (unlevered) equity returns, compared with levered stock returns", "strategy": "Unlever stock returns in two ways: data-driven scaling by (1 - leverage), and a parametric Merton or Leland-Toft structural model. Then run portfolio sorts and Fama-MacBeth regressions on the unlevered returns.", "key_assumptions": ["Linear asset pricing models allow levered risk to be decomposed into asset risk plus leverage-related risk", "The structural credit model (Merton/Leland-Toft) correctly maps equity returns to asset returns", "Book liabilities and assumed debt maturity proxy adequately for the debt structure"], "threats_addressed": ["Model dependence of the unlevering (checked with Leland-Toft and the scaling approach)", "Debt maturity and face-value measurement choices", "Portfolio weighting choices", "Nonlinearity of leverage that regression controls fail to capture (heteroskedasticity)"], "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": "tension_or_puzzle"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "robustness_mention"}], "contribution_paragraph_index": 0, "contribution_types": ["new_evidence", "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": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"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": false}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "alternative_measure", "sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "Panels report value-weighted returns for 25 double-sorted or 5 single-sorted portfolios, followed by Fama-MacBeth regression panels with t-statistics in parentheses. Notes define the leverage measure and its timing, and give the simulation parameters.", "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": ["'in part' qualifiers", "'suggest'", "'virtually disappear' / 'weakened'", "'consistent with theoretical predictions'"]}, "claim_evidence": [{"claim": "Unlevered market beta is more strongly priced than levered beta, even after controlling for size and BTM", "evidence_type": "table", "evidence_ref": "Tables I, III, IV; Figure 2"}, {"claim": "The value premium virtually disappears and the size effect weakens for unlevered returns", "evidence_type": "table", "evidence_ref": "Tables III, IV"}, {"claim": "Leverage induces heteroskedasticity in stock returns that unlevering removes", "evidence_type": "figure", "evidence_ref": "Figure 1, Figure 3"}, {"claim": "Controlling for leverage in regressions does not adequately address its nonlinear effect", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "The volatility puzzle holds for levered returns but not for unlevered returns", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Results are robust to debt maturity, face value of debt, the Leland-Toft model and weighting choices", "evidence_type": "table", "evidence_ref": "Tables VI, VII"}, {"claim": "Derivative of expected levered returns with respect to levered volatility is negative", "evidence_type": "equation_or_proof", "evidence_ref": "Appendix, Equation (3)"}], "structural_summary": "The paper opens with a question and motivates it with well-known anomalies, then positions itself as the first to test hypotheses on unlevered returns. A theory section derives predictions for asset beta, size, BTM and volatility. After the data section, the empirical work unlevers returns twice: first by data-driven leverage scaling, then with the Merton model, and shows why leverage controls in regressions are inadequate. A dedicated robustness section varies the model and measurement choices, and a separate section treats volatility. The conclusion restates the contribution, and an appendix contains a proof.", "writing_profile_keywords": ["unlevered returns", "anomalies", "structural credit models", "Fama-MacBeth", "portfolio sorts", "theory-derived hypotheses", "parsimonious factor model"], "confidence": 0.8, "notes": "The introduction excerpt is truncated, so a roadmap may exist but is not visible. The standard-error type is not stated in the visible excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.34, "usage": {"input": 6351, "output": 2125, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068698, "request_id": "req_011CggqMveUkwtMKV3FMVLW2", "cached": false, "ts": "2026-11-03T23:07:31+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d13be9aa500e00f972f4e640925da0aa86757f6043e2f91bc0c4185009a2f460.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"market_microstructure\",\"subfield_secondary\":\"asset_pricing; behavioral/fraud and manipulation in cryptocurrency markets\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Price impact of Tether-for-Bitcoin flows, especially those of one large Bitfinex account, on subsequent hourly Bitcoin and other cryptocurrency returns.\",\"strategy\":\"Blockchain clustering algorithms identify exchange and account flows. These feed lagged-flow predictive OLS regressions, conditional analyses after negative returns and Tether printing, bootstrap and placebo simulations, and a 2SLS that instruments flows with a dummy for prices just below round $500 thresholds. End-of-month return tests and Tether-USD rate tests discriminate between pushed and pulled hypotheses.\",\"key_assumptions\":[\"The clustering algorithm correctly assigns wallets to exchanges and entities.\",\"Prices falling just below round thresholds affect returns only through Tether flows (exclusion restriction).\",\"Lagged flows are not driven by anticipated future returns, i.e., no clairvoyant timing.\"],\"threats_addressed\":[\"Random chance, addressed by 10,000-iteration bootstrap simulations\",\"Flows to other exchanges, addressed with placebo tests\",\"Reverse causality or demand-driven flows, addressed by the round-threshold IV and the pulled-hypothesis tests\",\"Verification of exchange identification by cross-checking the two blockchains\"],\"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\":\"mechanism_or_theory_sketch\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"robustness_mention\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"identification_defense\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"separate_section\"},\"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\":[\"placebo\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the regression equation explicitly, define the dependent variable (average three-hour or hourly returns), describe how flows are constructed, give the sample window, and identify panels by estimator (OLS, 2SLS first and second stage).\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'most consistent with' the supply-based hypothesis\",\"'suggests' insufficient reserves\",\"'consistent with partial reserve backing'\",\"'provide support for the view'\",\"'could have' a negative impact\"]},\"claim_evidence\":[{\"claim\":\"Tether moves from Bitfinex to Poloniex and Bittrex after negative Bitcoin returns, and positive net flows predict higher Bitcoin returns over the next three hours.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"The top 1% of high-flow hours account for about 59% of Bitcoin's and 64.5% of other major coins' compounded returns, beyond what chance would produce.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 5-6; bootstrap simulations\"},{\"claim\":\"One large Bitfinex player's flows drive the price effect and the reversal of negative returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 7\"},{\"claim\":\"Purchases cluster just below multiples of $500, and instrumenting with this discontinuity shows flows cause positive returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V-VI; Figure 8\"},{\"claim\":\"Negative end-of-month Bitcoin returns in months with high Tether issuance point to insufficient reserves.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figure 9\"},{\"claim\":\"Cross-exchange Bitcoin flows match Tether flows, which validates the exchange identification algorithm.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1, 3\"}],\"structural_summary\":\"Opens with the historical link between bubbles and fraud and casts crypto as a bubble case. Sets out competing pulled (demand) and pushed (supply) hypotheses for Tether, each with testable implications. Builds blockchain-clustered flow data, then documents predictive price impact, concentration in a single entity, and aggregate magnitude via bootstrap and placebo tests. It then discriminates between the hypotheses using round-threshold clustering (also used as an IV), end-of-month returns, and Tether-USD rates. Concludes that the evidence favors supply-driven manipulation and draws regulatory implications.\",\"writing_profile_keywords\":[\"competing hypotheses\",\"forensic finance\",\"blockchain data\",\"historical motivation\",\"price manipulation\",\"bootstrap simulation\",\"instrumental variable\",\"regulatory implications\"],\"confidence\":0.8,\"notes\":\"No explicit contribution paragraph is visible, so contribution_paragraph_index is -1. The standard error type is not visible in the table excerpts. Literature is reviewed in Section I alongside the hypotheses. The Section III and IV roles were reassigned to results.\"}", "json": {"genre": "empirical", "subfield": "market_microstructure", "subfield_secondary": "asset_pricing; behavioral/fraud and manipulation in cryptocurrency markets", "primary_method": "reduced_form_regression", "identification": {"estimand": "Price impact of Tether-for-Bitcoin flows, especially those of one large Bitfinex account, on subsequent hourly Bitcoin and other cryptocurrency returns.", "strategy": "Blockchain clustering algorithms identify exchange and account flows. These feed lagged-flow predictive OLS regressions, conditional analyses after negative returns and Tether printing, bootstrap and placebo simulations, and a 2SLS that instruments flows with a dummy for prices just below round $500 thresholds. End-of-month return tests and Tether-USD rate tests discriminate between pushed and pulled hypotheses.", "key_assumptions": ["The clustering algorithm correctly assigns wallets to exchanges and entities.", "Prices falling just below round thresholds affect returns only through Tether flows (exclusion restriction).", "Lagged flows are not driven by anticipated future returns, i.e., no clairvoyant timing."], "threats_addressed": ["Random chance, addressed by 10,000-iteration bootstrap simulations", "Flows to other exchanges, addressed with placebo tests", "Reverse causality or demand-driven flows, addressed by the round-threshold IV and the pulled-hypothesis tests", "Verification of exchange identification by cross-checking the two blockchains"], "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": "mechanism_or_theory_sketch"}, {"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": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "robustness_mention"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "identification_defense"}, {"paragraph_index": 8, "move": "roadmap"}], "contribution_paragraph_index": -1, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "separate_section"}, "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": ["placebo", "falsification", "other"]}, "tables": {"convention_notes": "Table notes state the regression equation explicitly, define the dependent variable (average three-hour or hourly returns), describe how flows are constructed, give the sample window, and identify panels by estimator (OLS, 2SLS first and second stage).", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'most consistent with' the supply-based hypothesis", "'suggests' insufficient reserves", "'consistent with partial reserve backing'", "'provide support for the view'", "'could have' a negative impact"]}, "claim_evidence": [{"claim": "Tether moves from Bitfinex to Poloniex and Bittrex after negative Bitcoin returns, and positive net flows predict higher Bitcoin returns over the next three hours.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "The top 1% of high-flow hours account for about 59% of Bitcoin's and 64.5% of other major coins' compounded returns, beyond what chance would produce.", "evidence_type": "figure", "evidence_ref": "Figures 5-6; bootstrap simulations"}, {"claim": "One large Bitfinex player's flows drive the price effect and the reversal of negative returns.", "evidence_type": "table", "evidence_ref": "Table IV; Figure 7"}, {"claim": "Purchases cluster just below multiples of $500, and instrumenting with this discontinuity shows flows cause positive returns.", "evidence_type": "table", "evidence_ref": "Tables V-VI; Figure 8"}, {"claim": "Negative end-of-month Bitcoin returns in months with high Tether issuance point to insufficient reserves.", "evidence_type": "table", "evidence_ref": "Table VII; Figure 9"}, {"claim": "Cross-exchange Bitcoin flows match Tether flows, which validates the exchange identification algorithm.", "evidence_type": "figure", "evidence_ref": "Figures 1, 3"}], "structural_summary": "Opens with the historical link between bubbles and fraud and casts crypto as a bubble case. Sets out competing pulled (demand) and pushed (supply) hypotheses for Tether, each with testable implications. Builds blockchain-clustered flow data, then documents predictive price impact, concentration in a single entity, and aggregate magnitude via bootstrap and placebo tests. It then discriminates between the hypotheses using round-threshold clustering (also used as an IV), end-of-month returns, and Tether-USD rates. Concludes that the evidence favors supply-driven manipulation and draws regulatory implications.", "writing_profile_keywords": ["competing hypotheses", "forensic finance", "blockchain data", "historical motivation", "price manipulation", "bootstrap simulation", "instrumental variable", "regulatory implications"], "confidence": 0.8, "notes": "No explicit contribution paragraph is visible, so contribution_paragraph_index is -1. The standard error type is not visible in the table excerpts. Literature is reviewed in Section I alongside the hypotheses. The Section III and IV roles were reassigned to results."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.0, "usage": {"input": 8925, "output": 2317, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082834, "request_id": "req_011Cggqrr4iQ754gB4KZGVAs", "cached": false, "ts": "2026-11-03T23:14:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d48376b438e39e925f28934eb867195090cc119d3dc0fa803c5b2ec2bf0d5d33.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"asset_pricing (contingent capital pricing, equilibrium uniqueness)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical correction)\",\"strategy\":\"not applicable; analytical counterexample within the original continuous-time framework\",\"key_assumptions\":[\"Model consistent with SW's continuous-time framework\",\"Conversion triggered by the stock price\",\"Stochastic continuous-time setting\"],\"threats_addressed\":[],\"causal_language_strength\":1,\"uses_causal_claim\":false},\"abstract_moves\":{\"question\":{\"present\":false,\"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\":true,\"sentence_index\":3}},\"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\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"replication_or_correction\",\"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\"},{\"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\":\"No tables. Results are stated as lemmas, propositions, a theorem and a corollary, with proofs in the appendix.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"modal 'can exist' for uniqueness under broader terms\",\"explicit scope conditions (e.g., when conversion terms penalize shareholders)\",\"'not necessarily' when stating design implications\"]},\"claim_evidence\":[{\"claim\":\"SW's Theorem 1 is invalid because its pricing restriction is needed only at conversion, not before\",\"equation_or_proof\":null,\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II; Appendix proofs\"},{\"claim\":\"A parametric model consistent with SW yields closed-form stock and CC prices with a unique equilibrium under value-transferring terms\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I, Corollary 1; Appendix E\"},{\"claim\":\"Uniqueness requires that conversion occur when the post-conversion stock price equals the trigger\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 1; Appendix B\"},{\"claim\":\"Multiple equilibria can arise in static or deterministic settings but not in the continuous-time stochastic one, even with sunspots\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III\"},{\"claim\":\"Concurrent work also finds uniqueness when conversion transfers value to CC investors\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Glasserman and Nouri (2016)\"}],\"structural_summary\":\"A short corrigendum. The abstract and a single introductory paragraph flag an error in a prior theorem, state the corrected result, and distinguish the paper from concurrent work. Section I builds a counterexample model with closed-form prices. Section II pinpoints the faulty step in the original proof. Section III reconciles the findings with static, deterministic and sunspot settings where multiple equilibria can occur. The conclusion restates the correction and draws an implication for contingent-capital design. Formal proofs are collected in an appendix.\",\"writing_profile_keywords\":[\"corrigendum\",\"counterexample\",\"closed-form solutions\",\"continuous-time model\",\"equilibrium uniqueness\",\"contingent capital\",\"proof in appendix\",\"concise\"],\"confidence\":0.85,\"notes\":\"The abstract's final sentence appears to be an extraction artifact that duplicates the opening of the introduction. The `claim_evidence` array contains a malformed key (`equation_or_proof: null`) in its first entry, which is not allowed by the schema. Roles for Section I were set to theory although the role_rule labeled it 'other'.\"}", "json": {"genre": "replication_or_correction", "subfield": "financial_intermediation_banking", "subfield_secondary": "asset_pricing (contingent capital pricing, equilibrium uniqueness)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical correction)", "strategy": "not applicable; analytical counterexample within the original continuous-time framework", "key_assumptions": ["Model consistent with SW's continuous-time framework", "Conversion triggered by the stock price", "Stochastic continuous-time setting"], "threats_addressed": [], "causal_language_strength": 1, "uses_causal_claim": false}, "abstract_moves": {"question": {"present": false, "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": true, "sentence_index": 3}}, "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": "related_literature_positioning"}, {"paragraph_index": 0, "move": "contribution_statement"}], "contribution_paragraph_index": 0, "contribution_types": ["replication_or_correction", "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"}, {"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": "No tables. Results are stated as lemmas, propositions, a theorem and a corollary, with proofs in the appendix.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["modal 'can exist' for uniqueness under broader terms", "explicit scope conditions (e.g., when conversion terms penalize shareholders)", "'not necessarily' when stating design implications"]}, "claim_evidence": [{"claim": "SW's Theorem 1 is invalid because its pricing restriction is needed only at conversion, not before", "equation_or_proof": null, "evidence_type": "equation_or_proof", "evidence_ref": "Section II; Appendix proofs"}, {"claim": "A parametric model consistent with SW yields closed-form stock and CC prices with a unique equilibrium under value-transferring terms", "evidence_type": "equation_or_proof", "evidence_ref": "Section I, Corollary 1; Appendix E"}, {"claim": "Uniqueness requires that conversion occur when the post-conversion stock price equals the trigger", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 1; Appendix B"}, {"claim": "Multiple equilibria can arise in static or deterministic settings but not in the continuous-time stochastic one, even with sunspots", "evidence_type": "equation_or_proof", "evidence_ref": "Section III"}, {"claim": "Concurrent work also finds uniqueness when conversion transfers value to CC investors", "evidence_type": "prior_literature", "evidence_ref": "Glasserman and Nouri (2016)"}], "structural_summary": "A short corrigendum. The abstract and a single introductory paragraph flag an error in a prior theorem, state the corrected result, and distinguish the paper from concurrent work. Section I builds a counterexample model with closed-form prices. Section II pinpoints the faulty step in the original proof. Section III reconciles the findings with static, deterministic and sunspot settings where multiple equilibria can occur. The conclusion restates the correction and draws an implication for contingent-capital design. Formal proofs are collected in an appendix.", "writing_profile_keywords": ["corrigendum", "counterexample", "closed-form solutions", "continuous-time model", "equilibrium uniqueness", "contingent capital", "proof in appendix", "concise"], "confidence": 0.85, "notes": "The abstract's final sentence appears to be an extraction artifact that duplicates the opening of the introduction. The `claim_evidence` array contains a malformed key (`equation_or_proof: null`) in its first entry, which is not allowed by the schema. Roles for Section I were set to theory although the role_rule labeled it 'other'."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 15.3, "usage": {"input": 3142, "output": 1720, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.047762, "request_id": "req_011CggqP8hJc42PebUm37ZTR", "cached": false, "ts": "2026-11-03T23:07:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d4d4982bee0fd149333ed70c69aaf49c1dee22ade5f50263e4a6782878081d67.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"economic history / banking regulation\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of state deposit insurance on insured banks' deposits, loans, cash and capital ratios, and the sensitivity of deposit flows to bank risk (market discipline)\",\"strategy\":\"Within-state comparison of insured state banks with uninsured national banks, plus uninsured banks in adjacent non-insurance states, before and after the laws passed. Uses bank fixed effects in a panel, a border closest-neighbor-pair sample, and cross-county variation in the WWI agricultural price boom.\",\"key_assumptions\":[\"Insured and uninsured banks in the same and neighboring states would have followed parallel trends absent insurance\",\"Adoption of insurance was not driven by idiosyncratic banking shocks (states where it may have been are excluded)\",\"Adjacent non-insurance states form a valid comparison group\"],\"threats_addressed\":[\"Omitted factors coinciding with deposit insurance adoption, the weakness of cross-country studies\",\"Pre-existing trends, tested with state-level pretrend regressions for 1900-1908\",\"Geographic heterogeneity, addressed with border closest-neighbor pairs\",\"Endogenous adoption, addressed by excluding Mississippi and Washington\",\"Voluntary versus mandatory insurance differences\"],\"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\":2}},\"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_results\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"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\":\"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\":\"within_results\",\"types\":[\"subsample\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Notes define the dependent variable (given in the column titles) and the DI and DI Area indicators, and state the sample states and data sources. Most tables report linear regressions at the state or bank level.\",\"units_stated_in_notes\":false,\"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\":[\"seemingly\",\"apparently\",\"presumably\",\"it is conceivable that\",\"framing results as a conservative measure\"]},\"claim_evidence\":[{\"claim\":\"Deposit insurance raised insured banks' deposits and loans and lowered their cash and capital ratios\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables V, VI\"},{\"claim\":\"State-level aggregates show deposits shifting toward insured state banks after adoption\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 2, 4, 6; Table IV\"},{\"claim\":\"No differential pretrends existed before deposit insurance\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Insurance removed market discipline: deposit growth at insured banks was less sensitive to risk measures\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"Results hold in a closest-neighbor border sample\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"After the price collapse, banks fled the insurance systems by converting to national charters\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"The WWI price boom explains one- to two-thirds of the deposit insurance effects\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Section III.D.2 (stated in conclusion)\"}],\"structural_summary\":\"The introduction sets broad policy support for deposit insurance against empirical evidence of moral hazard. It argues that cross-country studies are confounded and positions within-state historical variation as a cleaner test. It previews findings on risk-taking, loss of market discipline, and the WWI agricultural boom and bust, then gives a roadmap. The body covers institutional history with aggregate trends, then data, then bank-level difference-style regressions with pretrend, border-sample and voluntary-versus-mandatory extensions. A deposit-flow market discipline analysis follows, and finally the post-1920 collapse. The conclusion restates the three findings, draws methodological lessons on time-varying risk-taking opportunities, and offers policy and political-economy implications.\",\"writing_profile_keywords\":[\"economic history\",\"natural experiment\",\"moral hazard\",\"market discipline\",\"bank-level panel\",\"policy implication\",\"state deposit insurance\"],\"confidence\":0.8,\"notes\":\"No standard error or significance details are visible in the excerpts. Theory is in the Internet Appendix. The strategy is DiD-like, combining insured state banks with national banks and neighboring states.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "economic history / banking regulation", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of state deposit insurance on insured banks' deposits, loans, cash and capital ratios, and the sensitivity of deposit flows to bank risk (market discipline)", "strategy": "Within-state comparison of insured state banks with uninsured national banks, plus uninsured banks in adjacent non-insurance states, before and after the laws passed. Uses bank fixed effects in a panel, a border closest-neighbor-pair sample, and cross-county variation in the WWI agricultural price boom.", "key_assumptions": ["Insured and uninsured banks in the same and neighboring states would have followed parallel trends absent insurance", "Adoption of insurance was not driven by idiosyncratic banking shocks (states where it may have been are excluded)", "Adjacent non-insurance states form a valid comparison group"], "threats_addressed": ["Omitted factors coinciding with deposit insurance adoption, the weakness of cross-country studies", "Pre-existing trends, tested with state-level pretrend regressions for 1900-1908", "Geographic heterogeneity, addressed with border closest-neighbor pairs", "Endogenous adoption, addressed by excluding Mississippi and Washington", "Voluntary versus mandatory insurance differences"], "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": 2}}, "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_results"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "roadmap"}], "contribution_paragraph_index": 1, "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": "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": "within_results", "types": ["subsample", "alternative_specification"]}, "tables": {"convention_notes": "Notes define the dependent variable (given in the column titles) and the DI and DI Area indicators, and state the sample states and data sources. Most tables report linear regressions at the state or bank level.", "units_stated_in_notes": false, "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": ["seemingly", "apparently", "presumably", "it is conceivable that", "framing results as a conservative measure"]}, "claim_evidence": [{"claim": "Deposit insurance raised insured banks' deposits and loans and lowered their cash and capital ratios", "evidence_type": "table", "evidence_ref": "Tables V, VI"}, {"claim": "State-level aggregates show deposits shifting toward insured state banks after adoption", "evidence_type": "figure", "evidence_ref": "Figures 2, 4, 6; Table IV"}, {"claim": "No differential pretrends existed before deposit insurance", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Insurance removed market discipline: deposit growth at insured banks was less sensitive to risk measures", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "Results hold in a closest-neighbor border sample", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "After the price collapse, banks fled the insurance systems by converting to national charters", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "The WWI price boom explains one- to two-thirds of the deposit insurance effects", "evidence_type": "none_stated", "evidence_ref": "Section III.D.2 (stated in conclusion)"}], "structural_summary": "The introduction sets broad policy support for deposit insurance against empirical evidence of moral hazard. It argues that cross-country studies are confounded and positions within-state historical variation as a cleaner test. It previews findings on risk-taking, loss of market discipline, and the WWI agricultural boom and bust, then gives a roadmap. The body covers institutional history with aggregate trends, then data, then bank-level difference-style regressions with pretrend, border-sample and voluntary-versus-mandatory extensions. A deposit-flow market discipline analysis follows, and finally the post-1920 collapse. The conclusion restates the three findings, draws methodological lessons on time-varying risk-taking opportunities, and offers policy and political-economy implications.", "writing_profile_keywords": ["economic history", "natural experiment", "moral hazard", "market discipline", "bank-level panel", "policy implication", "state deposit insurance"], "confidence": 0.8, "notes": "No standard error or significance details are visible in the excerpts. Theory is in the Internet Appendix. The strategy is DiD-like, combining insured state banks with national banks and neighboring states."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.91, "usage": {"input": 7685, "output": 2124, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074014, "request_id": "req_011CggqH7KnVn4Q3EjH2b4Gu", "cached": false, "ts": "2026-11-03T23:06:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d5323c916cc6337665f07f082c9cc93a2ae091a44cd1865fb54877ee205dec20.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"financial_intermediation_banking\",\"subfield_secondary\":\"economic history / law and finance; household finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of granting married women property rights (end of coverture) on household portfolio shares of moveable assets, credit supply (interest rates, deposits, loans), and sectoral labor allocation\",\"strategy\":\"Staggered adoption of married women's property laws across U.S. states (1850-1920) used in a difference-in-differences / event-time framework, plus border-county-pair 'difference-in-discontinuities' designs and randomization (placebo timing) tests\",\"key_assumptions\":[\"Timing of women's rights is not driven by omitted factors that also move financial and economic outcomes\",\"No reverse causality from economic outcomes to adoption of rights\",\"Openness of capital markets across states, so local supply shocks affect local credit outcomes\",\"Neighboring border counties are comparable apart from the change in rights\"],\"threats_addressed\":[\"omitted variable bias\",\"reverse causality\",\"geographic variation in economic conditions (border pairs)\",\"concurrent banking regulation (usury laws, reserve requirements, double liability)\",\"pre-trends (event-time leads)\",\"rights affecting marriage composition (balancing test)\",\"chance findings (randomization of adoption dates)\"],\"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\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":5,\"move\":\"motivation\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"identification_defense\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":10,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":9,\"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\":\"empirical_design\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"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\":\"mixed\",\"types\":[\"placebo\",\"alternative_specification\",\"subsample\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Notes open with the clustering level and the *, **, *** p-value thresholds, then define variables, fixed effects and weights; clustering varies by design (state-year, state, border pair); summary tables report mean, median, SD and 10th/90th percentiles; event-time leads are shown as coefficient rows.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'we argue that' framing of causal interpretation\",\"'our estimates suggest'\",\"reporting ranges of estimates across specifications\",\"'may well have' in historical discussion\"]},\"claim_evidence\":[{\"claim\":\"After rights were granted, households shifted their portfolios toward moveable assets, by roughly 1 to 6 percentage points.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III and IV\"},{\"claim\":\"Granting rights did not alter marriage outcomes, supporting the balance of the sample composition.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Rights were followed by a positive credit supply shock: lower interest rates and higher deposits and loans.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3, Table VI\"},{\"claim\":\"Credit-market effects are not reproduced when adoption dates are randomly assigned.\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4\"},{\"claim\":\"Rights led to a dynamic reallocation of male labor toward nonagriculture, with no significant pre-trends.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII and VIII\"},{\"claim\":\"Within nonagriculture, employment shifted toward capital-intensive industries.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"Opens with a historical parliamentary quotation and frames the end of coverture as a major extension of property rights and investor protection. A brief theory section derives four testable predictions. Long data and empirical-strategy sections then defend the use of staggered state adoption as exogenous, drawing on the historical record. Results proceed sequentially through portfolios, credit markets, sectoral labor and capital intensity, combining nationwide and border-pair designs with randomization tests. A short conclusion restates the findings and suggests broader applicability to groups still denied economic rights.\",\"writing_profile_keywords\":[\"economic history\",\"law and finance\",\"staggered adoption\",\"border-pair design\",\"prediction-by-prediction results\",\"historical narrative\",\"investor protection\"],\"confidence\":0.8,\"notes\":\"Section idx 5 is a parsing artifact: it continues the Results section, covering credit markets and subsections C-D. Table VI is referenced but not excerpted. The intro uses 'difference-in-discontinuities' for the border exercise. Whether figures show confidence bands is not visible in the captions.\"}", "json": {"genre": "empirical", "subfield": "financial_intermediation_banking", "subfield_secondary": "economic history / law and finance; household finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of granting married women property rights (end of coverture) on household portfolio shares of moveable assets, credit supply (interest rates, deposits, loans), and sectoral labor allocation", "strategy": "Staggered adoption of married women's property laws across U.S. states (1850-1920) used in a difference-in-differences / event-time framework, plus border-county-pair 'difference-in-discontinuities' designs and randomization (placebo timing) tests", "key_assumptions": ["Timing of women's rights is not driven by omitted factors that also move financial and economic outcomes", "No reverse causality from economic outcomes to adoption of rights", "Openness of capital markets across states, so local supply shocks affect local credit outcomes", "Neighboring border counties are comparable apart from the change in rights"], "threats_addressed": ["omitted variable bias", "reverse causality", "geographic variation in economic conditions (border pairs)", "concurrent banking regulation (usury laws, reserve requirements, double liability)", "pre-trends (event-time leads)", "rights affecting marriage composition (balancing test)", "chance findings (randomization of adoption dates)"], "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": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "other"}, {"paragraph_index": 5, "move": "motivation"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "identification_defense"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 10, "move": "roadmap"}], "contribution_paragraph_index": 9, "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": "empirical_design"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "results"}, {"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": "mixed", "types": ["placebo", "alternative_specification", "subsample", "falsification"]}, "tables": {"convention_notes": "Notes open with the clustering level and the *, **, *** p-value thresholds, then define variables, fixed effects and weights; clustering varies by design (state-year, state, border pair); summary tables report mean, median, SD and 10th/90th percentiles; event-time leads are shown as coefficient rows.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": false, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'we argue that' framing of causal interpretation", "'our estimates suggest'", "reporting ranges of estimates across specifications", "'may well have' in historical discussion"]}, "claim_evidence": [{"claim": "After rights were granted, households shifted their portfolios toward moveable assets, by roughly 1 to 6 percentage points.", "evidence_type": "table", "evidence_ref": "Tables III and IV"}, {"claim": "Granting rights did not alter marriage outcomes, supporting the balance of the sample composition.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Rights were followed by a positive credit supply shock: lower interest rates and higher deposits and loans.", "evidence_type": "figure", "evidence_ref": "Figure 3, Table VI"}, {"claim": "Credit-market effects are not reproduced when adoption dates are randomly assigned.", "evidence_type": "figure", "evidence_ref": "Figure 4"}, {"claim": "Rights led to a dynamic reallocation of male labor toward nonagriculture, with no significant pre-trends.", "evidence_type": "table", "evidence_ref": "Tables VII and VIII"}, {"claim": "Within nonagriculture, employment shifted toward capital-intensive industries.", "evidence_type": "table", "evidence_ref": "Table IX"}], "structural_summary": "Opens with a historical parliamentary quotation and frames the end of coverture as a major extension of property rights and investor protection. A brief theory section derives four testable predictions. Long data and empirical-strategy sections then defend the use of staggered state adoption as exogenous, drawing on the historical record. Results proceed sequentially through portfolios, credit markets, sectoral labor and capital intensity, combining nationwide and border-pair designs with randomization tests. A short conclusion restates the findings and suggests broader applicability to groups still denied economic rights.", "writing_profile_keywords": ["economic history", "law and finance", "staggered adoption", "border-pair design", "prediction-by-prediction results", "historical narrative", "investor protection"], "confidence": 0.8, "notes": "Section idx 5 is a parsing artifact: it continues the Results section, covering credit markets and subsections C-D. Table VI is referenced but not excerpted. The intro uses 'difference-in-discontinuities' for the border exercise. Whether figures show confidence bands is not visible in the captions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.17, "usage": {"input": 6571, "output": 2179, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070658, "request_id": "req_011CggqZq7gDxtEMcy7VJ9W2", "cached": false, "ts": "2026-11-03T23:10:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d639df714a90625b71cfd43ce7ff9b4cc31c304138f043bea91a99ab3aaea003.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\":\"Difference in the predictive ability of insider trades for future returns (purchase minus sale returns) between politically connected and unconnected insiders, by crisis subperiod; abnormal pre-infusion trading around TARP announcements\",\"strategy\":\"Pooled and within-firm regressions of future returns on trade direction interacted with political connections across pre-crisis, crisis-pre-TARP and TARP periods, with fixed effects. These are complemented by an individual-level network map, an event study of trades 30 days before staggered TARP infusions with falsification tests, and FOIA application data.\",\"key_assumptions\":[\"Firm-level connection measures capture information sharing among board members; this is relaxed with an individual-level network map\",\"Short-window announcement returns proxy for the surprise content of TARP infusions\",\"Absent private information, connected insiders' trades would not differ around infusion dates relative to other dates and events\"],\"threats_addressed\":[\"Firm-specific characteristics and changing market conditions, via fixed effect structures\",\"Differential effects of market conditions on connected versus unconnected firms\",\"Spurious timing, via nonannouncement dates and other corporate events such as earnings announcements\",\"Firms that did not receive TARP funds used as a comparison\",\"Validity of the firm-level connection measure, via individual-level connections\"],\"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\":3}},\"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\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":4,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"contribution_statement\"},{\"paragraph_index\":7,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":8,\"move\":\"other\"},{\"paragraph_index\":8,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":8,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":7,\"contribution_types\":[\"new_evidence\"],\"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\":\"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_specification\",\"falsification\",\"placebo\",\"subsample\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes define variables, describe panels and their partitions (period, connection status, trade direction), and reference equation numbers. The event-study tables are paired with dedicated falsification and comparison-event 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 the notion that'\",\"'suggest' / 'can facilitate'\",\"explicit caveats on legality and generalizability\",\"acknowledging difficulty of identifying information source\",\"'important (but not the exclusive)' qualifiers\"],\"claim_evidence\":[{\"claim\":\"Insider trades do not predict returns, and connections do not matter, before the crisis or before TARP's creation\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-IV; Figures 2-3\"},{\"claim\":\"During TARP disbursement, the purchase-sale return spread is 8.89% for connected insiders versus 2.81% for unconnected insiders\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 3\"},{\"claim\":\"Connected insiders outperform unconnected peers within the same firm\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The information advantage is concentrated among insiders with recent, direct connections; board peers of these insiders gain modestly\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figure 4\"},{\"claim\":\"Connected insiders trade abnormally 30 days before TARP infusions, anticipating the market reaction\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII; Figures 5-6\"},{\"claim\":\"No similar pattern appears on nonannouncement dates, at earnings announcements, or in non-TARP firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII-IX\"},{\"claim\":\"Pre-infusion purchases by connected insiders rise with infusion size relative to the applied amount\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"}],\"structural_summary\":\"The paper opens with an established literature on the shareholder benefits of political connections and identifies scarce evidence on insider opportunism. It frames the 2007-2009 bailout as a powerful setting. Results are previewed in period sequence: null effects before TARP, then a sharp connected-insider advantage during disbursement, with magnitudes. A source-of-information discussion motivates the network-map and event-study/FOIA analyses. The introduction closes with explicit caveats on legality and generalizability plus a monitoring recommendation. After separate literature and data sections, three results sections escalate identification: regressions, network attributes, then event study with falsification tests. The conclusion restates the findings.\",\"writing_profile_keywords\":[\"insider trading\",\"political connections\",\"TARP\",\"event study\",\"falsification tests\",\"explicit caveats\",\"period-contrast design\",\"FOIA data\"],\"confidence\":0.8,\"notes\":\"The roadmap lists Section VI as alternative explanations/robustness and Section VII as the conclusion, but the excerpted sections show VI as the conclusion; the robustness section may be missing from the excerpts or the roadmap may be outdated. No statistical-significance notation or standard-error details are visible.\"}", "json": null, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 29.18, "usage": {"input": 8283, "output": 3287, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.099666, "request_id": "req_011Cggqr8XPZZEhkwmruqTQt", "cached": false, "ts": "2026-11-03T23:14:06+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/d827e65197d93e7c2147ee703cbedc05cd52b47baff323086aa250993dc945fa.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Predictive relation between trader-type net position changes or hedging pressure and subsequent commodity futures excess returns\",\"strategy\":\"Fama-MacBeth cross-sectional predictive regressions and portfolio sorts on weekly CFTC position changes, following the Kaniel–Saar–Titman approach; the direction of predictability is used to infer who provides liquidity. Hedging pressure is split into smoothed (long-term) and short-term components.\",\"key_assumptions\":[\"The sign of short-horizon return predictability reveals liquidity provision versus consumption\",\"Smoothed hedging pressure captures hedging demand that evolves slowly with physical production decisions\",\"Controls adequately capture expected returns\"],\"threats_addressed\":[\"Private information of commercials as an alternative explanation\",\"Momentum-driven trading by noncommercials\",\"Measurement of trader categories (DCOT data)\",\"Cross-sectional versus time-series predictability\",\"Major versus minor commodity heterogeneity\"],\"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\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"roadmap\"}],\"contribution_paragraph_index\":4,\"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\":\"results\"},{\"idx\":5,\"role\":\"results\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\"]},\"tables\":{\"convention_notes\":\"Table notes describe Fama-MacBeth regression setups: the dependent variable, the predictors (position changes scaled by open interest, hedging pressure) and the controls. Portfolio sort tables describe quintile and double-sort construction. The visible excerpts do not describe standard errors or significance markers.\",\"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\":[\"we conjecture\",\"suggests\",\"more likely reflects\",\"attempt to\"]},\"claim_evidence\":[{\"claim\":\"Noncommercial position changes are positively related to contemporaneous and past returns, while commercials trade as contrarians\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Commodities bought by noncommercials subsequently underperform, consistent with commercials providing liquidity\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III-V\"},{\"claim\":\"After controlling for short-term position changes, smoothed hedging pressure positively predicts futures returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VIII\"},{\"claim\":\"The cost of liquidity rises with commodity implied volatility, capital constraints and position imbalance, but not with the VIX\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Commercials lose on momentum-driven trades but gain on the larger nonmomentum component\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X, Figure 3\"},{\"claim\":\"Liquidity gains offset much of the insurance premium commercials pay\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"}],\"structural_summary\":\"The introduction frames the Keynesian normal-backwardation view as incomplete. It then previews a liquidity-provision test based on short-horizon predictability and conjectures that hedging pressure has a short-term and a long-term component. Results sections proceed in order: liquidity provision evidence, hedging pressure revisited with a smoothed measure, risk-environment and private-information checks, and a profit decomposition explaining commercials' trading against momentum. A robustness section and a conclusion that restates the two-premium finding close the paper.\",\"writing_profile_keywords\":[\"two-premium decomposition\",\"Fama-MacBeth\",\"CFTC COT data\",\"liquidity provision\",\"normal backwardation\",\"portfolio sorts\",\"puzzle resolution\"],\"confidence\":0.82,\"notes\":\"The data section contains several equations. Figure 2 is missing from the excerpts. The internet appendix is mentioned only in the conclusion.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure", "primary_method": "reduced_form_regression", "identification": {"estimand": "Predictive relation between trader-type net position changes or hedging pressure and subsequent commodity futures excess returns", "strategy": "Fama-MacBeth cross-sectional predictive regressions and portfolio sorts on weekly CFTC position changes, following the Kaniel–Saar–Titman approach; the direction of predictability is used to infer who provides liquidity. Hedging pressure is split into smoothed (long-term) and short-term components.", "key_assumptions": ["The sign of short-horizon return predictability reveals liquidity provision versus consumption", "Smoothed hedging pressure captures hedging demand that evolves slowly with physical production decisions", "Controls adequately capture expected returns"], "threats_addressed": ["Private information of commercials as an alternative explanation", "Momentum-driven trading by noncommercials", "Measurement of trader categories (DCOT data)", "Cross-sectional versus time-series predictability", "Major versus minor commodity heterogeneity"], "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": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"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": "roadmap"}], "contribution_paragraph_index": 4, "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": "results"}, {"idx": 5, "role": "results"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "main_text_section", "types": ["alternative_specification", "subsample", "alternative_measure"]}, "tables": {"convention_notes": "Table notes describe Fama-MacBeth regression setups: the dependent variable, the predictors (position changes scaled by open interest, hedging pressure) and the controls. Portfolio sort tables describe quintile and double-sort construction. The visible excerpts do not describe standard errors or significance markers.", "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": ["we conjecture", "suggests", "more likely reflects", "attempt to"]}, "claim_evidence": [{"claim": "Noncommercial position changes are positively related to contemporaneous and past returns, while commercials trade as contrarians", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Commodities bought by noncommercials subsequently underperform, consistent with commercials providing liquidity", "evidence_type": "table", "evidence_ref": "Tables III-V"}, {"claim": "After controlling for short-term position changes, smoothed hedging pressure positively predicts futures returns", "evidence_type": "table", "evidence_ref": "Tables VI-VIII"}, {"claim": "The cost of liquidity rises with commodity implied volatility, capital constraints and position imbalance, but not with the VIX", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Commercials lose on momentum-driven trades but gain on the larger nonmomentum component", "evidence_type": "table", "evidence_ref": "Table X, Figure 3"}, {"claim": "Liquidity gains offset much of the insurance premium commercials pay", "evidence_type": "table", "evidence_ref": "Table XI"}], "structural_summary": "The introduction frames the Keynesian normal-backwardation view as incomplete. It then previews a liquidity-provision test based on short-horizon predictability and conjectures that hedging pressure has a short-term and a long-term component. Results sections proceed in order: liquidity provision evidence, hedging pressure revisited with a smoothed measure, risk-environment and private-information checks, and a profit decomposition explaining commercials' trading against momentum. A robustness section and a conclusion that restates the two-premium finding close the paper.", "writing_profile_keywords": ["two-premium decomposition", "Fama-MacBeth", "CFTC COT data", "liquidity provision", "normal backwardation", "portfolio sorts", "puzzle resolution"], "confidence": 0.82, "notes": "The data section contains several equations. Figure 2 is missing from the excerpts. The internet appendix is mentioned only in the conclusion."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.29, "usage": {"input": 7582, "output": 2043, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071982, "request_id": "req_011Cggqe1Fk5NcN8jbtr1djP", "cached": false, "ts": "2026-11-03T23:11:08+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/dba159a809061963be8cbbc6111cdbb4998d565468b6932e981a30eba1d1ede9.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"macro_finance; numerical methods for recursive preferences\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical existence/uniqueness conditions for Epstein-Zin utility recursion)\",\"strategy\":\"not applicable; fixed-point theorem (Du 1990) and spectral radius arguments on a valuation operator\",\"key_assumptions\":[\"homothetic Epstein-Zin preferences\",\"Markov state driving consumption growth\",\"compact state space for sharpest results (relaxed under eventual weak compactness)\"],\"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\":3},\"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\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":8,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":10,\"move\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"preview_of_approach\"},{\"paragraph_index\":12,\"move\":\"preview_of_approach\"},{\"paragraph_index\":13,\"move\":\"preview_of_results\"},{\"paragraph_index\":14,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":15,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":16,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_theory_or_model\",\"new_method\"],\"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\":\"literature\"},{\"idx\":5,\"role\":\"theory\"},{\"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\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"sensitivity_to_parameters\",\"other\"]},\"tables\":{\"convention_notes\":\"Tables report calibration parameter values drawn from prior studies and numerical accuracy checks. Monte Carlo tables give means and standard deviations across 1,000 draws, with true values stated for comparison. One table reports wealth-consumption ratios over a parameter grid.\",\"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\":\"low\",\"typical_devices\":[\"explicit scope restrictions (compact state space)\",\"acknowledgment of more limited results in the unbounded case\",\"conjecture language about future research\"]},\"claim_evidence\":[{\"claim\":\"Test value below one is necessary and sufficient for existence and uniqueness of utility solutions in the compact state case\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Theorem 1, Section II, Appendix B-C\"},{\"claim\":\"Risk-adjusted long-run growth equals a power of the spectral radius of a valuation operator\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"eq. (5), Appendix B\"},{\"claim\":\"Monte Carlo and discretization methods compute the test value accurately\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, AI-AIII\"},{\"claim\":\"The new condition is weaker than probability-one bound and contraction conditions from prior work\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"Empirically calibrated long-run risk parameterizations lie close to the stability boundary\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2, Table III\"}],\"structural_summary\":\"The paper motivates the problem through the widespread use of recursive preferences and the risk that utility recursions have no solution or no unique one. It then defines a single test value and states the main if-and-only-if result, followed by an economic interpretation and a spectral-radius representation. Two computational methods are introduced, and related conditions in the literature are positioned within the introduction. The body builds the consumption setup and numerical methods, proves the main theorem, and works through applications of increasing realism. It closes by comparing against alternative conditions and extending to unbounded state spaces, with proofs and numerical details in the appendices.\",\"writing_profile_keywords\":[\"theorem-driven\",\"necessary and sufficient conditions\",\"economic interpretation of a test statistic\",\"applications to calibrated models\",\"numerical methods\",\"literature comparison by sharpness\"],\"confidence\":0.85,\"notes\":\"Theoretical paper whose numerical applications use calibrations taken from the literature. Section IV, flagged by the role rule as robustness, is better described as a comparison with prior literature. The abstract text is truncated with stray introduction text appended.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "macro_finance; numerical methods for recursive preferences", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical existence/uniqueness conditions for Epstein-Zin utility recursion)", "strategy": "not applicable; fixed-point theorem (Du 1990) and spectral radius arguments on a valuation operator", "key_assumptions": ["homothetic Epstein-Zin preferences", "Markov state driving consumption growth", "compact state space for sharpest results (relaxed under eventual weak compactness)"], "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": 3}, "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": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 8, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 10, "move": "preview_of_results"}, {"paragraph_index": 11, "move": "preview_of_approach"}, {"paragraph_index": 12, "move": "preview_of_approach"}, {"paragraph_index": 13, "move": "preview_of_results"}, {"paragraph_index": 14, "move": "related_literature_positioning"}, {"paragraph_index": 15, "move": "related_literature_positioning"}, {"paragraph_index": 16, "move": "related_literature_positioning"}], "contribution_paragraph_index": 1, "contribution_types": ["new_theory_or_model", "new_method"], "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": "literature"}, {"idx": 5, "role": "theory"}, {"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": false}, "robustness": {"location": "mixed", "types": ["sensitivity_to_parameters", "other"]}, "tables": {"convention_notes": "Tables report calibration parameter values drawn from prior studies and numerical accuracy checks. Monte Carlo tables give means and standard deviations across 1,000 draws, with true values stated for comparison. One table reports wealth-consumption ratios over a parameter grid.", "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": "low", "typical_devices": ["explicit scope restrictions (compact state space)", "acknowledgment of more limited results in the unbounded case", "conjecture language about future research"]}, "claim_evidence": [{"claim": "Test value below one is necessary and sufficient for existence and uniqueness of utility solutions in the compact state case", "evidence_type": "equation_or_proof", "evidence_ref": "Theorem 1, Section II, Appendix B-C"}, {"claim": "Risk-adjusted long-run growth equals a power of the spectral radius of a valuation operator", "evidence_type": "equation_or_proof", "evidence_ref": "eq. (5), Appendix B"}, {"claim": "Monte Carlo and discretization methods compute the test value accurately", "evidence_type": "table", "evidence_ref": "Tables III, AI-AIII"}, {"claim": "The new condition is weaker than probability-one bound and contraction conditions from prior work", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV"}, {"claim": "Empirically calibrated long-run risk parameterizations lie close to the stability boundary", "evidence_type": "figure", "evidence_ref": "Figure 2, Table III"}], "structural_summary": "The paper motivates the problem through the widespread use of recursive preferences and the risk that utility recursions have no solution or no unique one. It then defines a single test value and states the main if-and-only-if result, followed by an economic interpretation and a spectral-radius representation. Two computational methods are introduced, and related conditions in the literature are positioned within the introduction. The body builds the consumption setup and numerical methods, proves the main theorem, and works through applications of increasing realism. It closes by comparing against alternative conditions and extending to unbounded state spaces, with proofs and numerical details in the appendices.", "writing_profile_keywords": ["theorem-driven", "necessary and sufficient conditions", "economic interpretation of a test statistic", "applications to calibrated models", "numerical methods", "literature comparison by sharpness"], "confidence": 0.85, "notes": "Theoretical paper whose numerical applications use calibrations taken from the literature. Section IV, flagged by the role rule as robustness, is better described as a comparison with prior literature. The abstract text is truncated with stray introduction text appended."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.94, "usage": {"input": 8832, "output": 2045, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077022, "request_id": "req_011CggqoYyU8sPtgRuprwhGr", "cached": false, "ts": "2026-11-03T23:13:19+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/dbe4c8ee57f43543dcb47f0a8895612362f763ebba25a4bddc4911c434341b6c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"structural_estimation\",\"identification\":{\"estimand\":\"The fraction of analysts (and brokerages) whose true recommendation-revision abnormal returns are nonzero or high-type, and the distribution of the magnitude of that skill.\",\"strategy\":\"Model the cross-section of estimated analyst abnormal returns as a finite mixture of skill distributions that separates true skill dispersion from estimation error. Fit constrained and unconstrained variants. Split revisions by whether a RavenPack news event occurs concurrently. Let mixing probabilities depend on analyst characteristics through a logistic function.\",\"key_assumptions\":[\"True skill within each component follows a parametric distribution (normal; lognormal as an alternative).\",\"Estimation error variance is known or estimable per analyst.\",\"Buy-and-hold abnormal returns properly benchmark recommendation revision performance.\"],\"threats_addressed\":[\"luck versus skill (estimation noise)\",\"piggybacking on concurrent news versus new information production\",\"negative true ability under the normal assumption (lognormal variant)\",\"selection of the number of components\"],\"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\":0},\"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\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"tension_or_puzzle\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":8,\"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\":\"empirical_design\"},{\"idx\":2,\"role\":\"data\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"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\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Notes define the mixture parameters (π, μ, σ, σij) and the sample filter (at least 20 revisions, 1993–2015). They also explain how the long-upgrade/short-downgrade abnormal return is constructed. Tables are organized in panels contrasting constrained and unconstrained models and subsamples.\",\"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\":[\"'results suggest'\",\"'consistent with'\",\"'appear to be'\",\"'it is likely the case'\"]},\"claim_evidence\":[{\"claim\":\"Most analysts (about 94%) belong to the nonzero-skill component in the constrained mixture model.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Analysts are heterogeneous, with high- and low-type components that both earn positive abnormal returns.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"All analysts can process news, and most (93%) also produce new information.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Experience, sell share and industry breadth raise, and recommendation volume lowers, the probability of being high-type.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"A median analyst's pick is worth roughly $40,000 to an institutional client.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"86% of revisions are informative, versus 12% classified as influential in prior work.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII, Table VIII\"},{\"claim\":\"A large share of brokerages are skilled, but fewer than at the analyst level; larger brokers are more likely high-type.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX, Table X\"}],\"structural_summary\":\"The paper opens with the question of whether professionals can pick mispriced stocks and notes a gap: how widespread analyst skill is remains unknown. It then frames the problem of separating luck from skill. The method is a mixture-of-distributions model, set out before the data section. Results proceed from analyst-level skill fractions to news versus non-news decomposition, characteristic correlates, time variation and dollar value. The paper then reconciles its findings with recommendation-level 'influential' classifications and extends the analysis to brokerages. A short conclusion restates the findings, and an appendix gives a lognormal robustness variant.\",\"writing_profile_keywords\":[\"skill versus luck\",\"mixture model\",\"analyst recommendations\",\"information production versus processing\",\"dollar value\",\"contrast with mutual fund literature\"],\"confidence\":0.8,\"notes\":\"Primary method is classified as structural_estimation because the mixture-model distributional estimation is the core approach. The method section is labeled 'theory' by the rule, but its role is empirical design. Standard error reporting is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_intermediation_banking", "primary_method": "structural_estimation", "identification": {"estimand": "The fraction of analysts (and brokerages) whose true recommendation-revision abnormal returns are nonzero or high-type, and the distribution of the magnitude of that skill.", "strategy": "Model the cross-section of estimated analyst abnormal returns as a finite mixture of skill distributions that separates true skill dispersion from estimation error. Fit constrained and unconstrained variants. Split revisions by whether a RavenPack news event occurs concurrently. Let mixing probabilities depend on analyst characteristics through a logistic function.", "key_assumptions": ["True skill within each component follows a parametric distribution (normal; lognormal as an alternative).", "Estimation error variance is known or estimable per analyst.", "Buy-and-hold abnormal returns properly benchmark recommendation revision performance."], "threats_addressed": ["luck versus skill (estimation noise)", "piggybacking on concurrent news versus new information production", "negative true ability under the normal assumption (lognormal variant)", "selection of the number of components"], "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": 0}, "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": "literature_gap"}, {"paragraph_index": 1, "move": "policy_or_broader_implication"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "tension_or_puzzle"}, {"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": "preview_of_results"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 8, "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": "empirical_design"}, {"idx": 2, "role": "data"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"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": "mixed", "types": ["alternative_specification", "subsample", "alternative_sample_period"]}, "tables": {"convention_notes": "Notes define the mixture parameters (π, μ, σ, σij) and the sample filter (at least 20 revisions, 1993–2015). They also explain how the long-upgrade/short-downgrade abnormal return is constructed. Tables are organized in panels contrasting constrained and unconstrained models and subsamples.", "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": ["'results suggest'", "'consistent with'", "'appear to be'", "'it is likely the case'"]}, "claim_evidence": [{"claim": "Most analysts (about 94%) belong to the nonzero-skill component in the constrained mixture model.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Analysts are heterogeneous, with high- and low-type components that both earn positive abnormal returns.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "All analysts can process news, and most (93%) also produce new information.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Experience, sell share and industry breadth raise, and recommendation volume lowers, the probability of being high-type.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "A median analyst's pick is worth roughly $40,000 to an institutional client.", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "86% of revisions are informative, versus 12% classified as influential in prior work.", "evidence_type": "table", "evidence_ref": "Table VII, Table VIII"}, {"claim": "A large share of brokerages are skilled, but fewer than at the analyst level; larger brokers are more likely high-type.", "evidence_type": "table", "evidence_ref": "Table IX, Table X"}], "structural_summary": "The paper opens with the question of whether professionals can pick mispriced stocks and notes a gap: how widespread analyst skill is remains unknown. It then frames the problem of separating luck from skill. The method is a mixture-of-distributions model, set out before the data section. Results proceed from analyst-level skill fractions to news versus non-news decomposition, characteristic correlates, time variation and dollar value. The paper then reconciles its findings with recommendation-level 'influential' classifications and extends the analysis to brokerages. A short conclusion restates the findings, and an appendix gives a lognormal robustness variant.", "writing_profile_keywords": ["skill versus luck", "mixture model", "analyst recommendations", "information production versus processing", "dollar value", "contrast with mutual fund literature"], "confidence": 0.8, "notes": "Primary method is classified as structural_estimation because the mixture-model distributional estimation is the core approach. The method section is labeled 'theory' by the rule, but its role is empirical design. Standard error reporting is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.07, "usage": {"input": 8884, "output": 2261, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08155, "request_id": "req_011CggqpomzBCxHbu6L25jrB", "cached": false, "ts": "2026-11-03T23:13:39+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/de7ac4a1b2d1048ad0b72f8d19a51646b95c35d4a1f4edda8f3c1c68c936d968.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of relaxing short-sale constraints on monthly long-short anomaly portfolio returns (and on the long and short legs separately) for pilot versus nonpilot stocks during the pilot period\",\"strategy\":\"Natural experiment from the Regulation SHO pilot program, which suspended price tests for a quasi-randomly chosen set of stocks. Difference-in-differences regressions of portfolio returns use time fixed effects and a Pilot×During interaction, with a Pilot×Post term for the postpilot period.\",\"key_assumptions\":[\"Assignment of pilot stocks is quasi-random\",\"The regulatory shift alters only permitted short-selling and is uncorrelated with shifts in factor risk\",\"Absent the program, pilot and nonpilot anomaly returns would have evolved in parallel\"],\"threats_addressed\":[\"Proxies for limits to arbitrage correlated with risk\",\"Balance of anomaly variables between pilot and nonpilot firms\",\"Spurious timing (placebo test with a fictitious pilot period)\",\"Risk adjustment with CAPM and Fama-French factors\",\"Low power from a short pilot period\",\"Sensitivity to sample period, exchange, and portfolio weighting\"],\"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\":2},\"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\":\"preview_of_approach\"},{\"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\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":7,\"move\":\"literature_gap\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":9,\"move\":\"identification_defense\"}],\"contribution_paragraph_index\":9,\"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\":\"data\"},{\"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\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"alternative_sample_period\",\"placebo\",\"falsification\",\"subsample\",\"alternative_measure\",\"alternative_specification\"]},\"tables\":{\"convention_notes\":\"Table notes write out the difference-in-differences regression equation, define the variables, and state the sample (NYSE/Amex pilot and nonpilot stocks) and the sample period. Coefficients are reported for each of the 11 anomalies and in aggregate. Return units are stated as monthly raw returns and alphas.\",\"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\":[\"'suggests' when interpreting results as evidence of mispricing\",\"'consistent with' the hypotheses\",\"'quasi-random' qualifier on the assignment\",\"'unlikely to be correlated' with factor risk\"]},\"claim_evidence\":[{\"claim\":\"Anomalies weakened for pilot-stock portfolios during the pilot; combined long-short returns fell by about 72 bp per month\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The effect comes from the short legs; long legs show no significant change\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"The pilot-nonpilot difference vanished after the pilot program ended\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Pilot short legs underperformed at the start of the pilot and outperformed at its end, relative to nonpilot short legs\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"A fictitious 2001-2003 pilot period had no effect on the anomalies\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Pilot and nonpilot firms were similar in their anomaly variables before the program\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Shorting activity differed between pilot and nonpilot short-leg portfolios over the program\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"Results survive adjustment with CAPM and Fama-French three-factor benchmarks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Appendix B\"}],\"structural_summary\":\"The introduction frames the risk-versus-mispricing debate over anomalies and argues that existing limits-to-arbitrage proxies may capture risk. It proposes the Reg SHO pilot as a natural experiment and states two main hypotheses (weaker anomalies for pilot stocks; effect concentrated in short legs), each followed by results. Two auxiliary hypotheses (postpilot reversal; start/end dynamics) and robustness checks follow, then literature positioning and the contribution claim. The body runs from regulatory background and data to a balance check, the main DiD results, postpilot and dynamics results, and shorting activity. A dedicated robustness section covers sample periods, placebo tests, and NASDAQ stocks. Appendices give variable definitions, benchmark-adjusted returns, and a power analysis.\",\"writing_profile_keywords\":[\"natural experiment\",\"difference-in-differences\",\"hypothesis-driven\",\"Regulation SHO\",\"short-sale constraints\",\"anomalies\",\"mispricing vs risk\",\"placebo test\"],\"confidence\":0.82,\"notes\":\"Section I (regulatory background) was auto-tagged as literature; it is assigned 'other' here. Section IV was tagged as data but is the robustness section. Standard error type is not visible in the excerpts. Tables IX-XI (size split, uptick-rule restrictiveness split, value-weighted returns) suggest further subsample and alternative-specification analyses.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of relaxing short-sale constraints on monthly long-short anomaly portfolio returns (and on the long and short legs separately) for pilot versus nonpilot stocks during the pilot period", "strategy": "Natural experiment from the Regulation SHO pilot program, which suspended price tests for a quasi-randomly chosen set of stocks. Difference-in-differences regressions of portfolio returns use time fixed effects and a Pilot×During interaction, with a Pilot×Post term for the postpilot period.", "key_assumptions": ["Assignment of pilot stocks is quasi-random", "The regulatory shift alters only permitted short-selling and is uncorrelated with shifts in factor risk", "Absent the program, pilot and nonpilot anomaly returns would have evolved in parallel"], "threats_addressed": ["Proxies for limits to arbitrage correlated with risk", "Balance of anomaly variables between pilot and nonpilot firms", "Spurious timing (placebo test with a fictitious pilot period)", "Risk adjustment with CAPM and Fama-French factors", "Low power from a short pilot period", "Sensitivity to sample period, exchange, and portfolio weighting"], "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": 2}, "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": "preview_of_approach"}, {"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": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"paragraph_index": 7, "move": "literature_gap"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 9, "move": "identification_defense"}], "contribution_paragraph_index": 9, "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": "data"}, {"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": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "main_text_section", "types": ["alternative_sample_period", "placebo", "falsification", "subsample", "alternative_measure", "alternative_specification"]}, "tables": {"convention_notes": "Table notes write out the difference-in-differences regression equation, define the variables, and state the sample (NYSE/Amex pilot and nonpilot stocks) and the sample period. Coefficients are reported for each of the 11 anomalies and in aggregate. Return units are stated as monthly raw returns and alphas.", "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": ["'suggests' when interpreting results as evidence of mispricing", "'consistent with' the hypotheses", "'quasi-random' qualifier on the assignment", "'unlikely to be correlated' with factor risk"]}, "claim_evidence": [{"claim": "Anomalies weakened for pilot-stock portfolios during the pilot; combined long-short returns fell by about 72 bp per month", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The effect comes from the short legs; long legs show no significant change", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "The pilot-nonpilot difference vanished after the pilot program ended", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Pilot short legs underperformed at the start of the pilot and outperformed at its end, relative to nonpilot short legs", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "A fictitious 2001-2003 pilot period had no effect on the anomalies", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Pilot and nonpilot firms were similar in their anomaly variables before the program", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Shorting activity differed between pilot and nonpilot short-leg portfolios over the program", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "Results survive adjustment with CAPM and Fama-French three-factor benchmarks", "evidence_type": "table", "evidence_ref": "Appendix B"}], "structural_summary": "The introduction frames the risk-versus-mispricing debate over anomalies and argues that existing limits-to-arbitrage proxies may capture risk. It proposes the Reg SHO pilot as a natural experiment and states two main hypotheses (weaker anomalies for pilot stocks; effect concentrated in short legs), each followed by results. Two auxiliary hypotheses (postpilot reversal; start/end dynamics) and robustness checks follow, then literature positioning and the contribution claim. The body runs from regulatory background and data to a balance check, the main DiD results, postpilot and dynamics results, and shorting activity. A dedicated robustness section covers sample periods, placebo tests, and NASDAQ stocks. Appendices give variable definitions, benchmark-adjusted returns, and a power analysis.", "writing_profile_keywords": ["natural experiment", "difference-in-differences", "hypothesis-driven", "Regulation SHO", "short-sale constraints", "anomalies", "mispricing vs risk", "placebo test"], "confidence": 0.82, "notes": "Section I (regulatory background) was auto-tagged as literature; it is assigned 'other' here. Section IV was tagged as data but is the robustness section. Standard error type is not visible in the excerpts. Tables IX-XI (size split, uptick-rule restrictiveness split, value-weighted returns) suggest further subsample and alternative-specification analyses."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.69, "usage": {"input": 8908, "output": 2602, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088466, "request_id": "req_011CggqxfhmwhyZrar8D4Yv4", "cached": false, "ts": "2026-11-03T23:15:26+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/dfffda13a509ab3c11be43005b0cb64b3e942626f884949540cb9858c1f1820e.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure; experimental finance\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Treatment effects of exogenous information-network structure (none, complete, circle, core-periphery) on signal acquisition, volume, liquidity, information aggregation, price informativeness and profits\",\"strategy\":\"Between-session laboratory treatments that randomly vary the information network, analyzed with ordered probit and OLS regressions on treatment dummies, using standard errors clustered at subject level\",\"key_assumptions\":[\"Network structure is exogenous and assigned by design\",\"Information sharing is automatic, perfect and nonstrategic\",\"Subjects in different sessions are comparable\"],\"threats_addressed\":[\"Endogenous network formation (removed by imposing exogenous networks)\",\"Within-subject dependence (handled by clustering)\"],\"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\":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\":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\":\"tension_or_puzzle\"},{\"paragraph_index\":4,\"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\":\"empirical_design\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"other\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Tables pair period-average descriptive statistics, reported with standard deviations, with regression tables. The regressions use treatment dummies against a nonnetworked baseline and include a period control. Standard errors are clustered by subject and reported in parentheses, with stars marking significance. Notes define each dependent variable precisely.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"on average\",\"typically\",\"explicit design simplifications acknowledged\",\"might not be as strong\"]},\"claim_evidence\":[{\"claim\":\"Signal acquisition probability and quantity fall as a trader's number of network connections rises.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III, IV; Figure 2\"},{\"claim\":\"Information sharing raises trading volume and improves liquidity.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Prices aggregate more of the available information with sharing, and aggregation rises with network density.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figures 3-4\"},{\"claim\":\"Price informativeness about the true state does not improve with social communication.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX; Figures 5-6\"},{\"claim\":\"Net earnings are higher under sharing because of savings on signal costs.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables X, XIII\"},{\"claim\":\"Prices support semi-strong but not strong-form efficiency.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Discussion; Page and Siemroth (2018)\"}],\"structural_summary\":\"The introduction opens with the rational-expectations literature on costly information and flags the absence of social networks as an open question. It then describes the exogenous-network setup and frames a trade-off between richer information sets and free-riding. Results are previewed sequentially: crowding out of acquisition, then liquidity, aggregation, informativeness and earnings, with a counter-to-conventional-wisdom point on forecasting quality. A design section presents the lab procedures and the four network treatments. A long results section is organized by outcome, using regressions on treatment dummies. A discussion section links the findings to market efficiency, and a short conclusion restates the findings and lists extensions.\",\"writing_profile_keywords\":[\"laboratory experiment\",\"information networks\",\"free-riding\",\"treatment comparisons\",\"price efficiency\",\"outcome-by-outcome results\"],\"confidence\":0.82,\"notes\":\"Experimental asset-market paper with no equations shown. No explicit robustness section is visible in the excerpts. An Internet Appendix exists, but its contents are not shown.\"}", "json": {"genre": "empirical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure; experimental finance", "primary_method": "survey_or_experiment", "identification": {"estimand": "Treatment effects of exogenous information-network structure (none, complete, circle, core-periphery) on signal acquisition, volume, liquidity, information aggregation, price informativeness and profits", "strategy": "Between-session laboratory treatments that randomly vary the information network, analyzed with ordered probit and OLS regressions on treatment dummies, using standard errors clustered at subject level", "key_assumptions": ["Network structure is exogenous and assigned by design", "Information sharing is automatic, perfect and nonstrategic", "Subjects in different sessions are comparable"], "threats_addressed": ["Endogenous network formation (removed by imposing exogenous networks)", "Within-subject dependence (handled by clustering)"], "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": 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": 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": "tension_or_puzzle"}, {"paragraph_index": 4, "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": "empirical_design"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "other"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Tables pair period-average descriptive statistics, reported with standard deviations, with regression tables. The regressions use treatment dummies against a nonnetworked baseline and include a period control. Standard errors are clustered by subject and reported in parentheses, with stars marking significance. Notes define each dependent variable precisely.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["on average", "typically", "explicit design simplifications acknowledged", "might not be as strong"]}, "claim_evidence": [{"claim": "Signal acquisition probability and quantity fall as a trader's number of network connections rises.", "evidence_type": "table", "evidence_ref": "Tables II, III, IV; Figure 2"}, {"claim": "Information sharing raises trading volume and improves liquidity.", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Prices aggregate more of the available information with sharing, and aggregation rises with network density.", "evidence_type": "table", "evidence_ref": "Table VI; Figures 3-4"}, {"claim": "Price informativeness about the true state does not improve with social communication.", "evidence_type": "table", "evidence_ref": "Table IX; Figures 5-6"}, {"claim": "Net earnings are higher under sharing because of savings on signal costs.", "evidence_type": "table", "evidence_ref": "Tables X, XIII"}, {"claim": "Prices support semi-strong but not strong-form efficiency.", "evidence_type": "prior_literature", "evidence_ref": "Discussion; Page and Siemroth (2018)"}], "structural_summary": "The introduction opens with the rational-expectations literature on costly information and flags the absence of social networks as an open question. It then describes the exogenous-network setup and frames a trade-off between richer information sets and free-riding. Results are previewed sequentially: crowding out of acquisition, then liquidity, aggregation, informativeness and earnings, with a counter-to-conventional-wisdom point on forecasting quality. A design section presents the lab procedures and the four network treatments. A long results section is organized by outcome, using regressions on treatment dummies. A discussion section links the findings to market efficiency, and a short conclusion restates the findings and lists extensions.", "writing_profile_keywords": ["laboratory experiment", "information networks", "free-riding", "treatment comparisons", "price efficiency", "outcome-by-outcome results"], "confidence": 0.82, "notes": "Experimental asset-market paper with no equations shown. No explicit robustness section is visible in the excerpts. An Internet Appendix exists, but its contents are not shown."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.39, "usage": {"input": 6629, "output": 1881, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06493, "request_id": "req_011CggqSHKnsFwpsf8BVeY35", "cached": false, "ts": "2026-11-03T23:08:29+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e064d5a52c6b6c1f8d998f12ca5ee039fbacc1bb9935cd9369977e6f07e4b62c.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of FHA streamline-refinance policy changes (employment documentation and out-of-pocket closing costs) on the monthly probability that an FHA loan refinances\",\"strategy\":\"Event study around the 2009 FHA SLR policy change, then difference-in-differences using conventional loans as controls, then triple differences across county unemployment changes and borrower equity\",\"key_assumptions\":[\"Conventional refinancing would have tracked FHA refinancing absent the policy (parallel trends)\",\"No concurrent shocks differentially affecting FHA borrowers\",\"County unemployment changes proxy for the likelihood that a borrower is unemployed\",\"The two policy channels affect observably distinct borrower groups\"],\"threats_addressed\":[\"Concurrent macroeconomic shocks\",\"Timing of effects (flexible month-by-month estimates)\",\"Confounding between the employment-documentation and closing-cost channels (positive-equity subsample, flexible equity controls)\",\"Correlation between unemployment and negative equity\"],\"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\":4}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"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\":\"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\":\"contribution_statement\"}],\"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\":\"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\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Notes state the sample, the outcome (monthly refinance indicator) and that each column is a separate regression; summary statistics report means with standard deviations in parentheses. The table progression runs from the policy rules to descriptives, then event study, DiD and triple-difference estimates.\",\"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\":[\"may/likely modals\",\"'suggests' and 'imply' framing\",\"acknowledging the event study cannot fully rule out confounds\",\"'plausibly' trace out effects\",\"scope disclaimers (beyond the scope of this paper)\"]},\"claim_evidence\":[{\"claim\":\"The FHA policy change sharply reduced FHA refinancing relative to conventional loans, by about 0.7 pp per month, over half the baseline rate\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"FHA refinancing fell discretely at the policy date while conventional refinancing showed no break\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2, Figure 3\"},{\"claim\":\"A one pp higher county unemployment change lowers post-policy monthly FHA refinancing by about 0.05 pp\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Borrowers newly facing out-of-pocket closing costs reduced refinancing more\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Event study estimates show a large post-policy drop in FHA refinancing\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"}],\"structural_summary\":\"The paper motivates household refinancing frictions as impediments to recession policy and identifies two overlooked frictions. It describes the institutional details of the FHA streamline program, then the data. Combined policy effects are estimated via an event study and a DiD with conventional loans as controls, plus flexible timing checks and a cost-of-not-refinancing calculation. A mechanisms section uses triple differences to separate the unemployment and closing-cost channels, interprets magnitudes and aggregate effects, and distinguishes optimality from liquidity. The conclusion draws distributional and monetary-policy implications and notes limits of scope.\",\"writing_profile_keywords\":[\"policy natural experiment\",\"layered identification\",\"event study to DiD to triple difference\",\"monetary policy pass-through\",\"distributional implications\",\"mechanism decomposition\"],\"confidence\":0.82,\"notes\":\"Section I is institutional background, labeled 'other' rather than literature. The Mechanisms section (labeled robustness by the rule) is treated as results. Standard error type is not visible in the excerpts.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "macro_finance", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of FHA streamline-refinance policy changes (employment documentation and out-of-pocket closing costs) on the monthly probability that an FHA loan refinances", "strategy": "Event study around the 2009 FHA SLR policy change, then difference-in-differences using conventional loans as controls, then triple differences across county unemployment changes and borrower equity", "key_assumptions": ["Conventional refinancing would have tracked FHA refinancing absent the policy (parallel trends)", "No concurrent shocks differentially affecting FHA borrowers", "County unemployment changes proxy for the likelihood that a borrower is unemployed", "The two policy channels affect observably distinct borrower groups"], "threats_addressed": ["Concurrent macroeconomic shocks", "Timing of effects (flexible month-by-month estimates)", "Confounding between the employment-documentation and closing-cost channels (positive-equity subsample, flexible equity controls)", "Correlation between unemployment and negative equity"], "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": 4}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 1, "move": "identification_defense"}, {"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": "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": "contribution_statement"}], "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": "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": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "subsample"]}, "tables": {"convention_notes": "Notes state the sample, the outcome (monthly refinance indicator) and that each column is a separate regression; summary statistics report means with standard deviations in parentheses. The table progression runs from the policy rules to descriptives, then event study, DiD and triple-difference estimates.", "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": ["may/likely modals", "'suggests' and 'imply' framing", "acknowledging the event study cannot fully rule out confounds", "'plausibly' trace out effects", "scope disclaimers (beyond the scope of this paper)"]}, "claim_evidence": [{"claim": "The FHA policy change sharply reduced FHA refinancing relative to conventional loans, by about 0.7 pp per month, over half the baseline rate", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "FHA refinancing fell discretely at the policy date while conventional refinancing showed no break", "evidence_type": "figure", "evidence_ref": "Figure 2, Figure 3"}, {"claim": "A one pp higher county unemployment change lowers post-policy monthly FHA refinancing by about 0.05 pp", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Borrowers newly facing out-of-pocket closing costs reduced refinancing more", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Event study estimates show a large post-policy drop in FHA refinancing", "evidence_type": "table", "evidence_ref": "Table III"}], "structural_summary": "The paper motivates household refinancing frictions as impediments to recession policy and identifies two overlooked frictions. It describes the institutional details of the FHA streamline program, then the data. Combined policy effects are estimated via an event study and a DiD with conventional loans as controls, plus flexible timing checks and a cost-of-not-refinancing calculation. A mechanisms section uses triple differences to separate the unemployment and closing-cost channels, interprets magnitudes and aggregate effects, and distinguishes optimality from liquidity. The conclusion draws distributional and monetary-policy implications and notes limits of scope.", "writing_profile_keywords": ["policy natural experiment", "layered identification", "event study to DiD to triple difference", "monetary policy pass-through", "distributional implications", "mechanism decomposition"], "confidence": 0.82, "notes": "Section I is institutional background, labeled 'other' rather than literature. The Mechanisms section (labeled robustness by the rule) is treated as results. Standard error type is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.36, "usage": {"input": 7332, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070362, "request_id": "req_011Cggqv7w2QUBzgfrXX4GMa", "cached": false, "ts": "2026-11-03T23:14:48+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e4337005b997fd6aea5d553c61822d8e2dff1f84a34203c1758996ca7a7ba51d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"difference_in_differences\",\"identification\":{\"estimand\":\"Effect of an exogenous rise in the relative cost of very short-term bank debt on firms' short-term borrowing, cash, trade credit, and investment, and how this effect differs with industry access to trade credit\",\"strategy\":\"Difference-in-differences around Colombia's 2011 BAD tax change. Treated firms are those in the top quintile of pre-2011 very short-term debt use; the bottom two quintiles are controls. Matching is applied, the sample is restricted to firms above a sales threshold, and a triple interaction uses industry trade-credit access measured from U.S. Compustat firms.\",\"key_assumptions\":[\"Treated and control firms would have followed parallel trends absent the tax change\",\"The tax shock affects very short-term bank credit without spilling over into banks' long-term credit role or into investment opportunities\",\"Trade-credit access measured from U.S. Compustat firms captures inherent industry access in Colombia\"],\"threats_addressed\":[\"Pre-trend differences, examined with year-coefficient plots\",\"Confounds from broad credit shocks such as the financial crisis\",\"Tax avoidance behavior\",\"Covariate imbalance, addressed with a matched sample\"],\"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\":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\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":1,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":3,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":3,\"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\":\"results\"},{\"idx\":3,\"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\":\"Each regression table note restates the difference-in-differences setup: the Treatment × Post 2011 interaction of interest, how treatment and control groups are defined, and outcomes scaled by beginning-of-year assets. Table I compares prematch and postmatch samples.\",\"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\":[\"our results suggest\",\"likely occurs\",\"plausibly exogenous\",\"explicit acknowledgment of external validity limits\"]},\"claim_evidence\":[{\"claim\":\"Exposed firms sharply reduce very short-term bank borrowing after the 2011 tax change\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II; Figure 6\"},{\"claim\":\"Exposed firms increase both cash and net accounts payable\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III, IV\"},{\"claim\":\"Some treated firms cut capital expenditures, R&D, and inventory investment\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"In high trade-credit-access industries, firms substitute into payables with no change in cash or investment; in low-access industries, cash rises and investment falls\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VIII\"},{\"claim\":\"The tax raised the cost of short-maturity loans more than that of long-maturity loans\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"}],\"structural_summary\":\"The paper opens with a broad question about how firms manage liquidity and previews the findings. It then sketches a two-regime tax mechanism and positions itself against the trade-credit and bank-liquidity literatures inside the introduction. A dedicated institutional section explains why the Colombian BAD tax provides a clean shock. One combined section covers design, data, and four numbered results, ending with heterogeneity by industry trade-credit access. A short conclusion restates the contribution and acknowledges external validity limits.\",\"writing_profile_keywords\":[\"natural experiment\",\"tax shock\",\"difference-in-differences\",\"trade credit\",\"liquidity management\",\"emerging market\",\"numbered results\",\"institutional detail\"],\"confidence\":0.8,\"notes\":\"No robustness section is visible in the excerpts. Matching and pre-trend plots appear within the design. The standard errors clustering level is not visible.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "financial_intermediation_banking", "primary_method": "difference_in_differences", "identification": {"estimand": "Effect of an exogenous rise in the relative cost of very short-term bank debt on firms' short-term borrowing, cash, trade credit, and investment, and how this effect differs with industry access to trade credit", "strategy": "Difference-in-differences around Colombia's 2011 BAD tax change. Treated firms are those in the top quintile of pre-2011 very short-term debt use; the bottom two quintiles are controls. Matching is applied, the sample is restricted to firms above a sales threshold, and a triple interaction uses industry trade-credit access measured from U.S. Compustat firms.", "key_assumptions": ["Treated and control firms would have followed parallel trends absent the tax change", "The tax shock affects very short-term bank credit without spilling over into banks' long-term credit role or into investment opportunities", "Trade-credit access measured from U.S. Compustat firms captures inherent industry access in Colombia"], "threats_addressed": ["Pre-trend differences, examined with year-coefficient plots", "Confounds from broad credit shocks such as the financial crisis", "Tax avoidance behavior", "Covariate imbalance, addressed with a matched sample"], "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": 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": 0, "move": "preview_of_results"}, {"paragraph_index": 1, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 3, "move": "related_literature_positioning"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "roadmap"}], "contribution_paragraph_index": 3, "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": "results"}, {"idx": 3, "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": "Each regression table note restates the difference-in-differences setup: the Treatment × Post 2011 interaction of interest, how treatment and control groups are defined, and outcomes scaled by beginning-of-year assets. Table I compares prematch and postmatch samples.", "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": ["our results suggest", "likely occurs", "plausibly exogenous", "explicit acknowledgment of external validity limits"]}, "claim_evidence": [{"claim": "Exposed firms sharply reduce very short-term bank borrowing after the 2011 tax change", "evidence_type": "table", "evidence_ref": "Table II; Figure 6"}, {"claim": "Exposed firms increase both cash and net accounts payable", "evidence_type": "table", "evidence_ref": "Tables III, IV"}, {"claim": "Some treated firms cut capital expenditures, R&D, and inventory investment", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "In high trade-credit-access industries, firms substitute into payables with no change in cash or investment; in low-access industries, cash rises and investment falls", "evidence_type": "table", "evidence_ref": "Tables VI-VIII"}, {"claim": "The tax raised the cost of short-maturity loans more than that of long-maturity loans", "evidence_type": "figure", "evidence_ref": "Figure 2"}], "structural_summary": "The paper opens with a broad question about how firms manage liquidity and previews the findings. It then sketches a two-regime tax mechanism and positions itself against the trade-credit and bank-liquidity literatures inside the introduction. A dedicated institutional section explains why the Colombian BAD tax provides a clean shock. One combined section covers design, data, and four numbered results, ending with heterogeneity by industry trade-credit access. A short conclusion restates the contribution and acknowledges external validity limits.", "writing_profile_keywords": ["natural experiment", "tax shock", "difference-in-differences", "trade credit", "liquidity management", "emerging market", "numbered results", "institutional detail"], "confidence": 0.8, "notes": "No robustness section is visible in the excerpts. Matching and pre-trend plots appear within the design. The standard errors clustering level is not visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.61, "usage": {"input": 6704, "output": 1934, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06629, "request_id": "req_011CggqZ5VbKfALxaf53ubk2", "cached": false, "ts": "2026-11-03T23:10:02+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e54b86304f1bf53643ceea67eac2515c02888b8bb4eee539323759abf03a06de.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"asset_pricing\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Sensitivity of net equity repurchases and net debt issuance to lagged debt- and equity-market expected excess returns (valuation measures)\",\"strategy\":\"Firm-level panel and aggregate time-series regressions of financing flows on lagged valuation proxies with fundamental controls; logit models for simultaneous issuance/repurchase; return-forecasting regressions of bond returns on firm actions controlling for hedge-ratio-adjusted stock returns; additional tests using market-specific sentiment and supply shocks (e.g., forecast errors)\",\"key_assumptions\":[\"Valuation proxies capture expected excess returns rather than fundamentals\",\"Control variables adequately absorb fundamental drivers of financing such as cash flows, investment demand and the business cycle\",\"Hedge-ratio benchmarks correctly capture the comovement of bonds and stocks implied by fundamentals\"],\"threats_addressed\":[\"Fundamentals-driven financing, addressed through controls\",\"Alternative definitions of simultaneous issuance (above-average thresholds)\",\"Firm heterogeneity, addressed through sorts by size, profitability and KZ index\"],\"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\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"robustness_mention\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"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\":\"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_measure\",\"subsample\"]},\"tables\":{\"convention_notes\":\"Table notes write out the estimating equation (panel, logit or time-series) and define the dependent variables, which are normalized by lagged assets. Panels split results by outcome or by firm group. The motivating table uses a firm example with units stated in millions of dollars.\",\"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\":[\"'to the extent that' qualifiers\",\"'helps account for' rather than 'explains'\",\"'tend to'\",\"framing returns as possibly understating firm gains\"]},\"claim_evidence\":[{\"claim\":\"Net equity repurchases and net debt issuance both rise when expected excess debt returns are low or expected equity returns are high\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Simultaneous issuance in one market and repurchase in the other is more likely under the corresponding relative valuations\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Cross-market spillovers are strongest among large, unconstrained firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"After firms issue debt and buy back equity, debt subsequently underperforms hedge-ratio-adjusted equity by about 0.5-0.6 pp per year\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Aggregate financing flows show the same cross-market valuation patterns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"A large share of financing comes from firms simultaneously issuing in one market and repurchasing in another\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1, Table I\"},{\"claim\":\"Market-specific valuation shocks, such as bond-yield forecast errors, drive financing decisions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"}],\"structural_summary\":\"The introduction opens with established single-market timing evidence and pivots to a multi-market gap. It then previews firm-level and aggregate results, including return magnitudes, positions the paper briefly against prior work and closes with a roadmap. The body moves from motivating stylized facts to a conceptual framework that generates predictions. Firm-level tests come next (baseline regressions, simultaneous-issuance logits, heterogeneity, return forecasting), followed by the aggregate analogue and additional evidence on sentiment and supply shocks. A short conclusion restates the findings and points to future research.\",\"writing_profile_keywords\":[\"market timing\",\"cross-market arbitrage\",\"capital structure\",\"debt-equity substitution\",\"firm-level and aggregate\",\"prediction-guided empirics\",\"return predictability\"],\"confidence\":0.8,\"notes\":\"Standard-error and significance conventions are not visible in the excerpts. The conceptual framework is verbal or organizing rather than formal; equations appear only in the empirical sections.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "asset_pricing", "primary_method": "reduced_form_regression", "identification": {"estimand": "Sensitivity of net equity repurchases and net debt issuance to lagged debt- and equity-market expected excess returns (valuation measures)", "strategy": "Firm-level panel and aggregate time-series regressions of financing flows on lagged valuation proxies with fundamental controls; logit models for simultaneous issuance/repurchase; return-forecasting regressions of bond returns on firm actions controlling for hedge-ratio-adjusted stock returns; additional tests using market-specific sentiment and supply shocks (e.g., forecast errors)", "key_assumptions": ["Valuation proxies capture expected excess returns rather than fundamentals", "Control variables adequately absorb fundamental drivers of financing such as cash flows, investment demand and the business cycle", "Hedge-ratio benchmarks correctly capture the comovement of bonds and stocks implied by fundamentals"], "threats_addressed": ["Fundamentals-driven financing, addressed through controls", "Alternative definitions of simultaneous issuance (above-average thresholds)", "Firm heterogeneity, addressed through sorts by size, profitability and KZ index"], "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": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"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": "preview_of_results"}, {"paragraph_index": 4, "move": "robustness_mention"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"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": "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_measure", "subsample"]}, "tables": {"convention_notes": "Table notes write out the estimating equation (panel, logit or time-series) and define the dependent variables, which are normalized by lagged assets. Panels split results by outcome or by firm group. The motivating table uses a firm example with units stated in millions of dollars.", "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": ["'to the extent that' qualifiers", "'helps account for' rather than 'explains'", "'tend to'", "framing returns as possibly understating firm gains"]}, "claim_evidence": [{"claim": "Net equity repurchases and net debt issuance both rise when expected excess debt returns are low or expected equity returns are high", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Simultaneous issuance in one market and repurchase in the other is more likely under the corresponding relative valuations", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Cross-market spillovers are strongest among large, unconstrained firms", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "After firms issue debt and buy back equity, debt subsequently underperforms hedge-ratio-adjusted equity by about 0.5-0.6 pp per year", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Aggregate financing flows show the same cross-market valuation patterns", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "A large share of financing comes from firms simultaneously issuing in one market and repurchasing in another", "evidence_type": "figure", "evidence_ref": "Figure 1, Table I"}, {"claim": "Market-specific valuation shocks, such as bond-yield forecast errors, drive financing decisions", "evidence_type": "table", "evidence_ref": "Table VIII"}], "structural_summary": "The introduction opens with established single-market timing evidence and pivots to a multi-market gap. It then previews firm-level and aggregate results, including return magnitudes, positions the paper briefly against prior work and closes with a roadmap. The body moves from motivating stylized facts to a conceptual framework that generates predictions. Firm-level tests come next (baseline regressions, simultaneous-issuance logits, heterogeneity, return forecasting), followed by the aggregate analogue and additional evidence on sentiment and supply shocks. A short conclusion restates the findings and points to future research.", "writing_profile_keywords": ["market timing", "cross-market arbitrage", "capital structure", "debt-equity substitution", "firm-level and aggregate", "prediction-guided empirics", "return predictability"], "confidence": 0.8, "notes": "Standard-error and significance conventions are not visible in the excerpts. The conceptual framework is verbal or organizing rather than formal; equations appear only in the empirical sections."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.85, "usage": {"input": 6634, "output": 2158, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07049, "request_id": "req_011CggqaJqYiE3XnKPqZDv5j", "cached": false, "ts": "2026-11-03T23:10:20+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/e73f90c2d1c768f9cce3e7be8885bba27f4723aaf1b1260942b2244898245a34.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Return predictability (factor-model alphas) of the gap between structural model-implied equity value and market equity value, interpreted as mispricing of the default option\",\"strategy\":\"Implement a structural real-option equity valuation model with firm-specific inputs, sort stocks on model-to-market value, and test long-short alphas. Double-sorts by default-related characteristics, a counterfactual model with the default option shut down, earnings-announcement returns, Fama-MacBeth regressions, and an international out-of-sample test.\",\"key_assumptions\":[\"Model inputs (industry leverage refinancing, constant parameters) adequately capture fundamental value\",\"Factor models adequately control for risk so that alphas reflect mispricing\",\"Return spreads concentrated around information events indicate mispricing correction rather than risk\"],\"threats_addressed\":[\"value premium in disguise (double-sorts on market-to-book and residual income value)\",\"alternative factor models (FF5, q-factor)\",\"model specification concerns (shutting down default option)\",\"out-of-sample validity (international markets)\",\"time variation and subperiods\"],\"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\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"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\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"robustness_mention\"},{\"paragraph_index\":6,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":7,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"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\":\"theory\"},{\"idx\":3,\"role\":\"data\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"robustness\"},{\"idx\":6,\"role\":\"results\"},{\"idx\":7,\"role\":\"conclusion\"},{\"idx\":8,\"role\":\"appendix\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"alternative_sample_period\",\"falsification\"]},\"tables\":{\"convention_notes\":\"Notes define the sort procedure: decile 1 is most overvalued and decile 10 most undervalued, with value-weighted one-month holding. Later tables refer back to the Table II sort. Alphas are reported in percent per month with t-statistics. Factor models are named explicitly, and international tables are organized by country.\",\"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\":[\"'at least in part'\",\"'suggests'\",\"'we conjecture'\",\"explicit disclaimers of what is not claimed (e.g., not resolving the distress puzzle)\"]},\"claim_evidence\":[{\"claim\":\"Long-short strategy on model-implied relative value earns roughly 11% annual four-factor alpha in U.S. stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Model performance is stronger among distressed and highly volatile stocks\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Removing the default option sharply weakens the alphas, implicating the option as the driver\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Return spread is larger on earnings announcement days, consistent with mispricing correction\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Results are not the value premium in disguise\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Model value predicts returns in several of nine developed markets\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XV-XVI\"},{\"claim\":\"Investors may lack the skills to value options correctly\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Poteshman and Serbin (2003); Benartzi and Thaler (2001); Hirshleifer and Teoh (2003)\"}],\"structural_summary\":\"The paper motivates equity misvaluation and proposes an unrecognized source: investors fail to price the default option. It builds a structural real-option valuation model and applies it monthly to the full cross section. It then shows portfolio-sort alphas, rules out a value-premium explanation, and ties performance to the default option via characteristic double-sorts and a shut-down counterfactual. Mispricing-consistent tests (earnings announcements, transparency, time variation) and an international out-of-sample test follow. The paper closes with a brief restatement of findings.\",\"writing_profile_keywords\":[\"structural valuation model\",\"portfolio sorts\",\"factor alphas\",\"mispricing\",\"default option\",\"out-of-sample international\",\"counterfactual model\"],\"confidence\":0.8,\"notes\":\"Primary method is labeled calibration_simulation because the paper implements a calibrated structural model whose outputs feed portfolio-sort tests; it could arguably be classified as reduced_form_regression or other. Section II (Model Descriptives) is labeled data/descriptive even though its rule role was theory. Standard-error treatment is not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "behavioral_finance", "primary_method": "calibration_simulation", "identification": {"estimand": "Return predictability (factor-model alphas) of the gap between structural model-implied equity value and market equity value, interpreted as mispricing of the default option", "strategy": "Implement a structural real-option equity valuation model with firm-specific inputs, sort stocks on model-to-market value, and test long-short alphas. Double-sorts by default-related characteristics, a counterfactual model with the default option shut down, earnings-announcement returns, Fama-MacBeth regressions, and an international out-of-sample test.", "key_assumptions": ["Model inputs (industry leverage refinancing, constant parameters) adequately capture fundamental value", "Factor models adequately control for risk so that alphas reflect mispricing", "Return spreads concentrated around information events indicate mispricing correction rather than risk"], "threats_addressed": ["value premium in disguise (double-sorts on market-to-book and residual income value)", "alternative factor models (FF5, q-factor)", "model specification concerns (shutting down default option)", "out-of-sample validity (international markets)", "time variation and subperiods"], "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": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "literature_gap"}, {"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": "identification_defense"}, {"paragraph_index": 5, "move": "robustness_mention"}, {"paragraph_index": 6, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 7, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "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": "theory"}, {"idx": 3, "role": "data"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "robustness"}, {"idx": 6, "role": "results"}, {"idx": 7, "role": "conclusion"}, {"idx": 8, "role": "appendix"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "mixed", "types": ["alternative_specification", "subsample", "alternative_measure", "alternative_sample_period", "falsification"]}, "tables": {"convention_notes": "Notes define the sort procedure: decile 1 is most overvalued and decile 10 most undervalued, with value-weighted one-month holding. Later tables refer back to the Table II sort. Alphas are reported in percent per month with t-statistics. Factor models are named explicitly, and international tables are organized by country.", "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": ["'at least in part'", "'suggests'", "'we conjecture'", "explicit disclaimers of what is not claimed (e.g., not resolving the distress puzzle)"]}, "claim_evidence": [{"claim": "Long-short strategy on model-implied relative value earns roughly 11% annual four-factor alpha in U.S. stocks", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Model performance is stronger among distressed and highly volatile stocks", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Removing the default option sharply weakens the alphas, implicating the option as the driver", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Return spread is larger on earnings announcement days, consistent with mispricing correction", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Results are not the value premium in disguise", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Model value predicts returns in several of nine developed markets", "evidence_type": "table", "evidence_ref": "Tables XV-XVI"}, {"claim": "Investors may lack the skills to value options correctly", "evidence_type": "prior_literature", "evidence_ref": "Poteshman and Serbin (2003); Benartzi and Thaler (2001); Hirshleifer and Teoh (2003)"}], "structural_summary": "The paper motivates equity misvaluation and proposes an unrecognized source: investors fail to price the default option. It builds a structural real-option valuation model and applies it monthly to the full cross section. It then shows portfolio-sort alphas, rules out a value-premium explanation, and ties performance to the default option via characteristic double-sorts and a shut-down counterfactual. Mispricing-consistent tests (earnings announcements, transparency, time variation) and an international out-of-sample test follow. The paper closes with a brief restatement of findings.", "writing_profile_keywords": ["structural valuation model", "portfolio sorts", "factor alphas", "mispricing", "default option", "out-of-sample international", "counterfactual model"], "confidence": 0.8, "notes": "Primary method is labeled calibration_simulation because the paper implements a calibrated structural model whose outputs feed portfolio-sort tests; it could arguably be classified as reduced_form_regression or other. Section II (Model Descriptives) is labeled data/descriptive even though its rule role was theory. Standard-error treatment is not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.42, "usage": {"input": 9665, "output": 2285, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.085154, "request_id": "req_011CggqJSWyBNF1hAkmG3TQX", "cached": false, "ts": "2026-11-03T23:06:45+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/eb69a89ff0952fd932969fc30399d3b362d1b5664412953f1bab31247a4d7c26.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_econometrics\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"Conditional expected (excess and excess-of-market) returns on individual stocks implied by option-based risk-neutral variances\",\"strategy\":\"Derive a parameter-free formula from the growth-optimal SDF, then test predicted coefficients in pooled and fixed-effects panel forecasting regressions and in out-of-sample R2 comparisons against benchmarks\",\"key_assumptions\":[\"Theoretical conditions linking the growth-optimal return to option-implied variances hold approximately\",\"Option prices accurately measure risk-neutral variance\",\"Bootstrap inference is valid given overlapping observations and cross-sectional correlation\"],\"threats_addressed\":[\"Data snooping / free parameters (formula has none)\",\"Overlapping returns and panel dependence in standard errors (block bootstrap)\",\"Omitted characteristics (beta, size, book-to-market, momentum)\",\"Out-of-sample validity\"],\"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\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"related_literature_positioning\"},{\"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\":\"roadmap\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"},{\"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\"}],\"contribution_paragraph_index\":2,\"contribution_types\":[\"new_theory_or_model\",\"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\":\"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\":\"mixed\",\"types\":[\"subsample\",\"alternative_specification\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes describe the sample (S&P 100/500 firms, monthly data from 1996 to 2014), the regression specification and its panels. Coefficient tests are framed as hypotheses on theory-implied parameter values. Out-of-sample R2 is defined explicitly in the notes.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"bootstrap\"},\"figures\":{\"uncertainty_shown\":\"unclear\",\"figure_note\":\"Figure 7 reports subsample regression estimates, so uncertainty may be shown there; the captions are truncated and do not confirm it.\"},\"conclusion\":{\"bounded_claims\":false,\"policy_implications\":false,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"'suggest' in abstract\",\"'in principle'\",\"quoted 'loses' acknowledging a competitor win\"]},\"claim_evidence\":[{\"claim\":\"Expected stock return can be expressed via market risk-neutral variance and stock excess risk-neutral variance with no free parameters\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section I\"},{\"claim\":\"Formula's panel forecasting coefficients are consistent with theory-implied values\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II-V\"},{\"claim\":\"Formula beats out-of-sample competitors at horizons from 3 to 24 months\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Formula mostly beats competitors that use in-sample information on characteristics\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table X\"},{\"claim\":\"Model implies more time-series and cross-sectional variation in expected returns than CAPM\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 1, 9\"}],\"structural_summary\":\"The paper opens by stating a derived formula and its option-based inputs. It then lists distinctive features and contrasts them with the CAPM and factor models. The introduction previews in-sample and out-of-sample results, illustrates with firm examples, and positions the work against the idiosyncratic-volatility and option-forecasting literatures. The theory section is followed by variance measurement, tests of the theory-implied coefficients, a check against characteristics, out-of-sample R2 comparisons, and a conclusion restating the distinctive features. Appendices cover implied correlation and the block bootstrap.\",\"writing_profile_keywords\":[\"parameter-free theory\",\"option-implied variance\",\"cross-section of expected returns\",\"out-of-sample R2\",\"CAPM comparison\",\"confident tone\"],\"confidence\":0.8,\"notes\":\"The roadmap and the related-literature discussion are merged in the same introduction paragraph. Significance-test conventions are unclear from the excerpts. Robustness evidence is spread across subsample estimates (Figure 7), characteristic controls (Section IV), and the appendix and Internet Appendix.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "financial_econometrics", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "Conditional expected (excess and excess-of-market) returns on individual stocks implied by option-based risk-neutral variances", "strategy": "Derive a parameter-free formula from the growth-optimal SDF, then test predicted coefficients in pooled and fixed-effects panel forecasting regressions and in out-of-sample R2 comparisons against benchmarks", "key_assumptions": ["Theoretical conditions linking the growth-optimal return to option-implied variances hold approximately", "Option prices accurately measure risk-neutral variance", "Bootstrap inference is valid given overlapping observations and cross-sectional correlation"], "threats_addressed": ["Data snooping / free parameters (formula has none)", "Overlapping returns and panel dependence in standard errors (block bootstrap)", "Omitted characteristics (beta, size, book-to-market, momentum)", "Out-of-sample validity"], "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": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "related_literature_positioning"}, {"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": "roadmap"}, {"paragraph_index": 6, "move": "related_literature_positioning"}, {"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"}], "contribution_paragraph_index": 2, "contribution_types": ["new_theory_or_model", "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": "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": "mixed", "types": ["subsample", "alternative_specification", "other"]}, "tables": {"convention_notes": "Notes describe the sample (S&P 100/500 firms, monthly data from 1996 to 2014), the regression specification and its panels. Coefficient tests are framed as hypotheses on theory-implied parameter values. Out-of-sample R2 is defined explicitly in the notes.", "units_stated_in_notes": false, "standard_errors_description": "bootstrap"}, "figures": {"uncertainty_shown": "unclear", "figure_note": "Figure 7 reports subsample regression estimates, so uncertainty may be shown there; the captions are truncated and do not confirm it."}, "conclusion": {"bounded_claims": false, "policy_implications": false, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["'suggest' in abstract", "'in principle'", "quoted 'loses' acknowledging a competitor win"]}, "claim_evidence": [{"claim": "Expected stock return can be expressed via market risk-neutral variance and stock excess risk-neutral variance with no free parameters", "evidence_type": "equation_or_proof", "evidence_ref": "Section I"}, {"claim": "Formula's panel forecasting coefficients are consistent with theory-implied values", "evidence_type": "table", "evidence_ref": "Tables II-V"}, {"claim": "Formula beats out-of-sample competitors at horizons from 3 to 24 months", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Formula mostly beats competitors that use in-sample information on characteristics", "evidence_type": "table", "evidence_ref": "Table X"}, {"claim": "Model implies more time-series and cross-sectional variation in expected returns than CAPM", "evidence_type": "figure", "evidence_ref": "Figures 1, 9"}], "structural_summary": "The paper opens by stating a derived formula and its option-based inputs. It then lists distinctive features and contrasts them with the CAPM and factor models. The introduction previews in-sample and out-of-sample results, illustrates with firm examples, and positions the work against the idiosyncratic-volatility and option-forecasting literatures. The theory section is followed by variance measurement, tests of the theory-implied coefficients, a check against characteristics, out-of-sample R2 comparisons, and a conclusion restating the distinctive features. Appendices cover implied correlation and the block bootstrap.", "writing_profile_keywords": ["parameter-free theory", "option-implied variance", "cross-section of expected returns", "out-of-sample R2", "CAPM comparison", "confident tone"], "confidence": 0.8, "notes": "The roadmap and the related-literature discussion are merged in the same introduction paragraph. Significance-test conventions are unclear from the excerpts. Robustness evidence is spread across subsample estimates (Figure 7), characteristic controls (Section IV), and the appendix and Internet Appendix."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.18, "usage": {"input": 8562, "output": 2160, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078242, "request_id": "req_011CggqRYnguqXyWf2iMEUMR", "cached": false, "ts": "2026-11-03T23:08:20+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/eb979d6eda3363630fa2240fea8f0f643b6c16507aa523174acf354329fd56ad.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"macro_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Effect of city-level local government debt on investment and financing constraints of private versus state-owned manufacturing firms\",\"strategy\":\"Exploits geographic segmentation of China's credit market. Uses city-level and firm-level panel regressions, heterogeneity by firm distance to city border, and Rajan-Zingales external-dependence interactions with city-year, industry-year and industry-city fixed effects. Adds cash-flow sensitivity tests with exogenous and endogenous (switching regression) sample splits.\",\"key_assumptions\":[\"Credit markets are geographically segmented across cities\",\"Industry external financial dependence is technologically determined and comparable across cities\",\"Cash-flow sensitivity of investment indicates financing constraints\",\"Fixed effects absorb confounding local demand shocks\"],\"threats_addressed\":[\"reverse causality\",\"omitted variables\",\"endogeneity of industry financing needs to local conditions\",\"misclassification from exogenous sample splits\"],\"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\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"preview_of_results\"},{\"paragraph_index\":0,\"move\":\"identification_defense\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":3,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"}],\"contribution_paragraph_index\":4,\"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\":\"data\"},{\"idx\":2,\"role\":\"empirical_design\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"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\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define the dependent variable and how it is computed, then list the explanatory variables with their abbreviations. A separate variable-definition table gives data sources (ASIF, ATS). Regression tables move from city-level to firm-level to industry-level specifications. The switching regression table reports separate selection and outcome panels.\",\"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\":[\"'our estimates suggest'\",\"'is likely to'\",\"'could undercut'\",\"'consistent with'\",\"calling correlational evidence 'preliminary'\",\"calling estimates 'conservative'\"]},\"claim_evidence\":[{\"claim\":\"Cities with more local public debt show lower manufacturing investment\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III, Table V\"},{\"claim\":\"Local debt is associated with lower capital productivity in city manufacturing\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Firms near city borders, with access to banks in other cities, are less affected by crowding-out\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Local debt reduces investment more for financially dependent private firms but not SOEs\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII, Figure 3\"},{\"claim\":\"Local debt raises investment–cash flow sensitivity for private and small firms\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX, Table X, Figure 4\"},{\"claim\":\"In a switching regression, debt affects cash-flow sensitivity only for constrained firms; private and small firms are more often constrained\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"},{\"claim\":\"The purpose-built local debt data line up with official NAO aggregates\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"China's credit market is geographically segmented\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"}],\"structural_summary\":\"The paper opens with the post-2008 surge in Chinese local debt and states the crowding-out finding upfront. It argues that geographic credit-market segmentation makes China a good test setting, then previews three escalating identification approaches before positioning itself against the public debt–growth and Chinese stimulus literatures. The body proceeds in order: novel debt data, documentation of segmentation, preliminary city-level and firm-level correlations, border-distance heterogeneity, Rajan-Zingales industry interactions, and cash-flow sensitivity tests with exogenous and then endogenous splits. Each section motivates itself by the weakness of the previous approach. The conclusion revisits the stimulus and warns of costs from capital misallocation.\",\"writing_profile_keywords\":[\"institutional setting as identification\",\"novel hand-collected data\",\"layered identification strategies\",\"heterogeneity tests\",\"financing constraints\",\"state-owned vs private firms\",\"policy relevance\"],\"confidence\":0.8,\"notes\":\"Standard error treatment and significance conventions are not visible in the excerpted table notes. No explicit roadmap paragraph appears in the excerpts. The intro excerpt omits the description of the first approach, which is likely city-level regressions.\"}", "json": {"genre": "empirical", "subfield": "corporate_finance", "subfield_secondary": "macro_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "Effect of city-level local government debt on investment and financing constraints of private versus state-owned manufacturing firms", "strategy": "Exploits geographic segmentation of China's credit market. Uses city-level and firm-level panel regressions, heterogeneity by firm distance to city border, and Rajan-Zingales external-dependence interactions with city-year, industry-year and industry-city fixed effects. Adds cash-flow sensitivity tests with exogenous and endogenous (switching regression) sample splits.", "key_assumptions": ["Credit markets are geographically segmented across cities", "Industry external financial dependence is technologically determined and comparable across cities", "Cash-flow sensitivity of investment indicates financing constraints", "Fixed effects absorb confounding local demand shocks"], "threats_addressed": ["reverse causality", "omitted variables", "endogeneity of industry financing needs to local conditions", "misclassification from exogenous sample splits"], "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": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "preview_of_results"}, {"paragraph_index": 0, "move": "identification_defense"}, {"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": "related_literature_positioning"}, {"paragraph_index": 3, "move": "literature_gap"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 4, "move": "contribution_statement"}], "contribution_paragraph_index": 4, "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": "data"}, {"idx": 2, "role": "empirical_design"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"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": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["alternative_specification", "alternative_measure", "other"]}, "tables": {"convention_notes": "Table notes define the dependent variable and how it is computed, then list the explanatory variables with their abbreviations. A separate variable-definition table gives data sources (ASIF, ATS). Regression tables move from city-level to firm-level to industry-level specifications. The switching regression table reports separate selection and outcome panels.", "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": ["'our estimates suggest'", "'is likely to'", "'could undercut'", "'consistent with'", "calling correlational evidence 'preliminary'", "calling estimates 'conservative'"]}, "claim_evidence": [{"claim": "Cities with more local public debt show lower manufacturing investment", "evidence_type": "table", "evidence_ref": "Table III, Table V"}, {"claim": "Local debt is associated with lower capital productivity in city manufacturing", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Firms near city borders, with access to banks in other cities, are less affected by crowding-out", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Local debt reduces investment more for financially dependent private firms but not SOEs", "evidence_type": "table", "evidence_ref": "Table VIII, Figure 3"}, {"claim": "Local debt raises investment–cash flow sensitivity for private and small firms", "evidence_type": "table", "evidence_ref": "Table IX, Table X, Figure 4"}, {"claim": "In a switching regression, debt affects cash-flow sensitivity only for constrained firms; private and small firms are more often constrained", "evidence_type": "table", "evidence_ref": "Table XII"}, {"claim": "The purpose-built local debt data line up with official NAO aggregates", "evidence_type": "figure", "evidence_ref": "Figure 1"}, {"claim": "China's credit market is geographically segmented", "evidence_type": "figure", "evidence_ref": "Figure 2"}], "structural_summary": "The paper opens with the post-2008 surge in Chinese local debt and states the crowding-out finding upfront. It argues that geographic credit-market segmentation makes China a good test setting, then previews three escalating identification approaches before positioning itself against the public debt–growth and Chinese stimulus literatures. The body proceeds in order: novel debt data, documentation of segmentation, preliminary city-level and firm-level correlations, border-distance heterogeneity, Rajan-Zingales industry interactions, and cash-flow sensitivity tests with exogenous and then endogenous splits. Each section motivates itself by the weakness of the previous approach. The conclusion revisits the stimulus and warns of costs from capital misallocation.", "writing_profile_keywords": ["institutional setting as identification", "novel hand-collected data", "layered identification strategies", "heterogeneity tests", "financing constraints", "state-owned vs private firms", "policy relevance"], "confidence": 0.8, "notes": "Standard error treatment and significance conventions are not visible in the excerpted table notes. No explicit roadmap paragraph appears in the excerpts. The intro excerpt omits the description of the first approach, which is likely city-level regressions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.44, "usage": {"input": 7392, "output": 2284, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076042, "request_id": "req_011CggqzcB1oPyYyFVaknH6d", "cached": false, "ts": "2026-11-03T23:15:51+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ef78eb0321e85eecdb5b3f406f1b91ac1c8250766920cb82c3461c94d5b14023.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"derivatives pricing / option implied volatility\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (pricing relation linking fair implied volatility to risk-neutral moments of implied volatility changes; empirical part tests forecasting and trading implications)\",\"strategy\":\"Analytical derivation from BMS-based P&L attribution plus no dynamic arbitrage; empirical illustration on SPX options with forecasting regressions and out-of-sample trading strategies\",\"key_assumptions\":[\"continuous dynamics of underlying and implied volatility\",\"zero interest rates and carrying costs for simplicity\",\"local or global commonality assumptions on implied volatility moments for cross-sectional pricing\"],\"threats_addressed\":[\"out-of-sample evaluation of trading strategies\",\"comparison of cross-sectional vs time-series moment estimators\"],\"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\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"other\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":5,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":6,\"move\":\"preview_of_approach\"},{\"paragraph_index\":7,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":8,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":9,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":3,\"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\":\"theory\"},{\"idx\":3,\"role\":\"theory\"},{\"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\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Table notes spell out the regression equation and the statistics reported (mean, SD, autocorrelation, R2). Forecasting tables give absolute Newey-West t-values in parentheses. Trading results are annualized and arranged on maturity-by-moneyness grids.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"newey_west\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"framing the new approach as complementary to traditional models\",\"explicit simplifying assumptions\",\"a dedicated limitations subsection\"]},\"claim_evidence\":[{\"claim\":\"Fair implied volatility equals underlying volatility adjusted for the implied volatility's risk-neutral drift, variance, and covariance with the underlying return\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II, equation (9)\"},{\"claim\":\"Implied volatility change correlations decline with distance in moneyness and maturity, indicating local commonality\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table I, Figure 1\"},{\"claim\":\"Implied volatility movements across the surface show a global factor structure\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 2\"},{\"claim\":\"The term structure slope of at-the-money implied variance helps predict implied volatility changes\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Variance and covariance rates extracted cross-sectionally from the smile forecast realized variance and covariance alongside time-series estimators\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VI-VII\"},{\"claim\":\"Trading strategies based on the framework generate out-of-sample option P&L, reported with information ratios\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX, XI\"}],\"structural_summary\":\"Opens with an epigraph and a critique: traditional option pricing achieves internal consistency through full dynamics specifications, but those dynamics are often wrong. Proposes a top-down, localized framework that prices each option from its own P&L attribution and moment conditions. Contrasts it with bottom-up approaches and positions it against vanna-volga models and market models of implied volatility. Theory sections derive the single-contract pricing relation and then cross-sectional implications under commonality assumptions. A long empirical section on SPX options documents factor structure, extracts moment rates, runs forecasting regressions, and tests out-of-sample trading strategies. The paper closes with implications, limitations, a risk representation discussion, and a conclusion pointing to future research.\",\"writing_profile_keywords\":[\"option pricing\",\"implied volatility\",\"P&L attribution\",\"top-down valuation\",\"theory with empirical illustration\",\"SPX options\",\"trading strategies\",\"epigraph opening\"],\"confidence\":0.8,\"notes\":\"No explicit roadmap paragraph is visible. Literature positioning is woven into the introduction. Robustness checks are not visible in the excerpts. The Table XI notes appear to reuse the Table IX wording, likely a typo in the original. The magnitudes reported for trading results are not visible in the excerpts.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "asset_pricing", "subfield_secondary": "derivatives pricing / option implied volatility", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (pricing relation linking fair implied volatility to risk-neutral moments of implied volatility changes; empirical part tests forecasting and trading implications)", "strategy": "Analytical derivation from BMS-based P&L attribution plus no dynamic arbitrage; empirical illustration on SPX options with forecasting regressions and out-of-sample trading strategies", "key_assumptions": ["continuous dynamics of underlying and implied volatility", "zero interest rates and carrying costs for simplicity", "local or global commonality assumptions on implied volatility moments for cross-sectional pricing"], "threats_addressed": ["out-of-sample evaluation of trading strategies", "comparison of cross-sectional vs time-series moment estimators"], "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": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "other"}, {"paragraph_index": 1, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 5, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 6, "move": "preview_of_approach"}, {"paragraph_index": 7, "move": "related_literature_positioning"}, {"paragraph_index": 8, "move": "related_literature_positioning"}, {"paragraph_index": 9, "move": "related_literature_positioning"}], "contribution_paragraph_index": 3, "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": "theory"}, {"idx": 3, "role": "theory"}, {"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": "stars_and_tstats", "magnitude_statements_in_intro": false}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Table notes spell out the regression equation and the statistics reported (mean, SD, autocorrelation, R2). Forecasting tables give absolute Newey-West t-values in parentheses. Trading results are annualized and arranged on maturity-by-moneyness grids.", "units_stated_in_notes": true, "standard_errors_description": "newey_west"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["framing the new approach as complementary to traditional models", "explicit simplifying assumptions", "a dedicated limitations subsection"]}, "claim_evidence": [{"claim": "Fair implied volatility equals underlying volatility adjusted for the implied volatility's risk-neutral drift, variance, and covariance with the underlying return", "evidence_type": "equation_or_proof", "evidence_ref": "Section II, equation (9)"}, {"claim": "Implied volatility change correlations decline with distance in moneyness and maturity, indicating local commonality", "evidence_type": "table", "evidence_ref": "Table I, Figure 1"}, {"claim": "Implied volatility movements across the surface show a global factor structure", "evidence_type": "figure", "evidence_ref": "Figure 2"}, {"claim": "The term structure slope of at-the-money implied variance helps predict implied volatility changes", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Variance and covariance rates extracted cross-sectionally from the smile forecast realized variance and covariance alongside time-series estimators", "evidence_type": "table", "evidence_ref": "Tables VI-VII"}, {"claim": "Trading strategies based on the framework generate out-of-sample option P&L, reported with information ratios", "evidence_type": "table", "evidence_ref": "Tables IX, XI"}], "structural_summary": "Opens with an epigraph and a critique: traditional option pricing achieves internal consistency through full dynamics specifications, but those dynamics are often wrong. Proposes a top-down, localized framework that prices each option from its own P&L attribution and moment conditions. Contrasts it with bottom-up approaches and positions it against vanna-volga models and market models of implied volatility. Theory sections derive the single-contract pricing relation and then cross-sectional implications under commonality assumptions. A long empirical section on SPX options documents factor structure, extracts moment rates, runs forecasting regressions, and tests out-of-sample trading strategies. The paper closes with implications, limitations, a risk representation discussion, and a conclusion pointing to future research.", "writing_profile_keywords": ["option pricing", "implied volatility", "P&L attribution", "top-down valuation", "theory with empirical illustration", "SPX options", "trading strategies", "epigraph opening"], "confidence": 0.8, "notes": "No explicit roadmap paragraph is visible. Literature positioning is woven into the introduction. Robustness checks are not visible in the excerpts. The Table XI notes appear to reuse the Table IX wording, likely a typo in the original. The magnitudes reported for trading results are not visible in the excerpts."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.49, "usage": {"input": 8265, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076654, "request_id": "req_011Cggquqxr4Fi42x58Qjv5v", "cached": false, "ts": "2026-11-03T23:14:45+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f0fe699c8517548b5f055b43fa830d7f310a3b31f8917d7d92db15124f0c2876.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"corporate_finance\",\"subfield_secondary\":\"information economics / strategic voting and corporate governance\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical model); model outcome is firm value, defined as the probability of a correct vote decision, with versus without a proxy advisor\",\"strategy\":\"not applicable; strategic voting model with endogenous information acquisition and a monopolistic information seller, solved by backward induction\",\"key_assumptions\":[\"shareholders have aligned (common) preferences\",\"monopolistic advisor sets its fee to maximize profit\",\"private signals have independent (or imperfectly correlated) errors\",\"the advisor's signal is common to all who buy it\",\"cost of private information exceeds a cutoff in parts of the analysis\"],\"threats_addressed\":[\"equilibrium multiplicity and equilibrium selection\",\"correlated mistakes in private signals\",\"free access to recommendations\",\"communication or sale of information among shareholders\",\"acquisition of both signals in equilibrium\"],\"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\":0,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":1,\"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\":\"related_literature_positioning\"},{\"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\":\"theory\"},{\"idx\":4,\"role\":\"other\"},{\"idx\":5,\"role\":\"theory\"},{\"idx\":6,\"role\":\"robustness\"},{\"idx\":7,\"role\":\"other\"},{\"idx\":8,\"role\":\"conclusion\"},{\"idx\":9,\"role\":\"appendix\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"mixed\",\"types\":[\"other\"]},\"tables\":{\"convention_notes\":\"No tables. Results are stated as propositions and lemmas, and proofs sit in Appendix A and the Internet Appendix. Numerical illustrations appear in figures plotting equilibrium objects against the fee and against signal precision.\",\"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\":[\"conditional statements tied to parameter regions (only if precision is sufficiently high)\",\"'can have negative effects' / 'may' modality\",\"acknowledging the model is stylized\",\"in our model qualifiers\"]},\"claim_evidence\":[{\"claim\":\"The advisor raises firm value only if its information is sufficiently precise, for any fee.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 3; Appendix A\"},{\"claim\":\"With imprecise advisor information, shareholders over-rely on it and private research is crowded out.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II-III; Figure 2\"},{\"claim\":\"With precise information, the monopolist rations recommendations, so the information is underused.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Figure 3\"},{\"claim\":\"Reducing advisor market power or litigation pressure can lower decision quality.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section IV\"},{\"claim\":\"Proxy advisor recommendations strongly affect voting outcomes.\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction footnote 1\"}],\"structural_summary\":\"The introduction opens with institutional motivation (growth of proxy advisors, regulatory attention) and a policy debate. It then sketches the externality mechanism: correlated mistakes when shareholders follow a common signal. Next it previews the main threshold result and the inefficiency created by monopoly pricing, and positions the paper against corporate voting, strategic voting, sale-of-information, and information-externality literatures. The body builds a strategic voting model with a benchmark, solves it by backward induction for voting and acquisition given a fee, endogenizes the fee, and evaluates regulatory proposals. Later sections endogenize advisor quality, discuss robustness of assumptions, and derive empirical implications. The conclusion restates the results and the policy findings; proofs are in the appendix.\",\"writing_profile_keywords\":[\"applied theory\",\"strategic voting\",\"information sale\",\"policy evaluation\",\"backward induction\",\"propositions and proofs\",\"institutional motivation\",\"externalities\"],\"confidence\":0.85,\"notes\":\"Pure theory paper with no data. Robustness is handled verbally in Section VI and in the Internet Appendix. There is an empirical implications section but no tests. No roadmap paragraph is visible in the excerpted introduction.\"}", "json": {"genre": "theoretical", "subfield": "corporate_finance", "subfield_secondary": "information economics / strategic voting and corporate governance", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical model); model outcome is firm value, defined as the probability of a correct vote decision, with versus without a proxy advisor", "strategy": "not applicable; strategic voting model with endogenous information acquisition and a monopolistic information seller, solved by backward induction", "key_assumptions": ["shareholders have aligned (common) preferences", "monopolistic advisor sets its fee to maximize profit", "private signals have independent (or imperfectly correlated) errors", "the advisor's signal is common to all who buy it", "cost of private information exceeds a cutoff in parts of the analysis"], "threats_addressed": ["equilibrium multiplicity and equilibrium selection", "correlated mistakes in private signals", "free access to recommendations", "communication or sale of information among shareholders", "acquisition of both signals in equilibrium"], "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": 0, "move": "policy_or_broader_implication"}, {"paragraph_index": 1, "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": "related_literature_positioning"}, {"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": "theory"}, {"idx": 4, "role": "other"}, {"idx": 5, "role": "theory"}, {"idx": 6, "role": "robustness"}, {"idx": 7, "role": "other"}, {"idx": 8, "role": "conclusion"}, {"idx": 9, "role": "appendix"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": false}, "robustness": {"location": "mixed", "types": ["other"]}, "tables": {"convention_notes": "No tables. Results are stated as propositions and lemmas, and proofs sit in Appendix A and the Internet Appendix. Numerical illustrations appear in figures plotting equilibrium objects against the fee and against signal precision.", "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": ["conditional statements tied to parameter regions (only if precision is sufficiently high)", "'can have negative effects' / 'may' modality", "acknowledging the model is stylized", "in our model qualifiers"]}, "claim_evidence": [{"claim": "The advisor raises firm value only if its information is sufficiently precise, for any fee.", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 3; Appendix A"}, {"claim": "With imprecise advisor information, shareholders over-rely on it and private research is crowded out.", "evidence_type": "equation_or_proof", "evidence_ref": "Section II-III; Figure 2"}, {"claim": "With precise information, the monopolist rations recommendations, so the information is underused.", "evidence_type": "equation_or_proof", "evidence_ref": "Section III; Figure 3"}, {"claim": "Reducing advisor market power or litigation pressure can lower decision quality.", "evidence_type": "equation_or_proof", "evidence_ref": "Section IV"}, {"claim": "Proxy advisor recommendations strongly affect voting outcomes.", "evidence_type": "prior_literature", "evidence_ref": "Introduction footnote 1"}], "structural_summary": "The introduction opens with institutional motivation (growth of proxy advisors, regulatory attention) and a policy debate. It then sketches the externality mechanism: correlated mistakes when shareholders follow a common signal. Next it previews the main threshold result and the inefficiency created by monopoly pricing, and positions the paper against corporate voting, strategic voting, sale-of-information, and information-externality literatures. The body builds a strategic voting model with a benchmark, solves it by backward induction for voting and acquisition given a fee, endogenizes the fee, and evaluates regulatory proposals. Later sections endogenize advisor quality, discuss robustness of assumptions, and derive empirical implications. The conclusion restates the results and the policy findings; proofs are in the appendix.", "writing_profile_keywords": ["applied theory", "strategic voting", "information sale", "policy evaluation", "backward induction", "propositions and proofs", "institutional motivation", "externalities"], "confidence": 0.85, "notes": "Pure theory paper with no data. Robustness is handled verbally in Section VI and in the Internet Appendix. There is an empirical implications section but no tests. No roadmap paragraph is visible in the excerpted introduction."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.42, "usage": {"input": 6454, "output": 2039, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06739, "request_id": "req_011CggqVT8jCnp6CQwYbpJjv", "cached": false, "ts": "2026-11-03T23:09:14+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f1a99b7136bece4df6d2d0c5765c8095d2a771474c414d367847555c170cc26a.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"financial_intermediation_banking\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between quarterly changes in aggregate DB pension underfunding ratio (conditional on underfunded status) and changes in 30-year swap spreads\",\"strategy\":\"OLS time-series regressions of swap spread changes on changes in the underfunding ratio, with controls; 2SLS that instruments underfunding with excess stock market returns; cross-maturity and cross-country comparisons\",\"key_assumptions\":[\"Stock returns affect pension funding status through the asset side but have no direct link to swap spreads (exclusion restriction)\",\"The control variables capture other known drivers of swap spreads\",\"Dealer balance sheet constraints bind, so supply of long-dated swaps is limited\"],\"threats_addressed\":[\"The level of the yield curve jointly driving swap spreads and underfunding\",\"Omitted controls such as Treasury CDS premia and corporate bond issuance\",\"Time-period dependence of the relationship\"],\"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\":2},\"results\":{\"present\":true,\"sentence_index\":4},\"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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":1,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"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\",\"new_data_or_measure\",\"new_theory_or_model\",\"resolution_of_puzzle\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"other\"},{\"idx\":2,\"role\":\"other\"},{\"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\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":false},\"robustness\":{\"location\":\"main_text_section\",\"types\":[\"instrumental_variables_like_other\",\"alternative_specification\",\"alternative_sample_period\",\"alternative_measure\",\"subsample\"].filter?[]:[\"alternative_specification\",\"alternative_sample_period\",\"alternative_measure\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes define regression variables and state that the dependent variable is the quarterly change in swap spreads. Separate tables report OLS, 2SLS second-stage, cross-maturity, robustness and international results. A descriptive balance-sheet table states its units (billions of dollars) and gives its source.\",\"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\":[\"'helps explain'\",\"'can drive'\",\"'illustrative model'\",\"'support the view'\",\"'fairly consistent evidence'\",\"'possible reasons'\"]},\"claim_evidence\":[{\"claim\":\"The 30-year US swap spread turned negative after Lehman's default and has stayed negative\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"},{\"claim\":\"Underfunded pension plans optimally take receive-fixed positions in long-dated swaps; dealer constraints yield negative spreads\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section III; Appendix: Model and Proofs\"},{\"claim\":\"Changes in underfunding explain changes in 30-year swap spreads when plans are underfunded, controlling for other factors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Funding status has no explanatory power for swap spreads at 2–10 year maturities\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"The relationship survives 2SLS using stock returns as an instrument for underfunding\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Results are robust to added controls, sample periods and an alternative UFR measure\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Pension underfunding is linked to swap spreads in Japan and the Netherlands\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"}],\"structural_summary\":\"The paper opens by presenting a persistent pricing anomaly, negative 30-year swap spreads, and contrasting it with conventional credit-risk and convenience-yield views. It proposes a demand-driven hypothesis and documents stylized facts. Institutional background on pension duration needs and on swap supply follows. A model-free argument and a simple model link underfunding and dealer constraints to negative spreads. US regressions use a constructed underfunding measure, regime splits and 2SLS, followed by robustness tests and international replication for Japan and the Netherlands. A short conclusion restates the mechanism and the evidence.\",\"writing_profile_keywords\":[\"anomaly-driven\",\"demand-based explanation\",\"illustrative model\",\"limits to arbitrage\",\"pension funds\",\"instrumental variables\",\"international evidence\"],\"confidence\":0.75,\"notes\":\"Robustness types assigned: alternative_specification (added controls), alternative_sample_period (different time periods), alternative_measure (modified UFR), and other (the IV/2SLS test). 'subsample' is not assigned because sample splits beyond time periods are not visible. The method of reporting statistical significance is not visible in the excerpts, so it is coded unclear.\"}", "json": null, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.53, "usage": {"input": 7487, "output": 2347, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077682, "request_id": "req_011CggqH1ssBDghq937Byzkm", "cached": false, "ts": "2026-11-03T23:06:27+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f28c8fd8596d1a04492680e430fe5c26df4bec9721cd8f9b1f213c5b087689d5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"theoretical\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (theoretical paper)\",\"strategy\":\"not applicable; analytical characterization of equilibrium trade in a multiperiod exchange economy\",\"key_assumptions\":[\"complete market with respect to endowment states\",\"rational agents with common priors\",\"securities measurable to the information filtration\",\"heterogeneous time-inseparable preferences for trade to arise\"],\"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\":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\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"mechanism_or_theory_sketch\"}],\"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\":\"other\"},{\"idx\":4,\"role\":\"conclusion\"},{\"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 single table is schematic rather than numerical. It lays out the state-contingent securities available in the baseline model and in a two-goods variant, by endowment state.\",\"units_stated_in_notes\":false,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"no_figures\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"conditions stated as necessary and sufficient\",\"explicit statement of when the no-trade result still holds\",\"'may provide new insights' in future research\"]},\"claim_evidence\":[{\"claim\":\"A necessary and sufficient condition for information-based trade exists in a multiperiod consumption economy\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II, Appendix A\"},{\"claim\":\"Equilibrium exists, is unique, interior and fully revealing\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Lemma 1, Appendix A\"},{\"claim\":\"No-trade persists without aggregate shocks or with homogeneous or time-separable preferences\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Section II.C\"},{\"claim\":\"Multiperiod extensions are commonly believed to be isomorphic to single-period no-trade settings\",\"evidence_type\":\"prior_literature\",\"evidence_ref\":\"Introduction\"}],\"structural_summary\":\"Short theory paper. It opens by contrasting observed trading volume with the no-trade theorem, then argues against the presumption that dynamic consumption extensions are trivial. A three-date, two-agent exchange model follows, and the paper derives existence and full revelation of equilibrium plus a necessary and sufficient condition for trade. An example and preference-based cases illustrate the condition. A discussion covers filtration, preferences and welfare, and a brief conclusion restates the result and suggests future work. Proofs are in the appendix.\",\"writing_profile_keywords\":[\"no-trade theorem\",\"first-person singular\",\"lemma-proof\",\"stylized model\",\"intertemporal substitution\",\"short paper\"],\"confidence\":0.85,\"notes\":\"Single-author theoretical note. Section III 'Discussions' was rule-tagged as conclusion but is labeled 'other' here because it extends and interprets the model. Section II also blends theory with an illustrative example.\"}", "json": {"genre": "theoretical", "subfield": "asset_pricing", "subfield_secondary": "market_microstructure", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (theoretical paper)", "strategy": "not applicable; analytical characterization of equilibrium trade in a multiperiod exchange economy", "key_assumptions": ["complete market with respect to endowment states", "rational agents with common priors", "securities measurable to the information filtration", "heterogeneous time-inseparable preferences for trade to arise"], "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": 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": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "mechanism_or_theory_sketch"}], "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": "other"}, {"idx": 4, "role": "conclusion"}, {"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 single table is schematic rather than numerical. It lays out the state-contingent securities available in the baseline model and in a two-goods variant, by endowment state.", "units_stated_in_notes": false, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "no_figures"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "low", "typical_devices": ["conditions stated as necessary and sufficient", "explicit statement of when the no-trade result still holds", "'may provide new insights' in future research"]}, "claim_evidence": [{"claim": "A necessary and sufficient condition for information-based trade exists in a multiperiod consumption economy", "evidence_type": "equation_or_proof", "evidence_ref": "Section II, Appendix A"}, {"claim": "Equilibrium exists, is unique, interior and fully revealing", "evidence_type": "equation_or_proof", "evidence_ref": "Lemma 1, Appendix A"}, {"claim": "No-trade persists without aggregate shocks or with homogeneous or time-separable preferences", "evidence_type": "equation_or_proof", "evidence_ref": "Section II.C"}, {"claim": "Multiperiod extensions are commonly believed to be isomorphic to single-period no-trade settings", "evidence_type": "prior_literature", "evidence_ref": "Introduction"}], "structural_summary": "Short theory paper. It opens by contrasting observed trading volume with the no-trade theorem, then argues against the presumption that dynamic consumption extensions are trivial. A three-date, two-agent exchange model follows, and the paper derives existence and full revelation of equilibrium plus a necessary and sufficient condition for trade. An example and preference-based cases illustrate the condition. A discussion covers filtration, preferences and welfare, and a brief conclusion restates the result and suggests future work. Proofs are in the appendix.", "writing_profile_keywords": ["no-trade theorem", "first-person singular", "lemma-proof", "stylized model", "intertemporal substitution", "short paper"], "confidence": 0.85, "notes": "Single-author theoretical note. Section III 'Discussions' was rule-tagged as conclusion but is labeled 'other' here because it extends and interprets the model. Section II also blends theory with an illustrative example."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 12.13, "usage": {"input": 2964, "output": 1547, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.04359, "request_id": "req_011CggqhWy1ESYELfnspoFdC", "cached": false, "ts": "2026-11-03T23:11:52+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f314565a9ff32af388ccdd1a3c03171210f15e636e91d0c0f8a265b09e5fcb1d.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"entrepreneurial_finance_private_equity\",\"subfield_secondary\":\"innovation and intellectual property economics\",\"primary_method\":\"instrumental_variables\",\"identification\":{\"estimand\":\"Causal effect of first patent approval (the legal right, separate from the underlying invention) on startup employment growth, sales growth, survival, follow-on innovation, and access to external finance\",\"strategy\":\"2SLS using the past approval rate (leniency) of the randomly assigned examiner within art unit-by-year as an instrument for patent approval\",\"key_assumptions\":[\"Applications are assigned to examiners quasi-randomly within art unit and year, orthogonal to application quality\",\"Examiner leniency affects outcomes only through the approval decision (exclusion restriction)\",\"Relevance: leniency strongly predicts approval\"],\"threats_addressed\":[\"Correlation of examiner leniency with applicant, application, or examiner characteristics (instrument validity tests)\",\"Review speed as an alternative channel\",\"External validity, given that compliers are marginal applications\"],\"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\":\"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\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":3,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"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\":\"within_results\",\"types\":[\"subsample\",\"falsification\",\"other\"]},\"tables\":{\"convention_notes\":\"Table titles are phrased as questions. Notes reference the numbered estimating equation, describe the 2SLS setup and the identification assumption, define dependent variables over 1–5 year horizons, and explain data availability. Separate tables report the first stage and instrument validity.\",\"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\":[\"'plausibly exogenous'\",\"explicit disclaimers about welfare optimality\",\"'suggest', 'raise the possibility'\",\"stated agnosticism about efficiency\"]},\"claim_evidence\":[{\"claim\":\"A first patent grant raises five-year employment growth by about 55 pp and sales growth by about 80 pp\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV; Figure 2\"},{\"claim\":\"A first patent increases the number and citation quality of subsequent patents\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Second patent approvals boost follow-on patenting but not growth\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Patents raise the likelihood of VC funding, IPOs, and patent-collateralized loans\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"},{\"claim\":\"The VC effect is stronger for inexperienced founders and crowded startup states, and weaker in biochemistry\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IX\"},{\"claim\":\"Examiner leniency strongly predicts approval and is unrelated to applicant characteristics\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III; Figure 1\"}],\"structural_summary\":\"The paper opens with the theoretical benefits and costs of patents and asks how valuable patents are to startups. It identifies the difficulty of separating a patent's value from the underlying invention and uses examiner leniency as an instrument. The introduction then previews magnitudes for growth and innovation, contrasts first versus later patents, previews the funding-channel mechanism, and lists contributions with welfare caveats. The body covers the institutional setting and data, then the identification and threats to it, the main real effects, financing channels, and heterogeneity explaining how patents ease access to capital. The conclusion restates the findings with bounded welfare claims and a policy warning.\",\"writing_profile_keywords\":[\"patent lottery\",\"examiner leniency IV\",\"startups\",\"entrepreneurial finance\",\"quantified magnitudes\",\"welfare caveats\",\"mechanism via financing\"],\"confidence\":0.85,\"notes\":\"Section I is tagged 'theory' by the role rule but is actually institutional setting and data. Standard error type and significance notation are not visible in the excerpts. No roadmap paragraph is visible.\"}", "json": {"genre": "empirical", "subfield": "entrepreneurial_finance_private_equity", "subfield_secondary": "innovation and intellectual property economics", "primary_method": "instrumental_variables", "identification": {"estimand": "Causal effect of first patent approval (the legal right, separate from the underlying invention) on startup employment growth, sales growth, survival, follow-on innovation, and access to external finance", "strategy": "2SLS using the past approval rate (leniency) of the randomly assigned examiner within art unit-by-year as an instrument for patent approval", "key_assumptions": ["Applications are assigned to examiners quasi-randomly within art unit and year, orthogonal to application quality", "Examiner leniency affects outcomes only through the approval decision (exclusion restriction)", "Relevance: leniency strongly predicts approval"], "threats_addressed": ["Correlation of examiner leniency with applicant, application, or examiner characteristics (instrument validity tests)", "Review speed as an alternative channel", "External validity, given that compliers are marginal applications"], "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": "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": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "policy_or_broader_implication"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "related_literature_positioning"}], "contribution_paragraph_index": 3, "contribution_types": ["new_evidence", "new_identification", "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": "within_results", "types": ["subsample", "falsification", "other"]}, "tables": {"convention_notes": "Table titles are phrased as questions. Notes reference the numbered estimating equation, describe the 2SLS setup and the identification assumption, define dependent variables over 1–5 year horizons, and explain data availability. Separate tables report the first stage and instrument validity.", "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": ["'plausibly exogenous'", "explicit disclaimers about welfare optimality", "'suggest', 'raise the possibility'", "stated agnosticism about efficiency"]}, "claim_evidence": [{"claim": "A first patent grant raises five-year employment growth by about 55 pp and sales growth by about 80 pp", "evidence_type": "table", "evidence_ref": "Table IV; Figure 2"}, {"claim": "A first patent increases the number and citation quality of subsequent patents", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Second patent approvals boost follow-on patenting but not growth", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Patents raise the likelihood of VC funding, IPOs, and patent-collateralized loans", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}, {"claim": "The VC effect is stronger for inexperienced founders and crowded startup states, and weaker in biochemistry", "evidence_type": "table", "evidence_ref": "Table IX"}, {"claim": "Examiner leniency strongly predicts approval and is unrelated to applicant characteristics", "evidence_type": "table", "evidence_ref": "Tables II, III; Figure 1"}], "structural_summary": "The paper opens with the theoretical benefits and costs of patents and asks how valuable patents are to startups. It identifies the difficulty of separating a patent's value from the underlying invention and uses examiner leniency as an instrument. The introduction then previews magnitudes for growth and innovation, contrasts first versus later patents, previews the funding-channel mechanism, and lists contributions with welfare caveats. The body covers the institutional setting and data, then the identification and threats to it, the main real effects, financing channels, and heterogeneity explaining how patents ease access to capital. The conclusion restates the findings with bounded welfare claims and a policy warning.", "writing_profile_keywords": ["patent lottery", "examiner leniency IV", "startups", "entrepreneurial finance", "quantified magnitudes", "welfare caveats", "mechanism via financing"], "confidence": 0.85, "notes": "Section I is tagged 'theory' by the role rule but is actually institutional setting and data. Standard error type and significance notation are not visible in the excerpts. No roadmap paragraph is visible."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.57, "usage": {"input": 7459, "output": 2133, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07329, "request_id": "req_011CggqhwHEavj9GppuCvxb9", "cached": false, "ts": "2026-11-03T23:12:03+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f627ff7fb9747401fd6b1840f956d6b2c377a1f4b2cc94f9a766b2aa36048e28.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"asset_pricing; behavioral_finance\",\"primary_method\":\"survey_or_experiment\",\"identification\":{\"estimand\":\"Self-reported importance of theorized factors and beliefs (shares of respondents rating factors as very/extremely important; distributions of beliefs about funds and the cross-section of returns)\",\"strategy\":\"Direct survey of a nationally representative panel (RAND ALP, N=1,013), with questions mapped to specific theories; descriptive proportions split by participation, wealth and education; PCA; OLS correlates\",\"key_assumptions\":[\"Respondents can accurately introspect about and report their decision drivers\",\"Question wording isolates the targeted theory and excludes others\",\"The sample is representative of U.S. individuals\"],\"threats_addressed\":[\"Observational data cannot cleanly separate competing theories with similar predictions\",\"Comprehensibility of questions to laypeople\",\"Heterogeneity across wealth, education and participation subgroups\"],\"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\":\"preview_of_approach\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"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\":\"contribution_statement\"},{\"paragraph_index\":5,\"move\":\"policy_or_broader_implication\"},{\"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\":\"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\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Most tables report percentages of respondents rating each factor very or extremely important, for the full sample and for splits by participation, wealth (cutoff of $100,000) and bachelor's degree. Standard errors appear in parentheses below estimates. Notes quote the survey question wording and give subsample N. The PCA and OLS tables report loadings and coefficients.\",\"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\":[\"tend to believe\",\"roughly interpreted as\",\"appear to\",\"acknowledging that high-wealth investors are a small share of the sample\"]},\"claim_evidence\":[{\"claim\":\"Background risk, horizon, rare disasters, transactional factors and fixed participation costs receive strong support as drivers of equity share\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III, VII, XI, XII\"},{\"claim\":\"Six principal components explain about half the variance in importance ratings across 34 factors\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XIII\"},{\"claim\":\"Belief in higher active-fund returns and adviser recommendations are the main reasons for buying active funds\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XV\"},{\"claim\":\"Respondents view past fund outperformance as a skill signal but mostly reject diseconomies of scale, contrary to Berk and Green\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XV\"},{\"claim\":\"Value stocks are seen as safer without higher expected returns; high-momentum stocks are seen as riskier with higher returns\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables XVI, XVII\"}],\"structural_summary\":\"The introduction argues that observational tests struggle to discriminate between investor theories. It then proposes asking investors directly and previews descriptive survey findings in three domains: equity share, active funds, and value/momentum beliefs. It positions the work against corporate-finance survey studies, lists reasons why reported beliefs matter, and closes with a roadmap. A data section describes question design and the sample. Three results sections follow the survey's own order, each organized by family of theory, with percentage tables, subgroup splits, follow-up questions, PCA and regressions. A one-paragraph conclusion restates the findings.\",\"writing_profile_keywords\":[\"survey evidence\",\"theory testing via self-reports\",\"descriptive percentages\",\"many tables, no figures\",\"subgroup splits\",\"household portfolio choice\",\"beliefs about anomalies\"],\"confidence\":0.82,\"notes\":\"The excerpts show no figures and no equations. No robustness section is visible, so robustness checks may exist only in the Internet Appendix, which is not shown. Statistical significance is reported through standard errors in parentheses; whether stars are also used is not visible. The intro motivation is split between paragraphs 0 and 3–5 because of truncated text.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "asset_pricing; behavioral_finance", "primary_method": "survey_or_experiment", "identification": {"estimand": "Self-reported importance of theorized factors and beliefs (shares of respondents rating factors as very/extremely important; distributions of beliefs about funds and the cross-section of returns)", "strategy": "Direct survey of a nationally representative panel (RAND ALP, N=1,013), with questions mapped to specific theories; descriptive proportions split by participation, wealth and education; PCA; OLS correlates", "key_assumptions": ["Respondents can accurately introspect about and report their decision drivers", "Question wording isolates the targeted theory and excludes others", "The sample is representative of U.S. individuals"], "threats_addressed": ["Observational data cannot cleanly separate competing theories with similar predictions", "Comprehensibility of questions to laypeople", "Heterogeneity across wealth, education and participation subgroups"], "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": "preview_of_approach"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"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": "contribution_statement"}, {"paragraph_index": 5, "move": "policy_or_broader_implication"}, {"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": "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": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Most tables report percentages of respondents rating each factor very or extremely important, for the full sample and for splits by participation, wealth (cutoff of $100,000) and bachelor's degree. Standard errors appear in parentheses below estimates. Notes quote the survey question wording and give subsample N. The PCA and OLS tables report loadings and coefficients.", "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": ["tend to believe", "roughly interpreted as", "appear to", "acknowledging that high-wealth investors are a small share of the sample"]}, "claim_evidence": [{"claim": "Background risk, horizon, rare disasters, transactional factors and fixed participation costs receive strong support as drivers of equity share", "evidence_type": "table", "evidence_ref": "Tables II, III, VII, XI, XII"}, {"claim": "Six principal components explain about half the variance in importance ratings across 34 factors", "evidence_type": "table", "evidence_ref": "Table XIII"}, {"claim": "Belief in higher active-fund returns and adviser recommendations are the main reasons for buying active funds", "evidence_type": "table", "evidence_ref": "Table XV"}, {"claim": "Respondents view past fund outperformance as a skill signal but mostly reject diseconomies of scale, contrary to Berk and Green", "evidence_type": "table", "evidence_ref": "Table XV"}, {"claim": "Value stocks are seen as safer without higher expected returns; high-momentum stocks are seen as riskier with higher returns", "evidence_type": "table", "evidence_ref": "Tables XVI, XVII"}], "structural_summary": "The introduction argues that observational tests struggle to discriminate between investor theories. It then proposes asking investors directly and previews descriptive survey findings in three domains: equity share, active funds, and value/momentum beliefs. It positions the work against corporate-finance survey studies, lists reasons why reported beliefs matter, and closes with a roadmap. A data section describes question design and the sample. Three results sections follow the survey's own order, each organized by family of theory, with percentage tables, subgroup splits, follow-up questions, PCA and regressions. A one-paragraph conclusion restates the findings.", "writing_profile_keywords": ["survey evidence", "theory testing via self-reports", "descriptive percentages", "many tables, no figures", "subgroup splits", "household portfolio choice", "beliefs about anomalies"], "confidence": 0.82, "notes": "The excerpts show no figures and no equations. No robustness section is visible, so robustness checks may exist only in the Internet Appendix, which is not shown. Statistical significance is reported through standard errors in parentheses; whether stars are also used is not visible. The intro motivation is split between paragraphs 0 and 3–5 because of truncated text."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 18.04, "usage": {"input": 7208, "output": 2051, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070646, "request_id": "req_011CggqrsQaM56nkyLoMxWZ5", "cached": false, "ts": "2026-11-03T23:14:05+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/f7bea21d31e170ae50858c02f74a3528a0d73ef2d4134697f52e1dd93e53d80f.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"portfolio choice / covariance estimation (theoretical correction)\",\"primary_method\":\"theoretical_model_analytical\",\"identification\":{\"estimand\":\"not applicable (analytical correction of a proof)\",\"strategy\":\"not applicable; corrected KKT-based derivation of the adjusted covariance matrix and constrained MLE equivalence\",\"key_assumptions\":[\"returns i.i.d. jointly normal (for Proposition 2)\",\"adjusted covariance matrix nonsingular\",\"sample covariance matrix positive semidefinite\"],\"threats_addressed\":[],\"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\":false,\"sentence_index\":-1},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"contribution_statement\"},{\"paragraph_index\":8,\"move\":\"other\"},{\"paragraph_index\":25,\"move\":\"other\"},{\"paragraph_index\":57,\"move\":\"policy_or_broader_implication\"}],\"contribution_paragraph_index\":0,\"contribution_types\":[\"replication_or_correction\"],\"roadmap_present\":false,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"theory\"}],\"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; the note consists of an optimization setup, KKT conditions, restated propositions and step-by-step proofs with numbered 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\":\"low\",\"typical_devices\":[\"explicit scope condition: original proofs valid only under nonnegativity constraints\",\"stated assumptions (normality, nonsingularity) in propositions\"]},\"claim_evidence\":[{\"claim\":\"The original equation omitted a term, so the original proofs hold only with no-short-sale constraints and not with upper bounds on weights.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Corrected equation (10) vs. original equation (8)\"},{\"claim\":\"With the corrected adjustment, the shrunk covariance matrix is symmetric and PSD, and the constrained minimum variance portfolio is one of its GMV portfolios.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 1 and proof\"},{\"claim\":\"Under normality, the adjusted matrix is the constrained MLE whose implied GMV portfolio satisfies the weight constraints, and conversely.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Proposition 2 and proof; Morrison (2005); Zwiernik, Uhler, and Richards (2017)\"},{\"claim\":\"The omitted term shifts all variances and covariances equally, so the original shrinkage interpretation is unchanged.\",\"evidence_type\":\"equation_or_proof\",\"evidence_ref\":\"Closing discussion after Proposition 2\"}],\"structural_summary\":\"A short corrigendum in a single unnumbered section. It identifies the published propositions and pinpoints the omitted term in the original equation. It states the scope of the error (valid only without upper bounds), then restates the constrained minimum-variance problem and its KKT conditions. It re-proves Proposition 1 with the corrected adjusted covariance matrix and re-proves Proposition 2 on constrained-MLE equivalence. It closes by noting that the original shrinkage interpretation survives the correction.\",\"writing_profile_keywords\":[\"corrigendum\",\"proof correction\",\"KKT conditions\",\"portfolio constraints\",\"covariance shrinkage\",\"constrained MLE\",\"equation-dense\"],\"confidence\":0.85,\"notes\":\"Abstract is truncated and blends into the body; the body is mostly equations split into fragments that are treated as paragraphs. Paragraph indices for intro moves are approximate, given the fragmented equation text. There are no separate conclusion paragraphs; the conclusion fields are inferred from the closing remarks on the shrinkage interpretation.\"}", "json": {"genre": "replication_or_correction", "subfield": "asset_pricing", "subfield_secondary": "portfolio choice / covariance estimation (theoretical correction)", "primary_method": "theoretical_model_analytical", "identification": {"estimand": "not applicable (analytical correction of a proof)", "strategy": "not applicable; corrected KKT-based derivation of the adjusted covariance matrix and constrained MLE equivalence", "key_assumptions": ["returns i.i.d. jointly normal (for Proposition 2)", "adjusted covariance matrix nonsingular", "sample covariance matrix positive semidefinite"], "threats_addressed": [], "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": false, "sentence_index": -1}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "related_literature_positioning"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "contribution_statement"}, {"paragraph_index": 8, "move": "other"}, {"paragraph_index": 25, "move": "other"}, {"paragraph_index": 57, "move": "policy_or_broader_implication"}], "contribution_paragraph_index": 0, "contribution_types": ["replication_or_correction"], "roadmap_present": false, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "theory"}], "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; the note consists of an optimization setup, KKT conditions, restated propositions and step-by-step proofs with numbered 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": "low", "typical_devices": ["explicit scope condition: original proofs valid only under nonnegativity constraints", "stated assumptions (normality, nonsingularity) in propositions"]}, "claim_evidence": [{"claim": "The original equation omitted a term, so the original proofs hold only with no-short-sale constraints and not with upper bounds on weights.", "evidence_type": "equation_or_proof", "evidence_ref": "Corrected equation (10) vs. original equation (8)"}, {"claim": "With the corrected adjustment, the shrunk covariance matrix is symmetric and PSD, and the constrained minimum variance portfolio is one of its GMV portfolios.", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 1 and proof"}, {"claim": "Under normality, the adjusted matrix is the constrained MLE whose implied GMV portfolio satisfies the weight constraints, and conversely.", "evidence_type": "equation_or_proof", "evidence_ref": "Proposition 2 and proof; Morrison (2005); Zwiernik, Uhler, and Richards (2017)"}, {"claim": "The omitted term shifts all variances and covariances equally, so the original shrinkage interpretation is unchanged.", "evidence_type": "equation_or_proof", "evidence_ref": "Closing discussion after Proposition 2"}], "structural_summary": "A short corrigendum in a single unnumbered section. It identifies the published propositions and pinpoints the omitted term in the original equation. It states the scope of the error (valid only without upper bounds), then restates the constrained minimum-variance problem and its KKT conditions. It re-proves Proposition 1 with the corrected adjusted covariance matrix and re-proves Proposition 2 on constrained-MLE equivalence. It closes by noting that the original shrinkage interpretation survives the correction.", "writing_profile_keywords": ["corrigendum", "proof correction", "KKT conditions", "portfolio constraints", "covariance shrinkage", "constrained MLE", "equation-dense"], "confidence": 0.85, "notes": "Abstract is truncated and blends into the body; the body is mostly equations split into fragments that are treated as paragraphs. Paragraph indices for intro moves are approximate, given the fragmented equation text. There are no separate conclusion paragraphs; the conclusion fields are inferred from the closing remarks on the shrinkage interpretation."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 14.49, "usage": {"input": 6343, "output": 1646, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.059086, "request_id": "req_011CggqXZ8h5Z2aYB5aR2ZEW", "cached": false, "ts": "2026-11-03T23:09:39+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/fbfe3f396da58bb28542c34ba1de4ce0c098478bb4e0de15f5609947478f35a5.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"market_microstructure\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"Association between within-group (information) and cross-group (model) disagreement and abnormal trading volume; share of disagreement variance attributable to investment philosophy\",\"strategy\":\"Panel regressions of abnormal log volume on disagreement measures built from StockTwits sentiment, with date and firm fixed effects and lagged volume; ANOVA decomposition of sentiment by investment philosophy; earnings-announcement event windows; lead-lag between sophisticated and unsophisticated sentiment\",\"key_assumptions\":[\"Self-reported investment philosophies reflect actual investment models\",\"Sentiment classification, including maximum-entropy imputation of unclassified messages, measures beliefs accurately\",\"Disagreement measured before the market opens is not mechanically driven by same-day trading\"],\"threats_addressed\":[\"validity of self-reported philosophies, checked through textual analysis and event reactions\",\"confounding with investor attention and news coverage\",\"reverse timing, addressed using pre-market opinions\",\"external validity given overrepresentation of technical investors\"],\"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\":3}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":0,\"move\":\"literature_gap\"},{\"paragraph_index\":0,\"move\":\"preview_of_approach\"},{\"paragraph_index\":1,\"move\":\"identification_defense\"},{\"paragraph_index\":2,\"move\":\"preview_of_results\"},{\"paragraph_index\":2,\"move\":\"robustness_mention\"},{\"paragraph_index\":3,\"move\":\"preview_of_results\"},{\"paragraph_index\":3,\"move\":\"robustness_mention\"},{\"paragraph_index\":4,\"move\":\"contribution_statement\"},{\"paragraph_index\":4,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":5,\"move\":\"contribution_statement\"}],\"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\":\"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\":\"mixed\",\"types\":[\"alternative_specification\",\"subsample\",\"alternative_measure\",\"other\"]},\"tables\":{\"convention_notes\":\"Table notes state the regression equation, including fixed effects, lagged volume and controls, and define variables such as abnormal log volume relative to a 140-to-20 day window. Panels separate summary, validation and regression outputs. The excerpts do not describe standard-error treatment.\",\"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\",\"consistent with\",\"is associated with\",\"to the extent that\"]},\"claim_evidence\":[{\"claim\":\"Self-reported investment philosophies correspond to distinct language and to distinct reactions to fundamental versus technical events\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, III; Figure 1\"},{\"claim\":\"About half of overall disagreement comes from differences across investment philosophies\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"A one-SD rise in within-group disagreement relates to about four times the abnormal-volume increase of a one-SD rise in cross-group disagreement\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Sophisticated investors' sentiment leads unsophisticated investors' sentiment, consistent with slow information diffusion\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"Disagreement helps explain the spike in volume around earnings announcements\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX, X\"},{\"claim\":\"Disagreement is distinct from investor attention in explaining volume\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XI\"},{\"claim\":\"Results hold when technical investors are excluded from the measure\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table XII\"}],\"structural_summary\":\"The introduction opens with the theoretical centrality of disagreement for trading and notes that its sources are understudied. It introduces the social-media data, in which users self-report an investment approach, and defends the validity of those labels through textual and event-based checks. It then previews the volume decomposition, the sophistication and lead-lag evidence, and robustness, before positioning the paper against empirical and theoretical literatures. The body proceeds through data description, measure construction, a three-part empirical analysis (volume, sophistication, earnings announcements) and a short robustness section on attention and external validity. A one-paragraph conclusion restates the measurement contribution and the main split. An appendix covers an alternative measure and the classification method.\",\"writing_profile_keywords\":[\"social media data\",\"measurement validation\",\"disagreement decomposition\",\"trading volume\",\"textual analysis\",\"information diffusion\",\"fixed-effects panel regressions\"],\"confidence\":0.8,\"notes\":\"Statistical significance conventions and standard-error clustering are not visible in the excerpts. The introduction contains no visible roadmap paragraph.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "market_microstructure", "primary_method": "reduced_form_regression", "identification": {"estimand": "Association between within-group (information) and cross-group (model) disagreement and abnormal trading volume; share of disagreement variance attributable to investment philosophy", "strategy": "Panel regressions of abnormal log volume on disagreement measures built from StockTwits sentiment, with date and firm fixed effects and lagged volume; ANOVA decomposition of sentiment by investment philosophy; earnings-announcement event windows; lead-lag between sophisticated and unsophisticated sentiment", "key_assumptions": ["Self-reported investment philosophies reflect actual investment models", "Sentiment classification, including maximum-entropy imputation of unclassified messages, measures beliefs accurately", "Disagreement measured before the market opens is not mechanically driven by same-day trading"], "threats_addressed": ["validity of self-reported philosophies, checked through textual analysis and event reactions", "confounding with investor attention and news coverage", "reverse timing, addressed using pre-market opinions", "external validity given overrepresentation of technical investors"], "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": 3}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 0, "move": "literature_gap"}, {"paragraph_index": 0, "move": "preview_of_approach"}, {"paragraph_index": 1, "move": "identification_defense"}, {"paragraph_index": 2, "move": "preview_of_results"}, {"paragraph_index": 2, "move": "robustness_mention"}, {"paragraph_index": 3, "move": "preview_of_results"}, {"paragraph_index": 3, "move": "robustness_mention"}, {"paragraph_index": 4, "move": "contribution_statement"}, {"paragraph_index": 4, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "related_literature_positioning"}, {"paragraph_index": 5, "move": "contribution_statement"}], "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": "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": "mixed", "types": ["alternative_specification", "subsample", "alternative_measure", "other"]}, "tables": {"convention_notes": "Table notes state the regression equation, including fixed effects, lagged volume and controls, and define variables such as abnormal log volume relative to a 140-to-20 day window. Panels separate summary, validation and regression outputs. The excerpts do not describe standard-error treatment.", "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", "consistent with", "is associated with", "to the extent that"]}, "claim_evidence": [{"claim": "Self-reported investment philosophies correspond to distinct language and to distinct reactions to fundamental versus technical events", "evidence_type": "table", "evidence_ref": "Tables II, III; Figure 1"}, {"claim": "About half of overall disagreement comes from differences across investment philosophies", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "A one-SD rise in within-group disagreement relates to about four times the abnormal-volume increase of a one-SD rise in cross-group disagreement", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Sophisticated investors' sentiment leads unsophisticated investors' sentiment, consistent with slow information diffusion", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "Disagreement helps explain the spike in volume around earnings announcements", "evidence_type": "table", "evidence_ref": "Tables IX, X"}, {"claim": "Disagreement is distinct from investor attention in explaining volume", "evidence_type": "table", "evidence_ref": "Table XI"}, {"claim": "Results hold when technical investors are excluded from the measure", "evidence_type": "table", "evidence_ref": "Table XII"}], "structural_summary": "The introduction opens with the theoretical centrality of disagreement for trading and notes that its sources are understudied. It introduces the social-media data, in which users self-report an investment approach, and defends the validity of those labels through textual and event-based checks. It then previews the volume decomposition, the sophistication and lead-lag evidence, and robustness, before positioning the paper against empirical and theoretical literatures. The body proceeds through data description, measure construction, a three-part empirical analysis (volume, sophistication, earnings announcements) and a short robustness section on attention and external validity. A one-paragraph conclusion restates the measurement contribution and the main split. An appendix covers an alternative measure and the classification method.", "writing_profile_keywords": ["social media data", "measurement validation", "disagreement decomposition", "trading volume", "textual analysis", "information diffusion", "fixed-effects panel regressions"], "confidence": 0.8, "notes": "Statistical significance conventions and standard-error clustering are not visible in the excerpts. The introduction contains no visible roadmap paragraph."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 16.93, "usage": {"input": 8006, "output": 2149, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075798, "request_id": "req_011CggqcknaLYKyV5qaXGBa9", "cached": false, "ts": "2026-11-03T23:10:52+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/fd6f7cba94072fbdb0c270dbb4ed2922e6a72ec3bbab780eb52b2cf779c5f745.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"asset_pricing; household_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How differently investors' selling decisions, reinvestment choices, and analysts' price forecasts respond to the dividend component of performance versus the price-change component; time variation in aggregate dividend demand\",\"strategy\":\"Split position performance into a price-change part and a dividend part, then compare selling propensities (disposition and rank effects) under each measure. Benchmark observed holding changes against a mechanical full-reinvestment counterfactual. Regress analyst target-price errors and dividend interim returns on dividend yield, interest rates and past market returns.\",\"key_assumptions\":[\"Trading responses reveal how investors evaluate performance\",\"Absence of share increases matching the dividend amount indicates non-reinvestment\",\"The marketwide analysis relies on stronger assumptions drawn from prior literature, which the authors acknowledge\"],\"threats_addressed\":[\"Alternative rational explanations such as taxes and trading costs\",\"Alternative behavioral explanations such as self-control\",\"Investor heterogeneity, handled with account fixed effects and controls\",\"Split of investors into those showing a disposition effect and those not\"],\"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\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":6,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":7,\"move\":\"preview_of_approach\"},{\"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\":\"preview_of_results\"},{\"paragraph_index\":11,\"move\":\"contribution_statement\"},{\"paragraph_index\":12,\"move\":\"related_literature_positioning\"}],\"contribution_paragraph_index\":11,\"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\":\"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_measure\"]},\"tables\":{\"convention_notes\":\"Each table pits a price-change version against a returns-including-dividends version, with panels by investor type (individuals, mutual funds, institutions). The dependent variable is typically a Sell dummy. Notes define variables and sample periods.\",\"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\":[\"'trade as if' framing\",\"'consistent with'\",\"'appear to'\",\"'it seems more likely'\",\"explicit admission that the marketwide tests are less cleanly identified\"]},\"claim_evidence\":[{\"claim\":\"Investors' disposition effect is driven by price changes rather than by total returns including dividends.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Rank-based selling responds to price-change ranks but little to return ranks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Individuals judge reinvested positions' gain/loss status without including dividends.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Dividend payers are sold less often, and their sales are less sensitive to price changes.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V; Figure 1\"},{\"claim\":\"Funds and institutions rarely reinvest dividends, which causes portfolio weights of dividend payers to drift downward.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI; Figures 2-4\"},{\"claim\":\"Analysts' target prices are too optimistic for high-dividend-yield stocks.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VII\"},{\"claim\":\"Dividend demand rises when interest rates and past market returns are low.\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII; Figure 5\"}],\"structural_summary\":\"The paper opens with a press epigraph and a statement of Miller-Modigliani dividend irrelevance, then argues that the price-dividend tradeoff may not be salient to investors. It previews tests in which performance is split into price-change and dividend components. Results proceed through several sections: trading behavior (disposition, rank and rolled-disposition effects, plus price sensitivity), dividend reinvestment and portfolio drift, analyst forecast errors, and marketwide time-varying dividend demand. The conclusion restates the mental-accounting interpretation, argues rival theories cannot explain the findings together, and ends with an open question about investor education.\",\"writing_profile_keywords\":[\"mental accounting\",\"free dividends fallacy\",\"epigraph opening\",\"unifying explanation\",\"multiple investor types\",\"behavioral trading\"],\"confidence\":0.8,\"notes\":\"No roadmap paragraph is visible in the excerpts. Intro paragraphs 0-5 are fragments of the epigraph quote. Significance conventions and standard-error treatment are not visible in the excerpted table notes. The paper explicitly labels its marketwide analysis as less cleanly identified.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "asset_pricing; household_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "How differently investors' selling decisions, reinvestment choices, and analysts' price forecasts respond to the dividend component of performance versus the price-change component; time variation in aggregate dividend demand", "strategy": "Split position performance into a price-change part and a dividend part, then compare selling propensities (disposition and rank effects) under each measure. Benchmark observed holding changes against a mechanical full-reinvestment counterfactual. Regress analyst target-price errors and dividend interim returns on dividend yield, interest rates and past market returns.", "key_assumptions": ["Trading responses reveal how investors evaluate performance", "Absence of share increases matching the dividend amount indicates non-reinvestment", "The marketwide analysis relies on stronger assumptions drawn from prior literature, which the authors acknowledge"], "threats_addressed": ["Alternative rational explanations such as taxes and trading costs", "Alternative behavioral explanations such as self-control", "Investor heterogeneity, handled with account fixed effects and controls", "Split of investors into those showing a disposition effect and those not"], "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": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 6, "move": "tension_or_puzzle"}, {"paragraph_index": 7, "move": "preview_of_approach"}, {"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": "preview_of_results"}, {"paragraph_index": 11, "move": "contribution_statement"}, {"paragraph_index": 12, "move": "related_literature_positioning"}], "contribution_paragraph_index": 11, "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": "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_measure"]}, "tables": {"convention_notes": "Each table pits a price-change version against a returns-including-dividends version, with panels by investor type (individuals, mutual funds, institutions). The dependent variable is typically a Sell dummy. Notes define variables and sample periods.", "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": ["'trade as if' framing", "'consistent with'", "'appear to'", "'it seems more likely'", "explicit admission that the marketwide tests are less cleanly identified"]}, "claim_evidence": [{"claim": "Investors' disposition effect is driven by price changes rather than by total returns including dividends.", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Rank-based selling responds to price-change ranks but little to return ranks.", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Individuals judge reinvested positions' gain/loss status without including dividends.", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Dividend payers are sold less often, and their sales are less sensitive to price changes.", "evidence_type": "table", "evidence_ref": "Table V; Figure 1"}, {"claim": "Funds and institutions rarely reinvest dividends, which causes portfolio weights of dividend payers to drift downward.", "evidence_type": "table", "evidence_ref": "Table VI; Figures 2-4"}, {"claim": "Analysts' target prices are too optimistic for high-dividend-yield stocks.", "evidence_type": "table", "evidence_ref": "Table VII"}, {"claim": "Dividend demand rises when interest rates and past market returns are low.", "evidence_type": "table", "evidence_ref": "Table VIII; Figure 5"}], "structural_summary": "The paper opens with a press epigraph and a statement of Miller-Modigliani dividend irrelevance, then argues that the price-dividend tradeoff may not be salient to investors. It previews tests in which performance is split into price-change and dividend components. Results proceed through several sections: trading behavior (disposition, rank and rolled-disposition effects, plus price sensitivity), dividend reinvestment and portfolio drift, analyst forecast errors, and marketwide time-varying dividend demand. The conclusion restates the mental-accounting interpretation, argues rival theories cannot explain the findings together, and ends with an open question about investor education.", "writing_profile_keywords": ["mental accounting", "free dividends fallacy", "epigraph opening", "unifying explanation", "multiple investor types", "behavioral trading"], "confidence": 0.8, "notes": "No roadmap paragraph is visible in the excerpts. Intro paragraphs 0-5 are fragments of the epigraph quote. Significance conventions and standard-error treatment are not visible in the excerpted table notes. The paper explicitly labels its marketwide analysis as less cleanly identified."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 17.2, "usage": {"input": 7817, "output": 2138, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074822, "request_id": "req_011CggqTVVZVWmDnsKQRaApB", "cached": false, "ts": "2026-11-03T23:08:47+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/fe05d17bf6b7f7ae0ddee8c98b6ee9b0a9e0d3ce530855e42d2e9de5dab1a113.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"mixed_theory_empirical\",\"subfield\":\"macro_finance\",\"subfield_secondary\":\"asset_pricing (sovereign credit risk / CDS pricing)\",\"primary_method\":\"calibration_simulation\",\"identification\":{\"estimand\":\"Model-implied risk-adjusted fiscal default probabilities and CDS premiums, decomposed into expected loss and default risk premium, set against observed U.S. CDS premiums\",\"strategy\":\"Equilibrium macrofinance model with recursive preferences and fiscal/monetary policy rules, calibrated to post-2007 U.S. data and simulated; counterfactuals vary policy parameters\",\"key_assumptions\":[\"Exogenous consumption is unaffected by default\",\"Recursive (Epstein–Zin) preferences of a representative investor\",\"Fiscal default occurs at the Laffer-curve limit of tax revenue\",\"Fiscal and monetary rules do not change upon default\"],\"threats_addressed\":[\"Failure of standard CDS replication when Treasuries carry credit risk\",\"Model misspecifications acknowledged: downward-sloping real yield curve, constant price of inflation risk, countercyclical inflation\",\"Dynamic cross-correlations not quantitatively resolved\"],\"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\":false,\"sentence_index\":-1},\"contribution\":{\"present\":false,\"sentence_index\":-1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"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\"},{\"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\":\"policy_or_broader_implication\"},{\"paragraph_index\":6,\"move\":\"other\"},{\"paragraph_index\":6,\"move\":\"contribution_statement\"},{\"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\":11,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":6,\"contribution_types\":[\"new_theory_or_model\",\"synthesis\"],\"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\":\"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\":[\"sensitivity_to_parameters\",\"alternative_sample_period\"]},\"tables\":{\"convention_notes\":\"Tables compare data and model moments side by side. Notes state the data source and sample period, say moments are annualized, and describe the simulation design (1000 quarterly simulations over 10 or 70 years, averaged). Counterfactual tables show model moments across a grid of policy parameters.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"none\"},\"figures\":{\"uncertainty_shown\":\"mixed\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"'first step' framing\",\"'in the context of our model' qualifiers\",\"explicit listing of model simplifications and misspecifications\",\"'consistent with' rather than 'explains'\",\"deferring alternative explanations to future research\",\"'merely suggests' / 'likely' phrasing\"]},\"claim_evidence\":[{\"claim\":\"Calibrated model generates CDS premiums peaking near 100 bps, comparable to the U.S. crisis peak\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"High debt and default probability coincide with high marginal utility, so default risk premiums form a large share of CDS premiums\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III; Figure 3\"},{\"claim\":\"Less aggressive inflation response or less debt issuance lowers default probability yet raises CDS premiums through higher volatility\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VIII, IX\"},{\"claim\":\"Model broadly matches macro moments and the term structure of yields in the post-2007 sample\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables II, IV\"},{\"claim\":\"Dynamic cross-correlations with government expenditure overshoot the data\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figures 4, 5\"},{\"claim\":\"U.S. CDS premiums rose sharply in 2008 and remain elevated\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 1\"}],\"structural_summary\":\"The introduction opens with a puzzle: persistently high U.S. sovereign CDS premiums sit uneasily with the view that the U.S. can always tax or inflate away its debt. It frames the paper as a first-step quantitative benchmark, then sketches the marginal-utility mechanism and previews qualitative, quantitative and counterfactual results. Limitations are candidly listed before the authors position the paper against intensity-based, corporate-default and strategic-default literatures, ending with a roadmap. The body proceeds from an institutional primer on CDS to an equilibrium model and then a calibrated quantitative analysis. That analysis covers moment matching, the mechanism, term structures, dynamics and policy counterfactuals. The conclusion restates the findings and draws hedged policy implications.\",\"writing_profile_keywords\":[\"quantitative macrofinance\",\"calibrated equilibrium model\",\"recursive preferences\",\"sovereign CDS\",\"fiscal limits\",\"first-step framing\",\"candid limitations\",\"policy counterfactuals\"],\"confidence\":0.82,\"notes\":\"No regressions are reported, so standard errors appear only in the cross-correlation figures (Figures 4, 5), not in tables. Statistical significance language is therefore marked unclear and table standard errors as none. The long-run sample analysis (Tables VI, VII) serves as a counterfactual, recorded here as an alternative sample period. The conclusion calls the model 'partial equilibrium', despite the abstract's 'equilibrium' framing.\"}", "json": {"genre": "mixed_theory_empirical", "subfield": "macro_finance", "subfield_secondary": "asset_pricing (sovereign credit risk / CDS pricing)", "primary_method": "calibration_simulation", "identification": {"estimand": "Model-implied risk-adjusted fiscal default probabilities and CDS premiums, decomposed into expected loss and default risk premium, set against observed U.S. CDS premiums", "strategy": "Equilibrium macrofinance model with recursive preferences and fiscal/monetary policy rules, calibrated to post-2007 U.S. data and simulated; counterfactuals vary policy parameters", "key_assumptions": ["Exogenous consumption is unaffected by default", "Recursive (Epstein–Zin) preferences of a representative investor", "Fiscal default occurs at the Laffer-curve limit of tax revenue", "Fiscal and monetary rules do not change upon default"], "threats_addressed": ["Failure of standard CDS replication when Treasuries carry credit risk", "Model misspecifications acknowledged: downward-sloping real yield curve, constant price of inflation risk, countercyclical inflation", "Dynamic cross-correlations not quantitatively resolved"], "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": false, "sentence_index": -1}, "contribution": {"present": false, "sentence_index": -1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "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"}, {"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": "policy_or_broader_implication"}, {"paragraph_index": 6, "move": "other"}, {"paragraph_index": 6, "move": "contribution_statement"}, {"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": 11, "move": "roadmap"}], "contribution_paragraph_index": 6, "contribution_types": ["new_theory_or_model", "synthesis"], "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": "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": ["sensitivity_to_parameters", "alternative_sample_period"]}, "tables": {"convention_notes": "Tables compare data and model moments side by side. Notes state the data source and sample period, say moments are annualized, and describe the simulation design (1000 quarterly simulations over 10 or 70 years, averaged). Counterfactual tables show model moments across a grid of policy parameters.", "units_stated_in_notes": true, "standard_errors_description": "none"}, "figures": {"uncertainty_shown": "mixed"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["'first step' framing", "'in the context of our model' qualifiers", "explicit listing of model simplifications and misspecifications", "'consistent with' rather than 'explains'", "deferring alternative explanations to future research", "'merely suggests' / 'likely' phrasing"]}, "claim_evidence": [{"claim": "Calibrated model generates CDS premiums peaking near 100 bps, comparable to the U.S. crisis peak", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "High debt and default probability coincide with high marginal utility, so default risk premiums form a large share of CDS premiums", "evidence_type": "table", "evidence_ref": "Table III; Figure 3"}, {"claim": "Less aggressive inflation response or less debt issuance lowers default probability yet raises CDS premiums through higher volatility", "evidence_type": "table", "evidence_ref": "Tables VIII, IX"}, {"claim": "Model broadly matches macro moments and the term structure of yields in the post-2007 sample", "evidence_type": "table", "evidence_ref": "Tables II, IV"}, {"claim": "Dynamic cross-correlations with government expenditure overshoot the data", "evidence_type": "figure", "evidence_ref": "Figures 4, 5"}, {"claim": "U.S. CDS premiums rose sharply in 2008 and remain elevated", "evidence_type": "figure", "evidence_ref": "Figure 1"}], "structural_summary": "The introduction opens with a puzzle: persistently high U.S. sovereign CDS premiums sit uneasily with the view that the U.S. can always tax or inflate away its debt. It frames the paper as a first-step quantitative benchmark, then sketches the marginal-utility mechanism and previews qualitative, quantitative and counterfactual results. Limitations are candidly listed before the authors position the paper against intensity-based, corporate-default and strategic-default literatures, ending with a roadmap. The body proceeds from an institutional primer on CDS to an equilibrium model and then a calibrated quantitative analysis. That analysis covers moment matching, the mechanism, term structures, dynamics and policy counterfactuals. The conclusion restates the findings and draws hedged policy implications.", "writing_profile_keywords": ["quantitative macrofinance", "calibrated equilibrium model", "recursive preferences", "sovereign CDS", "fiscal limits", "first-step framing", "candid limitations", "policy counterfactuals"], "confidence": 0.82, "notes": "No regressions are reported, so standard errors appear only in the cross-correlation figures (Figures 4, 5), not in tables. Statistical significance language is therefore marked unclear and table standard errors as none. The long-run sample analysis (Tables VI, VII) serves as a counterfactual, recorded here as an alternative sample period. The conclusion calls the model 'partial equilibrium', despite the abstract's 'equilibrium' framing."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 20.06, "usage": {"input": 8134, "output": 2424, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08181, "request_id": "req_011CggqzCWDoqp7rp7xt9zkB", "cached": false, "ts": "2026-11-03T23:15:46+00:00", "effort": "medium"}
\ No newline at end of file
added data/annotations/_llm_cache/ff2aa4007251317f6ea95e62fe62444b86246bafe32bd6745867567c0eedbf83.json +1 −0
@@ -0,0 +1 @@
1 +{"text": "{\"genre\":\"replication_or_correction\",\"subfield\":\"asset_pricing\",\"subfield_secondary\":\"return predictability / learning models (corrigendum)\",\"primary_method\":\"none\",\"identification\":{\"estimand\":\"not applicable (correction of coding errors in prior paper's predictability estimates)\",\"strategy\":\"not applicable\",\"key_assumptions\":[],\"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\":false,\"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\":\"other\"},{\"paragraph_index\":5,\"move\":\"other\"},{\"paragraph_index\":12,\"move\":\"preview_of_results\"},{\"paragraph_index\":12,\"move\":\"other\"}],\"contribution_paragraph_index\":-1,\"contribution_types\":[\"replication_or_correction\"],\"roadmap_present\":false,\"literature_positioning_location\":\"none\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"}],\"results_presentation\":{\"economic_significance_discussed\":false,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"No tables are visible in the excerpts. The text says the corrected tables are reported in the erratum, with nontrivial changes underlined.\",\"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\":false},\"hedging\":{\"level\":\"low\",\"typical_devices\":[\"assertion that the corrections leave the original conclusions unchanged\",\"flagging which changes are nontrivial\"]},\"claim_evidence\":[{\"claim\":\"Corrected coding leaves the original paper's statements, findings and conclusions unchanged\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Introduction para 0 and 12\"},{\"claim\":\"Out-of-sample R2 of the dividend model falls after correction but stays statistically above competing models\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables III and V (updated)\"},{\"claim\":\"Return predictability R2 of the full learning model changes modestly overall, rising in expansions and falling in recessions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables IX and XI (updated)\"},{\"claim\":\"Data and code used to generate the results are made available\",\"evidence_type\":\"none_stated\",\"evidence_ref\":\"Internet Appendix\"}],\"structural_summary\":\"Short corrigendum with a single unnumbered section. It states that coding inconsistencies were found and claims the paper's conclusions are unaffected. It then lists four specific errors as bullets: mixed overlapping and nonoverlapping estimation, a mis-lagged regressor, an unflagged de-meaning in one figure, and a mis-timed inflation series. Next it reports the nontrivial revised R2 statistics, reaffirms that the main conclusions hold, and points to the Internet Appendix for data and code.\",\"writing_profile_keywords\":[\"erratum\",\"bulleted error list\",\"reassurance of unchanged conclusions\",\"revised statistics\",\"code transparency\"],\"confidence\":0.85,\"notes\":\"No abstract, conclusion, tables or figures appear in the excerpts, and paragraph segmentation reflects line breaks rather than true paragraphs. The original paper's methods (Kalman filter, learning model) are mentioned only in passing. Method fields are therefore set to none or not applicable.\"}", "json": {"genre": "replication_or_correction", "subfield": "asset_pricing", "subfield_secondary": "return predictability / learning models (corrigendum)", "primary_method": "none", "identification": {"estimand": "not applicable (correction of coding errors in prior paper's predictability estimates)", "strategy": "not applicable", "key_assumptions": [], "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": false, "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": "other"}, {"paragraph_index": 5, "move": "other"}, {"paragraph_index": 12, "move": "preview_of_results"}, {"paragraph_index": 12, "move": "other"}], "contribution_paragraph_index": -1, "contribution_types": ["replication_or_correction"], "roadmap_present": false, "literature_positioning_location": "none"}, "section_roles": [{"idx": 0, "role": "introduction"}], "results_presentation": {"economic_significance_discussed": false, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "No tables are visible in the excerpts. The text says the corrected tables are reported in the erratum, with nontrivial changes underlined.", "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": false}, "hedging": {"level": "low", "typical_devices": ["assertion that the corrections leave the original conclusions unchanged", "flagging which changes are nontrivial"]}, "claim_evidence": [{"claim": "Corrected coding leaves the original paper's statements, findings and conclusions unchanged", "evidence_type": "none_stated", "evidence_ref": "Introduction para 0 and 12"}, {"claim": "Out-of-sample R2 of the dividend model falls after correction but stays statistically above competing models", "evidence_type": "table", "evidence_ref": "Tables III and V (updated)"}, {"claim": "Return predictability R2 of the full learning model changes modestly overall, rising in expansions and falling in recessions", "evidence_type": "table", "evidence_ref": "Tables IX and XI (updated)"}, {"claim": "Data and code used to generate the results are made available", "evidence_type": "none_stated", "evidence_ref": "Internet Appendix"}], "structural_summary": "Short corrigendum with a single unnumbered section. It states that coding inconsistencies were found and claims the paper's conclusions are unaffected. It then lists four specific errors as bullets: mixed overlapping and nonoverlapping estimation, a mis-lagged regressor, an unflagged de-meaning in one figure, and a mis-timed inflation series. Next it reports the nontrivial revised R2 statistics, reaffirms that the main conclusions hold, and points to the Internet Appendix for data and code.", "writing_profile_keywords": ["erratum", "bulleted error list", "reassurance of unchanged conclusions", "revised statistics", "code transparency"], "confidence": 0.85, "notes": "No abstract, conclusion, tables or figures appear in the excerpts, and paragraph segmentation reflects line breaks rather than true paragraphs. The original paper's methods (Kalman filter, learning model) are mentioned only in passing. Method fields are therefore set to none or not applicable."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 11.47, "usage": {"input": 1240, "output": 1346, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.032674, "request_id": "req_011CggqTADwX9bNcbmhS1Tt5", "cached": false, "ts": "2026-11-03T23:08:37+00:00", "effort": "medium"}
\ No newline at end of file
modified data/annotations/annotations_table.csv +277 −40
@@ -1,40 +1,277 @@
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,
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
2 +jof_12407,True,empirical,corporate_finance,instrumental_variables,4,True,False,False,True,True,False,True,-1,-1,1,0,-1,3,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,-1,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,7,7,0.8,4,6,855,False,False,False,False,1.0,0.06673,claude-opus-5-5
3 +jof_12432,True,empirical,corporate_finance,reduced_form_regression,3,True,True,True,False,True,False,True,1,0,-1,2,-1,1,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,12,11,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,7,7,0.8,5,14,1227,False,False,False,False,1.0,0.080506,claude-opus-5-5
4 +jof_12434,True,empirical,corporate_finance,difference_in_differences,4,True,False,False,True,True,False,True,-1,-1,2,1,-1,0,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,13,7,new_evidence|new_identification,False,False,inside_introduction,True,unclear,True,mixed,placebo|alternative_specification|subsample|falsification,unclear,True,unclear,True,True,False,moderate,7,7,0.85,3,9,1304,False,True,False,False,0.75,0.079782,claude-opus-5-5
5 +jof_12438,True,mixed_theory_empirical,asset_pricing,structural_estimation,2,False,True,False,True,True,False,True,0,-1,1,2,-1,0,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,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,5,5,0.8,4,8,1290,False,True,False,False,1.0,0.082186,claude-opus-5-5
6 +jof_12439,True,empirical,corporate_finance,difference_in_differences,4,True,True,False,True,True,True,True,0,-1,2,2,2,0,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,new_evidence|new_identification,True,True,inside_introduction,True,unclear,True,mixed,placebo|alternative_specification|falsification,unclear,True,unclear,True,True,False,moderate,6,6,0.85,6,6,1025,True,True,False,False,0.5,0.066598,claude-opus-5-5
7 +jof_12464,True,mixed_theory_empirical,entrepreneurial_finance_private_equity,reduced_form_regression,4,True,True,False,True,True,False,True,0,-1,3,1,-1,5,motivation|related_literature_positioning|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|robustness_mention|preview_of_results|identification_defense|preview_of_results|contribution_statement|related_literature_positioning|roadmap,13,6,new_evidence|new_theory_or_model,True,False,inside_introduction,True,mixed,True,within_results,alternative_measure|alternative_specification,mixed,True,no,True,False,True,low,8,8,0.8,8,8,1160,False,True,False,False,1.0,0.107268,claude-opus-5-5
8 +jof_12465,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,3,True,True,False,True,True,False,False,0,-1,4,1,-1,-1,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,11,8,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,6,6,0.8,5,9,1485,False,True,False,False,1.0,0.073646,claude-opus-5-5
9 +jof_12466,True,empirical,asset_pricing,reduced_form_regression,3,True,False,False,True,True,False,True,-1,-1,0,0,-1,2,motivation|related_literature_positioning|literature_gap|mechanism_or_theory_sketch|identification_defense|preview_of_results|preview_of_results|preview_of_results|contribution_statement|roadmap,10,4,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,8,7,0.8,3,6,872,False,True,False,False,1.0,0.075834,claude-opus-5-5
10 +jof_12467,True,empirical,financial_intermediation_banking,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,1,2,-1,0,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,11,4,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,6,6,0.8,5,6,1076,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,True,False,False,True,True,False,True,-1,-1,1,2,-1,0,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,10,3,new_evidence|new_identification,False,False,inside_introduction,True,unclear,True,within_results,subsample|alternative_specification,unclear,False,yes,True,False,True,moderate,6,5,0.85,4,4,752,False,True,False,False,1.0,0.062382,claude-opus-5-5
12 +jof_12469,True,empirical,household_finance,natural_experiment_event_study,4,True,True,False,True,True,False,True,0,-1,1,2,-1,1,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,18,1,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,7,7,0.85,6,15,1483,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,True,True,False,True,True,False,True,0,-1,1,2,-1,4,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,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,5,5,0.85,5,13,1529,False,True,False,False,0.6,0.076878,claude-opus-5-5
14 +jof_12471,True,mixed_theory_empirical,asset_pricing,structural_estimation,2,False,True,False,True,True,False,True,0,-1,1,3,-1,0,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,11,1,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,8,8,0.8,6,6,936,False,True,False,False,0.9,0.086406,claude-opus-5-5
15 +jof_12472,True,empirical,market_microstructure,reduced_form_regression,2,False,True,False,True,True,True,True,0,-1,0,1,2,7,motivation|preview_of_approach|related_literature_positioning|preview_of_approach|other|preview_of_results|preview_of_results|identification_defense|contribution_statement|roadmap,10,4,new_evidence|new_data_or_measure,True,True,both,True,unclear,True,mixed,placebo|subsample|alternative_measure,unclear,True,unclear,True,False,True,moderate,7,7,0.82,8,5,779,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,False,True,False,True,True,False,True,0,-1,0,1,-1,3,motivation|literature_gap|preview_of_results|robustness_mention|identification_defense|contribution_statement|mechanism_or_theory_sketch|related_literature_positioning|roadmap,9,3,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,6,6,0.8,4,7,1009,False,True,False,False,0.857,0.071746,claude-opus-5-5
17 +jof_12482,True,empirical,household_finance,reduced_form_regression,3,True,True,False,False,True,False,True,0,-1,-1,2,-1,4,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,new_evidence,False,False,both,True,unclear,False,mixed,subsample|alternative_measure|alternative_specification|falsification|other,unclear,True,unclear,False,True,True,moderate,8,8,0.8,5,6,938,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,True,True,False,False,True,False,False,0,-1,-1,1,-1,-1,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,10,3,new_evidence,False,False,both,True,unclear,False,mixed,placebo|alternative_specification|alternative_measure,unclear,True,no_figures,True,False,False,moderate,7,7,0.8,6,4,860,False,True,False,False,1.0,0.067506,claude-opus-5-5
19 +jof_12484,True,mixed_theory_empirical,household_finance,structural_estimation,2,False,False,False,True,True,False,True,-1,-1,1,0,-1,0,contribution_statement|preview_of_approach|related_literature_positioning|tension_or_puzzle|preview_of_approach|identification_defense|preview_of_results|literature_gap,8,0,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,5,4,0.8,2,4,752,False,True,False,False,1.0,0.071706,claude-opus-5-5
20 +jof_12485,True,empirical,market_microstructure,natural_experiment_event_study,3,True,True,False,False,True,False,True,0,-1,-1,1,-1,3,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,13,6,new_evidence,False,False,inside_introduction,True,unclear,True,mixed,alternative_specification|alternative_measure|subsample|other,unclear,False,no,False,True,False,moderate,7,7,0.8,6,7,1343,False,True,False,False,1.0,0.07733,claude-opus-5-5
21 +jof_12486,True,mixed_theory_empirical,asset_pricing,structural_estimation,2,False,True,True,True,True,False,True,0,1,2,2,-1,3,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,14,2,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,7,7,0.8,4,12,2034,False,False,False,False,1.0,0.10605,claude-opus-5-5
22 +jof_12487,True,empirical,corporate_finance,reduced_form_regression,2,False,False,False,True,True,False,False,-1,-1,0,1,-1,-1,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,-1,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,7,7,0.8,3,6,1027,False,False,False,True,0.9,0.084146,claude-opus-5-5
23 +jof_12488,True,mixed_theory_empirical,asset_pricing,structural_estimation,1,False,False,False,True,True,False,True,-1,-1,0,2,-1,1,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,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,7,6,0.85,3,9,1406,False,True,False,False,1.0,0.076602,claude-opus-5-5
24 +jof_12489,True,mixed_theory_empirical,household_finance,structural_estimation,3,True,True,True,True,True,True,False,0,1,2,4,3,-1,motivation|literature_gap|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|policy_or_broader_implication|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,12,3,new_theory_or_model|new_evidence,True,True,inside_introduction,True,unclear,True,internet_appendix,sensitivity_to_parameters,none,True,unclear,True,True,True,moderate,6,6,0.85,6,7,1105,True,True,False,False,1.0,0.073518,claude-opus-5-5
25 +jof_12490,True,mixed_theory_empirical,corporate_finance,instrumental_variables,3,True,True,False,True,True,True,True,2,-1,2,3,1,4,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,14,6,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,8,8,0.8,6,8,1027,True,False,False,False,1.0,0.08025,claude-opus-5-5
26 +jof_12492,True,empirical,corporate_finance,instrumental_variables,3,True,True,False,True,True,True,False,0,-1,1,2,2,-1,motivation|motivation|literature_gap|tension_or_puzzle|contribution_statement|preview_of_approach|preview_of_approach|identification_defense|related_literature_positioning|related_literature_positioning|other,11,6,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,9,9,0.8,6,12,1457,False,True,False,False,0.667,0.088078,claude-opus-5-5
27 +jof_12493,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,1,False,False,False,True,True,False,True,-1,-1,1,0,-1,0,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,14,1,new_theory_or_model,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,False,False,moderate,5,5,0.85,5,9,1642,False,True,False,False,1.0,0.06903,claude-opus-5-5
28 +jof_12494,True,mixed_theory_empirical,financial_intermediation_banking,instrumental_variables,3,True,True,False,True,True,False,True,0,-1,1,2,-1,0,motivation|literature_gap|contribution_statement|preview_of_approach|identification_defense|robustness_mention|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|identification_defense,11,0,new_evidence|new_theory_or_model,False,False,inside_introduction,True,unclear,True,within_results,alternative_specification|subsample|other,unclear,True,no,False,True,True,moderate,7,7,0.8,6,4,761,False,False,False,False,1.0,0.069382,claude-opus-5-5
29 +jof_12495,True,empirical,entrepreneurial_finance_private_equity,instrumental_variables,4,True,True,True,True,True,False,False,1,0,1,2,-1,-1,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,11,1,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,8,8,0.8,5,7,1049,False,True,False,False,1.0,0.072966,claude-opus-5-5
30 +jof_12496,True,empirical,market_microstructure,reduced_form_regression,3,True,False,False,False,True,False,True,-1,-1,-1,2,-1,3,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,10,2,new_evidence,True,True,inside_introduction,True,unclear,False,none_seen,,unclear,True,unclear,True,True,False,moderate,7,7,0.8,5,6,800,False,True,False,False,1.0,0.073094,claude-opus-5-5
31 +jof_12497,True,mixed_theory_empirical,international_finance,difference_in_differences,4,True,True,False,True,True,True,False,0,-1,1,1,3,-1,motivation|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|identification_defense|robustness_mention|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning,10,3,new_evidence,False,False,inside_introduction,True,unclear,True,mixed,alternative_measure|alternative_specification|subsample|other,unclear,False,unclear,False,False,False,moderate,8,8,0.8,4,5,831,True,True,False,False,1.0,0.07177,claude-opus-5-5
32 +jof_12498,True,empirical,market_microstructure,reduced_form_regression,2,False,True,False,True,True,False,False,0,-1,2,2,-1,-1,motivation|motivation|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|related_literature_positioning|preview_of_results|contribution_statement|roadmap,10,45,new_evidence,True,True,inside_introduction,True,unclear,True,none_seen,,unclear,True,unclear,True,True,False,moderate,6,6,0.72,6,46,1247,False,True,False,False,0.5,0.062926,claude-opus-5-5
33 +jof_12499,True,mixed_theory_empirical,international_finance,natural_experiment_event_study,2,False,False,False,True,True,False,True,-1,-1,2,0,-1,2,motivation|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|robustness_mention|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,16,6,new_evidence|new_theory_or_model,True,True,inside_introduction,True,stars_and_tstats,True,mixed,alternative_measure|falsification|other|alternative_specification,unclear,True,mixed,True,False,False,moderate,7,7,0.85,4,9,1341,False,True,False,False,1.0,0.08645,claude-opus-5-5
34 +jof_12500,True,mixed_theory_empirical,household_finance,instrumental_variables,4,True,True,False,True,True,True,True,0,-1,1,2,3,0,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,13,1,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,6,6,0.8,4,7,1045,True,True,False,False,0.923,0.073714,claude-opus-5-5
35 +jof_12501,True,mixed_theory_empirical,asset_pricing,structural_estimation,2,False,True,False,True,True,False,True,0,-1,0,1,-1,4,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,17,9,new_evidence|new_theory_or_model,True,True,inside_introduction,True,confidence_intervals,True,appendix,alternative_specification,none,True,yes,True,False,False,moderate,6,6,0.8,6,11,1501,False,True,False,False,1.0,0.08133,claude-opus-5-5
36 +jof_12504,True,mixed_theory_empirical,asset_pricing,calibration_simulation,2,False,True,False,True,True,True,True,0,-1,2,1,1,0,motivation|preview_of_approach|preview_of_results|preview_of_approach|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,13,1,new_evidence|new_theory_or_model,True,True,inside_introduction,True,stars_and_tstats,False,mixed,alternative_specification|alternative_measure|sensitivity_to_parameters|subsample,unclear,True,no,False,False,True,low,7,7,0.8,5,11,1440,True,True,False,False,0.778,0.09952,claude-opus-5-5
37 +jof_12505,True,empirical,corporate_finance,reduced_form_regression,3,True,False,False,False,True,True,True,-1,-1,-1,0,0,2,motivation|motivation|literature_gap|identification_defense|preview_of_approach|preview_of_results|identification_defense|robustness_mention|identification_defense|preview_of_results|mechanism_or_theory_sketch|contribution_statement,12,16,new_evidence,False,False,both,True,unclear,True,within_results,alternative_sample_period|falsification|alternative_measure|subsample|alternative_specification,unclear,False,no_figures,True,False,False,moderate,6,5,0.8,3,17,947,False,False,False,False,1.0,0.065374,claude-opus-5-5
38 +jof_12506,True,empirical,corporate_finance,reduced_form_regression,3,True,True,False,False,True,True,False,3,-1,-1,2,2,-1,motivation|mechanism_or_theory_sketch|preview_of_results|identification_defense|robustness_mention|preview_of_results|preview_of_results|preview_of_results|preview_of_results|policy_or_broader_implication|roadmap,11,-1,new_evidence,True,True,inside_introduction,True,unclear,True,main_text_section,sensitivity_to_parameters|alternative_measure,unclear,True,no,True,True,False,moderate,8,8,0.8,6,8,956,True,False,False,False,1.0,0.076298,claude-opus-5-5
39 +jof_12507,True,empirical,market_microstructure,regression_discontinuity,4,True,True,False,True,True,False,True,0,-1,2,3,-1,1,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,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,7,7,0.85,5,5,831,False,False,False,False,0.8,0.066786,claude-opus-5-5
40 +jof_12508,True,mixed_theory_empirical,asset_pricing,calibration_simulation,3,True,False,True,True,True,False,True,-1,4,4,0,-1,1,motivation|tension_or_puzzle|literature_gap|contribution_statement|preview_of_results|related_literature_positioning|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|robustness_mention|preview_of_approach|mechanism_or_theory_sketch|related_literature_positioning|other|roadmap,16,0,new_evidence|new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,True,mixed,True,within_results,alternative_specification|alternative_measure|other,newey_west,False,mixed,False,False,True,moderate,7,7,0.8,5,9,1270,False,False,False,False,1.0,0.082302,claude-opus-5-5
41 +jof_12509,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,False,False,True,True,False,False,-1,-1,0,2,-1,-1,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,13,6,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,7,7,0.85,8,10,1171,False,False,False,False,1.0,0.076466,claude-opus-5-5
42 +jof_12510,True,empirical,market_microstructure,difference_in_differences,4,True,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|tension_or_puzzle|literature_gap|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|contribution_statement|policy_or_broader_implication|related_literature_positioning|roadmap,11,2,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,False,mixed,falsification|subsample,unclear,True,unclear,True,True,True,moderate,5,5,0.8,5,5,849,False,False,False,False,1.0,0.06787,claude-opus-5-5
43 +jof_12511,True,empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,True,False,0,-1,2,1,1,-1,motivation|tension_or_puzzle|related_literature_positioning|literature_gap|preview_of_results|related_literature_positioning|preview_of_results|related_literature_positioning|identification_defense|related_literature_positioning|contribution_statement|roadmap,12,5,new_evidence|new_identification,True,True,inside_introduction,True,unclear,True,main_text_section,alternative_measure|subsample|alternative_specification,unclear,False,no,True,True,True,moderate,5,5,0.82,4,6,909,True,True,False,False,1.0,0.069882,claude-opus-5-5
44 +jof_12512,True,mixed_theory_empirical,financial_intermediation_banking,reduced_form_regression,3,True,False,False,True,True,True,False,-1,-1,3,1,0,-1,motivation|literature_gap|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|identification_defense|identification_defense|identification_defense|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|contribution_statement|related_literature_positioning,14,6,new_evidence|new_theory_or_model,False,False,inside_introduction,True,unclear,True,mixed,subsample|alternative_measure|alternative_specification|other,unclear,True,unclear,True,True,False,moderate,8,8,0.8,5,8,1343,True,True,False,False,0.857,0.089202,claude-opus-5-5
45 +jof_12513,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,False,False,True,True,False,True,-1,-1,1,0,-1,4,preview_of_approach|preview_of_results|motivation|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|tension_or_puzzle|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,13,8,new_evidence|new_theory_or_model,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|alternative_measure|other,unclear,False,mixed,False,False,False,low,7,7,0.8,6,10,781,False,False,False,False,1.0,0.064622,claude-opus-5-5
46 +jof_12514,True,empirical,household_finance,reduced_form_regression,3,True,False,False,True,True,True,False,-1,-1,0,1,4,-1,motivation|literature_gap|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|other|identification_defense|preview_of_results|contribution_statement|roadmap,11,4,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,within_results,other,unclear,True,mixed,False,False,False,moderate,7,7,0.8,5,6,797,True,True,False,False,1.0,0.068798,claude-opus-5-5
47 +jof_12518,True,empirical,asset_pricing,reduced_form_regression,1,False,True,True,False,True,False,True,0,0,-1,0,-1,2,motivation|tension_or_puzzle|other|other|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement,8,4,new_evidence|resolution_of_puzzle,False,False,inside_introduction,True,unclear,False,main_text_section,alternative_measure|alternative_sample_period|alternative_specification,mixed,True,unclear,True,False,False,moderate,5,5,0.8,3,5,757,False,True,False,False,1.0,0.08846,claude-opus-5-5
48 +jof_12519,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,1,False,True,False,True,True,False,False,0,-1,1,3,-1,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|policy_or_broader_implication|mechanism_or_theory_sketch|policy_or_broader_implication|policy_or_broader_implication|contribution_statement|related_literature_positioning|literature_gap,12,8,new_theory_or_model,False,False,inside_introduction,False,unclear,False,appendix,other,none,False,no,True,True,False,moderate,6,6,0.85,7,9,1552,False,True,False,False,0.9,0.069166,claude-opus-5-5
49 +jof_12520,True,empirical,financial_intermediation_banking,reduced_form_regression,2,False,True,False,False,True,False,True,0,-1,-1,1,-1,5,motivation|related_literature_positioning|literature_gap|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|identification_defense|preview_of_results|other|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,16,6,new_evidence,True,True,inside_introduction,True,unclear,True,mixed,alternative_measure|alternative_specification|subsample|other,unclear,True,unclear,True,True,False,moderate,6,6,0.8,6,8,1384,False,True,False,False,0.857,0.084218,claude-opus-5-5
50 +jof_12521,True,empirical,household_finance,reduced_form_regression,3,True,False,False,True,True,False,True,-1,-1,1,2,-1,3,motivation|tension_or_puzzle|policy_or_broader_implication|preview_of_approach|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning|literature_gap|related_literature_positioning,10,1,new_evidence|new_identification,False,False,inside_introduction,True,unclear,False,mixed,subsample|alternative_specification|alternative_sample_period|other,unclear,True,no_figures,True,False,True,moderate,6,6,0.82,6,5,746,False,False,False,False,0.667,0.08409,claude-opus-5-5
51 +jof_12522,True,empirical,household_finance,regression_discontinuity,4,True,True,False,True,True,True,False,0,-1,1,1,3,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|identification_defense|preview_of_results|related_literature_positioning|contribution_statement|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning|policy_or_broader_implication|roadmap,13,4,new_evidence|new_identification,True,True,inside_introduction,True,stars_and_tstats,True,within_results,placebo|alternative_specification|falsification,unclear,True,unclear,True,True,False,moderate,8,8,0.85,6,8,1114,True,False,False,True,1.0,0.074438,claude-opus-5-5
52 +jof_12523,True,empirical,household_finance,structural_estimation,3,True,True,False,True,True,False,True,1,-1,1,1,-1,4,motivation|preview_of_approach|preview_of_results|literature_gap|contribution_statement|related_literature_positioning|tension_or_puzzle|policy_or_broader_implication|preview_of_approach|robustness_mention|identification_defense|mechanism_or_theory_sketch|policy_or_broader_implication|roadmap,14,1,new_evidence|new_method,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|subsample|other,unclear,True,unclear,True,False,True,moderate,6,6,0.8,6,10,1349,False,False,False,False,0.75,0.077334,claude-opus-5-5
53 +jof_12524,True,empirical,asset_pricing,other,2,False,True,False,True,True,False,False,0,-1,1,1,-1,-1,motivation|literature_gap|preview_of_approach|preview_of_approach|preview_of_results|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,9,2,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,appendix,alternative_specification,unclear,True,unclear,True,False,True,low,6,6,0.8,5,5,646,False,True,False,False,1.0,0.065194,claude-opus-5-5
54 +jof_12525,True,theoretical,market_microstructure,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,4,1,-1,0,motivation|preview_of_approach|motivation|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|policy_or_broader_implication|related_literature_positioning|roadmap,12,0,new_theory_or_model,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,True,False,moderate,6,6,0.88,5,10,978,False,False,False,False,1.0,0.064646,claude-opus-5-5
55 +jof_12526,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,True,False,True,True,False,True,2,-1,0,2,-1,0,preview_of_approach|mechanism_or_theory_sketch|preview_of_results,3,0,new_theory_or_model,False,False,inside_introduction,False,unclear,False,main_text_section,other,none,False,no,True,False,False,moderate,5,4,0.8,4,1,228,False,False,False,False,0.833,0.052926,claude-opus-5-5
56 +jof_12530,True,survey_or_address,financial_econometrics,other,2,False,True,True,True,True,False,True,4,3,5,1,-1,6,motivation|tension_or_puzzle|mechanism_or_theory_sketch|related_literature_positioning|policy_or_broader_implication|roadmap,6,2,new_method|synthesis,True,False,both,False,mixed,True,none_seen,,none,True,no,True,True,True,moderate,7,7,0.8,16,3,600,True,True,False,False,0.75,0.072698,claude-opus-5-5
57 +jof_12539,True,theoretical,macro_finance,calibration_simulation,4,True,True,False,True,True,False,True,0,-1,1,4,-1,0,motivation|tension_or_puzzle|preview_of_approach|mechanism_or_theory_sketch|other|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|preview_of_results|policy_or_broader_implication|preview_of_approach|preview_of_results|related_literature_positioning|roadmap,16,0,new_theory_or_model,True,True,separate_section,True,unclear,False,none_seen,,none,False,no,False,True,True,low,8,8,0.85,6,9,1288,False,False,False,False,1.0,0.076238,claude-opus-5-5
58 +jof_12540,True,theoretical,macro_finance,calibration_simulation,4,True,True,False,True,True,False,True,0,-1,0,2,-1,4,motivation|preview_of_approach|mechanism_or_theory_sketch|mechanism_or_theory_sketch|motivation|preview_of_results|preview_of_results|related_literature_positioning|preview_of_results|preview_of_results|policy_or_broader_implication|policy_or_broader_implication|related_literature_positioning|contribution_statement|robustness_mention,15,9,new_theory_or_model,False,False,inside_introduction,True,unclear,False,internet_appendix,other,none,False,no,False,True,False,low,7,7,0.85,5,10,1468,False,True,False,False,1.0,0.082542,claude-opus-5-5
59 +jof_12541,True,theoretical,market_microstructure,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,1,1,-1,4,motivation|preview_of_approach,2,-1,new_theory_or_model,False,False,inside_introduction,False,unclear,False,appendix,other,none,False,no,True,True,False,moderate,6,5,0.75,5,1,234,False,False,False,False,0.778,0.06161,claude-opus-5-5
60 +jof_12542,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,4,True,True,False,True,True,True,True,0,-1,0,1,1,0,contribution_statement|motivation|preview_of_results|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|other|roadmap,8,0,new_evidence|new_theory_or_model,True,True,both,True,unclear,True,mixed,alternative_measure|sensitivity_to_parameters|alternative_specification,unclear,True,unclear,True,False,True,moderate,7,7,0.8,4,4,658,True,False,False,False,1.0,0.07305,claude-opus-5-5
61 +jof_12544,True,mixed_theory_empirical,asset_pricing,theoretical_model_analytical,3,True,True,False,True,True,False,True,0,-1,1,3,-1,6,motivation|tension_or_puzzle|literature_gap|mechanism_or_theory_sketch|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning|roadmap,10,0,new_theory_or_model|new_evidence,True,True,inside_introduction,False,stars_and_tstats,False,none_seen,,mixed,False,mixed,True,False,True,moderate,6,6,0.8,8,6,870,False,False,False,True,1.0,0.065926,claude-opus-5-5
62 +jof_12545,True,empirical,corporate_finance,instrumental_variables,4,True,True,True,True,True,True,False,0,1,2,3,3,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_approach|identification_defense|preview_of_approach|identification_defense|preview_of_results|contribution_statement|related_literature_positioning,10,5,new_identification|new_evidence,False,False,inside_introduction,True,unclear,True,mixed,placebo|alternative_specification|subsample|falsification|other,unclear,True,no,True,True,False,moderate,6,6,0.85,5,6,1248,True,True,False,False,1.0,0.070834,claude-opus-5-5
63 +jof_12546,True,mixed_theory_empirical,international_finance,theoretical_model_analytical,2,False,False,False,True,True,False,True,-1,-1,2,3,-1,2,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning,7,0,new_theory_or_model|new_evidence,False,False,inside_introduction,True,unclear,False,within_results,subsample|alternative_specification,unclear,True,unclear,True,False,False,moderate,7,7,0.8,7,4,708,False,False,False,False,1.0,0.07009,claude-opus-5-5
64 +jof_12547,True,empirical,behavioral_finance,reduced_form_regression,2,False,True,False,True,True,False,False,0,-1,0,1,-1,-1,motivation|tension_or_puzzle|literature_gap|preview_of_approach|identification_defense|preview_of_results|preview_of_results|preview_of_results|policy_or_broader_implication|roadmap,10,-1,new_evidence|new_data_or_measure,True,True,inside_introduction,True,stars_and_tstats,True,appendix,subsample|alternative_specification,robust,False,no,True,True,True,moderate,7,7,0.8,6,7,1115,False,True,False,True,0.692,0.078794,claude-opus-5-5
65 +jof_12548,True,replication_or_correction,market_microstructure,theoretical_model_analytical,1,False,False,True,False,True,False,True,-1,1,-1,0,-1,3,motivation|tension_or_puzzle|contribution_statement|preview_of_results|roadmap,5,0,replication_or_correction,True,False,none,False,unclear,False,none_seen,,none,False,no_figures,False,False,False,low,4,3,0.88,4,1,188,False,False,False,False,1.0,0.042282,claude-opus-5-5
66 +jof_12583,True,mixed_theory_empirical,corporate_finance,theoretical_model_analytical,2,False,True,False,True,True,False,True,0,-1,0,2,-1,1,motivation|tension_or_puzzle|preview_of_approach|contribution_statement|preview_of_results|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,14,6,new_theory_or_model|new_method|new_evidence,True,True,inside_introduction,False,unclear,False,none_seen,,unclear,True,no,True,False,True,moderate,7,7,0.85,5,8,1506,False,True,False,False,1.0,0.075218,claude-opus-5-5
67 +jof_12584,True,theoretical,asset_pricing,theoretical_model_analytical,2,False,True,False,True,True,False,True,0,-1,1,2,-1,4,motivation|related_literature_positioning|tension_or_puzzle|preview_of_approach|mechanism_or_theory_sketch|preview_of_results,6,2,new_theory_or_model|resolution_of_puzzle,False,False,both,False,unclear,False,main_text_section,alternative_specification|sensitivity_to_parameters,none,False,no,True,True,False,moderate,5,5,0.85,5,3,624,False,False,False,False,0.9,0.062194,claude-opus-5-5
68 +jof_12585,True,empirical,financial_intermediation_banking,instrumental_variables,4,True,True,False,True,True,True,False,0,-1,4,1,2,-1,motivation|tension_or_puzzle|literature_gap|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|contribution_statement|identification_defense|related_literature_positioning|contribution_statement|roadmap,13,2,new_evidence|new_data_or_measure|new_identification|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,subsample|falsification|alternative_specification,unclear,True,unclear,True,True,False,moderate,7,7,0.8,5,6,974,False,False,False,False,0.833,0.080458,claude-opus-5-5
69 +jof_12586,True,mixed_theory_empirical,macro_finance,theoretical_model_analytical,2,False,False,False,True,True,False,True,-1,-1,0,4,-1,3,motivation|preview_of_approach|related_literature_positioning|motivation|mechanism_or_theory_sketch|contribution_statement|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning|literature_gap,11,3,new_theory_or_model|synthesis|new_evidence,False,False,inside_introduction,False,unclear,False,none_seen,,unclear,False,no,True,False,False,moderate,5,5,0.8,6,6,966,False,False,False,False,1.0,0.063046,claude-opus-5-5
70 +jof_12587,True,mixed_theory_empirical,international_finance,reduced_form_regression,2,False,False,False,True,True,True,True,-1,-1,0,1,3,4,motivation|literature_gap|preview_of_approach|preview_of_results|preview_of_results|contribution_statement|tension_or_puzzle|mechanism_or_theory_sketch|mechanism_or_theory_sketch|related_literature_positioning|related_literature_positioning,11,2,new_evidence|new_theory_or_model,False,False,inside_introduction,True,unclear,True,main_text_section,alternative_sample_period|subsample|alternative_measure|other,unclear,True,unclear,True,True,True,moderate,6,6,0.8,5,7,1230,True,True,False,False,1.0,0.071482,claude-opus-5-5
71 +jof_12588,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,False,False,False,True,False,True,-1,-1,-1,1,-1,3,contribution_statement|preview_of_approach|mechanism_or_theory_sketch|related_literature_positioning|preview_of_results|preview_of_results|preview_of_approach|policy_or_broader_implication,8,0,new_theory_or_model,False,False,both,False,unclear,False,none_seen,,none,False,no,True,True,False,low,6,6,0.85,5,5,830,False,False,False,False,0.833,0.064122,claude-opus-5-5
72 +jof_12589,True,empirical,financial_intermediation_banking,instrumental_variables,4,True,True,False,True,True,True,False,0,-1,0,1,1,-1,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,14,6,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,6,6,0.8,4,9,1408,True,True,False,False,1.0,0.09826,claude-opus-5-5
73 +jof_12590,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,False,False,True,True,False,True,-1,-1,0,3,-1,4,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,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,6,6,0.88,5,8,1553,False,False,False,False,1.0,0.078502,claude-opus-5-5
74 +jof_12591,True,mixed_theory_empirical,financial_intermediation_banking,reduced_form_regression,1,False,True,False,True,True,True,True,0,-1,0,3,1,4,motivation|literature_gap|related_literature_positioning|preview_of_approach|preview_of_results|preview_of_results|policy_or_broader_implication|related_literature_positioning|contribution_statement|preview_of_results|roadmap,11,3,new_data_or_measure|new_theory_or_model|new_evidence,True,True,inside_introduction,True,unclear,True,within_results,alternative_measure|sensitivity_to_parameters|alternative_sample_period,unclear,True,no,True,True,True,moderate,6,6,0.8,7,5,884,False,True,False,False,1.0,0.070734,claude-opus-5-5
75 +jof_12592,True,empirical,financial_intermediation_banking,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,0,1,-1,5,motivation|policy_or_broader_implication|preview_of_approach|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_results|identification_defense|preview_of_results|mechanism_or_theory_sketch|contribution_statement|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,14,6,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,False,mixed,alternative_measure|subsample|falsification,unclear,True,unclear,True,True,True,moderate,8,8,0.8,9,9,1320,False,True,False,False,0.833,0.082222,claude-opus-5-5
76 +jof_12593,True,empirical,household_finance,survey_or_experiment,4,True,True,False,True,True,False,True,0,-1,1,1,-1,5,motivation|motivation|tension_or_puzzle|mechanism_or_theory_sketch|policy_or_broader_implication|literature_gap|motivation|preview_of_approach|preview_of_approach|identification_defense|mechanism_or_theory_sketch|identification_defense|preview_of_results|preview_of_results|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,20,14,new_evidence,True,True,inside_introduction,True,mixed,True,mixed,subsample|alternative_specification|alternative_sample_period|other,robust,False,mixed,True,True,True,moderate,7,7,0.85,9,18,1855,True,False,False,True,0.714,0.088026,claude-opus-5-5
77 +jof_12600,True,replication_or_correction,corporate_finance,difference_in_differences,3,True,False,True,True,True,False,True,-1,1,3,3,-1,0,motivation|related_literature_positioning|preview_of_approach|preview_of_approach|contribution_statement|preview_of_results|preview_of_results|robustness_mention|related_literature_positioning|roadmap|policy_or_broader_implication,11,2,replication_or_correction|new_data_or_measure|new_evidence,True,True,both,True,unclear,False,mixed,subsample|alternative_specification|other,unclear,True,no,True,False,True,moderate,6,5,0.8,4,8,1255,False,False,False,False,0.75,0.083442,claude-opus-5-5
78 +jof_12601,True,empirical,asset_pricing,reduced_form_regression,3,True,True,False,False,True,False,True,1,-1,-1,2,-1,1,motivation|motivation|literature_gap|contribution_statement|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results,11,5,new_evidence|new_data_or_measure|resolution_of_puzzle,False,False,both,True,unclear,True,none_seen,,unclear,True,unclear,True,False,False,moderate,5,5,0.78,3,8,638,False,False,False,False,1.0,0.07053,claude-opus-5-5
79 +jof_12602,True,empirical,household_finance,difference_in_differences,4,True,True,False,True,True,True,True,0,-1,1,2,2,4,motivation|literature_gap|contribution_statement|preview_of_approach|identification_defense|preview_of_approach|identification_defense|preview_of_results|roadmap,9,0,new_evidence|new_identification|new_data_or_measure,True,True,inside_introduction,True,unclear,True,within_results,placebo|alternative_specification|subsample|other,unclear,True,unclear,True,False,False,moderate,6,6,0.85,6,5,1014,False,False,False,False,0.833,0.07731,claude-opus-5-5
80 +jof_12603,True,empirical,asset_pricing,reduced_form_regression,3,True,False,False,False,True,False,True,-1,-1,-1,0,-1,2,motivation|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|robustness_mention|preview_of_results|preview_of_approach|preview_of_results|contribution_statement|policy_or_broader_implication|policy_or_broader_implication|related_literature_positioning|roadmap,13,7,new_evidence|resolution_of_puzzle,True,True,inside_introduction,True,mixed,True,mixed,alternative_specification|subsample|alternative_measure|alternative_sample_period,newey_west,True,yes,True,True,False,moderate,7,7,0.85,4,12,1329,False,False,False,False,1.0,0.075274,claude-opus-5-5
81 +jof_12604,True,empirical,financial_intermediation_banking,reduced_form_regression,2,False,True,False,False,True,False,True,0,-1,-1,1,-1,2,motivation|literature_gap|preview_of_results|contribution_statement|policy_or_broader_implication|preview_of_approach|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_results|other|related_literature_positioning|contribution_statement,12,5,new_evidence,False,False,inside_introduction,True,unclear,True,within_results,alternative_specification,unclear,True,unclear,True,True,False,moderate,6,6,0.8,3,6,979,False,True,False,False,0.8,0.071398,claude-opus-5-5
82 +jof_12605,True,empirical,household_finance,reduced_form_regression,2,False,False,False,True,True,True,False,-1,-1,0,2,1,-1,motivation|literature_gap|preview_of_approach|contribution_statement|preview_of_results|mechanism_or_theory_sketch|preview_of_results|tension_or_puzzle|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning,12,2,new_evidence|new_data_or_measure,False,False,inside_introduction,True,stars_and_tstats,True,mixed,subsample|alternative_specification|other,robust,True,mixed,True,True,False,moderate,6,5,0.85,4,7,930,True,True,False,True,1.0,0.067546,claude-opus-5-5
83 +jof_12606,True,theoretical,asset_pricing,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,2,3,-1,0,preview_of_approach|motivation|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,11,4,new_theory_or_model,True,True,inside_introduction,False,unclear,False,appendix,other,none,False,mixed,False,False,False,moderate,5,5,0.85,5,9,828,False,True,False,False,1.0,0.060654,claude-opus-5-5
84 +jof_12607,True,methodological_econometric,asset_pricing,other,1,False,False,False,True,True,False,True,-1,-1,0,2,-1,1,motivation|literature_gap|contribution_statement|related_literature_positioning|literature_gap|preview_of_approach|preview_of_results|preview_of_approach|preview_of_approach|preview_of_results,10,0,new_method|new_evidence,False,False,inside_introduction,True,mixed,True,mixed,sensitivity_to_parameters|alternative_sample_period|other,unclear,True,no,True,False,True,moderate,7,6,0.8,3,5,964,False,False,False,False,0.75,0.080822,claude-opus-5-5
85 +jof_12609,True,empirical,financial_intermediation_banking,reduced_form_regression,3,True,True,False,True,True,False,True,2,-1,1,3,-1,3,motivation|identification_defense|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|policy_or_broader_implication|related_literature_positioning|contribution_statement|related_literature_positioning|contribution_statement|roadmap,12,3,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,False,within_results,alternative_measure|subsample|alternative_specification|other,unclear,True,no_figures,True,False,False,moderate,6,6,0.82,4,6,1024,False,True,False,False,1.0,0.067874,claude-opus-5-5
86 +jof_12610,True,mixed_theory_empirical,corporate_finance,instrumental_variables,3,True,True,False,True,True,False,True,0,-1,2,1,-1,1,motivation|literature_gap|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|identification_defense|related_literature_positioning|contribution_statement|related_literature_positioning|policy_or_broader_implication,13,4,new_evidence|new_theory_or_model,False,False,inside_introduction,True,unclear,True,within_results,subsample|falsification|other,unclear,True,unclear,True,False,False,moderate,9,9,0.8,6,6,1139,False,True,False,False,1.0,0.07587,claude-opus-5-5
87 +jof_12611,True,empirical,asset_pricing,reduced_form_regression,2,False,True,True,False,True,False,False,1,1,-1,2,-1,-1,literature_gap|tension_or_puzzle|preview_of_results|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|contribution_statement,13,7,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,False,mixed,alternative_specification|subsample|alternative_measure|falsification|other,unclear,True,unclear,True,False,False,moderate,8,7,0.8,5,8,1529,False,True,False,False,1.0,0.079906,claude-opus-5-5
88 +jof_12612,True,mixed_theory_empirical,asset_pricing,theoretical_model_analytical,2,False,True,False,True,True,False,True,0,-1,1,2,-1,0,motivation|tension_or_puzzle|related_literature_positioning|contribution_statement|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|preview_of_approach|preview_of_results|policy_or_broader_implication|related_literature_positioning|related_literature_positioning|roadmap,14,8,new_theory_or_model|new_evidence|resolution_of_puzzle,True,True,inside_introduction,True,unclear,False,within_results,alternative_sample_period|alternative_measure,newey_west,True,no,True,False,True,moderate,6,6,0.8,4,16,1543,False,True,False,False,1.0,0.080594,claude-opus-5-5
89 +jof_12613,True,empirical,financial_intermediation_banking,natural_experiment_event_study,4,True,False,False,True,True,False,True,-1,-1,1,3,-1,0,motivation|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|identification_defense|preview_of_results|robustness_mention|mechanism_or_theory_sketch|identification_defense|related_literature_positioning|contribution_statement|policy_or_broader_implication|contribution_statement|policy_or_broader_implication|roadmap,15,5,new_evidence|new_data_or_measure|new_identification,True,True,inside_introduction,True,unclear,True,mixed,placebo|alternative_specification|subsample|falsification|other,unclear,True,unclear,False,True,False,low,8,6,0.82,8,7,1198,False,True,False,False,1.0,0.079778,claude-opus-5-5
90 +jof_12614,True,theoretical,macro_finance,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|preview_of_approach|motivation|related_literature_positioning|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|policy_or_broader_implication|preview_of_results|preview_of_results|contribution_statement|roadmap,12,6,new_theory_or_model,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,True,True,moderate,6,6,0.85,5,8,1188,False,False,False,False,0.714,0.068854,claude-opus-5-5
91 +jof_12615,True,methodological_econometric,asset_pricing,calibration_simulation,1,False,False,False,False,True,False,True,-1,-1,-1,0,-1,0,motivation|related_literature_positioning|tension_or_puzzle|contribution_statement|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|policy_or_broader_implication|preview_of_results|policy_or_broader_implication|roadmap,12,0,new_evidence|replication_or_correction,True,True,inside_introduction,True,unclear,True,mixed,sensitivity_to_parameters|alternative_specification|other,none,True,no,True,False,False,low,6,6,0.8,4,4,717,False,True,False,False,0.857,0.100634,claude-opus-5-5
92 +jof_12616,True,empirical,asset_pricing,reduced_form_regression,4,True,True,False,True,True,False,True,1,-1,0,2,-1,3,motivation|mechanism_or_theory_sketch|literature_gap|preview_of_approach|preview_of_results|robustness_mention|tension_or_puzzle|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,11,1,new_evidence|new_data_or_measure,True,True,inside_introduction,False,unclear,False,within_results,alternative_specification|alternative_measure|subsample|other,unclear,False,unclear,True,False,True,moderate,7,7,0.82,5,6,901,False,False,False,False,1.0,0.074482,claude-opus-5-5
93 +jof_12617,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,False,False,True,True,False,True,-1,-1,0,1,-1,4,motivation|tension_or_puzzle|literature_gap|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|robustness_mention|identification_defense|contribution_statement|related_literature_positioning|related_literature_positioning|contribution_statement,12,5,new_evidence|new_data_or_measure|new_theory_or_model|resolution_of_puzzle,False,False,inside_introduction,True,stars_and_tstats,True,mixed,subsample|alternative_measure|alternative_specification,unclear,False,unclear,True,False,False,moderate,8,8,0.82,5,7,1325,False,True,False,False,1.0,0.08487,claude-opus-5-5
94 +jof_12618,True,theoretical,asset_pricing,theoretical_model_analytical,3,False,False,False,True,True,False,True,-1,-1,0,2,-1,4,motivation|tension_or_puzzle|literature_gap|preview_of_approach|contribution_statement,5,0,new_theory_or_model|resolution_of_puzzle,False,False,inside_introduction,False,unclear,False,main_text_section,other,none,False,no,True,False,False,moderate,6,6,0.85,5,1,229,False,False,False,False,1.0,0.061754,claude-opus-5-5
95 +jof_12619,True,empirical,behavioral_finance,survey_or_experiment,3,True,True,False,True,True,False,False,0,-1,0,4,-1,-1,motivation|related_literature_positioning|motivation|literature_gap|preview_of_approach|identification_defense|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|roadmap,10,2,new_evidence,True,True,both,False,unclear,False,none_seen,,clustered,True,no,True,False,False,moderate,6,6,0.8,5,7,775,False,False,True,False,1.0,0.061318,claude-opus-5-5
96 +jof_12620,True,empirical,international_finance,reduced_form_regression,4,True,False,True,False,True,False,True,-1,1,-1,0,-1,2,motivation|tension_or_puzzle|preview_of_results|mechanism_or_theory_sketch|identification_defense|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|literature_gap,14,0,new_evidence|resolution_of_puzzle,False,False,inside_introduction,True,unclear,True,none_seen,,unclear,True,no,True,True,True,moderate,7,7,0.8,5,8,1295,False,False,False,False,1.0,0.080938,claude-opus-5-5
97 +jof_12684,True,mixed_theory_empirical,asset_pricing,theoretical_model_analytical,2,False,True,False,True,True,True,True,0,-1,3,4,4,2,motivation|contribution_statement|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|roadmap|literature_gap|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning,12,0,new_theory_or_model|new_data_or_measure|new_evidence,True,True,inside_introduction,True,unclear,True,within_results,alternative_specification|other,unclear,True,no,True,False,True,low,6,6,0.8,5,8,1245,True,True,False,False,1.0,0.06673,claude-opus-5-5
98 +jof_12685,True,empirical,behavioral_finance,reduced_form_regression,4,True,True,True,True,True,False,True,0,1,3,2,-1,1,other|motivation|mechanism_or_theory_sketch|preview_of_results|preview_of_results|identification_defense|preview_of_results|contribution_statement|related_literature_positioning,9,13,new_evidence,False,False,inside_introduction,True,stars_and_tstats,True,main_text_section,alternative_measure|subsample|alternative_specification|falsification|placebo,clustered,True,unclear,True,True,True,moderate,7,7,0.8,5,14,1315,False,True,True,True,1.0,0.093224,claude-opus-5-5
99 +jof_12687,True,empirical,household_finance,difference_in_differences,4,True,False,False,True,True,True,False,-1,-1,2,2,3,-1,motivation|literature_gap|contribution_statement|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning|roadmap,10,0,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,appendix,alternative_measure|subsample|alternative_specification,unclear,True,unclear,True,False,False,moderate,8,8,0.8,5,7,1063,True,True,False,False,0.9,0.080826,claude-opus-5-5
100 +jof_12688,True,methodological_econometric,entrepreneurial_finance_private_equity,structural_estimation,2,False,False,False,True,True,False,True,-1,-1,0,3,-1,1,motivation|literature_gap|contribution_statement|related_literature_positioning|preview_of_approach|robustness_mention|contribution_statement|preview_of_results|policy_or_broader_implication|roadmap,10,1,new_method|new_data_or_measure|new_evidence,True,True,inside_introduction,True,mixed,False,mixed,sensitivity_to_parameters|alternative_specification|alternative_measure|other,mixed,True,unclear,True,True,False,moderate,5,5,0.8,4,5,520,False,True,False,False,1.0,0.06077,claude-opus-5-5
101 +jof_12689,True,theoretical,asset_pricing,theoretical_model_analytical,1,False,False,False,True,True,False,False,-1,-1,0,1,-1,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|policy_or_broader_implication|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning,15,5,new_theory_or_model,False,False,inside_introduction,False,unclear,False,none_seen,,none,False,no_figures,True,False,False,low,6,6,0.85,6,8,1174,False,True,False,False,1.0,0.06273,claude-opus-5-5
102 +jof_12690,True,empirical,financial_intermediation_banking,difference_in_differences,3,True,True,True,True,True,True,True,1,0,3,3,3,4,motivation|literature_gap|preview_of_approach|identification_defense|identification_defense|robustness_mention|robustness_mention|preview_of_results|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|contribution_statement|policy_or_broader_implication|roadmap,14,8,new_evidence|new_data_or_measure|new_identification,True,True,inside_introduction,True,stars_and_tstats,True,mixed,alternative_sample_period|subsample|alternative_specification|placebo|other,clustered,True,unclear,True,True,True,moderate,6,6,0.85,5,10,1250,True,True,True,True,0.75,0.075822,claude-opus-5-5
103 +jof_12691,True,mixed_theory_empirical,corporate_finance,instrumental_variables,4,True,True,False,True,True,True,False,0,-1,1,2,2,-1,motivation|literature_gap|preview_of_results|mechanism_or_theory_sketch|identification_defense|identification_defense|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|contribution_statement,12,8,new_evidence|new_identification,False,False,inside_introduction,True,unclear,True,main_text_section,alternative_measure|alternative_specification|other,unclear,True,no,True,True,True,moderate,8,8,0.85,5,9,1366,False,True,False,False,1.0,0.08721,claude-opus-5-5
104 +jof_12692,True,empirical,corporate_finance,regression_discontinuity,4,True,False,False,False,True,True,True,-1,-1,-1,0,0,3,motivation|related_literature_positioning|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|other|tension_or_puzzle|contribution_statement|related_literature_positioning,11,4,new_evidence,False,False,inside_introduction,True,unclear,False,mixed,alternative_specification|placebo|alternative_measure|sensitivity_to_parameters|falsification,unclear,False,mixed,True,False,False,moderate,7,7,0.8,4,5,808,True,True,False,False,0.889,0.074722,claude-opus-5-5
105 +jof_12694,True,empirical,market_microstructure,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|tension_or_puzzle|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|identification_defense|preview_of_results|policy_or_broader_implication|roadmap,10,5,new_evidence|new_data_or_measure,True,True,both,True,p_values,False,mixed,subsample|alternative_measure|other,newey_west,True,no,True,True,True,moderate,5,5,0.82,6,7,948,False,True,False,False,1.0,0.076454,claude-opus-5-5
106 +jof_12695,True,mixed_theory_empirical,asset_pricing,survey_or_experiment,2,False,True,True,True,True,False,True,2,3,4,6,-1,3,motivation|literature_gap|contribution_statement|literature_gap|related_literature_positioning|related_literature_positioning|related_literature_positioning,7,0,new_evidence|new_theory_or_model,False,False,inside_introduction,True,p_values,False,none_seen,,none,False,no,True,False,True,moderate,4,4,0.8,7,5,1046,False,False,False,False,1.0,0.066766,claude-opus-5-5
107 +jof_12696,True,theoretical,asset_pricing,theoretical_model_analytical,1,False,False,False,True,True,False,False,-1,-1,0,1,-1,-1,motivation|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|tension_or_puzzle|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,11,7,new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,appendix,other,none,False,no,False,False,False,low,6,6,0.85,4,9,1121,False,False,False,False,1.0,0.066666,claude-opus-5-5
108 +jof_12697,True,mixed_theory_empirical,household_finance,calibration_simulation,2,False,True,False,True,True,True,False,0,-1,2,1,3,-1,motivation|contribution_statement|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|robustness_mention,10,0,new_evidence|new_data_or_measure|new_theory_or_model,False,False,inside_introduction,True,unclear,True,within_results,sensitivity_to_parameters|alternative_specification|subsample,unclear,True,unclear,True,True,True,moderate,5,5,0.8,5,4,852,True,True,False,False,1.0,0.07425,claude-opus-5-5
109 +jof_12698,True,empirical,corporate_finance,difference_in_differences,4,True,False,False,True,True,True,True,-1,-1,2,2,1,3,motivation|tension_or_puzzle|literature_gap|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|identification_defense|robustness_mention|identification_defense|policy_or_broader_implication|roadmap,12,-1,new_evidence|new_identification,True,True,separate_section,True,unclear,True,within_results,alternative_specification|placebo|alternative_measure|subsample|other,unclear,False,unclear,True,True,True,moderate,7,7,0.8,16,6,907,False,False,False,False,0.714,0.08379,claude-opus-5-5
110 +jof_12699,True,empirical,asset_pricing,reduced_form_regression,3,True,False,False,True,True,True,True,-1,-1,1,0,2,1,motivation|literature_gap|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,11,5,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,main_text_section,alternative_measure|alternative_specification,unclear,True,no_figures,False,False,False,moderate,8,8,0.82,5,8,1205,True,True,False,False,1.0,0.075354,claude-opus-5-5
111 +jof_12700,True,mixed_theory_empirical,asset_pricing,other,1,False,True,False,True,True,True,True,1,-1,0,2,2,3,motivation|literature_gap|preview_of_approach|contribution_statement|preview_of_results|preview_of_results|mechanism_or_theory_sketch|roadmap,8,1,new_evidence,True,True,inside_introduction,True,unclear,True,within_results,subsample|alternative_measure|alternative_specification|sensitivity_to_parameters,unclear,True,unclear,True,False,False,moderate,6,6,0.75,6,4,622,True,True,False,False,0.75,0.07601,claude-opus-5-5
112 +jof_12701,True,empirical,behavioral_finance,survey_or_experiment,4,True,True,True,True,True,False,False,1,0,1,2,-1,-1,motivation|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|tension_or_puzzle|robustness_mention|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning|policy_or_broader_implication,13,5,new_evidence,False,False,inside_introduction,False,unclear,False,main_text_section,alternative_specification|subsample|other,unclear,True,unclear,True,True,True,moderate,6,6,0.8,8,16,1330,False,True,False,False,1.0,0.070982,claude-opus-5-5
113 +jof_12702,True,empirical,financial_intermediation_banking,reduced_form_regression,3,True,True,False,True,True,True,False,0,-1,0,2,1,-1,motivation|literature_gap|mechanism_or_theory_sketch|preview_of_approach|identification_defense|preview_of_results|identification_defense|preview_of_results|tension_or_puzzle|preview_of_results|related_literature_positioning,11,-1,new_evidence,False,False,inside_introduction,True,unclear,False,mixed,alternative_measure|placebo|alternative_specification|other,unclear,True,unclear,True,False,False,moderate,9,9,0.8,7,4,710,True,False,False,False,0.857,0.07619,claude-opus-5-5
114 +jof_12703,True,empirical,corporate_finance,reduced_form_regression,3,True,True,False,True,True,False,False,0,-1,1,3,-1,-1,motivation|identification_defense|preview_of_results|preview_of_results|related_literature_positioning|literature_gap|contribution_statement|contribution_statement|policy_or_broader_implication|roadmap,10,4,new_data_or_measure|new_evidence,True,True,inside_introduction,True,unclear,True,main_text_section,alternative_specification|subsample|alternative_measure,unclear,True,no,True,False,True,moderate,6,6,0.8,6,7,1249,False,True,False,False,0.5,0.070314,claude-opus-5-5
115 +jof_12704,True,empirical,corporate_finance,reduced_form_regression,3,True,False,False,False,True,True,False,-1,-1,-1,1,0,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|roadmap|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|identification_defense|mechanism_or_theory_sketch|related_literature_positioning|policy_or_broader_implication|related_literature_positioning|policy_or_broader_implication,17,15,new_evidence|new_data_or_measure,True,False,inside_introduction,True,mixed,True,mixed,alternative_specification|subsample|alternative_measure|alternative_sample_period,unclear,True,unclear,True,False,True,moderate,6,6,0.82,4,17,1216,False,True,False,False,1.0,0.079654,claude-opus-5-5
116 +jof_12705,True,mixed_theory_empirical,asset_pricing,calibration_simulation,2,False,True,False,True,True,False,False,0,-1,3,3,-1,-1,motivation|motivation|literature_gap|preview_of_approach|related_literature_positioning|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_results|preview_of_results|contribution_statement|policy_or_broader_implication,11,4,new_theory_or_model|new_evidence|resolution_of_puzzle,False,False,inside_introduction,True,unclear,True,within_results,sensitivity_to_parameters|alternative_specification,none,True,unclear,True,False,True,moderate,7,7,0.82,4,5,869,False,False,False,False,1.0,0.072306,claude-opus-5-5
117 +jof_12706,True,empirical,financial_intermediation_banking,reduced_form_regression,3,True,False,False,False,True,False,False,-1,-1,-1,0,-1,-1,motivation|literature_gap|preview_of_approach|preview_of_results|robustness_mention|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_results|identification_defense|contribution_statement|related_literature_positioning|related_literature_positioning,13,6,new_evidence,False,False,inside_introduction,True,unclear,True,mixed,subsample|alternative_specification|alternative_measure,unclear,True,unclear,True,False,False,moderate,6,6,0.8,3,8,1391,False,True,False,False,1.0,0.077806,claude-opus-5-5
118 +jof_12710,True,mixed_theory_empirical,macro_finance,theoretical_model_analytical,2,False,True,False,True,True,False,True,0,-1,2,3,-1,2,motivation|motivation|literature_gap|policy_or_broader_implication|preview_of_approach|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,13,8,new_theory_or_model|new_evidence,True,True,inside_introduction,False,stars_and_tstats,False,main_text_section,alternative_specification|other,robust,True,yes,True,True,True,moderate,6,6,0.8,4,10,1315,False,True,False,False,1.0,0.082902,claude-opus-5-5
119 +jof_12717,True,theoretical,market_microstructure,theoretical_model_analytical,1,False,True,False,True,True,False,False,1,-1,1,4,-1,-1,motivation|preview_of_approach|contribution_statement|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning,10,6,new_theory_or_model,False,False,inside_introduction,False,unclear,False,none_seen,,none,False,no,False,False,False,low,4,4,0.85,6,7,891,False,True,False,False,0.833,0.05727,claude-opus-5-5
120 +jof_12718,True,empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,True,True,1,-1,2,3,0,2,motivation|tension_or_puzzle|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|identification_defense|mechanism_or_theory_sketch|preview_of_results|identification_defense|preview_of_results|related_literature_positioning|contribution_statement|contribution_statement|contribution_statement|roadmap,19,10,new_evidence|new_identification,True,True,inside_introduction,True,unclear,True,within_results,alternative_specification|subsample|falsification|other,unclear,True,unclear,True,False,True,moderate,8,8,0.85,4,14,1713,True,False,False,False,0.8,0.08015,claude-opus-5-5
121 +jof_12719,True,mixed_theory_empirical,macro_finance,theoretical_model_analytical,2,False,True,True,False,True,False,True,0,3,-1,2,-1,4,motivation|literature_gap|preview_of_approach|mechanism_or_theory_sketch|robustness_mention|tension_or_puzzle|related_literature_positioning|preview_of_results|contribution_statement|other|roadmap,11,4,new_theory_or_model|new_evidence|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,main_text_section,other,none,True,unclear,True,False,True,moderate,6,6,0.8,5,6,917,False,False,False,False,0.833,0.067898,claude-opus-5-5
122 +jof_12720,True,mixed_theory_empirical,international_finance,calibration_simulation,2,False,False,False,True,True,False,True,-1,-1,1,0,-1,3,motivation|related_literature_positioning|preview_of_approach|contribution_statement|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning,11,0,new_evidence|new_theory_or_model,False,False,inside_introduction,True,mixed,False,mixed,alternative_specification|other,unclear,True,mixed,True,False,True,moderate,5,5,0.8,4,6,840,False,False,False,False,1.0,0.066438,claude-opus-5-5
123 +jof_12721,True,empirical,corporate_finance,reduced_form_regression,2,False,True,False,False,True,False,True,0,-1,-1,2,-1,3,motivation|literature_gap|preview_of_results|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning,7,3,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,mixed,alternative_measure|alternative_specification,unclear,True,unclear,True,False,False,moderate,6,5,0.8,4,4,808,False,False,False,False,1.0,0.068346,claude-opus-5-5
124 +jof_12722,True,mixed_theory_empirical,asset_pricing,survey_or_experiment,4,True,True,False,True,True,True,False,0,-1,1,2,4,-1,motivation|related_literature_positioning|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|tension_or_puzzle|preview_of_results|roadmap,8,-1,new_evidence|new_theory_or_model,True,True,inside_introduction,True,p_values,False,within_results,subsample|sensitivity_to_parameters,none,False,no,False,False,False,low,7,7,0.85,5,4,724,False,False,False,False,1.0,0.07691,claude-opus-5-5
125 +jof_12723,True,empirical,behavioral_finance,reduced_form_regression,4,True,False,False,True,True,False,False,-1,-1,0,0,-1,-1,motivation|literature_gap|preview_of_results|preview_of_approach|related_literature_positioning|preview_of_results|preview_of_results|mechanism_or_theory_sketch|contribution_statement|contribution_statement|related_literature_positioning|related_literature_positioning|contribution_statement,13,10,new_evidence|new_data_or_measure|resolution_of_puzzle,False,False,inside_introduction,True,unclear,True,main_text_section,subsample|falsification|alternative_specification|other,unclear,True,no_figures,False,True,False,low,7,7,0.8,4,11,1286,False,True,False,False,1.0,0.074842,claude-opus-5-5
126 +jof_12724,True,empirical,household_finance,reduced_form_regression,3,True,True,False,False,True,False,True,0,-1,-1,1,-1,3,motivation|preview_of_approach|related_literature_positioning|mechanism_or_theory_sketch|identification_defense|robustness_mention|preview_of_results|preview_of_results|other|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,13,5,new_evidence,True,True,inside_introduction,True,stars_and_tstats,True,main_text_section,alternative_specification|alternative_measure|subsample|other,clustered,True,mixed,True,True,False,moderate,9,9,0.82,4,8,1267,False,True,True,False,0.8,0.097166,claude-opus-5-5
127 +jof_12725,True,empirical,financial_intermediation_banking,reduced_form_regression,3,True,False,False,False,True,True,False,-1,-1,-1,1,2,-1,motivation|mechanism_or_theory_sketch|preview_of_results|identification_defense|preview_of_results|contribution_statement|robustness_mention|roadmap,8,3,new_evidence,True,True,inside_introduction,True,unclear,True,main_text_section,placebo|alternative_sample_period|alternative_specification|subsample|falsification,unclear,False,no_figures,True,False,True,moderate,6,6,0.8,5,5,809,False,True,False,False,0.8,0.072706,claude-opus-5-5
128 +jof_12726,True,empirical,financial_intermediation_banking,difference_in_differences,4,True,False,False,True,True,True,False,-1,-1,1,0,2,-1,motivation|literature_gap|mechanism_or_theory_sketch|preview_of_approach|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|identification_defense|identification_defense|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|roadmap,14,5,new_evidence,True,False,inside_introduction,True,unclear,True,within_results,falsification|alternative_measure|other,unclear,True,unclear,False,False,False,moderate,7,7,0.8,4,6,1071,False,True,False,False,1.0,0.078954,claude-opus-5-5
129 +jof_12727,True,empirical,asset_pricing,instrumental_variables,3,True,True,False,True,True,True,False,1,-1,2,2,4,-1,motivation|mechanism_or_theory_sketch|preview_of_results|identification_defense|preview_of_results|tension_or_puzzle|preview_of_approach|preview_of_results|related_literature_positioning|contribution_statement|roadmap,11,5,new_evidence,True,True,inside_introduction,True,unclear,True,mixed,subsample|alternative_sample_period|alternative_specification,unclear,True,no,True,False,True,moderate,7,7,0.8,5,7,999,True,True,False,False,0.8,0.071978,claude-opus-5-5
130 +jof_12728,True,empirical,market_microstructure,reduced_form_regression,2,False,True,False,True,True,True,True,0,-1,1,2,2,4,motivation|literature_gap|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|other|identification_defense|preview_of_results|robustness_mention|identification_defense|preview_of_results|related_literature_positioning|contribution_statement,13,5,new_evidence|new_data_or_measure,False,False,both,True,unclear,True,main_text_section,alternative_specification|subsample|alternative_measure|falsification|other,unclear,True,no,False,True,False,moderate,7,7,0.8,5,6,1281,False,True,False,False,0.889,0.086522,claude-opus-5-5
131 +jof_12729,True,theoretical,behavioral_finance,theoretical_model_analytical,1,False,True,False,True,True,False,False,0,-1,0,1,-1,-1,motivation|tension_or_puzzle|related_literature_positioning|preview_of_results|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|preview_of_results|preview_of_results|related_literature_positioning,12,1,new_theory_or_model,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no_figures,True,False,False,moderate,7,7,0.85,5,8,1109,False,False,False,False,0.667,0.058134,claude-opus-5-5
132 +jof_12730,True,mixed_theory_empirical,asset_pricing,calibration_simulation,2,False,False,False,True,True,True,False,-1,-1,1,0,1,-1,motivation|related_literature_positioning|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|tension_or_puzzle|robustness_mention|related_literature_positioning|contribution_statement|related_literature_positioning|policy_or_broader_implication|other|roadmap,15,8,new_evidence,True,True,inside_introduction,True,unclear,True,main_text_section,alternative_specification|sensitivity_to_parameters,none,True,no,True,False,True,moderate,6,6,0.8,3,13,1123,False,True,False,False,0.857,0.077994,claude-opus-5-5
133 +jof_12731,True,mixed_theory_empirical,asset_pricing,structural_estimation,2,False,False,False,True,True,True,True,-1,-1,0,1,0,2,motivation|tension_or_puzzle|related_literature_positioning|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results,7,-1,new_theory_or_model|new_evidence|new_method,False,False,inside_introduction,True,mixed,True,mixed,alternative_specification|alternative_measure|subsample|other,mixed,False,unclear,False,False,False,moderate,5,5,0.8,3,3,607,True,True,False,False,0.9,0.079402,claude-opus-5-5
134 +jof_12732,True,mixed_theory_empirical,asset_pricing,structural_estimation,1,False,True,False,True,True,False,False,0,-1,0,1,-1,-1,motivation|tension_or_puzzle|literature_gap|related_literature_positioning|mechanism_or_theory_sketch|related_literature_positioning|related_literature_positioning|preview_of_approach|contribution_statement|preview_of_approach,10,4,new_evidence|new_theory_or_model,False,False,inside_introduction,True,unclear,True,within_results,alternative_specification|alternative_sample_period,robust,True,no,True,False,False,moderate,6,6,0.8,5,5,938,False,True,False,False,1.0,0.07557,claude-opus-5-5
135 +jof_12733,True,empirical,asset_pricing,machine_learning_textual,2,False,False,False,True,True,False,True,-1,-1,0,1,-1,3,motivation|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|identification_defense|other|preview_of_results|related_literature_positioning|related_literature_positioning|related_literature_positioning|contribution_statement|related_literature_positioning|other,15,10,new_evidence|new_method,False,False,inside_introduction,True,mixed,True,mixed,placebo|alternative_specification|subsample|other,clustered,True,mixed,True,False,False,moderate,6,6,0.8,4,14,1536,False,False,True,False,0.833,0.082154,claude-opus-5-5
136 +jof_12734,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,3,3,-1,0,motivation|related_literature_positioning|motivation|preview_of_approach|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement,7,2,new_theory_or_model|new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,False,main_text_section,placebo|alternative_specification|other,unclear,True,unclear,True,False,False,moderate,8,8,0.8,4,3,624,False,False,False,False,0.857,0.068098,claude-opus-5-5
137 +jof_12735,True,empirical,financial_intermediation_banking,reduced_form_regression,4,True,True,False,True,True,False,True,0,-1,0,1,-1,2,motivation|policy_or_broader_implication|preview_of_approach|identification_defense|preview_of_approach|identification_defense|preview_of_results|preview_of_results|preview_of_results|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning|roadmap,13,7,new_identification|new_evidence,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|alternative_sample_period,unclear,True,no,True,True,True,moderate,6,5,0.8,7,9,1393,False,True,False,False,1.0,0.07699,claude-opus-5-5
138 +jof_12736,True,empirical,asset_pricing,survey_or_experiment,4,True,True,False,True,True,False,False,0,-1,1,1,-1,-1,motivation|literature_gap|preview_of_approach|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|related_literature_positioning|preview_of_results|contribution_statement,12,5,new_evidence,False,False,both,True,mixed,False,main_text_section,other,mixed,True,no,True,False,False,moderate,7,7,0.8,5,6,822,False,True,False,True,0.857,0.09151,claude-opus-5-5
139 +jof_12737,True,mixed_theory_empirical,asset_pricing,other,1,False,False,False,True,True,False,True,-1,-1,0,1,-1,0,motivation|literature_gap|related_literature_positioning|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|contribution_statement|contribution_statement|roadmap,12,1,new_method|new_evidence,True,True,inside_introduction,True,unclear,True,within_results,alternative_specification|falsification,unclear,False,unclear,True,False,False,moderate,6,6,0.72,3,8,970,False,False,False,False,1.0,0.079714,claude-opus-5-5
140 +jof_12738,True,empirical,asset_pricing,reduced_form_regression,2,False,False,False,True,True,False,True,-1,-1,0,1,-1,0,preview_of_approach|preview_of_results|motivation|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning|preview_of_results|contribution_statement|related_literature_positioning|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|contribution_statement,14,1,new_data_or_measure|new_evidence,False,False,inside_introduction,True,stars_and_tstats,True,internet_appendix,alternative_specification|subsample|alternative_measure|other,unclear,False,no_figures,True,False,True,moderate,7,7,0.85,5,5,888,False,True,False,False,1.0,0.07219,claude-opus-5-5
141 +jof_12739,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,1,False,True,False,False,True,False,True,0,-1,-1,1,-1,5,preview_of_results|preview_of_results|mechanism_or_theory_sketch|tension_or_puzzle|literature_gap|contribution_statement|preview_of_approach|preview_of_results|related_literature_positioning|policy_or_broader_implication|roadmap,11,4,new_theory_or_model|resolution_of_puzzle,True,True,both,True,unclear,True,mixed,sensitivity_to_parameters,none,True,no,True,True,True,moderate,8,8,0.82,6,7,797,False,False,False,False,1.0,0.084754,claude-opus-5-5
142 +jof_12740,True,empirical,corporate_finance,instrumental_variables,4,True,True,False,True,True,False,True,0,-1,1,1,-1,3,motivation|literature_gap|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|identification_defense|preview_of_approach|preview_of_results|tension_or_puzzle|preview_of_results|mechanism_or_theory_sketch|preview_of_results|contribution_statement|related_literature_positioning|roadmap,15,10,new_evidence|new_data_or_measure|new_identification,True,True,inside_introduction,True,unclear,True,mixed,subsample|alternative_specification|placebo,unclear,True,unclear,True,False,False,moderate,7,7,0.85,4,11,1419,False,True,False,False,0.857,0.080978,claude-opus-5-5
143 +jof_12743,True,empirical,behavioral_finance,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|tension_or_puzzle|motivation|preview_of_approach|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,14,5,new_evidence|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,alternative_measure|subsample|alternative_specification|other,unclear,True,unclear,True,False,True,moderate,5,5,0.8,5,8,1148,False,True,False,False,1.0,0.074502,claude-opus-5-5
144 +jof_12744,True,theoretical,asset_pricing,calibration_simulation,1,False,False,False,True,True,False,True,-1,-1,0,3,-1,1,motivation|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|preview_of_results|mechanism_or_theory_sketch|contribution_statement|policy_or_broader_implication|related_literature_positioning|related_literature_positioning|roadmap,13,5,new_theory_or_model|new_method,True,True,inside_introduction,True,unclear,False,within_results,sensitivity_to_parameters|other,none,True,no,True,False,False,moderate,6,6,0.8,16,9,1265,False,True,False,False,1.0,0.09693,claude-opus-5-5
145 +jof_12745,True,empirical,corporate_finance,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,1,1,-1,4,motivation|tension_or_puzzle|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|identification_defense|preview_of_results|preview_of_results|preview_of_results|policy_or_broader_implication|related_literature_positioning|contribution_statement|roadmap,14,7,new_evidence|new_theory_or_model,True,True,inside_introduction,True,unclear,False,none_seen,,unclear,True,unclear,True,False,False,moderate,7,7,0.8,5,8,1334,False,True,False,False,0.8,0.071142,claude-opus-5-5
146 +jof_12746,True,mixed_theory_empirical,asset_pricing,theoretical_model_analytical,2,False,True,True,True,True,False,True,0,1,1,2,-1,3,motivation|related_literature_positioning|tension_or_puzzle|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|policy_or_broader_implication,7,2,new_data_or_measure|new_theory_or_model|new_evidence,False,False,inside_introduction,True,stars_and_tstats,True,within_results,alternative_measure|subsample|alternative_specification|other,unclear,True,no,False,False,False,low,6,6,0.8,7,4,510,False,False,False,False,1.0,0.064662,claude-opus-5-5
147 +jof_12747,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,3,True,False,True,True,True,False,True,-1,0,2,1,-1,2,tension_or_puzzle|preview_of_results|preview_of_approach|preview_of_results|robustness_mention|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_approach|robustness_mention|identification_defense|preview_of_results|identification_defense|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|contribution_statement|policy_or_broader_implication|roadmap,20,16,new_evidence|new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,alternative_measure|subsample|alternative_specification,unclear,True,unclear,False,False,True,moderate,7,7,0.8,5,19,2771,False,True,False,False,1.0,0.095538,claude-opus-5-5
148 +jof_12748,True,mixed_theory_empirical,asset_pricing,calibration_simulation,2,False,True,False,True,True,True,False,0,-1,0,1,2,-1,motivation|literature_gap|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|identification_defense|robustness_mention|mechanism_or_theory_sketch|mechanism_or_theory_sketch|related_literature_positioning|roadmap,12,1,new_evidence|new_data_or_measure,True,True,inside_introduction,True,stars_and_tstats,True,mixed,alternative_specification|subsample|alternative_measure|alternative_sample_period|falsification,unclear,True,no_figures,True,False,False,moderate,7,7,0.8,6,10,1622,True,True,False,False,0.778,0.085154,claude-opus-5-5
149 +jof_12749,True,empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,0,2,-1,1,motivation|literature_gap|other|preview_of_results|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|tension_or_puzzle,12,6,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,within_results,placebo|alternative_specification|other,unclear,True,no_figures,True,False,False,moderate,7,7,0.8,7,8,1332,False,True,False,False,0.833,0.07761,claude-opus-5-5
150 +jof_12750,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,3,True,True,True,True,True,False,True,0,1,2,4,-1,1,motivation|preview_of_approach|tension_or_puzzle|mechanism_or_theory_sketch|identification_defense|robustness_mention|preview_of_results|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,11,5,new_evidence|new_data_or_measure|new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,False,unclear,False,main_text_section,alternative_specification|alternative_measure|alternative_sample_period|subsample|falsification,unclear,True,no,True,False,False,moderate,7,7,0.8,7,7,1096,False,True,False,False,1.0,0.07187,claude-opus-5-5
151 +jof_12751,True,empirical,corporate_finance,difference_in_differences,4,True,False,False,True,True,False,True,-1,-1,0,2,-1,0,motivation|literature_gap|tension_or_puzzle|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|tension_or_puzzle|policy_or_broader_implication|preview_of_approach|preview_of_results|robustness_mention|identification_defense|contribution_statement|related_literature_positioning,14,5,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,within_results,placebo|alternative_measure|alternative_specification|falsification|other,unclear,True,no,True,True,False,moderate,6,6,0.8,4,6,1128,False,True,False,False,0.833,0.07623,claude-opus-5-5
152 +jof_12752,True,empirical,corporate_finance,structural_estimation,2,False,True,False,True,True,True,True,0,-1,1,2,2,4,motivation|tension_or_puzzle|literature_gap|preview_of_approach|preview_of_results|roadmap,6,0,new_method|new_evidence|new_data_or_measure,True,True,separate_section,True,unclear,False,none_seen,,unclear,True,no_figures,False,False,False,low,6,6,0.75,5,2,328,True,False,False,False,0.857,0.062994,claude-opus-5-5
153 +jof_12753,True,empirical,financial_intermediation_banking,difference_in_differences,4,True,True,False,True,True,False,True,0,-1,2,3,-1,2,motivation|tension_or_puzzle|literature_gap|preview_of_results|contribution_statement|mechanism_or_theory_sketch|preview_of_results|roadmap,8,1,new_evidence|new_identification|new_data_or_measure,True,True,inside_introduction,True,unclear,False,within_results,subsample|alternative_specification,unclear,False,unclear,True,True,False,moderate,7,6,0.8,7,4,570,False,True,False,False,1.0,0.074014,claude-opus-5-5
154 +jof_12754,True,empirical,behavioral_finance,reduced_form_regression,2,False,True,False,True,True,False,False,0,-1,1,1,-1,-1,motivation|tension_or_puzzle|preview_of_results|preview_of_results|preview_of_approach|identification_defense|policy_or_broader_implication|policy_or_broader_implication|related_literature_positioning,9,0,new_evidence,False,False,inside_introduction,True,p_values,False,mixed,alternative_measure|alternative_specification|subsample,unclear,True,no_figures,True,False,True,moderate,6,6,0.8,3,5,910,False,True,False,False,1.0,0.06587,claude-opus-5-5
155 +jof_12755,True,mixed_theory_empirical,international_finance,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,2,1,-1,0,motivation|tension_or_puzzle|literature_gap|preview_of_results|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|roadmap,13,5,new_evidence|new_theory_or_model|new_data_or_measure,True,True,inside_introduction,True,unclear,True,appendix,alternative_measure|subsample|alternative_sample_period|alternative_specification,unclear,True,unclear,False,False,False,low,6,6,0.8,5,8,1165,False,True,False,False,0.438,0.080014,claude-opus-5-5
156 +jof_12756,True,theoretical,entrepreneurial_finance_private_equity,theoretical_model_analytical,1,False,False,False,False,True,False,True,-1,-1,-1,0,-1,0,motivation|related_literature_positioning|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|other|contribution_statement|related_literature_positioning,8,2,new_theory_or_model,False,False,inside_introduction,False,unclear,False,main_text_section,alternative_specification|sensitivity_to_parameters,none,False,no,True,False,False,moderate,6,6,0.85,5,3,700,False,True,False,False,1.0,0.064618,claude-opus-5-5
157 +jof_12757,True,theoretical,asset_pricing,theoretical_model_analytical,1,False,False,False,True,True,False,False,-1,-1,0,2,-1,-1,motivation|preview_of_approach|contribution_statement|preview_of_results|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|preview_of_results|related_literature_positioning|related_literature_positioning,10,0,new_theory_or_model,False,False,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,False,True,moderate,6,6,0.85,6,6,1204,False,False,False,False,1.0,0.063122,claude-opus-5-5
158 +jof_12758,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|tension_or_puzzle|literature_gap|contribution_statement|mechanism_or_theory_sketch|preview_of_results|robustness_mention,7,0,new_evidence|new_theory_or_model|resolution_of_puzzle,False,False,inside_introduction,True,stars_and_tstats,False,main_text_section,alternative_specification|alternative_measure|sensitivity_to_parameters|other,unclear,True,unclear,True,False,False,moderate,7,7,0.8,5,2,446,False,False,False,False,0.857,0.068698,claude-opus-5-5
159 +jof_12759,True,empirical,market_microstructure,reduced_form_regression,2,False,True,False,False,True,False,False,1,-1,-1,3,-1,-1,motivation|literature_gap|preview_of_approach|preview_of_approach|preview_of_results|identification_defense|identification_defense|preview_of_results|related_literature_positioning|contribution_statement|related_literature_positioning|contribution_statement|other,13,20,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,internet_appendix,alternative_measure|placebo|other,unclear,True,unclear,True,True,False,moderate,7,7,0.82,6,24,1757,False,True,False,False,1.0,0.083134,claude-opus-5-5
160 +jof_12760,True,theoretical,macro_finance,theoretical_model_analytical,4,True,True,False,True,True,False,False,0,-1,0,1,-1,-1,motivation|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|policy_or_broader_implication|other|related_literature_positioning|related_literature_positioning|roadmap,11,6,new_theory_or_model,True,False,inside_introduction,False,unclear,False,main_text_section,sensitivity_to_parameters|alternative_specification,none,False,no_figures,False,True,False,moderate,7,7,0.85,7,9,1101,False,True,False,False,0.857,0.064086,claude-opus-5-5
161 +jof_12761,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,True,False,False,True,False,True,0,-1,-1,4,-1,2,motivation|contribution_statement|preview_of_results|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement|related_literature_positioning|mechanism_or_theory_sketch|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,19,0,new_theory_or_model,True,True,inside_introduction,True,unclear,False,main_text_section,sensitivity_to_parameters|other,none,True,no,True,False,True,low,6,6,0.85,7,14,1815,False,True,False,False,1.0,0.093426,claude-opus-5-5
162 +jof_12762,True,replication_or_correction,financial_intermediation_banking,theoretical_model_analytical,1,False,False,True,True,True,False,True,-1,0,1,2,-1,3,motivation|tension_or_puzzle|preview_of_results|related_literature_positioning|contribution_statement,5,0,replication_or_correction|new_theory_or_model,False,False,inside_introduction,False,unclear,False,none_seen,,none,False,no_figures,True,True,False,low,5,5,0.85,5,1,241,False,False,False,False,1.0,0.047762,claude-opus-5-5
163 +jof_12763,True,mixed_theory_empirical,household_finance,structural_estimation,3,True,True,False,True,True,True,False,0,-1,3,1,2,-1,motivation|policy_or_broader_implication|literature_gap|preview_of_approach|identification_defense|identification_defense|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|related_literature_positioning|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,14,7,new_data_or_measure|new_evidence|new_theory_or_model,True,True,inside_introduction,True,stars_and_tstats,True,none_seen,,unclear,True,unclear,True,False,False,moderate,6,6,0.8,5,10,1608,False,True,False,True,0.667,0.080046,claude-opus-5-5
164 +jof_12764,True,theoretical,macro_finance,theoretical_model_analytical,1,False,False,False,True,True,False,True,-1,-1,0,1,-1,0,motivation|tension_or_puzzle|preview_of_approach|mechanism_or_theory_sketch,4,0,new_theory_or_model,False,False,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,False,True,low,4,4,0.8,4,1,216,False,False,False,False,1.0,0.053138,claude-opus-5-5
165 +jof_12765,True,mixed_theory_empirical,asset_pricing,structural_estimation,2,False,True,True,True,True,False,True,0,0,1,2,-1,3,tension_or_puzzle|literature_gap|contribution_statement|preview_of_approach|preview_of_results|contribution_statement|preview_of_results|related_literature_positioning|mechanism_or_theory_sketch|roadmap,10,1,new_theory_or_model|new_evidence|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|sensitivity_to_parameters|other,unclear,True,unclear,True,False,False,moderate,6,6,0.8,6,7,1034,False,True,False,False,0.875,0.072226,claude-opus-5-5
166 +jof_12766,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,False,0,-1,1,4,-1,-1,motivation|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|identification_defense|contribution_statement|related_literature_positioning|roadmap,9,3,new_theory_or_model|new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,within_results,alternative_measure|alternative_specification|subsample,unclear,True,unclear,False,False,False,moderate,7,7,0.8,5,5,744,False,True,False,False,1.0,0.070214,claude-opus-5-5
167 +jof_12768,True,empirical,asset_pricing,survey_or_experiment,4,True,True,False,True,True,False,False,0,-1,0,1,-1,-1,motivation|literature_gap|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|tension_or_puzzle|related_literature_positioning,9,1,new_evidence,False,False,inside_introduction,True,stars_and_tstats,True,none_seen,,clustered,True,mixed,True,False,True,moderate,6,6,0.82,4,5,792,False,False,True,False,0.75,0.06493,claude-opus-5-5
168 +jof_12769,True,empirical,market_microstructure,other,2,False,True,False,False,True,False,False,0,-1,-1,2,-1,-1,motivation|preview_of_approach|literature_gap|preview_of_results|preview_of_results|tension_or_puzzle|related_literature_positioning|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|policy_or_broader_implication|related_literature_positioning|preview_of_results|roadmap,14,1,new_evidence,True,True,inside_introduction,True,unclear,True,within_results,alternative_specification|alternative_measure,unclear,True,no_figures,True,True,True,moderate,7,7,0.8,6,11,1643,False,True,False,False,1.0,0.076742,claude-opus-5-5
169 +jof_12770,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,0,1,-1,5,motivation|literature_gap|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|contribution_statement|preview_of_results|related_literature_positioning|related_literature_positioning|roadmap,11,3,new_theory_or_model,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,False,False,moderate,6,6,0.85,6,8,1265,False,False,False,False,0.857,0.07281,claude-opus-5-5
170 +jof_12771,True,mixed_theory_empirical,asset_pricing,structural_estimation,3,True,True,False,True,True,False,True,0,-1,1,2,-1,3,motivation|tension_or_puzzle|preview_of_results|related_literature_positioning|preview_of_approach|identification_defense|motivation|robustness_mention|preview_of_approach|preview_of_results|contribution_statement|related_literature_positioning|mechanism_or_theory_sketch|related_literature_positioning|roadmap,15,4,new_theory_or_model|new_method|new_evidence,True,True,inside_introduction,True,confidence_intervals,True,mixed,sensitivity_to_parameters|alternative_specification|other,unclear,True,mixed,True,True,True,moderate,8,7,0.8,4,7,1140,False,True,False,False,0.92,0.097342,claude-opus-5-5
171 +jof_12772,True,empirical,asset_pricing,reduced_form_regression,2,False,False,False,False,True,False,True,-1,-1,-1,0,-1,1,motivation|preview_of_results|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|related_literature_positioning|related_literature_positioning|other|roadmap,10,2,new_evidence,True,True,inside_introduction,True,mixed,False,within_results,alternative_specification|alternative_measure|subsample,mixed,True,unclear,True,False,False,moderate,7,6,0.8,2,7,956,False,False,False,True,0.8,0.068262,claude-opus-5-5
172 +jof_12773,True,mixed_theory_empirical,macro_finance,calibration_simulation,2,False,True,False,True,True,False,False,0,-1,2,3,-1,-1,motivation|tension_or_puzzle|related_literature_positioning|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning,10,6,new_evidence|new_theory_or_model,False,False,inside_introduction,True,unclear,True,appendix,alternative_specification|subsample|other,unclear,True,unclear,False,False,False,moderate,6,6,0.8,7,7,1292,False,True,False,False,1.0,0.078122,claude-opus-5-5
173 +jof_12775,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,3,True,True,False,True,True,True,True,0,-1,1,2,4,0,motivation|related_literature_positioning|mechanism_or_theory_sketch|preview_of_approach|related_literature_positioning|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|robustness_mention|preview_of_results|identification_defense|identification_defense|contribution_statement|related_literature_positioning|roadmap,17,12,new_evidence|new_data_or_measure|new_theory_or_model,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|subsample|alternative_measure|falsification|other,unclear,True,no,True,False,True,moderate,7,7,0.8,5,14,1618,True,True,False,False,0.875,0.100474,claude-opus-5-5
174 +jof_12776,True,empirical,asset_pricing,other,2,False,False,False,True,True,False,True,-1,-1,1,0,-1,3,motivation|literature_gap|preview_of_results|preview_of_approach|preview_of_results|contribution_statement|contribution_statement|robustness_mention|related_literature_positioning|mechanism_or_theory_sketch|related_literature_positioning|roadmap,12,6,new_evidence|new_method,True,True,inside_introduction,True,mixed,False,mixed,alternative_specification|alternative_sample_period|subsample|alternative_measure|other,mixed,False,unclear,True,False,False,moderate,7,7,0.8,4,15,1472,False,False,False,True,0.75,0.078554,claude-opus-5-5
175 +jof_12777,True,empirical,asset_pricing,reduced_form_regression,4,True,True,False,True,True,True,True,0,-1,0,1,1,3,motivation|mechanism_or_theory_sketch|related_literature_positioning|tension_or_puzzle|preview_of_results|preview_of_results|preview_of_results|preview_of_results,8,-1,new_evidence|new_data_or_measure,False,False,inside_introduction,True,stars_and_tstats,True,mixed,placebo|alternative_measure|subsample|alternative_sample_period|other,clustered,True,unclear,True,True,False,moderate,6,6,0.8,5,6,870,True,False,True,True,1.0,0.072766,claude-opus-5-5
176 +jof_12778,True,mixed_theory_empirical,asset_pricing,theoretical_model_analytical,1,False,False,False,True,True,False,True,-1,-1,0,2,-1,3,preview_of_approach|contribution_statement|preview_of_approach|contribution_statement|related_literature_positioning|preview_of_results|preview_of_results|preview_of_results|roadmap|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning,14,2,new_theory_or_model|new_evidence|new_data_or_measure,True,False,inside_introduction,True,unclear,False,mixed,subsample|alternative_specification|other,bootstrap,False,unclear,False,False,False,low,5,5,0.8,4,11,1371,False,False,False,False,0.714,0.078242,claude-opus-5-5
177 +jof_12779,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|policy_or_broader_implication|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|contribution_statement|related_literature_positioning,9,6,new_theory_or_model,False,False,inside_introduction,False,unclear,False,mixed,other,none,False,no,True,True,False,moderate,5,5,0.85,5,7,1238,False,True,False,False,1.0,0.06739,claude-opus-5-5
178 +jof_12780,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,3,True,True,False,True,True,True,True,0,-1,1,2,2,5,motivation|preview_of_approach|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|robustness_mention|preview_of_results|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,18,9,new_evidence|new_theory_or_model,True,True,inside_introduction,True,unclear,True,within_results,subsample|alternative_measure|alternative_sample_period|alternative_specification|sensitivity_to_parameters,mixed,True,unclear,True,False,True,moderate,7,7,0.8,6,14,1887,True,True,False,False,1.0,0.082926,claude-opus-5-5
179 +jof_12782,True,survey_or_address,corporate_finance,theoretical_model_analytical,1,False,True,True,False,True,False,True,0,1,-1,2,-1,3,tension_or_puzzle|preview_of_results|contribution_statement|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|contribution_statement|policy_or_broader_implication|motivation|preview_of_approach|preview_of_results|preview_of_results|roadmap,14,1,new_theory_or_model|synthesis,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,True,True,moderate,7,7,0.85,6,23,1452,False,False,False,False,1.0,0.076034,claude-opus-5-5
180 +jof_12783,True,empirical,entrepreneurial_finance_private_equity,other,3,True,True,False,True,True,True,False,0,-1,2,1,2,-1,motivation|literature_gap|motivation|preview_of_approach|preview_of_results|identification_defense|identification_defense|related_literature_positioning|contribution_statement,9,5,new_evidence|new_method,False,False,inside_introduction,True,unclear,True,mixed,alternative_specification|subsample|alternative_sample_period|other,unclear,False,unclear,True,False,False,moderate,6,6,0.8,5,6,933,False,False,False,False,0.833,0.089882,claude-opus-5-5
181 +jof_12784,True,replication_or_correction,asset_pricing,calibration_simulation,2,False,True,False,True,True,True,True,0,-1,1,1,2,4,motivation|related_literature_positioning|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_approach,8,0,replication_or_correction|new_evidence,False,False,inside_introduction,True,confidence_intervals,True,within_results,sensitivity_to_parameters|alternative_specification,none,True,unclear,True,False,False,moderate,6,6,0.85,6,4,599,True,True,False,False,1.0,0.056774,claude-opus-5-5
182 +jof_12785,True,empirical,behavioral_finance,reduced_form_regression,2,False,True,False,False,True,False,False,0,-1,-1,1,-1,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning,9,11,new_evidence|resolution_of_puzzle,False,False,inside_introduction,True,unclear,True,mixed,subsample|alternative_measure,unclear,True,unclear,True,True,True,moderate,7,7,0.8,5,13,1306,False,False,False,False,1.0,0.074822,claude-opus-5-5
183 +jof_12786,True,replication_or_correction,asset_pricing,none,1,False,False,False,False,False,False,False,-1,-1,-1,-1,-1,-1,motivation|other|other|preview_of_results|other,5,-1,replication_or_correction,False,False,none,False,unclear,True,none_seen,,none,False,no_figures,True,False,False,low,4,2,0.85,0,13,342,False,False,False,False,1.0,0.032674,claude-opus-5-5
184 +jof_12817,True,empirical,asset_pricing,other,2,False,False,True,True,True,False,True,-1,1,2,3,-1,4,motivation|literature_gap|mechanism_or_theory_sketch|preview_of_approach|related_literature_positioning|contribution_statement|preview_of_results|preview_of_results|robustness_mention|policy_or_broader_implication|roadmap,11,2,new_method|new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,False,mixed,alternative_measure|alternative_specification|subsample,unclear,False,no,False,False,True,moderate,6,6,0.8,6,8,1179,False,False,False,False,1.0,0.07297,claude-opus-5-5
185 +jof_12818,True,empirical,asset_pricing,reduced_form_regression,4,True,False,False,True,True,False,False,-1,-1,1,0,-1,-1,motivation|literature_gap|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|policy_or_broader_implication|preview_of_results,10,0,new_evidence,False,False,inside_introduction,True,stars_and_tstats,False,within_results,alternative_sample_period|subsample|other,robust,True,unclear,True,True,False,moderate,8,8,0.8,4,5,818,False,False,False,True,0.5,0.077002,claude-opus-5-5
186 +jof_12819,True,empirical,household_finance,reduced_form_regression,4,True,False,False,True,True,False,True,-1,-1,0,1,-1,0,motivation|literature_gap|preview_of_approach|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|contribution_statement|related_literature_positioning|policy_or_broader_implication,10,3,new_evidence|new_data_or_measure,False,False,inside_introduction,True,stars_and_tstats,True,mixed,alternative_measure|alternative_specification|subsample|other,clustered,True,unclear,False,False,False,moderate,7,7,0.8,6,4,755,False,True,True,True,1.0,0.083918,claude-opus-5-5
187 +jof_12820,True,mixed_theory_empirical,corporate_finance,reduced_form_regression,3,True,False,True,False,True,False,True,-1,0,-1,1,-1,3,motivation|literature_gap|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|identification_defense|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning,13,9,new_evidence|new_theory_or_model,False,False,inside_introduction,True,unclear,False,mixed,alternative_measure,unclear,True,unclear,True,False,False,moderate,7,7,0.8,4,10,1556,False,True,False,False,1.0,0.08271,claude-opus-5-5
188 +jof_12821,True,empirical,household_finance,instrumental_variables,4,True,False,False,True,True,True,False,-1,-1,0,1,1,-1,motivation|mechanism_or_theory_sketch|literature_gap|contribution_statement|preview_of_approach|identification_defense|preview_of_results|related_literature_positioning|preview_of_approach|identification_defense|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|roadmap,17,1,new_identification|new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,mixed,placebo|subsample|falsification|alternative_measure,unclear,True,unclear,True,True,False,moderate,8,8,0.8,4,12,1773,True,True,False,False,0.667,0.092546,claude-opus-5-5
189 +jof_12822,True,empirical,behavioral_finance,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,1,2,-1,0,motivation|preview_of_approach|other|preview_of_approach|identification_defense|literature_gap|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|contribution_statement|mechanism_or_theory_sketch|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,18,9,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,none_seen,,unclear,True,unclear,True,True,False,moderate,6,6,0.8,4,16,1610,False,True,False,False,0.667,0.074258,claude-opus-5-5
190 +jof_12823,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,4,True,False,False,True,True,False,True,-1,-1,1,2,-1,3,motivation|literature_gap|mechanism_or_theory_sketch|related_literature_positioning|preview_of_approach|preview_of_results|identification_defense|preview_of_results|robustness_mention|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,13,4,new_evidence|new_data_or_measure|resolution_of_puzzle,True,True,inside_introduction,False,unclear,False,mixed,placebo|alternative_measure|subsample|alternative_specification,unclear,False,no_figures,True,False,True,moderate,7,7,0.8,4,7,953,False,True,False,False,1.0,0.069442,claude-opus-5-5
191 +jof_12824,True,replication_or_correction,asset_pricing,theoretical_model_analytical,1,False,False,True,False,False,False,True,-1,0,-1,-1,-1,0,related_literature_positioning|literature_gap|contribution_statement|other|other|policy_or_broader_implication,6,0,replication_or_correction,False,False,inside_introduction,False,unclear,False,none_seen,,none,False,no_figures,True,False,False,low,4,4,0.85,2,58,2132,False,False,False,False,0.0,0.059086,claude-opus-5-5
192 +jof_12827,True,empirical,asset_pricing,reduced_form_regression,2,False,True,True,True,True,True,True,0,0,1,1,1,4,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|roadmap,12,0,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,mixed,alternative_sample_period|alternative_specification|alternative_measure,unclear,True,no,True,False,False,moderate,5,5,0.8,7,7,1121,True,False,False,False,0.75,0.067482,claude-opus-5-5
193 +jof_12828,True,mixed_theory_empirical,household_finance,calibration_simulation,3,True,True,False,True,True,True,False,0,-1,2,1,3,-1,motivation|motivation|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|identification_defense|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,15,9,new_evidence|new_data_or_measure|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,within_results,alternative_specification|other,unclear,True,unclear,True,True,True,moderate,6,6,0.8,6,11,1279,True,True,False,False,0.75,0.075002,claude-opus-5-5
194 +jof_12829,True,empirical,financial_intermediation_banking,difference_in_differences,4,True,True,False,True,True,False,True,0,-1,2,2,-1,3,other|motivation|preview_of_approach|mechanism_or_theory_sketch|preview_of_approach|identification_defense|preview_of_results|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,11,9,new_evidence,True,True,inside_introduction,True,stars_and_tstats,True,mixed,placebo|alternative_specification|subsample|falsification,clustered,True,unclear,False,True,False,low,6,6,0.8,5,11,1017,False,False,True,True,1.0,0.070658,claude-opus-5-5
195 +jof_12830,True,empirical,household_finance,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,1,1,-1,3,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|identification_defense|robustness_mention|mechanism_or_theory_sketch|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning|contribution_statement,14,4,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,within_results,subsample|alternative_specification|other,unclear,True,no_figures,True,False,False,moderate,6,6,0.8,6,7,1269,False,True,False,False,0.667,0.071222,claude-opus-5-5
196 +jof_12831,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,1,2,-1,1,tension_or_puzzle|related_literature_positioning|preview_of_approach|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|contribution_statement|preview_of_results|related_literature_positioning|policy_or_broader_implication|roadmap,11,3,new_theory_or_model,True,True,both,False,unclear,False,main_text_section,sensitivity_to_parameters|other,none,False,no,True,False,True,moderate,5,5,0.85,6,7,1071,False,False,False,False,1.0,0.065526,claude-opus-5-5
197 +jof_12832,True,empirical,corporate_finance,difference_in_differences,4,True,False,False,True,True,False,True,-1,-1,0,1,-1,4,motivation|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning|contribution_statement|literature_gap|identification_defense|roadmap,11,3,new_evidence|new_identification,True,True,inside_introduction,True,unclear,False,none_seen,,unclear,True,unclear,True,False,False,moderate,5,5,0.8,6,6,1120,False,True,False,False,0.667,0.06629,claude-opus-5-5
198 +jof_12833,True,mixed_theory_empirical,asset_pricing,structural_estimation,2,False,True,False,True,True,False,True,0,-1,1,3,-1,2,motivation|tension_or_puzzle|preview_of_approach|contribution_statement|preview_of_results|roadmap|preview_of_results|preview_of_results|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning,11,1,new_evidence|new_theory_or_model|resolution_of_puzzle,True,False,inside_introduction,True,unclear,False,none_seen,,unclear,True,unclear,True,False,True,moderate,7,7,0.8,5,6,791,False,False,False,True,1.0,0.06987,claude-opus-5-5
199 +jof_12834,True,theoretical,corporate_finance,theoretical_model_analytical,1,False,True,False,True,True,False,True,1,-1,1,2,-1,4,other|motivation|literature_gap|preview_of_results|policy_or_broader_implication|preview_of_approach,6,1,new_theory_or_model,False,False,inside_introduction,False,unclear,False,main_text_section,other,none,False,no,False,True,False,low,5,5,0.85,5,2,231,False,False,False,False,0.75,0.047858,claude-opus-5-5
200 +jof_12835,True,empirical,behavioral_finance,survey_or_experiment,4,True,True,False,True,True,False,False,1,-1,2,3,-1,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|policy_or_broader_implication|roadmap,8,0,new_evidence,True,True,none,True,mixed,True,mixed,alternative_specification|subsample|other,clustered,True,mixed,False,True,True,low,6,6,0.8,6,4,653,False,True,True,False,1.0,0.08355,claude-opus-5-5
201 +jof_12836,True,empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|tension_or_puzzle|contribution_statement|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|contribution_statement|other|roadmap,18,1,new_evidence,True,True,both,True,mixed,True,mixed,alternative_specification|alternative_measure|subsample,unclear,True,unclear,True,False,True,moderate,9,9,0.85,5,15,1777,False,False,False,False,0.833,0.086042,claude-opus-5-5
202 +jof_12837,True,empirical,corporate_finance,reduced_form_regression,3,True,False,False,False,True,False,True,-1,-1,-1,2,-1,0,motivation|literature_gap|contribution_statement|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|robustness_mention|preview_of_results|related_literature_positioning|roadmap,11,1,new_evidence,True,True,inside_introduction,True,unclear,True,within_results,alternative_measure|subsample,unclear,True,unclear,False,False,True,moderate,7,7,0.8,5,8,950,False,True,False,False,1.0,0.07049,claude-opus-5-5
203 +jof_12838,True,empirical,market_microstructure,other,2,False,True,False,True,True,False,False,0,-1,2,3,-1,-1,motivation|motivation|literature_gap|contribution_statement|preview_of_approach|related_literature_positioning|preview_of_approach|preview_of_results|identification_defense|preview_of_results|roadmap,11,1,new_evidence|new_method|new_data_or_measure,True,True,inside_introduction,False,unclear,False,none_seen,,unclear,True,unclear,True,False,True,moderate,5,5,0.75,7,5,860,False,True,False,False,1.0,0.08897,claude-opus-5-5
204 +jof_12839,True,theoretical,asset_pricing,calibration_simulation,1,False,True,False,True,True,False,True,0,-1,1,2,-1,3,motivation|literature_gap|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|other|preview_of_results|related_literature_positioning|related_literature_positioning|roadmap,11,0,new_theory_or_model,True,True,inside_introduction,True,unclear,True,none_seen,,none,True,unclear,True,False,False,moderate,5,5,0.8,4,14,919,False,False,False,False,1.0,0.07523,claude-opus-5-5
205 +jof_12840,True,empirical,market_microstructure,reduced_form_regression,3,True,True,False,True,True,True,True,0,-1,1,2,3,4,motivation|mechanism_or_theory_sketch|preview_of_results|robustness_mention|identification_defense|identification_defense|contribution_statement|preview_of_results|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,12,5,new_evidence,True,True,inside_introduction,True,unclear,True,within_results,subsample|placebo|other,unclear,True,mixed,True,True,False,moderate,7,7,0.8,6,6,1087,True,True,False,False,1.0,0.071218,claude-opus-5-5
206 +jof_12841,True,empirical,behavioral_finance,natural_experiment_event_study,5,True,True,False,True,True,True,False,0,-1,0,0,0,-1,motivation|tension_or_puzzle|preview_of_results|identification_defense|preview_of_approach|identification_defense|robustness_mention|preview_of_results|related_literature_positioning|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning|related_literature_positioning|mechanism_or_theory_sketch|roadmap,17,5,new_evidence|new_identification,True,True,inside_introduction,True,unclear,True,within_results,placebo|alternative_specification|alternative_measure,unclear,False,mixed,False,False,False,moderate,6,6,0.85,4,8,1311,False,True,False,False,1.0,0.079458,claude-opus-5-5
207 +jof_12842,True,mixed_theory_empirical,household_finance,structural_estimation,3,True,False,False,True,True,False,True,-1,-1,1,0,-1,3,motivation|tension_or_puzzle|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning|preview_of_approach|robustness_mention|related_literature_positioning|roadmap,12,0,new_evidence|new_theory_or_model,True,True,inside_introduction,True,unclear,False,within_results,sensitivity_to_parameters|alternative_specification,newey_west,True,no,True,True,False,moderate,6,6,0.8,4,7,1035,False,False,False,False,1.0,0.076254,claude-opus-5-5
208 +jof_12843,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,False,0,-1,1,2,-1,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_results|identification_defense|robustness_mention|related_literature_positioning|contribution_statement,12,5,new_evidence|new_theory_or_model,False,False,inside_introduction,True,stars_and_tstats,True,within_results,alternative_specification|alternative_measure|alternative_sample_period|subsample|other,clustered,True,no,True,False,False,moderate,5,5,0.82,6,6,881,False,True,True,True,1.0,0.07253,claude-opus-5-5
209 +jof_12844,True,mixed_theory_empirical,market_microstructure,natural_experiment_event_study,3,True,True,True,True,True,False,True,1,1,1,2,-1,3,motivation|identification_defense|mechanism_or_theory_sketch|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|roadmap,10,4,new_evidence|new_theory_or_model|new_identification,True,True,inside_introduction,False,unclear,False,none_seen,,unclear,True,unclear,True,False,False,moderate,6,6,0.8,4,6,1015,False,True,False,False,0.75,0.061738,claude-opus-5-5
210 +jof_12845,True,empirical,asset_pricing,reduced_form_regression,3,True,True,False,False,True,False,False,0,-1,-1,1,-1,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|identification_defense|related_literature_positioning|contribution_statement|roadmap,11,4,new_evidence|resolution_of_puzzle,True,True,inside_introduction,True,unclear,False,main_text_section,alternative_specification|subsample|alternative_measure,unclear,True,unclear,False,False,False,moderate,6,6,0.82,4,6,943,False,True,False,False,1.0,0.071982,claude-opus-5-5
211 +jof_12846,True,empirical,corporate_finance,difference_in_differences,4,True,False,False,True,True,False,False,-1,-1,1,1,-1,-1,other|motivation|literature_gap|preview_of_approach|preview_of_results|identification_defense|preview_of_results|preview_of_approach|preview_of_results|related_literature_positioning|preview_of_results|policy_or_broader_implication|preview_of_results|contribution_statement|related_literature_positioning,15,9,new_evidence,False,False,inside_introduction,True,unclear,True,mixed,placebo|falsification|subsample|alternative_specification|other,unclear,True,no,False,True,False,moderate,7,7,0.8,16,11,1282,False,True,False,False,0.8,0.108242,claude-opus-5-5
212 +jof_12847,True,empirical,household_finance,difference_in_differences,4,True,True,False,True,True,True,True,0,-1,1,1,1,4,motivation|literature_gap|preview_of_approach|preview_of_results|preview_of_approach|identification_defense|robustness_mention|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,11,6,new_evidence|new_identification|new_data_or_measure,True,True,inside_introduction,True,stars_and_tstats,True,mixed,alternative_specification|placebo|subsample|other,clustered,True,unclear,True,True,False,moderate,6,6,0.85,7,8,1137,True,False,True,True,0.8,0.075346,claude-opus-5-5
213 +jof_12848,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|literature_gap|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|policy_or_broader_implication|preview_of_results|policy_or_broader_implication|contribution_statement|related_literature_positioning|contribution_statement|related_literature_positioning,13,4,new_theory_or_model|resolution_of_puzzle,False,False,inside_introduction,False,unclear,False,mixed,alternative_specification|other,none,False,no,True,True,False,low,6,6,0.85,5,7,1227,False,False,False,False,1.0,0.06361,claude-opus-5-5
214 +jof_12849,True,empirical,financial_intermediation_banking,structural_estimation,2,False,True,False,True,True,True,True,0,-1,1,2,3,4,related_literature_positioning|literature_gap|preview_of_results|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|tension_or_puzzle|preview_of_approach|preview_of_results|robustness_mention|identification_defense|policy_or_broader_implication|other|roadmap,14,0,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|alternative_measure|sensitivity_to_parameters|falsification|other,unclear,False,unclear,True,False,True,moderate,7,7,0.8,6,9,1336,True,True,False,False,1.0,0.075526,claude-opus-5-5
215 +jof_12851,True,empirical,behavioral_finance,difference_in_differences,5,True,True,False,True,True,True,False,0,-1,1,2,2,-1,motivation|tension_or_puzzle|literature_gap|preview_of_approach|identification_defense|related_literature_positioning|contribution_statement|related_literature_positioning|contribution_statement|roadmap,10,3,new_evidence|new_identification,True,True,inside_introduction,True,unclear,False,mixed,alternative_specification|placebo|alternative_measure|alternative_sample_period|subsample|other,clustered,True,unclear,False,False,False,low,8,8,0.85,5,5,671,True,True,True,False,0.8,0.070878,claude-opus-5-5
216 +jof_12852,True,empirical,behavioral_finance,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,1,2,-1,3,motivation|literature_gap|preview_of_approach|identification_defense|preview_of_results|robustness_mention|preview_of_results|robustness_mention|contribution_statement|related_literature_positioning|related_literature_positioning|contribution_statement,12,4,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,mixed,alternative_specification|subsample|alternative_measure|other,unclear,True,unclear,True,False,False,moderate,7,7,0.8,4,6,1197,False,False,False,False,1.0,0.075798,claude-opus-5-5
217 +jof_12853,True,empirical,asset_pricing,machine_learning_textual,2,False,True,False,True,True,False,False,0,-1,0,1,-1,-1,motivation|literature_gap|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|mechanism_or_theory_sketch|contribution_statement|related_literature_positioning|roadmap,12,6,new_evidence|new_data_or_measure,True,True,inside_introduction,True,unclear,True,mixed,alternative_measure|sensitivity_to_parameters,unclear,True,unclear,True,False,False,moderate,7,7,0.8,5,7,1038,False,False,False,False,0.9,0.077478,claude-opus-5-5
218 +jof_12854,True,replication_or_correction,financial_econometrics,calibration_simulation,1,False,True,True,True,True,False,True,0,1,3,2,-1,4,contribution_statement|tension_or_puzzle|preview_of_approach,3,0,replication_or_correction|new_method,False,False,inside_introduction,False,unclear,False,within_results,sensitivity_to_parameters|other,none,True,no_figures,False,False,False,low,5,4,0.8,5,1,236,False,False,False,False,1.0,0.068406,claude-opus-5-5
219 +jof_12855,True,methodological_econometric,asset_pricing,other,1,False,True,True,True,True,False,True,0,1,2,3,-1,4,motivation|tension_or_puzzle|related_literature_positioning|literature_gap|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|contribution_statement|other|roadmap,13,6,new_method|new_evidence|resolution_of_puzzle,True,True,inside_introduction,False,mixed,False,within_results,alternative_measure|alternative_sample_period|alternative_specification,mixed,False,mixed,True,False,True,moderate,6,6,0.8,6,8,1286,True,False,False,False,0.6,0.078374,claude-opus-5-5
220 +jof_12856,True,theoretical,corporate_finance,theoretical_model_analytical,2,False,True,False,True,True,False,True,0,-1,1,5,-1,0,motivation|tension_or_puzzle|literature_gap|preview_of_approach|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|policy_or_broader_implication|roadmap,14,0,new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,False,unclear,False,main_text_section,other,none,False,no_figures,True,False,False,moderate,6,6,0.85,6,9,1659,False,False,False,False,1.0,0.074054,claude-opus-5-5
221 +jof_12857,True,theoretical,asset_pricing,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,1,3,-1,1,motivation|tension_or_puzzle|preview_of_approach|literature_gap|related_literature_positioning|mechanism_or_theory_sketch,6,0,new_theory_or_model,False,False,inside_introduction,False,unclear,False,none_seen,,none,False,no_figures,True,False,True,low,4,4,0.85,5,1,213,False,False,False,False,0.8,0.04359,claude-opus-5-5
222 +jof_12862,True,methodological_econometric,corporate_finance,theoretical_model_analytical,2,False,False,False,True,True,True,True,-1,-1,0,1,2,3,motivation|motivation|literature_gap|preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|policy_or_broader_implication|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning,14,6,new_theory_or_model|new_method|new_identification,False,False,inside_introduction,True,unclear,True,main_text_section,alternative_specification|sensitivity_to_parameters,none,True,no,False,False,False,moderate,6,6,0.8,7,12,1591,True,False,False,False,1.0,0.103098,claude-opus-5-5
223 +jof_12863,True,empirical,market_microstructure,natural_experiment_event_study,3,True,True,False,True,True,False,True,0,-1,1,2,-1,3,literature_gap|motivation|preview_of_results|identification_defense|preview_of_results|robustness_mention|identification_defense|preview_of_results|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|roadmap,12,0,new_evidence|new_identification,True,True,separate_section,True,mixed,True,main_text_section,alternative_measure|alternative_specification|subsample|other,unclear,True,mixed,True,False,True,moderate,8,8,0.82,5,5,833,False,True,False,False,1.0,0.073118,claude-opus-5-5
224 +jof_12864,True,mixed_theory_empirical,market_microstructure,structural_estimation,2,False,True,False,True,True,True,True,0,-1,3,2,1,4,motivation|preview_of_approach|mechanism_or_theory_sketch|literature_gap|policy_or_broader_implication|literature_gap|preview_of_results|preview_of_results|policy_or_broader_implication|related_literature_positioning|related_literature_positioning|roadmap,12,5,new_evidence|new_theory_or_model,True,True,inside_introduction,True,stars_and_tstats,True,mixed,alternative_specification|sensitivity_to_parameters,mixed,True,mixed,True,True,False,moderate,6,6,0.8,6,8,1196,False,False,True,True,0.857,0.074786,claude-opus-5-5
225 +jof_12865,True,replication_or_correction,market_microstructure,none,1,False,False,False,False,False,False,False,-1,-1,-1,-1,-1,-1,other,1,-1,replication_or_correction,False,False,none,False,unclear,False,none_seen,,none,False,no_figures,False,False,False,low,1,0,0.95,0,2,25,False,False,False,False,0.0,0.020102,claude-opus-5-5
226 +jof_12866,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,0,1,-1,3,motivation|policy_or_broader_implication|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|policy_or_broader_implication,7,-1,new_theory_or_model,False,False,separate_section,False,unclear,False,none_seen,,none,False,no,True,True,False,moderate,6,6,0.85,4,4,651,False,False,False,False,1.0,0.062698,claude-opus-5-5
227 +jof_12867,True,empirical,entrepreneurial_finance_private_equity,instrumental_variables,4,True,True,False,True,True,True,False,0,-1,0,1,1,-1,motivation|tension_or_puzzle|literature_gap|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|contribution_statement|contribution_statement|policy_or_broader_implication|policy_or_broader_implication|related_literature_positioning,14,3,new_evidence|new_identification|new_data_or_measure,False,False,inside_introduction,True,unclear,True,within_results,subsample|falsification|other,unclear,True,unclear,True,True,False,moderate,6,6,0.85,5,7,1096,True,True,False,False,0.75,0.07329,claude-opus-5-5
228 +jof_12868,True,empirical,financial_intermediation_banking,difference_in_differences,4,True,True,False,True,True,False,True,0,-1,2,1,-1,5,motivation|literature_gap|preview_of_results|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|identification_defense|mechanism_or_theory_sketch|identification_defense|related_literature_positioning|tension_or_puzzle|roadmap,13,0,new_evidence|new_identification|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,placebo|alternative_measure|alternative_specification|subsample|other,unclear,True,unclear,True,True,False,moderate,7,7,0.8,6,7,974,False,False,False,False,0.833,0.075338,claude-opus-5-5
229 +jof_12869,True,empirical,macro_finance,instrumental_variables,4,True,True,False,True,True,False,True,0,-1,1,2,-1,1,motivation|literature_gap|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|identification_defense|robustness_mention|identification_defense|preview_of_results|roadmap,13,0,new_evidence|new_data_or_measure|new_method,True,True,both,False,mixed,False,mixed,alternative_specification|placebo|subsample|alternative_measure,robust,True,mixed,True,False,False,moderate,10,10,0.8,5,6,881,False,False,False,True,0.857,0.10301,claude-opus-5-5
230 +jof_12870,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,0,1,-1,3,mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|robustness_mention|identification_defense|preview_of_results|literature_gap|related_literature_positioning|contribution_statement,10,4,new_data_or_measure|new_evidence,False,False,inside_introduction,True,stars_and_tstats,False,mixed,alternative_specification|alternative_measure|subsample|alternative_sample_period|other,unclear,True,no,True,False,False,moderate,7,7,0.85,4,5,779,False,False,False,True,0.8,0.067146,claude-opus-5-5
231 +jof_12871,True,theoretical,asset_pricing,theoretical_model_analytical,2,False,False,True,True,True,False,True,-1,0,2,3,-1,4,motivation|related_literature_positioning|preview_of_approach|literature_gap|mechanism_or_theory_sketch|contribution_statement|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|roadmap,11,1,new_theory_or_model,True,True,inside_introduction,False,unclear,False,appendix,alternative_specification|other,none,False,no_figures,True,True,False,moderate,5,5,0.85,6,8,1357,False,True,False,False,1.0,0.069062,claude-opus-5-5
232 +jof_12876,True,empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,0,2,-1,4,motivation|tension_or_puzzle|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|robustness_mention|policy_or_broader_implication|preview_of_results|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,12,5,new_evidence|new_method,True,True,inside_introduction,True,unclear,True,main_text_section,alternative_specification|alternative_sample_period|alternative_measure|subsample|other,unclear,True,unclear,True,False,False,moderate,6,6,0.8,5,8,1122,False,True,False,False,1.0,0.07539,claude-opus-5-5
233 +jof_12877,True,theoretical,asset_pricing,theoretical_model_analytical,1,False,False,False,True,True,False,True,-1,-1,3,0,-1,2,motivation|tension_or_puzzle|contribution_statement|preview_of_approach|preview_of_results|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_approach|preview_of_results|related_literature_positioning|related_literature_positioning|related_literature_positioning,15,1,new_theory_or_model|new_method,False,False,inside_introduction,True,unclear,False,mixed,sensitivity_to_parameters|other,none,True,no,True,False,True,low,5,5,0.85,6,17,1764,False,True,False,False,0.833,0.077022,claude-opus-5-5
234 +jof_12878,True,empirical,asset_pricing,reduced_form_regression,2,False,True,False,False,True,False,False,0,-1,-1,2,-1,-1,motivation|literature_gap|preview_of_approach|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|contribution_statement,12,7,new_data_or_measure|new_evidence,False,False,inside_introduction,False,unclear,False,mixed,alternative_specification|other,unclear,False,no,True,True,True,moderate,7,7,0.8,5,8,1401,False,False,False,False,1.0,0.080282,claude-opus-5-5
235 +jof_12879,True,empirical,entrepreneurial_finance_private_equity,reduced_form_regression,2,False,True,True,True,True,True,False,1,0,3,3,0,-1,tension_or_puzzle|motivation|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|preview_of_approach|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,12,6,new_evidence,True,True,inside_introduction,True,unclear,True,none_seen,,unclear,False,unclear,True,False,False,moderate,6,6,0.8,5,8,1412,True,True,False,False,1.0,0.077842,claude-opus-5-5
236 +jof_12880,True,empirical,entrepreneurial_finance_private_equity,difference_in_differences,4,True,True,False,True,True,True,True,0,-1,1,2,5,6,other|other|motivation|tension_or_puzzle|preview_of_approach|preview_of_results|robustness_mention|identification_defense|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement|roadmap,13,5,new_evidence,True,True,inside_introduction,True,unclear,True,within_results,alternative_measure|alternative_sample_period|alternative_specification|other,unclear,True,unclear,True,True,False,moderate,7,6,0.85,7,7,770,False,True,False,False,0.833,0.074078,claude-opus-5-5
237 +jof_12882,True,theoretical,market_microstructure,theoretical_model_analytical,4,True,True,False,True,True,False,True,0,-1,0,1,-1,3,motivation|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|related_literature_positioning|preview_of_results|preview_of_approach|preview_of_results|related_literature_positioning|policy_or_broader_implication,10,2,new_theory_or_model,False,False,both,False,unclear,False,main_text_section,other,none,False,no,True,True,False,moderate,5,5,0.85,5,4,807,False,False,False,False,0.8,0.05785,claude-opus-5-5
238 +jof_12883,True,methodological_econometric,financial_econometrics,machine_learning_textual,1,False,True,True,True,True,False,True,0,1,0,3,-1,1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|preview_of_approach|robustness_mention|preview_of_approach|contribution_statement|mechanism_or_theory_sketch|preview_of_approach|other|related_literature_positioning|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,17,6,new_method|new_evidence,True,True,inside_introduction,False,stars_and_tstats,False,within_results,sensitivity_to_parameters|alternative_measure|alternative_specification|subsample,unclear,False,mixed,True,False,False,moderate,6,5,0.8,4,12,1777,False,True,False,False,1.0,0.06979,claude-opus-5-5
239 +jof_12884,True,empirical,corporate_finance,difference_in_differences,3,True,False,True,True,True,False,True,-1,0,1,1,-1,0,motivation|literature_gap|contribution_statement|preview_of_results|tension_or_puzzle|preview_of_approach|preview_of_approach|identification_defense|identification_defense|identification_defense|identification_defense|preview_of_results|preview_of_results|related_literature_positioning|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|robustness_mention|contribution_statement|related_literature_positioning|policy_or_broader_implication|roadmap,23,17,new_evidence|new_data_or_measure|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,mixed,subsample|alternative_measure|placebo|alternative_sample_period|alternative_specification,unclear,True,unclear,True,True,False,moderate,8,8,0.8,5,21,2129,False,True,False,False,0.75,0.087454,claude-opus-5-5
240 +jof_12885,True,empirical,asset_pricing,machine_learning_textual,2,False,False,False,True,True,True,True,-1,-1,0,0,1,3,motivation|literature_gap|contribution_statement|motivation|other|other|other|other|other|preview_of_approach|preview_of_results|identification_defense|preview_of_results|preview_of_results|mechanism_or_theory_sketch|robustness_mention|robustness_mention|mechanism_or_theory_sketch|preview_of_approach|roadmap,20,2,new_evidence|new_data_or_measure,True,True,both,True,stars_and_tstats,True,mixed,alternative_measure|alternative_specification|subsample|other,unclear,True,unclear,True,False,True,moderate,8,8,0.85,14,20,2701,True,False,False,False,0.8,0.096634,claude-opus-5-5
241 +jof_12888,True,theoretical,asset_pricing,theoretical_model_analytical,1,False,True,False,True,True,False,True,0,-1,1,4,-1,0,motivation|tension_or_puzzle|preview_of_approach|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement|related_literature_positioning|roadmap,12,6,new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,False,False,moderate,6,6,0.85,6,8,1325,False,True,False,False,1.0,0.063894,claude-opus-5-5
242 +jof_12889,True,empirical,corporate_finance,reduced_form_regression,1,False,True,False,True,True,False,True,0,-1,0,1,-1,7,motivation|literature_gap|preview_of_approach|preview_of_results|related_literature_positioning|related_literature_positioning|preview_of_approach|identification_defense|preview_of_results|contribution_statement|policy_or_broader_implication|roadmap,12,56,new_data_or_measure|new_evidence,True,True,inside_introduction,True,unclear,True,mixed,alternative_measure|subsample|alternative_specification|alternative_sample_period,unclear,True,unclear,True,False,True,moderate,7,7,0.8,9,64,1177,False,True,False,False,0.8,0.084382,claude-opus-5-5
243 +jof_12890,True,empirical,asset_pricing,structural_estimation,2,False,False,False,True,True,False,False,-1,-1,0,0,-1,-1,motivation|related_literature_positioning|literature_gap|policy_or_broader_implication|preview_of_approach|tension_or_puzzle|preview_of_results|preview_of_results|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning|mechanism_or_theory_sketch|roadmap,15,6,new_evidence,True,True,inside_introduction,True,unclear,True,mixed,alternative_specification|subsample|alternative_sample_period,unclear,False,unclear,False,False,False,moderate,7,7,0.8,7,9,1525,False,True,False,False,0.8,0.08155,claude-opus-5-5
244 +jof_12893,True,empirical,corporate_finance,reduced_form_regression,3,True,False,False,False,True,True,True,-1,-1,-1,1,2,1,motivation|literature_gap|preview_of_approach|preview_of_results|motivation|mechanism_or_theory_sketch|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|preview_of_results|policy_or_broader_implication|robustness_mention|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,19,19,new_evidence|new_data_or_measure,True,True,inside_introduction,True,stars_and_tstats,True,mixed,alternative_specification|other,clustered,False,unclear,True,False,True,moderate,6,6,0.85,3,22,2477,False,True,True,True,1.0,0.088154,claude-opus-5-5
245 +jof_12894,True,mixed_theory_empirical,asset_pricing,theoretical_model_analytical,1,False,False,False,True,True,False,True,-1,-1,1,2,-1,0,other|tension_or_puzzle|literature_gap|preview_of_results|contribution_statement|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_approach|related_literature_positioning|related_literature_positioning|related_literature_positioning,11,3,new_theory_or_model|new_evidence,False,False,inside_introduction,True,stars_and_tstats,False,none_seen,,newey_west,True,unclear,True,False,True,low,6,6,0.8,3,10,1416,False,False,False,False,1.0,0.076654,claude-opus-5-5
246 +jof_12895,True,empirical,household_finance,survey_or_experiment,2,False,True,False,True,True,False,False,0,-1,0,1,-1,-1,motivation|tension_or_puzzle|preview_of_approach|contribution_statement|preview_of_results|preview_of_results|preview_of_results|related_literature_positioning|contribution_statement|policy_or_broader_implication|roadmap,11,0,new_evidence|new_data_or_measure,True,True,inside_introduction,False,unclear,True,none_seen,,unclear,True,no_figures,False,False,False,moderate,5,5,0.82,5,7,948,False,False,False,False,1.0,0.070646,claude-opus-5-5
247 +jof_12896,True,empirical,financial_intermediation_banking,reduced_form_regression,4,True,False,False,True,True,False,True,-1,-1,1,0,-1,6,motivation|preview_of_results|preview_of_approach|identification_defense|preview_of_results|identification_defense|tension_or_puzzle|preview_of_results|mechanism_or_theory_sketch|policy_or_broader_implication,10,0,new_evidence,False,False,separate_section,True,stars_and_tstats,True,within_results,alternative_specification|alternative_measure|subsample|other,clustered,True,unclear,True,True,False,moderate,8,8,0.82,8,4,769,False,False,True,True,0.9,0.09503,claude-opus-5-5
248 +jof_12897,True,empirical,corporate_finance,reduced_form_regression,4,True,True,False,True,True,False,True,0,-1,0,2,-1,5,preview_of_approach|preview_of_approach|preview_of_results|preview_of_results|robustness_mention|related_literature_positioning|identification_defense|identification_defense|contribution_statement|contribution_statement|other|roadmap,12,7,new_evidence|new_data_or_measure|new_identification,True,True,inside_introduction,True,unclear,True,within_results,alternative_specification|subsample|placebo|falsification|alternative_measure,unclear,True,mixed,False,False,False,moderate,7,7,0.8,6,11,1183,False,True,False,False,0.75,0.076398,claude-opus-5-5
249 +jof_12898,True,empirical,asset_pricing,other,1,False,True,False,True,True,False,False,0,-1,1,2,-1,-1,motivation|other|preview_of_approach|related_literature_positioning|literature_gap|literature_gap|contribution_statement|preview_of_results|roadmap,9,3,new_evidence|new_method|new_data_or_measure,True,True,inside_introduction,True,unclear,False,mixed,subsample|alternative_measure|sensitivity_to_parameters,none,True,unclear,True,False,False,low,5,5,0.8,5,5,637,False,True,False,False,1.0,0.066314,claude-opus-5-5
250 +jof_12899,True,empirical,corporate_finance,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,0,1,-1,3,motivation|literature_gap|preview_of_approach|preview_of_results|preview_of_results|robustness_mention|mechanism_or_theory_sketch|identification_defense|robustness_mention|preview_of_results|contribution_statement|policy_or_broader_implication|roadmap,13,7,new_evidence|new_data_or_measure,True,True,separate_section,True,unclear,True,mixed,alternative_specification|falsification|placebo|subsample,unclear,True,unclear,True,False,False,moderate,7,7,0.8,5,9,1349,False,False,False,False,1.0,0.075254,claude-opus-5-5
251 +jof_12900,True,mixed_theory_empirical,financial_intermediation_banking,structural_estimation,3,True,True,False,True,True,False,True,0,-1,2,3,-1,2,motivation|policy_or_broader_implication|contribution_statement|preview_of_approach|preview_of_results|preview_of_approach|mechanism_or_theory_sketch|identification_defense|preview_of_results|tension_or_puzzle|related_literature_positioning|related_literature_positioning|related_literature_positioning,13,1,new_evidence|new_identification|new_method|new_theory_or_model,False,False,inside_introduction,True,confidence_intervals,False,mixed,subsample|alternative_measure|other,bootstrap,True,unclear,True,False,False,moderate,7,7,0.8,5,9,1297,False,True,False,False,0.75,0.074746,claude-opus-5-5
252 +jof_12901,True,empirical,asset_pricing,other,2,False,False,False,True,True,False,True,-1,-1,1,2,-1,0,motivation|literature_gap|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_approach|preview_of_results|roadmap|robustness_mention,10,-1,new_method|new_data_or_measure|new_evidence,True,True,inside_introduction,True,stars_and_tstats,False,mixed,alternative_specification|alternative_measure|subsample|other,unclear,False,unclear,False,False,False,moderate,7,7,0.75,5,15,893,False,False,False,False,0.75,0.103042,claude-opus-5-5
253 +jof_12902,True,empirical,financial_intermediation_banking,difference_in_differences,4,True,True,False,True,True,False,False,0,-1,2,2,-1,-1,motivation|literature_gap|preview_of_approach|contribution_statement|identification_defense|contribution_statement|mechanism_or_theory_sketch|preview_of_results|preview_of_results,9,1,new_evidence|new_data_or_measure|new_identification,False,False,inside_introduction,True,unclear,True,within_results,alternative_measure|other,unclear,True,unclear,True,True,True,moderate,5,5,0.8,6,5,918,False,True,False,False,0.6,0.065058,claude-opus-5-5
254 +jof_12903,True,empirical,market_microstructure,reduced_form_regression,4,True,True,False,True,True,False,True,0,-1,1,1,-1,3,motivation|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|robustness_mention|preview_of_results|identification_defense|roadmap,11,-1,new_evidence|new_data_or_measure,True,True,separate_section,True,unclear,True,within_results,placebo|falsification|other,unclear,True,mixed,True,True,False,moderate,6,6,0.8,6,9,1401,False,False,False,False,1.0,0.082834,claude-opus-5-5
255 +jof_12906,True,survey_or_address,behavioral_finance,theoretical_model_analytical,2,False,True,True,True,True,False,True,0,2,3,4,-1,3,motivation|literature_gap|related_literature_positioning|mechanism_or_theory_sketch|tension_or_puzzle|contribution_statement|preview_of_approach|preview_of_results|literature_gap|mechanism_or_theory_sketch|mechanism_or_theory_sketch|roadmap,12,5,new_theory_or_model|synthesis,True,False,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,False,True,moderate,6,5,0.8,5,10,966,False,False,False,False,1.0,0.057706,claude-opus-5-5
256 +jof_12908,True,empirical,corporate_finance,regression_discontinuity,4,True,True,False,True,True,False,False,0,-1,2,1,-1,-1,motivation|tension_or_puzzle|related_literature_positioning|preview_of_approach|identification_defense|preview_of_results|robustness_mention|preview_of_approach|preview_of_results|preview_of_results|other|policy_or_broader_implication|roadmap,13,0,new_evidence|new_identification,True,True,inside_introduction,True,unclear,True,mixed,alternative_measure|alternative_specification|subsample,unclear,True,unclear,True,False,False,moderate,6,6,0.85,5,7,1105,False,False,False,False,1.0,0.07211,claude-opus-5-5
257 +jof_12909,True,empirical,macro_finance,other,2,False,False,False,True,True,False,True,-1,-1,0,1,-1,0,motivation|related_literature_positioning|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|identification_defense|robustness_mention,9,1,new_data_or_measure|new_evidence,False,False,both,True,unclear,True,mixed,sensitivity_to_parameters|alternative_measure|subsample|other,clustered,True,unclear,False,False,False,moderate,6,6,0.8,4,6,1015,False,True,True,False,1.0,0.071354,claude-opus-5-5
258 +jof_12910,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,True,False,True,True,False,True,0,-1,1,2,-1,0,motivation|contribution_statement|mechanism_or_theory_sketch|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|tension_or_puzzle|related_literature_positioning,9,0,new_evidence|new_theory_or_model|resolution_of_puzzle|new_data_or_measure,False,False,inside_introduction,True,unclear,True,mixed,alternative_specification|alternative_measure|other,unclear,False,unclear,True,False,False,moderate,6,6,0.8,17,5,1055,False,False,False,False,0.667,0.096066,claude-opus-5-5
259 +jof_12947,True,empirical,asset_pricing,difference_in_differences,4,True,True,False,True,True,True,False,0,-1,0,1,2,-1,motivation|tension_or_puzzle|literature_gap|preview_of_approach|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|robustness_mention|related_literature_positioning|literature_gap|related_literature_positioning|related_literature_positioning|contribution_statement|identification_defense,16,9,new_evidence|new_identification,False,False,inside_introduction,True,unclear,True,main_text_section,alternative_sample_period|placebo|falsification|subsample|alternative_measure|alternative_specification,unclear,True,unclear,True,False,False,moderate,8,8,0.82,4,10,1364,True,True,False,False,0.778,0.088466,claude-opus-5-5
260 +jof_12948,True,mixed_theory_empirical,macro_finance,calibration_simulation,2,False,True,False,True,True,False,False,1,-1,2,3,-1,-1,motivation|tension_or_puzzle|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|policy_or_broader_implication|other|contribution_statement|related_literature_positioning|related_literature_positioning|related_literature_positioning|related_literature_positioning|roadmap,16,6,new_theory_or_model|synthesis,True,True,inside_introduction,True,unclear,True,within_results,sensitivity_to_parameters|alternative_sample_period,none,True,mixed,True,True,True,moderate,6,6,0.82,6,12,1472,False,True,False,False,1.0,0.08181,claude-opus-5-5
261 +jof_12949,True,mixed_theory_empirical,asset_pricing,reduced_form_regression,2,False,False,False,False,True,False,True,-1,-1,-1,0,-1,1,motivation|preview_of_results|mechanism_or_theory_sketch|preview_of_approach|mechanism_or_theory_sketch|tension_or_puzzle|preview_of_results|related_literature_positioning|contribution_statement|related_literature_positioning|contribution_statement|roadmap,12,12,new_evidence|new_theory_or_model,True,True,inside_introduction,True,unclear,False,mixed,alternative_specification|alternative_measure|other,unclear,True,unclear,True,False,True,moderate,7,7,0.8,5,14,889,False,True,False,False,0.75,0.068794,claude-opus-5-5
262 +jof_12950,True,empirical,asset_pricing,reduced_form_regression,3,True,False,False,False,True,False,True,-1,-1,-1,0,-1,3,motivation|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_approach|robustness_mention|preview_of_results|preview_of_results|preview_of_results|identification_defense|robustness_mention|contribution_statement|related_literature_positioning|roadmap,13,5,new_evidence|resolution_of_puzzle,True,True,inside_introduction,True,unclear,True,appendix,alternative_specification|subsample|alternative_measure|alternative_sample_period,unclear,True,unclear,True,False,False,moderate,7,7,0.82,4,8,1176,False,True,False,False,0.9,0.082114,claude-opus-5-5
263 +jof_12951,True,methodological_econometric,financial_econometrics,calibration_simulation,1,False,True,True,True,True,False,True,0,0,2,4,-1,1,motivation|literature_gap|tension_or_puzzle|preview_of_approach|contribution_statement|related_literature_positioning|preview_of_results,7,2,new_method|new_evidence,False,False,inside_introduction,True,unclear,False,within_results,sensitivity_to_parameters|subsample|other,none,False,unclear,True,False,False,moderate,4,4,0.8,5,5,910,True,False,False,False,1.0,0.061198,claude-opus-5-5
264 +jof_12952,True,empirical,household_finance,difference_in_differences,4,True,True,False,True,True,False,True,0,-1,2,3,-1,4,motivation|literature_gap|identification_defense|preview_of_results|preview_of_approach|preview_of_results|identification_defense|related_literature_positioning|contribution_statement|related_literature_positioning|contribution_statement,11,4,new_evidence,False,False,inside_introduction,True,unclear,True,within_results,alternative_specification|subsample,unclear,True,unclear,True,True,True,moderate,5,5,0.82,6,6,1168,False,True,False,False,0.6,0.070362,claude-opus-5-5
265 +jof_12953,True,empirical,market_microstructure,reduced_form_regression,2,False,True,False,False,True,False,True,0,-1,-1,1,-1,4,motivation|literature_gap|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|preview_of_results|preview_of_results|policy_or_broader_implication|related_literature_positioning|related_literature_positioning,12,9,new_evidence|new_data_or_measure,False,False,inside_introduction,True,unclear,True,none_seen,,unclear,True,unclear,True,True,False,moderate,6,6,0.8,5,10,1455,False,False,False,False,1.0,0.08019,claude-opus-5-5
266 +jof_12954,True,empirical,household_finance,difference_in_differences,4,True,True,False,True,True,False,False,0,-1,1,1,-1,-1,motivation|preview_of_approach|preview_of_results|preview_of_results|preview_of_results|mechanism_or_theory_sketch|robustness_mention|mechanism_or_theory_sketch|preview_of_results,9,-1,new_evidence,False,False,separate_section,True,mixed,True,within_results,subsample|alternative_specification|other,unclear,False,unclear,True,True,True,moderate,7,6,0.85,4,8,1429,False,False,False,False,0.833,0.074294,claude-opus-5-5
267 +jof_12955,True,empirical,corporate_finance,difference_in_differences,4,True,True,False,True,True,False,True,0,-1,0,1,-1,4,motivation|motivation|mechanism_or_theory_sketch|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning|policy_or_broader_implication|roadmap,9,3,new_evidence,True,True,inside_introduction,True,unclear,False,within_results,falsification|alternative_specification,unclear,True,unclear,True,True,False,moderate,6,6,0.8,6,7,926,False,True,False,False,1.0,0.060726,claude-opus-5-5
268 +jof_12958,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,1,False,False,False,False,True,False,True,-1,-1,-1,3,-1,4,motivation|tension_or_puzzle|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|mechanism_or_theory_sketch|contribution_statement|contribution_statement|related_literature_positioning|preview_of_results|related_literature_positioning|other|policy_or_broader_implication|roadmap,15,5,new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,True,unclear,False,mixed,sensitivity_to_parameters|alternative_specification,none,False,no,True,True,False,moderate,5,5,0.85,5,11,1050,False,True,False,False,0.857,0.071414,claude-opus-5-5
269 +jof_12959,True,mixed_theory_empirical,financial_intermediation_banking,reduced_form_regression,3,True,True,True,True,True,False,False,0,1,3,2,-1,-1,motivation|contribution_statement|mechanism_or_theory_sketch|preview_of_results|identification_defense|preview_of_results|related_literature_positioning|literature_gap|roadmap,9,3,new_theory_or_model|new_evidence,True,True,inside_introduction,True,unclear,True,none_seen,,unclear,True,unclear,False,False,False,low,5,5,0.8,4,5,833,False,True,False,False,1.0,0.06227,claude-opus-5-5
270 +jof_12961,True,theoretical,corporate_finance,theoretical_model_analytical,3,False,True,False,True,True,False,True,0,-1,0,3,-1,4,motivation|tension_or_puzzle|mechanism_or_theory_sketch|mechanism_or_theory_sketch|preview_of_results|preview_of_results|related_literature_positioning|related_literature_positioning|roadmap,9,5,new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,False,unclear,False,mixed,other,none,False,no,True,True,True,moderate,6,6,0.85,5,8,1573,False,False,False,False,1.0,0.069038,claude-opus-5-5
271 +jof_12962,True,replication_or_correction,corporate_finance,reduced_form_regression,3,True,True,True,True,True,False,True,0,1,1,2,-1,3,motivation|preview_of_approach|preview_of_results|mechanism_or_theory_sketch|mechanism_or_theory_sketch|tension_or_puzzle|mechanism_or_theory_sketch|other|related_literature_positioning|contribution_statement|roadmap,11,8,replication_or_correction|new_method,True,True,inside_introduction,True,stars_and_tstats,False,none_seen,,mixed,True,unclear,True,False,True,moderate,5,5,0.8,4,10,1171,False,False,True,False,1.0,0.061086,claude-opus-5-5
272 +jof_12964,True,empirical,financial_intermediation_banking,reduced_form_regression,3,True,True,False,True,True,False,True,0,-1,1,2,-1,3,motivation|literature_gap|preview_of_approach|preview_of_results|literature_gap|preview_of_approach|preview_of_results|policy_or_broader_implication|identification_defense|robustness_mention|mechanism_or_theory_sketch|related_literature_positioning|contribution_statement|roadmap,14,5,new_evidence|new_data_or_measure|new_identification,True,True,inside_introduction,True,unclear,False,mixed,alternative_specification|alternative_sample_period|other,clustered,True,unclear,True,True,False,moderate,6,6,0.82,6,7,994,False,True,False,False,1.0,0.07131,claude-opus-5-5
273 +jof_12965,True,mixed_theory_empirical,international_finance,calibration_simulation,2,False,False,False,True,True,False,False,-1,-1,1,0,-1,-1,motivation|tension_or_puzzle|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|preview_of_results|contribution_statement|related_literature_positioning|related_literature_positioning|roadmap,12,5,new_evidence|new_theory_or_model|resolution_of_puzzle,True,True,inside_introduction,True,unclear,False,within_results,alternative_measure|alternative_specification|sensitivity_to_parameters,unclear,True,unclear,True,True,False,moderate,5,5,0.8,4,9,1280,False,True,False,False,1.0,0.069258,claude-opus-5-5
274 +jof_12966,True,empirical,corporate_finance,reduced_form_regression,4,True,False,False,True,True,False,True,-1,-1,1,0,-1,1,motivation|preview_of_results|identification_defense|mechanism_or_theory_sketch|preview_of_approach|preview_of_results|preview_of_approach|preview_of_results|related_literature_positioning|literature_gap|related_literature_positioning|contribution_statement,12,4,new_evidence|new_data_or_measure|new_identification,False,False,inside_introduction,True,unclear,False,mixed,alternative_specification|alternative_measure|other,unclear,True,unclear,True,True,False,moderate,8,8,0.8,4,5,829,False,True,False,False,0.6,0.076042,claude-opus-5-5
275 +jof_12967,True,theoretical,financial_intermediation_banking,theoretical_model_analytical,3,False,True,False,True,True,False,False,0,-1,1,3,-1,-1,motivation|preview_of_approach|mechanism_or_theory_sketch|preview_of_results|preview_of_results|policy_or_broader_implication|mechanism_or_theory_sketch|preview_of_results|related_literature_positioning|roadmap,10,0,new_theory_or_model,True,True,inside_introduction,False,unclear,False,none_seen,,none,False,no,True,True,True,moderate,5,5,0.85,5,6,1030,False,False,False,False,1.0,0.058382,claude-opus-5-5
276 +jof_12968,True,empirical,household_finance,instrumental_variables,4,True,True,False,True,True,True,False,0,-1,1,2,2,-1,motivation|policy_or_broader_implication|literature_gap|identification_defense|preview_of_approach|identification_defense|preview_of_results|robustness_mention|mechanism_or_theory_sketch|related_literature_positioning|related_literature_positioning|contribution_statement|roadmap,13,6,new_evidence|new_identification|new_data_or_measure,True,True,both,True,unclear,True,main_text_section,alternative_specification|subsample|alternative_measure|other,unclear,True,no,True,False,False,moderate,8,8,0.8,5,7,1166,True,True,False,False,0.6,0.085874,claude-opus-5-5
277 +jof_12969,True,empirical,market_microstructure,natural_experiment_event_study,3,True,True,True,True,True,False,False,1,1,2,3,-1,-1,motivation|tension_or_puzzle|preview_of_approach|preview_of_results|preview_of_results|other|preview_of_approach|preview_of_results|related_literature_positioning|related_literature_positioning|identification_defense|contribution_statement|policy_or_broader_implication|roadmap,14,5,new_evidence|new_data_or_measure|new_identification,True,True,inside_introduction,True,unclear,True,mixed,alternative_measure|falsification|other,unclear,True,unclear,True,False,False,moderate,7,6,0.8,5,7,1020,False,True,False,False,0.8,0.069154,claude-opus-5-5
modified data/annotations/annotations_table.parquet +0 −0

Binary file not shown.

modified data/annotations/api_ledger.jsonl +144 −0
@@ -144,3 +144,147 @@
144 144 {"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"}
145 145 {"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"}
146 146 {"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"}
147 +{"ts": "2026-11-03T23:06:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7685, "output": 2124, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074014, "latency_s": 17.91, "request_id": "req_011CggqH7KnVn4Q3EjH2b4Gu", "stop_reason": "end_turn", "cache_key": "d4d4982bee0fd149333ed70c69aaf49c1dee22ade5f50263e4a6782878081d67", "code_version": "87c73cd-dirty", "doc_id": "jof_12753"}
148 +{"ts": "2026-11-03T23:06:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7487, "output": 2347, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077682, "latency_s": 19.53, "request_id": "req_011CggqH1ssBDghq937Byzkm", "stop_reason": "end_turn", "cache_key": "f1a99b7136bece4df6d2d0c5765c8095d2a771474c414d367847555c170cc26a", "code_version": "87c73cd-dirty", "doc_id": "jof_12750"}
149 +{"ts": "2026-11-03T23:06:29+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6612, "output": 2195, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071142, "latency_s": 18.1, "request_id": "req_011CggqHH9dcjTCJ3Jp2ymKg", "stop_reason": "end_turn", "cache_key": "3ffce7e2b1d262c25bdd53e011ded0f92152e560e0456f5d71118c6aed6d845f", "code_version": "87c73cd-dirty", "doc_id": "jof_12745"}
150 +{"ts": "2026-11-03T23:06:37+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 10054, "output": 2796, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.09693, "latency_s": 23.6, "request_id": "req_011CggqHRA69YfTqZ2qWpPsC", "stop_reason": "end_turn", "cache_key": "7a21a8e4b0bcc69196024989ad795d1bd784a26421878fe1f1971404ab839fa3", "code_version": "87c73cd-dirty", "doc_id": "jof_12744"}
151 +{"ts": "2026-11-03T23:06:44+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7927, "output": 2100, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074502, "latency_s": 17.25, "request_id": "req_011CggqJTkdqUhgQRaTD2TLv", "stop_reason": "end_turn", "cache_key": "885e16dd4123f8380aa168e299916677cc673f1c78a2a06d7c63aa3b61caf2f3", "code_version": "87c73cd-dirty", "doc_id": "jof_12743"}
152 +{"ts": "2026-11-03T23:06:45+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9665, "output": 2285, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.085154, "latency_s": 18.42, "request_id": "req_011CggqJSWyBNF1hAkmG3TQX", "stop_reason": "end_turn", "cache_key": "e73f90c2d1c768f9cce3e7be8885bba27f4723aaf1b1260942b2244898245a34", "code_version": "87c73cd-dirty", "doc_id": "jof_12748"}
153 +{"ts": "2026-11-03T23:06:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5907, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064662, "latency_s": 16.66, "request_id": "req_011CggqJcWV9Jv5pNpJxSFhb", "stop_reason": "end_turn", "cache_key": "2e7f7ee4ec3461e390ee7997fe53115493f9a24844b6d088ca5ec88892f731a8", "code_version": "87c73cd-dirty", "doc_id": "jof_12746"}
154 +{"ts": "2026-11-03T23:06:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6179, "output": 2018, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06587, "latency_s": 18.76, "request_id": "req_011CggqKAV5xYLQsCH9gaujz", "stop_reason": "end_turn", "cache_key": "47257b571304bb2e171c60026012a1a3442993d967783d8bbf4d54bdc96d279a", "code_version": "87c73cd-dirty", "doc_id": "jof_12754"}
155 +{"ts": "2026-11-03T23:07:01+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6182, "output": 1880, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.063122, "latency_s": 15.45, "request_id": "req_011CggqKoAxwiQXKFBm5JRfa", "stop_reason": "end_turn", "cache_key": "52851d7bd0d9124739586217bd63b0bddc7421cc0768630797fc4b76e1bdba14", "code_version": "87c73cd-dirty", "doc_id": "jof_12757"}
156 +{"ts": "2026-11-03T23:07:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9380, "output": 2241, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.083134, "latency_s": 18.69, "request_id": "req_011CggqKjnp2Bz2RepMqRRRb", "stop_reason": "end_turn", "cache_key": "3a15c72ab0f87b59a3fa6b1dd0c38cc64d0c7cd083ff40d473ecbb8aaeaacd8c", "code_version": "87c73cd-dirty", "doc_id": "jof_12759"}
157 +{"ts": "2026-11-03T23:07:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5360, "output": 2038, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062994, "latency_s": 17.44, "request_id": "req_011CggqKr2cigGf4YaUW4fBG", "stop_reason": "end_turn", "cache_key": "8309841fa059de2eb2c85989720a04cef0b89ae8bf2db5b56f3a7083c4f042a5", "code_version": "87c73cd-dirty", "doc_id": "jof_12752"}
158 +{"ts": "2026-11-03T23:07:14+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8663, "output": 2230, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080046, "latency_s": 18.59, "request_id": "req_011CggqLYrHGW6qRXytiHu9C", "stop_reason": "end_turn", "cache_key": "654fb9e305590f84219c10a164a06a8e0e992b7b6f3b3592b19ae1b1b9f012ac", "code_version": "87c73cd-dirty", "doc_id": "jof_12763"}
159 +{"ts": "2026-11-03T23:07:20+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4106, "output": 2370, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064618, "latency_s": 19.39, "request_id": "req_011CggqLwaX1ACqpRRCHpaDU", "stop_reason": "end_turn", "cache_key": "783a341cf7ecd522f0cd48a9db3e1a500a117b42347e800549d0c85097ecfcac", "code_version": "87c73cd-dirty", "doc_id": "jof_12756"}
160 +{"ts": "2026-11-03T23:07:22+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8390, "output": 2283, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080014, "latency_s": 18.66, "request_id": "req_011CggqM7zakNpsuj1KdsaEz", "stop_reason": "end_turn", "cache_key": "8540515f7cc1e4b1e05f9bf0718b5db203a7e2254ac31ce665a7029addf00db0", "code_version": "87c73cd-dirty", "doc_id": "jof_12755"}
161 +{"ts": "2026-11-03T23:07:30+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 11241, "output": 3417, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.114098, "latency_s": 27.01, "request_id": "req_011CggqM8rvEuFUd3t8bW1JQ", "stop_reason": "end_turn", "cache_key": "652a1a515d481af8b9a42a3f97d9bdf9a217bfcc143a4f32e81a1cd3e19e1f10", "code_version": "87c73cd-dirty", "doc_id": "jof_12761"}
162 +{"ts": "2026-11-03T23:07:31+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6351, "output": 2125, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068698, "latency_s": 17.34, "request_id": "req_011CggqMveUkwtMKV3FMVLW2", "stop_reason": "end_turn", "cache_key": "d00a527123aa739d540a9c652d017e00d716308ddc7acac385ebb16e70230870", "code_version": "87c73cd-dirty", "doc_id": "jof_12758"}
163 +{"ts": "2026-11-03T23:07:36+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4806, "output": 1656, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.053138, "latency_s": 14.08, "request_id": "req_011CggqNW87nM8jy5NKVkRHS", "stop_reason": "end_turn", "cache_key": "651a22bab9b94ab82ce0d816f19b3ba6f4d086d352c89bd663bd746a1f50bb4e", "code_version": "87c73cd-dirty", "doc_id": "jof_12764"}
164 +{"ts": "2026-11-03T23:07:38+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4963, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.064086, "latency_s": 18.24, "request_id": "req_011CggqNPAg7Jb6TyY4G9qb8", "stop_reason": "end_turn", "cache_key": "15431e3aa08ae0fd1f99b60fbd4a4a722783db3954ccbe5479963cbfe98cd419", "code_version": "87c73cd-dirty", "doc_id": "jof_12760"}
165 +{"ts": "2026-11-03T23:07:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 3142, "output": 1720, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.047762, "latency_s": 15.3, "request_id": "req_011CggqP8hJc42PebUm37ZTR", "stop_reason": "end_turn", "cache_key": "d48376b438e39e925f28934eb867195090cc119d3dc0fa803c5b2ec2bf0d5d33", "code_version": "87c73cd-dirty", "doc_id": "jof_12762"}
166 +{"ts": "2026-11-03T23:07:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8085, "output": 2145, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076034, "latency_s": 16.85, "request_id": "req_011CggqPEHuN4Fnve9xYkmT2", "stop_reason": "end_turn", "cache_key": "7a002050bd1e11dfc3aaa21fa1d2b4dcd0076eef8caa0ea7f936220ebad2aa63", "code_version": "87c73cd-dirty", "doc_id": "jof_12782"}
167 +{"ts": "2026-11-03T23:07:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8092, "output": 2179, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076742, "latency_s": 19.05, "request_id": "req_011CggqPYdiPkQ99kXXDFk76", "stop_reason": "end_turn", "cache_key": "9a853263dbbd0967af274478400265b59fb2c8e4e093dbac44749d816aba5edf", "code_version": "87c73cd-dirty", "doc_id": "jof_12769"}
168 +{"ts": "2026-11-03T23:08:00+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 10062, "output": 2815, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.097342, "latency_s": 21.76, "request_id": "req_011CggqPjU63RJeaRiqxi2n4", "stop_reason": "end_turn", "cache_key": "b08414a5d5c5f032f4057e9357372be0dee529718bf1344baecaa1b956922ead", "code_version": "87c73cd-dirty", "doc_id": "jof_12771"}
169 +{"ts": "2026-11-03T23:08:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8847, "output": 2097, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078122, "latency_s": 17.24, "request_id": "req_011CggqQGkkh38ZFtxi7NPsr", "stop_reason": "end_turn", "cache_key": "b5f17c509d784f17eefaa6904dccef60456c165983dad4d3d6c2731402242574", "code_version": "87c73cd-dirty", "doc_id": "jof_12773"}
170 +{"ts": "2026-11-03T23:08:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6192, "output": 2135, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068262, "latency_s": 17.68, "request_id": "req_011CggqQTyCQqugpGngRYxWr", "stop_reason": "end_turn", "cache_key": "ad2b078317f8b54857863ed2aa97503df085fec159a82397761a9c97eb2644e2", "code_version": "87c73cd-dirty", "doc_id": "jof_12772"}
171 +{"ts": "2026-11-03T23:08:13+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6410, "output": 2189, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070214, "latency_s": 17.86, "request_id": "req_011CggqQxUx2n31r4CQYqMnf", "stop_reason": "end_turn", "cache_key": "97cb73be2b9f18ccf2e44e1c250633c83d69d28ec48db0e4cca3867dfe56c7b5", "code_version": "87c73cd-dirty", "doc_id": "jof_12766"}
172 +{"ts": "2026-11-03T23:08:17+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7423, "output": 2087, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072226, "latency_s": 16.62, "request_id": "req_011CggqRLX1KiMN2ygSHLMHR", "stop_reason": "end_turn", "cache_key": "75f27005e9688ca4cfc2fa79e5eaa16c0a3c3c9df3ac2a5fad0ed2bacf1e2551", "code_version": "87c73cd-dirty", "doc_id": "jof_12765"}
173 +{"ts": "2026-11-03T23:08:20+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8562, "output": 2160, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078242, "latency_s": 17.18, "request_id": "req_011CggqRYnguqXyWf2iMEUMR", "stop_reason": "end_turn", "cache_key": "eb69a89ff0952fd932969fc30399d3b362d1b5664412953f1bab31247a4d7c26", "code_version": "87c73cd-dirty", "doc_id": "jof_12778"}
174 +{"ts": "2026-11-03T23:08:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8060, "output": 2276, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078554, "latency_s": 18.81, "request_id": "req_011CggqRmRTbEKkYxM768Y6K", "stop_reason": "end_turn", "cache_key": "c0167b441b0060e75bf38951919a4929908409eb84b379ec0ea10d157381fd63", "code_version": "87c73cd-dirty", "doc_id": "jof_12776"}
175 +{"ts": "2026-11-03T23:08:30+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6629, "output": 1881, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06493, "latency_s": 16.39, "request_id": "req_011CggqSHKnsFwpsf8BVeY35", "stop_reason": "end_turn", "cache_key": "dfffda13a509ab3c11be43005b0cb64b3e942626f884949540cb9858c1f1820e", "code_version": "87c73cd-dirty", "doc_id": "jof_12768"}
176 +{"ts": "2026-11-03T23:08:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8199, "output": 1961, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07281, "latency_s": 15.34, "request_id": "req_011CggqSZshvXubj6wdBKhxF", "stop_reason": "end_turn", "cache_key": "05a1bd71a760dcd19473dddacb6fd47a8073eae0e374a5526ff6409526922f8c", "code_version": "87c73cd-dirty", "doc_id": "jof_12770"}
177 +{"ts": "2026-11-03T23:08:37+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 1240, "output": 1346, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.032674, "latency_s": 11.47, "request_id": "req_011CggqTADwX9bNcbmhS1Tt5", "stop_reason": "end_turn", "cache_key": "ff2aa4007251317f6ea95e62fe62444b86246bafe32bd6745867567c0eedbf83", "code_version": "87c73cd-dirty", "doc_id": "jof_12786"}
178 +{"ts": "2026-11-03T23:08:39+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8844, "output": 2355, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08327, "latency_s": 18.41, "request_id": "req_011CggqSpDuxLiQRKv6SLfC9", "stop_reason": "end_turn", "cache_key": "5cbc8b8f878c546bc89f3b0a76e77f1a599afbfc02ff42993933bff677640195", "code_version": "87c73cd-dirty", "doc_id": "jof_12775"}
179 +{"ts": "2026-11-03T23:08:47+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7817, "output": 2138, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074822, "latency_s": 17.2, "request_id": "req_011CggqTVVZVWmDnsKQRaApB", "stop_reason": "end_turn", "cache_key": "fd6f7cba94072fbdb0c270dbb4ed2922e6a72ec3bbab780eb52b2cf779c5f745", "code_version": "87c73cd-dirty", "doc_id": "jof_12785"}
180 +{"ts": "2026-11-03T23:08:50+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8132, "output": 2184, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077002, "latency_s": 17.26, "request_id": "req_011CggqThnVdAkgQomuh7jLL", "stop_reason": "end_turn", "cache_key": "9d04601f4376306a0fd449122bcade4d09ac773b3da832824526c716ab2b21fd", "code_version": "87c73cd-dirty", "doc_id": "jof_12818"}
181 +{"ts": "2026-11-03T23:08:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 10593, "output": 2469, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.092546, "latency_s": 19.4, "request_id": "req_011CggqU1ZKgq8Gy1WyCxH3S", "stop_reason": "end_turn", "cache_key": "c4d7af1ca4a3f704b3a1cf28e9fc7681276f874e10ab7a56f049f113aa3275dd", "code_version": "87c73cd-dirty", "doc_id": "jof_12821"}
182 +{"ts": "2026-11-03T23:08:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7618, "output": 2075, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.072766, "latency_s": 17.31, "request_id": "req_011CggqUBFjLBYv7wWYtZ7Ei", "stop_reason": "end_turn", "cache_key": "7666c87ce830de1edd907db1b083478629d59037a4b154b10c2c470673e53a41", "code_version": "87c73cd-dirty", "doc_id": "jof_12777"}
183 +{"ts": "2026-11-03T23:09:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7671, "output": 2139, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074258, "latency_s": 19.31, "request_id": "req_011CggqUmKbfb79TRYA2e7De", "stop_reason": "end_turn", "cache_key": "704f54e7e61f0d97dc788c26131eacde6051657ec2eb8c0e12bef9eea32f1fa3", "code_version": "87c73cd-dirty", "doc_id": "jof_12822"}
184 +{"ts": "2026-11-03T23:09:09+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8773, "output": 2352, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082926, "latency_s": 19.31, "request_id": "req_011CggqUyzLMpPtf6uLRAVKr", "stop_reason": "end_turn", "cache_key": "a35d5711921df19327d84ec38609827824c189a4e1a81cabf03a260cb943bf06", "code_version": "87c73cd-dirty", "doc_id": "jof_12780"}
185 +{"ts": "2026-11-03T23:09:14+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6454, "output": 2039, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06739, "latency_s": 17.42, "request_id": "req_011CggqVT8jCnp6CQwYbpJjv", "stop_reason": "end_turn", "cache_key": "f0fe699c8517548b5f055b43fa830d7f310a3b31f8917d7d92db15124f0c2876", "code_version": "87c73cd-dirty", "doc_id": "jof_12779"}
186 +{"ts": "2026-11-03T23:09:15+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9421, "output": 2272, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.083918, "latency_s": 18.81, "request_id": "req_011CggqVTdFnMLSuv1RpheyD", "stop_reason": "end_turn", "cache_key": "251ceef6deb5062f6ed1c78c28cf0cae1cb8155ec24061e0a15f11980f809994", "code_version": "87c73cd-dirty", "doc_id": "jof_12819"}
187 +{"ts": "2026-11-03T23:09:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6147, "output": 2203, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069442, "latency_s": 18.4, "request_id": "req_011CggqWCDpcyqTTXrKNNrps", "stop_reason": "end_turn", "cache_key": "4bbab0c1538c5cd05b140f4d34c878f3b7905731a29e2768cec70a6d992e0791", "code_version": "87c73cd-dirty", "doc_id": "jof_12823"}
188 +{"ts": "2026-11-03T23:09:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7734, "output": 2062, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07297, "latency_s": 17.75, "request_id": "req_011CggqWQoNfKa6nntCsaErj", "stop_reason": "end_turn", "cache_key": "cc9bc1fb238a779ecd51580db2319875f4d025ed3ad960de667ecd81a8404234", "code_version": "87c73cd-dirty", "doc_id": "jof_12817"}
189 +{"ts": "2026-11-03T23:09:30+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4705, "output": 1858, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.056774, "latency_s": 15.16, "request_id": "req_011CggqWrPF5w93NWWWPqFq8", "stop_reason": "end_turn", "cache_key": "44b97a75d78ecde77bedbf36dc8cd24b25993c91a380fa77dd63152dd17331ac", "code_version": "87c73cd-dirty", "doc_id": "jof_12784"}
190 +{"ts": "2026-11-03T23:09:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9214, "output": 2253, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08271, "latency_s": 18.76, "request_id": "req_011CggqWjkPiLmRJbaqij4mP", "stop_reason": "end_turn", "cache_key": "a7ad10f0ce1a66b55d0ca5ddc23ed62f9a2ac9b011fb8bf1cc6a66042463c698", "code_version": "87c73cd-dirty", "doc_id": "jof_12820"}
191 +{"ts": "2026-11-03T23:09:39+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6343, "output": 1646, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.059086, "latency_s": 14.49, "request_id": "req_011CggqXZ8h5Z2aYB5aR2ZEW", "stop_reason": "end_turn", "cache_key": "f7bea21d31e170ae50858c02f74a3528a0d73ef2d4134697f52e1dd93e53d80f", "code_version": "87c73cd-dirty", "doc_id": "jof_12824"}
192 +{"ts": "2026-11-03T23:09:45+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6561, "output": 2209, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071218, "latency_s": 18.29, "request_id": "req_011CggqXjPqPp19BLMSZMvX8", "stop_reason": "end_turn", "cache_key": "1d90fd10a16c50f99e6d5152d7fa6bc981b6dd1b30553546753f49dfcf569b13", "code_version": "87c73cd-dirty", "doc_id": "jof_12840"}
193 +{"ts": "2026-11-03T23:09:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6904, "output": 2073, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06987, "latency_s": 15.98, "request_id": "req_011CggqYdTS7WFZxfDL1WwFb", "stop_reason": "end_turn", "cache_key": "07fb5481ac8e3909608462ba9c7774a92d927980ab5b24118f903f0e696b77f3", "code_version": "87c73cd-dirty", "doc_id": "jof_12833"}
194 +{"ts": "2026-11-03T23:09:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7356, "output": 2462, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.079458, "latency_s": 23.65, "request_id": "req_011CggqYL4ASoYxTq9jTgLev", "stop_reason": "end_turn", "cache_key": "b52d80ce5109b2e900d2f54a5b72d3df9b08820aa9447c8b98e3b39b43da8e1c", "code_version": "87c73cd-dirty", "doc_id": "jof_12841"}
195 +{"ts": "2026-11-03T23:09:57+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6824, "output": 3044, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08897, "latency_s": 26.51, "request_id": "req_011CggqXyTSCGQhDCwQ9nDa7", "stop_reason": "end_turn", "cache_key": "ce6960dc55d960a84a6a597149fe7c6595619d259e9603c79e80ae16d343f9ed", "code_version": "87c73cd-dirty", "doc_id": "jof_12838"}
196 +{"ts": "2026-11-03T23:10:02+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6704, "output": 1934, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06629, "latency_s": 16.61, "request_id": "req_011CggqZ5VbKfALxaf53ubk2", "stop_reason": "end_turn", "cache_key": "e4337005b997fd6aea5d553c61822d8e2dff1f84a34203c1758996ca7a7ba51d", "code_version": "87c73cd-dirty", "doc_id": "jof_12832"}
197 +{"ts": "2026-11-03T23:10:14+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7557, "output": 2199, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075002, "latency_s": 17.38, "request_id": "req_011CggqZshU4Ub4QVJn4becS", "stop_reason": "end_turn", "cache_key": "43b8aad31f3cc75f46a190612d80f4a883d5816862639025d6c2b4b3f0654b91", "code_version": "87c73cd-dirty", "doc_id": "jof_12828"}
198 +{"ts": "2026-11-03T23:10:15+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6571, "output": 2179, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070658, "latency_s": 19.17, "request_id": "req_011CggqZq7gDxtEMcy7VJ9W2", "stop_reason": "end_turn", "cache_key": "d5323c916cc6337665f07f082c9cc93a2ae091a44cd1865fb54877ee205dec20", "code_version": "87c73cd-dirty", "doc_id": "jof_12829"}
199 +{"ts": "2026-11-03T23:10:15+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7270, "output": 2185, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073574, "latency_s": 18.11, "request_id": "req_011CggqZwGmiz6AggyVuWXij", "stop_reason": "end_turn", "cache_key": "9571185376b3300c161d181c60541157ca0595a30f12534a6ea63dd002cc01f8", "code_version": "87c73cd-dirty", "doc_id": "jof_12835"}
200 +{"ts": "2026-11-03T23:10:20+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6634, "output": 2158, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07049, "latency_s": 17.85, "request_id": "req_011CggqaJqYiE3XnKPqZDv5j", "stop_reason": "end_turn", "cache_key": "e54b86304f1bf53643ceea67eac2515c02888b8bb4eee539323759abf03a06de", "code_version": "87c73cd-dirty", "doc_id": "jof_12837"}
201 +{"ts": "2026-11-03T23:10:33+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9727, "output": 2317, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.086042, "latency_s": 18.33, "request_id": "req_011CggqbFPiUXWGbNJTiJhvW", "stop_reason": "end_turn", "cache_key": "4fd1a8fbc5fb0f8a2a98a053acf07d33242f93e61efbf736db11a872927484a1", "code_version": "87c73cd-dirty", "doc_id": "jof_12836"}
202 +{"ts": "2026-11-03T23:10:35+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7427, "output": 1849, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067482, "latency_s": 15.15, "request_id": "req_011CggqbdrpRSgxehFaNq2LC", "stop_reason": "end_turn", "cache_key": "7e4b15d719f74f91511de4987b5567dcf374804b0cb684d38fe14449db6c6be4", "code_version": "87c73cd-dirty", "doc_id": "jof_12827"}
203 +{"ts": "2026-11-03T23:10:37+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5729, "output": 2576, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07523, "latency_s": 22.13, "request_id": "req_011CggqbGrGUfGpf6dcpqk1x", "stop_reason": "end_turn", "cache_key": "cb6cee47a4f2ab19e457837a07091a576eebea18113f0cd7a30b3a63b056c231", "code_version": "87c73cd-dirty", "doc_id": "jof_12839"}
204 +{"ts": "2026-11-03T23:10:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7232, "output": 3008, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.089882, "latency_s": 26.29, "request_id": "req_011CggqbAHP9NWTM8AiVrTbS", "stop_reason": "end_turn", "cache_key": "21d16eff8739ebf0b6e86bebade33e5c9f7ed8c6f3ef746efa167d288940a884", "code_version": "87c73cd-dirty", "doc_id": "jof_12783"}
205 +{"ts": "2026-11-03T23:10:50+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6594, "output": 2167, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07051, "latency_s": 17.35, "request_id": "req_011Cggqcb3gWzwYn9zemPWva", "stop_reason": "end_turn", "cache_key": "62ba81ca78173b2f46d062cc1d05de1fa006f42a4d130c2cf52b86eebfabf73c", "code_version": "87c73cd-dirty", "doc_id": "jof_12830"}
206 +{"ts": "2026-11-03T23:10:52+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8006, "output": 2149, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075798, "latency_s": 16.93, "request_id": "req_011CggqcknaLYKyV5qaXGBa9", "stop_reason": "end_turn", "cache_key": "fbfe3f396da58bb28542c34ba1de4ce0c098478bb4e0de15f5609947478f35a5", "code_version": "87c73cd-dirty", "doc_id": "jof_12852"}
207 +{"ts": "2026-11-03T23:10:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6546, "output": 2195, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070878, "latency_s": 18.29, "request_id": "req_011CggqcusGDKrgnTTLAnHxk", "stop_reason": "end_turn", "cache_key": "10b5cc1686c6b22206a5f90408ba4bc1c86973ffdbfa137c87ac6f5475a102cf", "code_version": "87c73cd-dirty", "doc_id": "jof_12851"}
208 +{"ts": "2026-11-03T23:10:57+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8168, "output": 2094, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075346, "latency_s": 17.45, "request_id": "req_011Cggqd75DEFJ99j2rFhYPr", "stop_reason": "end_turn", "cache_key": "02ecec4fe3bdb05c21ba0c4e7462575ada873e71c06c8395e12708d5d1fc63c4", "code_version": "87c73cd-dirty", "doc_id": "jof_12847"}
209 +{"ts": "2026-11-03T23:11:08+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8125, "output": 2148, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076254, "latency_s": 17.15, "request_id": "req_011Cggqdu7ee5y7fCCxfDcH1", "stop_reason": "end_turn", "cache_key": "31f696fc4109ce47f84d1b10256f9c21c5bf19871d1f54cb252c9ec66c3819d8", "code_version": "87c73cd-dirty", "doc_id": "jof_12842"}
210 +{"ts": "2026-11-03T23:11:09+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7582, "output": 2043, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071982, "latency_s": 16.29, "request_id": "req_011Cggqe1Fk5NcN8jbtr1djP", "stop_reason": "end_turn", "cache_key": "d827e65197d93e7c2147ee703cbedc05cd52b47baff323086aa250993dc945fa", "code_version": "87c73cd-dirty", "doc_id": "jof_12845"}
211 +{"ts": "2026-11-03T23:11:12+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6198, "output": 1997, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065526, "latency_s": 16.66, "request_id": "req_011CggqeGKrAH8p6avJDdciw", "stop_reason": "end_turn", "cache_key": "77d2144cdcf32212b09787f35e32d51b2d88149df44c076c432fd56928105240", "code_version": "87c73cd-dirty", "doc_id": "jof_12831"}
212 +{"ts": "2026-11-03T23:11:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5311, "output": 1985, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061738, "latency_s": 16.85, "request_id": "req_011CggqfDDcDsDd4KhZx1ghi", "stop_reason": "end_turn", "cache_key": "0d5a25ec8d5728fda0842f54eaccb4d20c4b17439266bb45769b8d3761fd530d", "code_version": "87c73cd-dirty", "doc_id": "jof_12844"}
213 +{"ts": "2026-11-03T23:11:25+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 10292, "output": 3314, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.108242, "latency_s": 27.97, "request_id": "req_011CggqePv1tCxVYwG3KSGDU", "stop_reason": "end_turn", "cache_key": "137c40d9432146deb120ef991664163035d2441483944a8c83ee16393b77e1fc", "code_version": "87c73cd-dirty", "doc_id": "jof_12846"}
214 +{"ts": "2026-11-03T23:11:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6680, "output": 2327, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074054, "latency_s": 19.45, "request_id": "req_011Cggqf9Fj9HKLBS3fthWkz", "stop_reason": "end_turn", "cache_key": "9240787a12c8f80016a47c7be0c41b1f821a66854185f4b05e28ca0675c868a7", "code_version": "87c73cd-dirty", "doc_id": "jof_12856"}
215 +{"ts": "2026-11-03T23:11:30+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8930, "output": 2093, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.078374, "latency_s": 17.7, "request_id": "req_011CggqfW2tNvDBH3UXmzuj5", "stop_reason": "end_turn", "cache_key": "447c7f5143d62deb469bb9b69fae8b495ec354b4554d9cf399312289b4920a65", "code_version": "87c73cd-dirty", "doc_id": "jof_12855"}
216 +{"ts": "2026-11-03T23:11:35+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 512, "output": 863, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.020102, "latency_s": 7.54, "request_id": "req_011Cggqgaq96smJZ3xDBorNf", "stop_reason": "end_turn", "cache_key": "778309e1b77f4cb82c470d31ccb4d0e86d64c266336bb13287d4a6d00b44c169", "code_version": "87c73cd-dirty", "doc_id": "jof_12865"}
217 +{"ts": "2026-11-03T23:11:40+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 3306, "output": 1692, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.047858, "latency_s": 14.25, "request_id": "req_011CggqgU1dCjbFbp3AVf6A9", "stop_reason": "end_turn", "cache_key": "ca2731f10192511f085eff32d161d0394967411500523c7c89867d637097321b", "code_version": "87c73cd-dirty", "doc_id": "jof_12834"}
218 +{"ts": "2026-11-03T23:11:41+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7108, "output": 1959, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068406, "latency_s": 15.65, "request_id": "req_011CggqgTbaG6ecUo7n2ka3q", "stop_reason": "end_turn", "cache_key": "0984ff4d68525b141809efdcbc389662955e8d6cbb598831b0d5e231de49487d", "code_version": "87c73cd-dirty", "doc_id": "jof_12854"}
219 +{"ts": "2026-11-03T23:11:45+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5931, "output": 1909, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.062698, "latency_s": 15.2, "request_id": "req_011CggqgpCou84kuePWsqUKJ", "stop_reason": "end_turn", "cache_key": "41743a0f5b51bbf404eb9b3835720b29545ea858acf37f807d7f5ac257725d28", "code_version": "87c73cd-dirty", "doc_id": "jof_12866"}
220 +{"ts": "2026-11-03T23:11:52+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 2964, "output": 1547, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.04359, "latency_s": 12.13, "request_id": "req_011CggqhWy1ESYELfnspoFdC", "stop_reason": "end_turn", "cache_key": "f28c8fd8596d1a04492680e430fe5c26df4bec9721cd8f9b1f213c5b087689d5", "code_version": "87c73cd-dirty", "doc_id": "jof_12857"}
221 +{"ts": "2026-11-03T23:11:53+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7211, "output": 2174, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.073118, "latency_s": 17.83, "request_id": "req_011Cggqh9GXt23tctuJLviJo", "stop_reason": "end_turn", "cache_key": "6ef11269a772779a65ab49b17e05c3e9b2cb34a54f6378275574123a6c63062f", "code_version": "87c73cd-dirty", "doc_id": "jof_12863"}
222 +{"ts": "2026-11-03T23:11:59+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7119, "output": 2249, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07425, "latency_s": 17.75, "request_id": "req_011CggqhcsTSD3PztSsA4YEH", "stop_reason": "end_turn", "cache_key": "a220807b8df84f59afa957ecaae7b06207a80a128d0517ce79478aacdfcb4b53", "code_version": "87c73cd-dirty", "doc_id": "jof_12868"}
223 +{"ts": "2026-11-03T23:12:03+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7459, "output": 2133, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07329, "latency_s": 17.57, "request_id": "req_011CggqhwHEavj9GppuCvxb9", "stop_reason": "end_turn", "cache_key": "f314565a9ff32af388ccdd1a3c03171210f15e636e91d0c0f8a265b09e5fcb1d", "code_version": "87c73cd-dirty", "doc_id": "jof_12867"}
224 +{"ts": "2026-11-03T23:12:09+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7923, "output": 2115, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074786, "latency_s": 17.26, "request_id": "req_011CggqiREFHhNFu9KFJbT7t", "stop_reason": "end_turn", "cache_key": "71b4085e8750e582f55ffcba0bae94d5e81c9f94d4ebcee643db7e29eedc5262", "code_version": "87c73cd-dirty", "doc_id": "jof_12864"}
225 +{"ts": "2026-11-03T23:12:11+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7129, "output": 2161, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07253, "latency_s": 18.47, "request_id": "req_011CggqiToH3haSNfn47w1Sx", "stop_reason": "end_turn", "cache_key": "5892cabec8710f58cffba0a4ab288b1da3a5a296465b0bda819175195e25657c", "code_version": "87c73cd-dirty", "doc_id": "jof_12843"}
226 +{"ts": "2026-11-03T23:12:19+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7811, "output": 2272, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077478, "latency_s": 19.81, "request_id": "req_011Cggqiw8qWKTKSQ6Ny3F78", "stop_reason": "end_turn", "cache_key": "70a64c5b162459911bae3d62993fda236cf6548ef1487398101d4b48e4342e5f", "code_version": "87c73cd-dirty", "doc_id": "jof_12853"}
227 +{"ts": "2026-11-03T23:12:26+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5809, "output": 1979, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06361, "latency_s": 16.32, "request_id": "req_011CggqjhRcLvVKUTfkJ2JEH", "stop_reason": "end_turn", "cache_key": "ced67225c925ba08dce1bfb2063b3fcd4f360f742fb69d037435b9b34eee76a1", "code_version": "87c73cd-dirty", "doc_id": "jof_12848"}
228 +{"ts": "2026-11-03T23:12:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8616, "output": 3392, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.103098, "latency_s": 28.67, "request_id": "req_011CggqjEbZWiv8dnnTe1ysK", "stop_reason": "end_turn", "cache_key": "bb6e92f688613c5d57e8f5b9df1f3ed7c76ae3a84f8b244baf08fdf603bfebf1", "code_version": "87c73cd-dirty", "doc_id": "jof_12862"}
229 +{"ts": "2026-11-03T23:12:37+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7033, "output": 2330, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075526, "latency_s": 25.91, "request_id": "req_011Cggqjq8xDi5ZWghgazdr1", "stop_reason": "end_turn", "cache_key": "75e4885abcf57a6944f3b27ced9f3a61fac6cf3cf60d48bcb308c510a6d2d49b", "code_version": "87c73cd-dirty", "doc_id": "jof_12849"}
230 +{"ts": "2026-11-03T23:12:42+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6742, "output": 2065, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069062, "latency_s": 16.22, "request_id": "req_011CggqkupG6grN4bUB7DvWi", "stop_reason": "end_turn", "cache_key": "a65cfd8c4ab55eb8cc233c8b12d96619ae6be0dadf2ba58f495cbd9498c0a83b", "code_version": "87c73cd-dirty", "doc_id": "jof_12871"}
231 +{"ts": "2026-11-03T23:12:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7959, "output": 3519, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.10301, "latency_s": 29.09, "request_id": "req_011CggqkQ6MCF6cYDcAhF5pn", "stop_reason": "end_turn", "cache_key": "710ba5233f728cc9638a6fbdef7d7645a6886e83a33269ea49126e84e460b1a1", "code_version": "87c73cd-dirty", "doc_id": "jof_12869"}
232 +{"ts": "2026-11-03T23:12:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9492, "output": 2281, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.084382, "latency_s": 23.05, "request_id": "req_011CggqmMSdP6J2v1avvVgLV", "stop_reason": "end_turn", "cache_key": "514cb7818a090ac577eed65bbe3edbf8bbcddfa50ff5bfb771f90a4145a8ab3b", "code_version": "87c73cd-dirty", "doc_id": "jof_12889"}
233 +{"ts": "2026-11-03T23:12:56+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6826, "output": 2299, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074078, "latency_s": 18.6, "request_id": "req_011CggqmkDaycHQHUhbeVViL", "stop_reason": "end_turn", "cache_key": "47138d806fe22becc97f0e9adca4d53513dc2dc0cca0f2ad28d07de979cd9b2e", "code_version": "87c73cd-dirty", "doc_id": "jof_12880"}
234 +{"ts": "2026-11-03T23:13:02+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9580, "output": 2417, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.087454, "latency_s": 19.55, "request_id": "req_011Cggqn7N5CSq1mWF9ZZBDr", "stop_reason": "end_turn", "cache_key": "2191044f4ec7264189d44b7d68a27bbcbfe72a8e1a410bd3349b46c1f0043856", "code_version": "87c73cd-dirty", "doc_id": "jof_12884"}
235 +{"ts": "2026-11-03T23:13:05+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6854, "output": 2079, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06979, "latency_s": 17.36, "request_id": "req_011CggqnYk3fjFve8PEGzYkU", "stop_reason": "end_turn", "cache_key": "7c19348481ee2de19f010bd4d19e7e63916952916535329b2c3409b0c6b13f18", "code_version": "87c73cd-dirty", "doc_id": "jof_12883"}
236 +{"ts": "2026-11-03T23:13:14+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7614, "output": 2207, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07539, "latency_s": 17.81, "request_id": "req_011Cggqo7yKSS5fC9wE3GLMq", "stop_reason": "end_turn", "cache_key": "42db43fd3bfced6b13f02ef74b8c1a12b7e94599990df16c6465e0fc78ffd467", "code_version": "87c73cd-dirty", "doc_id": "jof_12876"}
237 +{"ts": "2026-11-03T23:13:14+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 11785, "output": 2435, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.096634, "latency_s": 19.43, "request_id": "req_011Cggqo4KnaQ5nnwqRUkZHQ", "stop_reason": "end_turn", "cache_key": "36a4aa112068cd189c63c919702409b72f4a3ee5fa862e4a8ef19c8b36e040c2", "code_version": "87c73cd-dirty", "doc_id": "jof_12885"}
238 +{"ts": "2026-11-03T23:13:19+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8832, "output": 2045, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077022, "latency_s": 16.94, "request_id": "req_011CggqoYyU8sPtgRuprwhGr", "stop_reason": "end_turn", "cache_key": "dba159a809061963be8cbbc6111cdbb4998d565468b6932e981a30eba1d1ede9", "code_version": "87c73cd-dirty", "doc_id": "jof_12877"}
239 +{"ts": "2026-11-03T23:13:20+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5319, "output": 1789, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.05785, "latency_s": 14.92, "request_id": "req_011CggqoqFGSEiPyNuE7dLf4", "stop_reason": "end_turn", "cache_key": "0b589c503b552bf39df5549af335d894130707cd4c67422ba146a0b7b163cb62", "code_version": "87c73cd-dirty", "doc_id": "jof_12882"}
240 +{"ts": "2026-11-03T23:13:31+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8562, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.077842, "latency_s": 17.7, "request_id": "req_011CggqpSUa1DkNYCkcNFXGw", "stop_reason": "end_turn", "cache_key": "bafb10791c70c11ecc952851fe195a2769a1fac2933a717ed11f196b11416a1b", "code_version": "87c73cd-dirty", "doc_id": "jof_12879"}
241 +{"ts": "2026-11-03T23:13:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9312, "output": 2112, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.080282, "latency_s": 17.25, "request_id": "req_011CggqpVokERFULLTx3f2Dj", "stop_reason": "end_turn", "cache_key": "1cd0ccf36aca0c97f076fbee14ddb1e59c5c7c7fc4dfc47cf2979b0302fa7ea1", "code_version": "87c73cd-dirty", "doc_id": "jof_12878"}
242 +{"ts": "2026-11-03T23:13:37+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6588, "output": 2000, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.067146, "latency_s": 16.1, "request_id": "req_011CggqpwP8pEVScFKF8GxuW", "stop_reason": "end_turn", "cache_key": "7471f998a4ad89e8a145407df6d2914f9b688eeb7e29d9a9aee52ad981ad0e88", "code_version": "87c73cd-dirty", "doc_id": "jof_12870"}
243 +{"ts": "2026-11-03T23:13:39+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8884, "output": 2261, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08155, "latency_s": 20.07, "request_id": "req_011CggqpomzBCxHbu6L25jrB", "stop_reason": "end_turn", "cache_key": "dbe4c8ee57f43543dcb47f0a8895612362f763ebba25a4bddc4911c434341b6c", "code_version": "87c73cd-dirty", "doc_id": "jof_12890"}
244 +{"ts": "2026-11-03T23:13:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6115, "output": 1932, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.063894, "latency_s": 14.82, "request_id": "req_011CggqqkTbFju7sfosbPTfb", "stop_reason": "end_turn", "cache_key": "83a0bc298ba801da87d38c25cff633198201fc1bc6e1480e1d76a8a243b8d5f5", "code_version": "87c73cd-dirty", "doc_id": "jof_12888"}
245 +{"ts": "2026-11-03T23:13:47+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 4983, "output": 1849, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.057706, "latency_s": 14.83, "request_id": "req_011CggqqmysEjCvXEvLQrPdC", "stop_reason": "end_turn", "cache_key": "c764aa4df1930139041f995ceca5719d1bdc8e60695c98b84bf8211f49e9fcd5", "code_version": "87c73cd-dirty", "doc_id": "jof_12906"}
246 +{"ts": "2026-11-03T23:13:59+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7426, "output": 2295, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076398, "latency_s": 19.94, "request_id": "req_011CggqrHqDv5zh1vDTiNiU2", "stop_reason": "end_turn", "cache_key": "b8a092d7b206323d782b0ced46b1623f59824a9473877ed7489a87971ee3fae9", "code_version": "87c73cd-dirty", "doc_id": "jof_12897"}
247 +{"ts": "2026-11-03T23:14:05+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7208, "output": 2051, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070646, "latency_s": 18.04, "request_id": "req_011CggqrsQaM56nkyLoMxWZ5", "stop_reason": "end_turn", "cache_key": "f627ff7fb9747401fd6b1840f956d6b2c377a1f4b2cc94f9a766b2aa36048e28", "code_version": "87c73cd-dirty", "doc_id": "jof_12895"}
248 +{"ts": "2026-11-03T23:14:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8283, "output": 3287, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.099666, "latency_s": 29.18, "request_id": "req_011Cggqr8XPZZEhkwmruqTQt", "stop_reason": "end_turn", "cache_key": "d639df714a90625b71cfd43ce7ff9b4cc31c304138f043bea91a99ab3aaea003", "code_version": "87c73cd-dirty", "doc_id": "jof_12899"}
249 +{"ts": "2026-11-03T23:14:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8925, "output": 2317, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082834, "latency_s": 20.0, "request_id": "req_011Cggqrr4iQ754gB4KZGVAs", "stop_reason": "end_turn", "cache_key": "d13be9aa500e00f972f4e640925da0aa86757f6043e2f91bc0c4185009a2f460", "code_version": "87c73cd-dirty", "doc_id": "jof_12903"}
250 +{"ts": "2026-11-03T23:14:10+00:00", "purpose": "review:referee", "prompt_id": "review_referee", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "high", "usage": {"input": 16678, "output": 6000, "cache_read": 0, "cache_write": 0}, "cost_usd": 0.186712, "latency_s": 60.52, "request_id": "req_011Cggqp6AT4gJZUrb9VcrMw", "stop_reason": "max_tokens", "cache_key": "38d3f1d2e0611f94c242a86c0c68625ac3512ec5d60ae45bd2a69212eb2faf01", "code_version": "87c73cd-dirty", "paper": "/Users/simon-pierreboucher/Desktop/UQO/Recherche/Recherche-UQO/methodes/autonomous-research-claude-code"}
251 +{"ts": "2026-11-03T23:14:15+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6106, "output": 1992, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.065058, "latency_s": 16.33, "request_id": "req_011Cggqsme64xZsf5Pic2fBQ", "stop_reason": "end_turn", "cache_key": "812721c9d8310b09695447caf1d5f05c5f3fc031019d8a22d66a57c7e8f334d5", "code_version": "87c73cd-dirty", "doc_id": "jof_12902"}
252 +{"ts": "2026-11-03T23:14:24+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7173, "output": 2263, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074746, "latency_s": 19.23, "request_id": "req_011CggqtCrPoWtU55iEzQ87w", "stop_reason": "end_turn", "cache_key": "b6704e407393e5c0d02f04f9f1a4c690eaf7b6f0a44923d39a06e6ff1c88536b", "code_version": "87c73cd-dirty", "doc_id": "jof_12900"}
253 +{"ts": "2026-11-03T23:14:27+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 9895, "output": 2389, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088154, "latency_s": 21.07, "request_id": "req_011CggqtHtkgXUVBLjQotgL8", "stop_reason": "end_turn", "cache_key": "3ca11a3f110cb7c48365ada659a1f7c247b5f7f7a2dcfe24674affc8800825b6", "code_version": "87c73cd-dirty", "doc_id": "jof_12893"}
254 +{"ts": "2026-11-03T23:14:31+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6825, "output": 1911, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.066314, "latency_s": 15.44, "request_id": "req_011CggqtyUFnrkgoaGZqPnyW", "stop_reason": "end_turn", "cache_key": "078ed694a74223e3094ba6a11b5417aaa554f1d52bf8a873af15d4fb42771785", "code_version": "87c73cd-dirty", "doc_id": "jof_12898"}
255 +{"ts": "2026-11-03T23:14:33+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8029, "output": 3106, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.09503, "latency_s": 26.62, "request_id": "req_011CggqtLH7iAkwihjHXjd1Q", "stop_reason": "end_turn", "cache_key": "5e3d09497c89adf5c4da84d06f2af2e028694369e816704e5c50be96b71218ba", "code_version": "87c73cd-dirty", "doc_id": "jof_12896"}
256 +{"ts": "2026-11-03T23:14:45+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8265, "output": 2140, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076654, "latency_s": 17.49, "request_id": "req_011Cggquqxr4Fi42x58Qjv5v", "stop_reason": "end_turn", "cache_key": "ef78eb0321e85eecdb5b3f406f1b91ac1c8250766920cb82c3461c94d5b14023", "code_version": "87c73cd-dirty", "doc_id": "jof_12894"}
257 +{"ts": "2026-11-03T23:14:48+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7332, "output": 2012, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.070362, "latency_s": 17.36, "request_id": "req_011Cggqv7w2QUBzgfrXX4GMa", "stop_reason": "end_turn", "cache_key": "e064d5a52c6b6c1f8d998f12ca5ee039fbacc1bb9935cd9369977e6f07e4b62c", "code_version": "87c73cd-dirty", "doc_id": "jof_12952"}
258 +{"ts": "2026-11-03T23:14:52+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7585, "output": 2158, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074294, "latency_s": 18.83, "request_id": "req_011CggqvJ3V9ZPSocBR3U532", "stop_reason": "end_turn", "cache_key": "882a310eebfc9e638b5b603f29c2001b2f4dd484198e8e03f207c2f423429cb4", "code_version": "87c73cd-dirty", "doc_id": "jof_12954"}
259 +{"ts": "2026-11-03T23:14:55+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7832, "output": 3546, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.103042, "latency_s": 30.8, "request_id": "req_011CggqudMorJJbKh1iaEnRz", "stop_reason": "end_turn", "cache_key": "a31d50d30927029b05a279d9ade02c179b1f4e5dea2a256df7c35af9a7b6babe", "code_version": "87c73cd-dirty", "doc_id": "jof_12901"}
260 +{"ts": "2026-11-03T23:15:04+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6730, "output": 2182, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071354, "latency_s": 18.9, "request_id": "req_011Cggqw94mEByYhVxRLQPCg", "stop_reason": "end_turn", "cache_key": "2c0624905e39be8e60eab94902a3ad6d2310ca7a6600fd8beeb3e14e31491840", "code_version": "87c73cd-dirty", "doc_id": "jof_12909"}
261 +{"ts": "2026-11-03T23:15:05+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6785, "output": 2043, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.068794, "latency_s": 17.08, "request_id": "req_011CggqwQPFkSETN4MLwuy6e", "stop_reason": "end_turn", "cache_key": "82bf9ddad90a66cc44ee28e552d11876101b8ef17883a6a1a05903567309bf71", "code_version": "87c73cd-dirty", "doc_id": "jof_12949"}
262 +{"ts": "2026-11-03T23:15:07+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6346, "output": 1751, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061198, "latency_s": 14.6, "request_id": "req_011CggqwgtCH5zGkkKAEoqYZ", "stop_reason": "end_turn", "cache_key": "41db738e96ed032bfbaa7890b26fb1610e63ddf2625613b1570ab125100caf5e", "code_version": "87c73cd-dirty", "doc_id": "jof_12951"}
263 +{"ts": "2026-11-03T23:15:14+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8520, "output": 2362, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.082114, "latency_s": 18.68, "request_id": "req_011CggqwuLnwHm462RUKEHnj", "stop_reason": "end_turn", "cache_key": "49b8df40aa601f891206b51c199e0b8765acd0a3df0cc3f92104a9bb879d7b97", "code_version": "87c73cd-dirty", "doc_id": "jof_12950"}
264 +{"ts": "2026-11-03T23:15:21+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7364, "output": 2093, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07211, "latency_s": 17.25, "request_id": "req_011CggqxYeNUY2XBZ2cXfCcx", "stop_reason": "end_turn", "cache_key": "614fa4ac11b1090d97276848e17f5c5f9afd7155de42064cd6440669f74256bc", "code_version": "87c73cd-dirty", "doc_id": "jof_12908"}
265 +{"ts": "2026-11-03T23:15:26+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8908, "output": 2602, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.088466, "latency_s": 20.69, "request_id": "req_011CggqxfhmwhyZrar8D4Yv4", "stop_reason": "end_turn", "cache_key": "de7ac4a1b2d1048ad0b72f8d19a51646b95c35d4a1f4edda8f3c1c68c936d968", "code_version": "87c73cd-dirty", "doc_id": "jof_12947"}
266 +{"ts": "2026-11-03T23:15:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8854, "output": 2199, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08019, "latency_s": 17.85, "request_id": "req_011CggqyHYZJhdkNNtTh1ZE3", "stop_reason": "end_turn", "cache_key": "3b2893a637e63e56da9722ba3fb0c87c0fad1ef4d6b8ec1d1059c7df77d69a95", "code_version": "87c73cd-dirty", "doc_id": "jof_12953"}
267 +{"ts": "2026-11-03T23:15:32+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8508, "output": 3062, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.096066, "latency_s": 25.17, "request_id": "req_011CggqxmZzxhCNyonDuGrQr", "stop_reason": "end_turn", "cache_key": "21a1bbe9dd87153e56403f7016f29c8f78dced1668577681ea52dc547a8def80", "code_version": "87c73cd-dirty", "doc_id": "jof_12910"}
268 +{"ts": "2026-11-03T23:15:38+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6964, "output": 2133, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07131, "latency_s": 17.26, "request_id": "req_011CggqypdppARXsUNAuEEjC", "stop_reason": "end_turn", "cache_key": "4d250a68e2dae22f82812846220efc3f0079c6faec898686434b88e0520563dc", "code_version": "87c73cd-dirty", "doc_id": "jof_12964"}
269 +{"ts": "2026-11-03T23:15:46+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8134, "output": 2424, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08181, "latency_s": 20.06, "request_id": "req_011CggqzCWDoqp7rp7xt9zkB", "stop_reason": "end_turn", "cache_key": "fe05d17bf6b7f7ae0ddee8c98b6ee9b0a9e0d3ce530855e42d2e9de5dab1a113", "code_version": "87c73cd-dirty", "doc_id": "jof_12948"}
270 +{"ts": "2026-11-03T23:15:50+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6230, "output": 2172, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069154, "latency_s": 18.1, "request_id": "req_011CggqzdcpsN7TKQCHG7Uoy", "stop_reason": "end_turn", "cache_key": "cee6f622dfc9abc7ae159d8daed818a4276313752bdd31dc72d9510c8488df5b", "code_version": "87c73cd-dirty", "doc_id": "jof_12969"}
271 +{"ts": "2026-11-03T23:15:51+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7392, "output": 2284, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.076042, "latency_s": 19.44, "request_id": "req_011CggqzcB1oPyYyFVaknH6d", "stop_reason": "end_turn", "cache_key": "eb979d6eda3363630fa2240fea8f0f643b6c16507aa523174acf354329fd56ad", "code_version": "87c73cd-dirty", "doc_id": "jof_12966"}
272 +{"ts": "2026-11-03T23:15:54+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7151, "output": 1982, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069038, "latency_s": 16.04, "request_id": "req_011Cggr16XqKTQZDK4J6XPHg", "stop_reason": "end_turn", "cache_key": "56d424900734b5ad8813e1bf225b7b19a92e228583614c06de6ade8b8e6b49f0", "code_version": "87c73cd-dirty", "doc_id": "jof_12961"}
273 +{"ts": "2026-11-03T23:16:01+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 5307, "output": 1818, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.058382, "latency_s": 14.66, "request_id": "req_011Cggr1gWVWEqzHKu6HntmR", "stop_reason": "end_turn", "cache_key": "abfd93bd40dbe8e9ee87d0e9fe106d2fe250a52d78a9c7f5e114ecdfd4365a0a", "code_version": "87c73cd-dirty", "doc_id": "jof_12967"}
274 +{"ts": "2026-11-03T23:16:06+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6314, "output": 1811, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.06227, "latency_s": 15.14, "request_id": "req_011Cggr23eUojpAyvc4DvqGv", "stop_reason": "end_turn", "cache_key": "740371daffcefcfdced488b88d3c6e4dc878eeca32376465d489829e5c0065de", "code_version": "87c73cd-dirty", "doc_id": "jof_12959"}
275 +{"ts": "2026-11-03T23:16:08+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6725, "output": 2186, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071414, "latency_s": 17.34, "request_id": "req_011Cggr1yBqizZ1bPf8Jo8V9", "stop_reason": "end_turn", "cache_key": "88a18d057a327bcaf9a06e0c09495cb70dbc4a94345c617d8341761aff97cfe3", "code_version": "87c73cd-dirty", "doc_id": "jof_12958"}
276 +{"ts": "2026-11-03T23:16:11+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6826, "output": 2058, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.069258, "latency_s": 16.84, "request_id": "req_011Cggr2HTvcF7VStyWU41Q9", "stop_reason": "end_turn", "cache_key": "9b6cfe2f36badfe4b83636ced2568bfb69fd6ea40fd9df65b5a10a31b940a0f0", "code_version": "87c73cd-dirty", "doc_id": "jof_12965"}
277 +{"ts": "2026-11-03T23:16:15+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6013, "output": 1794, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.060726, "latency_s": 14.26, "request_id": "req_011Cggr2mbMXEvNQCvi428w1", "stop_reason": "end_turn", "cache_key": "5c42355212cf0c12a943d37aac5c10f8ad205a4951f77c186999f9d1cb0b2e69", "code_version": "87c73cd-dirty", "doc_id": "jof_12955"}
278 +{"ts": "2026-11-03T23:16:23+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6078, "output": 1799, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.061086, "latency_s": 15.4, "request_id": "req_011Cggr3Fzu93A4EptHLDviQ", "stop_reason": "end_turn", "cache_key": "2ae051de4a22914d2a665565c0e97a3971c0e5829b2ff14ee6168ba05caa3043", "code_version": "87c73cd-dirty", "doc_id": "jof_12962"}
279 +{"ts": "2026-11-03T23:16:28+00:00", "purpose": "annotation", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8375, "output": 2579, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.085874, "latency_s": 21.86, "request_id": "req_011Cggr3AcwbXweLSSq79Z7b", "stop_reason": "end_turn", "cache_key": "8bbc87c88e1496850ec660dad89c60398530d95d622de4a5858a90117f55f6a2", "code_version": "87c73cd-dirty", "doc_id": "jof_12968"}
280 +{"ts": "2026-11-03T23:26:27+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7805, "output": 1870, "cache_read": 0, "cache_write": 3968}, "cost_usd": 0.08846, "latency_s": 16.82, "request_id": "req_011Cggroet385FRJyFSQYwsc", "stop_reason": "end_turn", "cache_key": "d7eb13d3dd0906c7d93d245d1e62159e0ce0ba326e96d956ece25e1c367cd75e", "code_version": "87c73cd-dirty", "doc_id": "jof_12518"}
281 +{"ts": "2026-11-03T23:26:28+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8311, "output": 2007, "cache_read": 0, "cache_write": 3968}, "cost_usd": 0.093224, "latency_s": 17.62, "request_id": "req_011CggrofhQR8XspmamsZo1N", "stop_reason": "end_turn", "cache_key": "aee5142468d70483db5d982daec00f8a3fe2ef07bed9842bcec6cb011f7520b0", "code_version": "87c73cd-dirty", "doc_id": "jof_12685"}
282 +{"ts": "2026-11-03T23:26:28+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8980, "output": 2188, "cache_read": 0, "cache_write": 3968}, "cost_usd": 0.09952, "latency_s": 18.22, "request_id": "req_011CggroerJwhucDKfDQHN62", "stop_reason": "end_turn", "cache_key": "bc8e5e2209dd40d82f7e5d468fbdacc82abc523e52082f56c0ba3ef53c9599b4", "code_version": "87c73cd-dirty", "doc_id": "jof_12504"}
283 +{"ts": "2026-11-03T23:26:32+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8752, "output": 2621, "cache_read": 0, "cache_write": 3968}, "cost_usd": 0.107268, "latency_s": 22.24, "request_id": "req_011Cggroer4RLy4rHCy7FZKH", "stop_reason": "end_turn", "cache_key": "7ee97b1138d63831c6ee26750180e98f62ee71b96aff22eb2c2899c66ec35b8f", "code_version": "87c73cd-dirty", "doc_id": "jof_12464"}
284 +{"ts": "2026-11-03T23:26:44+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7529, "output": 2048, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07187, "latency_s": 17.29, "request_id": "req_011CggrpukGjUVLF7kFJooeJ", "stop_reason": "end_turn", "cache_key": "8d78d26afc93f1e05335adc59b517ea00ff67f40ce1aa0f54171e0d59bb7a533", "code_version": "87c73cd-dirty", "doc_id": "jof_12750"}
285 +{"ts": "2026-11-03T23:26:47+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 11283, "output": 2375, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.093426, "latency_s": 18.71, "request_id": "req_011CggrpyP4Ddma4wQ4m3gJf", "stop_reason": "end_turn", "cache_key": "27962a14cb7623d324764a2bc58ddafa78311e7aea69f907c9844ad87abfdd77", "code_version": "87c73cd-dirty", "doc_id": "jof_12761"}
286 +{"ts": "2026-11-03T23:26:52+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 6672, "output": 2187, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.071222, "latency_s": 19.22, "request_id": "req_011CggrqJqancvwEzDPnSa8D", "stop_reason": "end_turn", "cache_key": "8f398b816df9c9ca69153730f34af21b426b21a70e0a7d7fa25161c2954766e8", "code_version": "87c73cd-dirty", "doc_id": "jof_12830"}
287 +{"ts": "2026-11-03T23:26:54+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8945, "output": 3195, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.100474, "latency_s": 26.05, "request_id": "req_011Cggrq1N7mxu2wVfywSrJN", "stop_reason": "end_turn", "cache_key": "6ecd5e0c06a85e5d5851aa7980a181e2b5c100618be033546800504b90e355b7", "code_version": "87c73cd-dirty", "doc_id": "jof_12775"}
288 +{"ts": "2026-11-03T23:27:05+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7161, "output": 2295, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075338, "latency_s": 18.68, "request_id": "req_011CggrrNBHK6QoQyEYrAUjs", "stop_reason": "end_turn", "cache_key": "5bfd9e76918e64c56dc993f3939431e2dda7a9aee592061321310701f22ad707", "code_version": "87c73cd-dirty", "doc_id": "jof_12868"}
289 +{"ts": "2026-11-03T23:27:07+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 7359, "output": 2666, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.08355, "latency_s": 22.39, "request_id": "req_011CggrrBzrjTiphppSbjAMY", "stop_reason": "end_turn", "cache_key": "30b444ed602a6d01c600e8984ffc866d61478b28f8978467cf7708d7c6634c69", "code_version": "87c73cd-dirty", "doc_id": "jof_12835"}
290 +{"ts": "2026-11-03T23:27:09+00:00", "purpose": "annotation_retry", "prompt_id": "document_annotation", "prompt_version": "1.0", "model": "claude-opus-5-5", "effort": "medium", "usage": {"input": 8325, "output": 2058, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.075254, "latency_s": 16.76, "request_id": "req_011CggrrjmJhp2hjaEQ2qaoU", "stop_reason": "end_turn", "cache_key": "57e5a1125dae285dc78ea3d57e62167da44dc58754e4d20d3d88588c65db48fc", "code_version": "87c73cd-dirty", "doc_id": "jof_12899"}
modified data/annotations/documents/jof_12464.json +54 −69
@@ -1,48 +1,46 @@
1 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 − ],
2 + "validation_errors": [],
3 + "repairs": [],
7 4 "doc_id": "jof_12464",
8 5 "ontology_version": "1.0",
9 6 "prompt_version": "1.0",
10 7 "model": "claude-opus-5-5",
11 8 "effort": "medium",
12 − "annotated_at": "2026-11-03T22:55:46+00:00",
9 + "annotated_at": "2026-11-03T23:26:32+00:00",
13 10 "source_sha256": "2ca10cacc14b69a439286f84226831d7e5a8bfe7a7b838caf66f391b5f6c08a8",
14 11 "parser": "pymupdf-jof-heuristics-1.3",
15 12 "payload_chars": 24477,
16 13 "usage": {
17 − "input": 8616,
18 − "output": 2568,
19 − "cache_read": 3968,
20 − "cache_write": 0
14 + "input": 8752,
15 + "output": 2621,
16 + "cache_read": 0,
17 + "cache_write": 3968
21 18 },
22 − "cost_usd": 0.086618,
23 − "request_id": "req_011CggpThPZKLuuR58E9Q9sk",
19 + "cost_usd": 0.107268,
20 + "request_id": "req_011Cggroer4RLy4rHCy7FZKH",
24 21 "annotation": {
25 22 "genre": "mixed_theory_empirical",
26 23 "subfield": "entrepreneurial_finance_private_equity",
27 − "subfield_secondary": "corporate_finance; asset_pricing (time-varying discount rates)",
24 + "subfield_secondary": "asset_pricing; corporate_finance",
28 25 "primary_method": "reduced_form_regression",
29 26 "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.",
27 + "estimand": "Association between the aggregate equity risk premium (predicted excess market return) and buyout activity, at both aggregate and firm level, plus how firm characteristics shift that sensitivity.",
28 + "strategy": "Time-series regressions of quarterly buyout activity on estimated risk premium and credit factors with Newey-West SEs; firm panel deal-likelihood regressions with firm or time fixed effects; high-low quartile interaction tests that check model-implied cross-sectional predictions.",
32 29 "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"
30 + "Predictive regressions (D/P, cay, T-bill) and the VAR capture the expected market risk premium",
31 + "Credit-factor controls adequately span the competing credit-market explanations",
32 + "Cross-sectional proxies (governance, competition, cash flow, industry M&A/IPO activity) capture potential gains and illiquidity"
36 33 ],
37 34 "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 − ]
35 + "Credit market conditions as an alternative driver",
36 + "Debt-equity mispricing explanation",
37 + "Debt capacity story, tested through the contrasting predictions for systematic versus idiosyncratic risk",
38 + "Seasonality",
39 + "Persistence of regressors",
40 + "Time-varying firm risk correlated with discount rates, handled with time fixed effects"
41 + ],
42 + "causal_language_strength": 4,
43 + "uses_causal_claim": true
46 44 },
47 45 "abstract_moves": {
48 46 "question": {
@@ -76,10 +74,6 @@
76 74 "paragraph_index": 0,
77 75 "move": "motivation"
78 76 },
79 − {
80 − "paragraph_index": 0,
81 − "move": "tension_or_puzzle"
82 − },
83 77 {
84 78 "paragraph_index": 0,
85 79 "move": "related_literature_positioning"
@@ -106,10 +100,6 @@
106 100 },
107 101 {
108 102 "paragraph_index": 3,
109 − "move": "identification_defense"
110 − },
111 − {
112 − "paragraph_index": 4,
113 103 "move": "preview_of_results"
114 104 },
115 105 {
@@ -120,10 +110,6 @@
120 110 "paragraph_index": 5,
121 111 "move": "preview_of_results"
122 112 },
123 − {
124 − "paragraph_index": 5,
125 − "move": "robustness_mention"
126 − },
127 113 {
128 114 "paragraph_index": 6,
129 115 "move": "contribution_statement"
@@ -208,7 +194,7 @@
208 194 ]
209 195 },
210 196 "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.",
197 + "convention_notes": "Notes give the sample period (1982Q4-2011Q4, 117 quarters), dependent-variable definitions (deals scaled by public firms in bps, 2010 dollars) and risk-premium estimator definitions (OLS predictors vs. VAR). Panels separate volume- and value-based measures. Text states Newey-West SEs for time-series tables and two-way firm-quarter clustering for panel tables.",
212 198 "units_stated_in_notes": true,
213 199 "standard_errors_description": "mixed"
214 200 },
@@ -216,7 +202,7 @@
216 202 "uncertainty_shown": "no"
217 203 },
218 204 "conclusion": {
219 − "bounded_claims": false,
205 + "bounded_claims": true,
220 206 "policy_implications": false,
221 207 "open_questions": true,
222 208 "restates_contribution": true
@@ -224,68 +210,67 @@
224 210 "hedging": {
225 211 "level": "low",
226 212 "typical_devices": [
227 − "'consistent with' phrasing",
228 − "'broadly match' for extension results",
229 − "acknowledging the simplified view of fund participants"
213 + "consistent with",
214 + "suggest",
215 + "difficult to reconcile with",
216 + "we believe",
217 + "broadly match"
230 218 ]
231 219 },
232 220 "claim_evidence": [
233 221 {
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",
222 + "claim": "A high aggregate risk premium is a strong negative predictor of buyout volume and explains more variation than credit spreads.",
240 223 "evidence_type": "table",
241 224 "evidence_ref": "Table II"
242 225 },
243 226 {
244 − "claim": "Expected market growth is positively related to buyout activity",
227 + "claim": "The risk premium alone explains over 30% of buyout activity variation, versus about 10% for credit factors.",
228 + "evidence_type": "figure",
229 + "evidence_ref": "Figure 1; Table II"
230 + },
231 + {
232 + "claim": "Expected market growth is positively related to buyout activity in a Campbell-Shiller decomposition.",
245 233 "evidence_type": "table",
246 234 "evidence_ref": "Table IV"
247 235 },
248 236 {
249 − "claim": "Firms with high beta or idiosyncratic volatility are less likely buyout targets",
237 + "claim": "Firms with high beta or high idiosyncratic volatility are less likely buyout targets.",
250 238 "evidence_type": "table",
251 239 "evidence_ref": "Table VII"
252 240 },
253 241 {
254 − "claim": "Target beta distribution shifts lower in high risk-premium periods while idiosyncratic risk does not",
242 + "claim": "Buyout targets' beta shifts lower in high-risk-premium periods while idiosyncratic risk does not change.",
255 243 "evidence_type": "figure",
256 244 "evidence_ref": "Figure 3; Table VIII"
257 245 },
258 246 {
259 − "claim": "Poor-governance and free-cash-flow firms are more sensitive, and more liquid industries less sensitive, to the risk premium",
247 + "claim": "Poorly governed, high-cash-flow and less liquid-industry firms show greater sensitivity of buyout activity to the risk premium.",
260 248 "evidence_type": "table",
261 249 "evidence_ref": "Table IX"
262 250 },
263 251 {
264 − "claim": "Low risk premium raises the NPV of performance gains and lowers the cost of GP incentive risk",
252 + "claim": "The buyout decision trades off NPV of performance gains against the cost of GP risk-bearing, both worsening with the risk premium.",
265 253 "evidence_type": "equation_or_proof",
266 − "evidence_ref": "Section II model; Appendix A–B"
254 + "evidence_ref": "Section II; Appendix A-B"
267 255 },
268 256 {
269 − "claim": "M&A and IPO activity also co-move with the risk premium",
257 + "claim": "M&A and IPO activity also co-move with the risk premium broadly as the model predicts.",
270 258 "evidence_type": "table",
271 259 "evidence_ref": "Table X"
272 260 }
273 261 ],
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.",
262 + "structural_summary": "The abstract opens with a direct thesis that buyout booms respond to declines in the risk premium. The introduction contrasts this thesis with credit-centric literature, previews aggregate magnitudes, and sketches a two-channel mechanism (performance NPV and illiquidity/incentive cost). It then previews cross-sectional tests built to separate the thesis from a debt-capacity story, states the contribution and gives a roadmap. The body runs: aggregate time-series evidence, then a simple two-period model with predictions, then firm-panel tests of composition, then an extension to M&A and IPOs. The conclusion restates the findings, acknowledges simplifications about private equity fund heterogeneity, and proposes broader research.",
275 263 "writing_profile_keywords": [
276 264 "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"
265 + "horse race against credit explanations",
266 + "model-motivated cross-sectional tests",
267 + "Newey-West time series",
268 + "firm panel fixed effects",
269 + "confident thesis framing",
270 + "illiquidity channel"
282 271 ],
283 272 "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 − ]
273 + "notes": "Causal language strength is rated 4 because the evidence is predictive and associational (time-series and panel regressions with no exogenous variation), yet the framing is strongly causal: 'form in response to', 'primary determinant' and 'explain buyout waves'. Some section headings are equation fragments from parsing; they are assigned to the surrounding results or theory role. The abstract's 'A panel of U.S. buyouts confirms this view' is treated as the method sentence."
289 274 },
290 275 "deterministic_checks": {
291 276 "roadmap_regex": false,
@@ -314,6 +299,6 @@
314 299 "11": "appendix"
315 300 }
316 301 },
317 − "schema_valid": false,
318 − "code_version": "feb6c9e-dirty"
302 + "schema_valid": true,
303 + "code_version": "87c73cd-dirty"
319 304 }
\ No newline at end of file
modified data/annotations/documents/jof_12489.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('figures_note_placeholder' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.figures_note_placeholder"
4 5 ],
5 6 "doc_id": "jof_12489",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:56:55+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "891df134f35f992e0ff749ce882850fd37996ca9fdfbf8c8260046e8aaf503c0",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 20682,
@@ -169,8 +170,7 @@
169 170 "standard_errors_description": "none"
170 171 },
171 172 "figures": {
172 − "uncertainty_shown": "unclear",
173 − "figures_note_placeholder": null
173 + "uncertainty_shown": "unclear"
174 174 },
175 175 "conclusion": {
176 176 "bounded_claims": true,
@@ -253,6 +253,6 @@
253 253 "5": "conclusion"
254 254 }
255 255 },
256 − "schema_valid": false,
257 − "code_version": "feb6c9e-dirty"
256 + "schema_valid": true,
257 + "code_version": "87c73cd-dirty"
258 258 }
\ No newline at end of file
modified data/annotations/documents/jof_12494.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('figures_note_placeholder_removed' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.figures_note_placeholder_removed"
4 5 ],
5 6 "doc_id": "jof_12494",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:56:59+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "9d80b1fa06977d7cdb9c8a7496c73e9206a0ab277752c61c9c11d10ed1b30b47",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 17285,
@@ -166,8 +167,7 @@
166 167 "standard_errors_description": "unclear"
167 168 },
168 169 "figures": {
169 − "uncertainty_shown": "no",
170 − "figures_note_placeholder_removed": null
170 + "uncertainty_shown": "no"
171 171 },
172 172 "conclusion": {
173 173 "bounded_claims": false,
@@ -254,6 +254,6 @@
254 254 "5": "conclusion"
255 255 }
256 256 },
257 − "schema_valid": false,
258 − "code_version": "feb6c9e-dirty"
257 + "schema_valid": true,
258 + "code_version": "87c73cd-dirty"
259 259 }
\ No newline at end of file
modified data/annotations/documents/jof_12499.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('convention_notes_placeholder_removed' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.convention_notes_placeholder_removed"
4 5 ],
5 6 "doc_id": "jof_12499",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:57:21+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "4b2f62ba52a66448b39539a0d6864e9f14617c2e3e779fcd7b7fc526baca0750",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 24119,
@@ -203,8 +204,7 @@
203 204 "standard_errors_description": "unclear"
204 205 },
205 206 "figures": {
206 − "uncertainty_shown": "mixed",
207 − "convention_notes_placeholder_removed": null
207 + "uncertainty_shown": "mixed"
208 208 },
209 209 "conclusion": {
210 210 "bounded_claims": true,
@@ -295,6 +295,6 @@
295 295 "8": "appendix"
296 296 }
297 297 },
298 − "schema_valid": false,
299 − "code_version": "feb6c9e-dirty"
298 + "schema_valid": true,
299 + "code_version": "87c73cd-dirty"
300 300 }
\ No newline at end of file
modified data/annotations/documents/jof_12504.json +251 −14
@@ -1,25 +1,263 @@
1 1 {
2 − "validation_errors": [
3 − "no json"
4 − ],
2 + "validation_errors": [],
3 + "repairs": [],
5 4 "doc_id": "jof_12504",
6 5 "ontology_version": "1.0",
7 6 "prompt_version": "1.0",
8 7 "model": "claude-opus-5-5",
9 8 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:58:59+00:00",
9 + "annotated_at": "2026-11-03T23:26:28+00:00",
11 10 "source_sha256": "52d5b46ca492ea351913566bc1147928b95420efc304b364cee587e47e003f1e",
12 11 "parser": "pymupdf-jof-heuristics-1.3",
13 12 "payload_chars": 26051,
14 13 "usage": {
15 − "input": 8938,
16 − "output": 2405,
17 − "cache_read": 3968,
18 − "cache_write": 0
14 + "input": 8980,
15 + "output": 2188,
16 + "cache_read": 0,
17 + "cache_write": 3968
18 + },
19 + "cost_usd": 0.09952,
20 + "request_id": "req_011CggroerJwhucDKfDQHN62",
21 + "annotation": {
22 + "genre": "mixed_theory_empirical",
23 + "subfield": "asset_pricing",
24 + "subfield_secondary": "macro_finance",
25 + "primary_method": "calibration_simulation",
26 + "identification": {
27 + "estimand": "Cross-sectional return premium associated with firm return loadings on changes in aggregate labor market tightness",
28 + "strategy": "Rolling firm-level loadings estimated controlling for market return, followed by decile portfolio sorts with factor-model alphas and Fama-MacBeth regressions; a calibrated partial equilibrium search model interprets the pattern.",
29 + "key_assumptions": [
30 + "Rolling three-year loadings capture conditional exposure to matching efficiency shocks",
31 + "Exogenous pricing kernel in which both productivity and matching efficiency shocks are priced",
32 + "Labor market tightness and the market return span the unobserved aggregate shocks"
33 + ],
34 + "threats_addressed": [
35 + "Explanation by standard factor models (four-factor alphas)",
36 + "Overlap with known characteristics such as size, value, momentum, hiring, investment and asset growth",
37 + "Sensitivity to portfolio formation timing and other sorting choices"
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": true,
65 + "sentence_index": 0
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "motivation"
73 + },
74 + {
75 + "paragraph_index": 1,
76 + "move": "preview_of_approach"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 2,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 3,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 4,
92 + "move": "mechanism_or_theory_sketch"
93 + },
94 + {
95 + "paragraph_index": 5,
96 + "move": "mechanism_or_theory_sketch"
97 + },
98 + {
99 + "paragraph_index": 6,
100 + "move": "mechanism_or_theory_sketch"
101 + },
102 + {
103 + "paragraph_index": 7,
104 + "move": "related_literature_positioning"
105 + },
106 + {
107 + "paragraph_index": 8,
108 + "move": "related_literature_positioning"
109 + },
110 + {
111 + "paragraph_index": 9,
112 + "move": "related_literature_positioning"
113 + },
114 + {
115 + "paragraph_index": 10,
116 + "move": "related_literature_positioning"
117 + },
118 + {
119 + "paragraph_index": 10,
120 + "move": "roadmap"
121 + }
122 + ],
123 + "contribution_paragraph_index": 1,
124 + "contribution_types": [
125 + "new_evidence",
126 + "new_theory_or_model"
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": "data"
139 + },
140 + {
141 + "idx": 2,
142 + "role": "results"
143 + },
144 + {
145 + "idx": 3,
146 + "role": "theory"
147 + },
148 + {
149 + "idx": 4,
150 + "role": "results"
151 + },
152 + {
153 + "idx": 5,
154 + "role": "results"
155 + },
156 + {
157 + "idx": 6,
158 + "role": "conclusion"
159 + },
160 + {
161 + "idx": 7,
162 + "role": "appendix"
163 + },
164 + {
165 + "idx": 8,
166 + "role": "appendix"
167 + },
168 + {
169 + "idx": 9,
170 + "role": "appendix"
171 + }
172 + ],
173 + "results_presentation": {
174 + "economic_significance_discussed": true,
175 + "statistical_significance_language": "stars_and_tstats",
176 + "magnitude_statements_in_intro": false
177 + },
178 + "robustness": {
179 + "location": "mixed",
180 + "types": [
181 + "alternative_specification",
182 + "alternative_measure",
183 + "sensitivity_to_parameters",
184 + "subsample"
185 + ]
186 + },
187 + "tables": {
188 + "convention_notes": "Notes report returns and alphas in percent per month, give sample periods and data sources, and report t-statistics in bottom rows. Calibration tables list parameter values with symbols. Model and data are shown side by side.",
189 + "units_stated_in_notes": true,
190 + "standard_errors_description": "unclear"
191 + },
192 + "figures": {
193 + "uncertainty_shown": "no"
194 + },
195 + "conclusion": {
196 + "bounded_claims": false,
197 + "policy_implications": false,
198 + "open_questions": true,
199 + "restates_contribution": true
200 + },
201 + "hedging": {
202 + "level": "low",
203 + "typical_devices": [
204 + "we propose",
205 + "proxy for",
206 + "consistent with",
207 + "our results suggest",
208 + "agnostic about exact source"
209 + ]
210 + },
211 + "claim_evidence": [
212 + {
213 + "claim": "Firms with low loadings on labor market tightness outperform firms with high loadings by about 6% per year",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table III"
216 + },
217 + {
218 + "claim": "The return spread is not explained by four-factor models and survives alternative portfolio constructions",
219 + "evidence_type": "table",
220 + "evidence_ref": "Tables III, V"
221 + },
222 + {
223 + "claim": "Tightness loadings predict returns after controlling for firm characteristics",
224 + "evidence_type": "table",
225 + "evidence_ref": "Table VI"
226 + },
227 + {
228 + "claim": "Calibrated model replicates the decile return pattern as well as aggregate and firm-level labor moments",
229 + "evidence_type": "table",
230 + "evidence_ref": "Tables VIII, IX"
231 + },
232 + {
233 + "claim": "The pattern requires matching efficiency shocks to be priced",
234 + "evidence_type": "table",
235 + "evidence_ref": "Table X"
236 + },
237 + {
238 + "claim": "Portfolios sorted on loadings differ in the cyclicality of their labor characteristics, as the model predicts",
239 + "evidence_type": "table",
240 + "evidence_ref": "Table XII"
241 + },
242 + {
243 + "claim": "Labor market tightness is procyclical and persistent",
244 + "evidence_type": "figure",
245 + "evidence_ref": "Figure 1"
246 + }
247 + ],
248 + "structural_summary": "Opens with labor market flows as motivation and states the headline claim that search frictions matter for the cross-section of returns. Previews the empirical design (tightness loadings, sorts) and then a partial equilibrium search model, with an extended mechanism sketch contrasting cash flow and discount rate channels. Literature positioning spans several intro paragraphs and is followed by a roadmap. Body sequence: empirical facts, model, calibration with quantitative replication, mechanism check on micro labor data, and a conclusion that restates results and invites further research.",
249 + "writing_profile_keywords": [
250 + "production-based asset pricing",
251 + "labor search frictions",
252 + "portfolio sorts",
253 + "Fama-MacBeth",
254 + "calibrated heterogeneous-firm model",
255 + "two-factor pricing",
256 + "mechanism inspection"
257 + ],
258 + "confidence": 0.8,
259 + "notes": "Primary method is debatable between portfolio sorts and calibration; the paper combines both. The 6% magnitude is stated in the abstract and conclusion, but the truncated intro excerpt does not show it. Robustness appears within the results sections and in the Internet Appendix."
19 260 },
20 − "cost_usd": 0.084646,
21 − "request_id": "req_011Cggpi1Y7G1FaQpkW64PC5",
22 − "annotation": null,
23 261 "deterministic_checks": {
24 262 "roadmap_regex": true,
25 263 "contribution_regex_in_intro": true,
@@ -45,7 +283,6 @@
45 283 "9": "appendix"
46 284 }
47 285 },
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"
286 + "schema_valid": true,
287 + "code_version": "87c73cd-dirty"
51 288 }
\ No newline at end of file
modified data/annotations/documents/jof_12518.json +215 −14
@@ -1,25 +1,227 @@
1 1 {
2 − "validation_errors": [
3 − "no json"
4 − ],
2 + "validation_errors": [],
3 + "repairs": [],
5 4 "doc_id": "jof_12518",
6 5 "ontology_version": "1.0",
7 6 "prompt_version": "1.0",
8 7 "model": "claude-opus-5-5",
9 8 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:59:16+00:00",
9 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 10 "source_sha256": "f1de67b38763a6406a2a5f8ea6d7973c5b57e425764e66fe97346dd1ef029529",
12 11 "parser": "pymupdf-jof-heuristics-1.3",
13 12 "payload_chars": 20898,
14 13 "usage": {
15 − "input": 7763,
16 − "output": 2158,
17 − "cache_read": 3968,
18 − "cache_write": 0
14 + "input": 7805,
15 + "output": 1870,
16 + "cache_read": 0,
17 + "cache_write": 3968
18 + },
19 + "cost_usd": 0.08846,
20 + "request_id": "req_011Cggroet385FRJyFSQYwsc",
21 + "annotation": {
22 + "genre": "empirical",
23 + "subfield": "asset_pricing",
24 + "subfield_secondary": "value premium, cash-flow duration, accounting fundamentals",
25 + "primary_method": "reduced_form_regression",
26 + "identification": {
27 + "estimand": "Differences in future cash-flow (dividend, earnings) growth rates between low and high book-to-market portfolios",
28 + "strategy": "Descriptive portfolio sorts (buy-and-hold vs rebalanced), long-run growth extrapolation, Fama-MacBeth firm-level regressions with survivorship adjustments",
29 + "key_assumptions": [
30 + "Delisting proceeds reinvestment adequately corrects survivorship bias",
31 + "Terminal growth perpetuity assumption beyond year 20 for long-run growth"
32 + ],
33 + "threats_addressed": [
34 + "survivorship bias",
35 + "look-back bias",
36 + "rebalancing effect",
37 + "small-stock dominance in equal-weighted portfolios"
38 + ],
39 + "causal_language_strength": 1,
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": 0
50 + },
51 + "method": {
52 + "present": false,
53 + "sentence_index": -1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 0
58 + },
59 + "magnitude": {
60 + "present": false,
61 + "sentence_index": -1
62 + },
63 + "contribution": {
64 + "present": true,
65 + "sentence_index": 2
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": "other"
81 + },
82 + {
83 + "paragraph_index": 2,
84 + "move": "other"
85 + },
86 + {
87 + "paragraph_index": 3,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "mechanism_or_theory_sketch"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "related_literature_positioning"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "contribution_statement"
101 + }
102 + ],
103 + "contribution_paragraph_index": 4,
104 + "contribution_types": [
105 + "new_evidence",
106 + "resolution_of_puzzle"
107 + ],
108 + "roadmap_present": false,
109 + "literature_positioning_location": "inside_introduction"
110 + },
111 + "section_roles": [
112 + {
113 + "idx": 0,
114 + "role": "introduction"
115 + },
116 + {
117 + "idx": 1,
118 + "role": "data"
119 + },
120 + {
121 + "idx": 2,
122 + "role": "results"
123 + },
124 + {
125 + "idx": 3,
126 + "role": "results"
127 + },
128 + {
129 + "idx": 4,
130 + "role": "theory"
131 + },
132 + {
133 + "idx": 5,
134 + "role": "robustness"
135 + },
136 + {
137 + "idx": 6,
138 + "role": "conclusion"
139 + },
140 + {
141 + "idx": 7,
142 + "role": "appendix"
143 + },
144 + {
145 + "idx": 8,
146 + "role": "appendix"
147 + }
148 + ],
149 + "results_presentation": {
150 + "economic_significance_discussed": true,
151 + "statistical_significance_language": "unclear",
152 + "magnitude_statements_in_intro": false
153 + },
154 + "robustness": {
155 + "location": "main_text_section",
156 + "types": [
157 + "alternative_measure",
158 + "alternative_sample_period",
159 + "alternative_specification"
160 + ]
161 + },
162 + "tables": {
163 + "convention_notes": "Notes describe June sorting into value-weighted NYSE-breakpoint quintiles, the $100 investment normalization, CPI deflation and sample years; regression tables report Newey-West t-statistics in parentheses with winsorized variables.",
164 + "units_stated_in_notes": true,
165 + "standard_errors_description": "mixed"
166 + },
167 + "figures": {
168 + "uncertainty_shown": "unclear"
169 + },
170 + "conclusion": {
171 + "bounded_claims": true,
172 + "policy_implications": false,
173 + "open_questions": false,
174 + "restates_contribution": true
175 + },
176 + "hedging": {
177 + "level": "moderate",
178 + "typical_devices": [
179 + "'do not have substantially higher'",
180 + "'suggest'",
181 + "'unlikely to explain'",
182 + "'help explain'",
183 + "'appear to grow faster'"
184 + ]
185 + },
186 + "claim_evidence": [
187 + {
188 + "claim": "In buy-and-hold portfolios, growth stocks' dividends do not grow substantially faster than value stocks'",
189 + "evidence_type": "table",
190 + "evidence_ref": "Table I, Figure 1"
191 + },
192 + {
193 + "claim": "In rebalanced portfolios, value stocks' dividends grow faster than growth stocks'",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table II, Table VI"
196 + },
197 + {
198 + "claim": "After survivorship adjustment, book-to-market no longer predicts lower firm-level dividend growth",
199 + "evidence_type": "table",
200 + "evidence_ref": "Tables VIII-IX"
201 + },
202 + {
203 + "claim": "Fama-French ROE persistence implies lower initial earnings growth for growth stocks via efficiency growth",
204 + "evidence_type": "figure",
205 + "evidence_ref": "Figure 2"
206 + },
207 + {
208 + "claim": "Rebalancing raises value stocks' growth and lowers growth stocks' growth relative to buy-and-hold",
209 + "evidence_type": "equation_or_proof",
210 + "evidence_ref": "Section IV"
211 + }
212 + ],
213 + "structural_summary": "The paper opens by contrasting conventional wisdom with a puzzle tied to duration-based explanations of the value premium. It lists four reasons for the conventional view and previews rebuttals of each, then positions itself against prior growth-rate studies. Data are followed by portfolio evidence on dividend growth in buy-and-hold and rebalanced portfolios. A section then reconciles the findings with the conventional view through efficiency growth, survivorship bias and look-back bias. An analytical section derives the rebalancing effect, followed by additional tests and a short conclusion.",
214 + "writing_profile_keywords": [
215 + "challenging conventional wisdom",
216 + "portfolio sorts",
217 + "descriptive evidence",
218 + "reconciliation",
219 + "survivorship bias",
220 + "first-person singular"
221 + ],
222 + "confidence": 0.8,
223 + "notes": "Single-author paper. Statistical significance conventions for the portfolio tables are not visible in the excerpts. Table III's title appears truncated."
19 224 },
20 − "cost_usd": 0.075006,
21 − "request_id": "req_011CggpjJVqsfs4uEEvQbLBR",
22 − "annotation": null,
23 225 "deterministic_checks": {
24 226 "roadmap_regex": false,
25 227 "contribution_regex_in_intro": true,
@@ -44,7 +246,6 @@
44 246 "8": "appendix"
45 247 }
46 248 },
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"
249 + "schema_valid": true,
250 + "code_version": "87c73cd-dirty"
50 251 }
\ No newline at end of file
modified data/annotations/documents/jof_12519.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('typical_devices_note' was unexpected) @ hedging"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped hedging.typical_devices_note"
4 5 ],
5 6 "doc_id": "jof_12519",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:58:40+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "773110aa8eba11104516ee3a48ca9762debd2b6f03715e7e5a39842de147c771",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 22115,
@@ -203,8 +204,7 @@
203 204 "acknowledging the model is stylized",
204 205 "'consistent with' / 'in line with' when linking to evidence",
205 206 "scope limits on where policy implications apply"
206 − ],
207 − "typical_devices_note": null
207 + ]
208 208 },
209 209 "claim_evidence": [
210 210 {
@@ -279,6 +279,6 @@
279 279 "11": "appendix"
280 280 }
281 281 },
282 − "schema_valid": false,
283 − "code_version": "feb6c9e-dirty"
282 + "schema_valid": true,
283 + "code_version": "87c73cd-dirty"
284 284 }
\ No newline at end of file
modified data/annotations/documents/jof_12540.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('notes' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.notes"
4 5 ],
5 6 "doc_id": "jof_12540",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:59:18+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "fc0b38b62ff10b679833bc183e307748583d05c6ac8784fae1a24a2e2b2a5893",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 24070,
@@ -198,8 +199,7 @@
198 199 "standard_errors_description": "none"
199 200 },
200 201 "figures": {
201 − "uncertainty_shown": "no",
202 − "notes": ""
202 + "uncertainty_shown": "no"
203 203 },
204 204 "conclusion": {
205 205 "bounded_claims": false,
@@ -291,6 +291,6 @@
291 291 "10": "appendix"
292 292 }
293 293 },
294 − "schema_valid": false,
295 − "code_version": "feb6c9e-dirty"
294 + "schema_valid": true,
295 + "code_version": "87c73cd-dirty"
296 296 }
\ No newline at end of file
modified data/annotations/documents/jof_12542.json +11 −11
@@ -1,13 +1,18 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('claim_evidence', 'confidence', 'notes', 'structural_summary', 'writing_profile_keywords' were unexpected) @ hedging"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped hedging.claim_evidence",
5 + "dropped hedging.structural_summary",
6 + "dropped hedging.writing_profile_keywords",
7 + "dropped hedging.confidence",
8 + "dropped hedging.notes"
4 9 ],
5 10 "doc_id": "jof_12542",
6 11 "ontology_version": "1.0",
7 12 "prompt_version": "1.0",
8 13 "model": "claude-opus-5-5",
9 14 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:59:34+00:00",
15 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 16 "source_sha256": "b29a4217fefa017878b5e283eb9c6fe598903368d719ca608853fc3664f1368d",
12 17 "parser": "pymupdf-jof-heuristics-1.3",
13 18 "payload_chars": 20662,
@@ -194,12 +199,7 @@
194 199 "'consistent with' the hypothesis",
195 200 "explicit acknowledgment that behavioral explanations cannot be ruled out",
196 201 "limiting generalization to the mutual fund industry"
197 − ],
198 − "claim_evidence": [],
199 − "structural_summary": "",
200 − "writing_profile_keywords": [],
201 − "confidence": 0,
202 − "notes": ""
202 + ]
203 203 },
204 204 "claim_evidence": [
205 205 {
@@ -278,6 +278,6 @@
278 278 "11": "appendix"
279 279 }
280 280 },
281 − "schema_valid": false,
282 − "code_version": "feb6c9e-dirty"
281 + "schema_valid": true,
282 + "code_version": "87c73cd-dirty"
283 283 }
\ No newline at end of file
modified data/annotations/documents/jof_12544.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('convention_notes_placeholder' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.convention_notes_placeholder"
4 5 ],
5 6 "doc_id": "jof_12544",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:59:51+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "9650ee6f9da03c5eb8b6723eba2ae121b3cf5306df986b99a2d6c5a4768760fa",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 19398,
@@ -157,8 +158,7 @@
157 158 "standard_errors_description": "mixed"
158 159 },
159 160 "figures": {
160 − "uncertainty_shown": "mixed",
161 − "convention_notes_placeholder": null
161 + "uncertainty_shown": "mixed"
162 162 },
163 163 "conclusion": {
164 164 "bounded_claims": true,
@@ -240,6 +240,6 @@
240 240 "5": "conclusion"
241 241 }
242 242 },
243 − "schema_valid": false,
244 − "code_version": "feb6c9e-dirty"
243 + "schema_valid": true,
244 + "code_version": "87c73cd-dirty"
245 245 }
\ No newline at end of file
modified data/annotations/documents/jof_12546.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('note' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.note"
4 5 ],
5 6 "doc_id": "jof_12546",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T22:59:51+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "54a87712bbe3d47ae775768397244d621ba5ae9d1089237b69f32029755574d2",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 19588,
@@ -157,8 +158,7 @@
157 158 "standard_errors_description": "unclear"
158 159 },
159 160 "figures": {
160 − "uncertainty_shown": "unclear",
161 − "note": "Model figures plot comparative statics against relative productivity. Empirical figures are scatter and time-series plots."
161 + "uncertainty_shown": "unclear"
162 162 },
163 163 "conclusion": {
164 164 "bounded_claims": true,
@@ -246,6 +246,6 @@
246 246 "7": "appendix"
247 247 }
248 248 },
249 − "schema_valid": false,
250 − "code_version": "feb6c9e-dirty"
249 + "schema_valid": true,
250 + "code_version": "87c73cd-dirty"
251 251 }
\ No newline at end of file
modified data/annotations/documents/jof_12583.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('figures' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.figures"
4 5 ],
5 6 "doc_id": "jof_12583",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:00:27+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "074e6a8620dc1531236a29942c3e072a275fc5bc1f428dc8faaa2dfb0d20b1c0",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 23050,
@@ -172,8 +173,7 @@
172 173 "standard_errors_description": "unclear"
173 174 },
174 175 "figures": {
175 − "uncertainty_shown": "no",
176 − "figures": null
176 + "uncertainty_shown": "no"
177 177 },
178 178 "conclusion": {
179 179 "bounded_claims": true,
@@ -260,6 +260,6 @@
260 260 "5": "conclusion"
261 261 }
262 262 },
263 − "schema_valid": false,
264 − "code_version": "feb6c9e-dirty"
263 + "schema_valid": true,
264 + "code_version": "87c73cd-dirty"
265 265 }
\ No newline at end of file
modified data/annotations/documents/jof_12607.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('figures_note' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.figures_note"
4 5 ],
5 6 "doc_id": "jof_12607",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:01:04+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "e807a0f0dd6857e588e55355b5a1a286d2796b5a6aec069adcbca353f6f3825a",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 24073,
@@ -188,8 +189,7 @@
188 189 "standard_errors_description": "unclear"
189 190 },
190 191 "figures": {
191 − "uncertainty_shown": "no",
192 − "figures_note": "Figures plot posterior probabilities over recursive samples and as functions of the prior multiple; no confidence bands are evident."
192 + "uncertainty_shown": "no"
193 193 },
194 194 "conclusion": {
195 195 "bounded_claims": true,
@@ -282,6 +282,6 @@
282 282 "11": "appendix"
283 283 }
284 284 },
285 − "schema_valid": false,
286 − "code_version": "feb6c9e-dirty"
285 + "schema_valid": true,
286 + "code_version": "87c73cd-dirty"
287 287 }
\ No newline at end of file
modified data/annotations/documents/jof_12619.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('figures' was unexpected) @ tables"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped tables.figures"
4 5 ],
5 6 "doc_id": "jof_12619",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:01:36+00:00",
11 + "annotated_at": "2026-11-03T23:26:10+00:00",
11 12 "source_sha256": "034e718ad966ca13da4b9fbd3f0140a545b75993676a7dd9083c26ee5daff048",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 15389,
@@ -153,8 +154,7 @@
153 154 "tables": {
154 155 "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.",
155 156 "units_stated_in_notes": true,
156 − "standard_errors_description": "clustered",
157 − "figures": null
157 + "standard_errors_description": "clustered"
158 158 },
159 159 "figures": {
160 160 "uncertainty_shown": "no"
@@ -239,6 +239,6 @@
239 239 "5": "conclusion"
240 240 }
241 241 },
242 − "schema_valid": false,
243 − "code_version": "feb6c9e-dirty"
242 + "schema_valid": true,
243 + "code_version": "87c73cd-dirty"
244 244 }
\ No newline at end of file
modified data/annotations/documents/jof_12685.json +225 −13
@@ -1,25 +1,238 @@
1 1 {
2 − "validation_errors": [
3 − "no json"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.convention_notes_absent"
4 5 ],
5 6 "doc_id": "jof_12685",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:02:31+00:00",
11 + "annotated_at": "2026-11-03T23:26:28+00:00",
11 12 "source_sha256": "334f7d010f7f5af26c2405d2d81140f1c08a24989b39e6015f3a6937a71b35c5",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 25219,
14 15 "usage": {
15 − "input": 8269,
16 − "output": 2130,
17 − "cache_read": 3968,
18 − "cache_write": 0
16 + "input": 8311,
17 + "output": 2007,
18 + "cache_read": 0,
19 + "cache_write": 3968
20 + },
21 + "cost_usd": 0.093224,
22 + "request_id": "req_011CggrofhQR8XspmamsZo1N",
23 + "annotation": {
24 + "genre": "empirical",
25 + "subfield": "behavioral_finance",
26 + "subfield_secondary": "asset_pricing",
27 + "primary_method": "reduced_form_regression",
28 + "identification": {
29 + "estimand": "Effect of the previous trading day's salient (large-firm value-weighted) earnings surprise on the return reaction to a firm's own earnings announcement, holding its own surprise constant",
30 + "strategy": "Regressions of announcement returns on prior-day salient surprise, exploiting earnings dates scheduled weeks in advance as quasi-exogenous ordering; timing tests, unconditional return predictability, reversals and a trading strategy rule out information transmission",
31 + "key_assumptions": [
32 + "Announcement dates are set in advance and are uncorrelated with the fundamentals of the firm relative to prior-day surprises",
33 + "Prior-day surprises do not carry information relevant to the next-day announcer that rational markets would price",
34 + "Absence of strategic timing of announcements"
35 + ],
36 + "threats_addressed": [
37 + "information transmission from earlier announcements",
38 + "expectational errors versus perceptual errors",
39 + "risk changes, trading frictions and capital constraints",
40 + "strategic timing of earnings announcements",
41 + "bias or staleness of analyst forecasts"
42 + ],
43 + "causal_language_strength": 4,
44 + "uses_causal_claim": true
45 + },
46 + "abstract_moves": {
47 + "question": {
48 + "present": true,
49 + "sentence_index": 0
50 + },
51 + "gap": {
52 + "present": true,
53 + "sentence_index": 1
54 + },
55 + "method": {
56 + "present": true,
57 + "sentence_index": 3
58 + },
59 + "results": {
60 + "present": true,
61 + "sentence_index": 2
62 + },
63 + "magnitude": {
64 + "present": false,
65 + "sentence_index": -1
66 + },
67 + "contribution": {
68 + "present": true,
69 + "sentence_index": 1
70 + }
71 + },
72 + "introduction": {
73 + "moves": [
74 + {
75 + "paragraph_index": 0,
76 + "move": "other"
77 + },
78 + {
79 + "paragraph_index": 7,
80 + "move": "motivation"
81 + },
82 + {
83 + "paragraph_index": 8,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 9,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 10,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 11,
96 + "move": "identification_defense"
97 + },
98 + {
99 + "paragraph_index": 12,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 13,
104 + "move": "contribution_statement"
105 + },
106 + {
107 + "paragraph_index": 13,
108 + "move": "related_literature_positioning"
109 + }
110 + ],
111 + "contribution_paragraph_index": 13,
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": "data"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "results"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "robustness"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "robustness"
138 + },
139 + {
140 + "idx": 5,
141 + "role": "results"
142 + },
143 + {
144 + "idx": 6,
145 + "role": "conclusion"
146 + }
147 + ],
148 + "results_presentation": {
149 + "economic_significance_discussed": true,
150 + "statistical_significance_language": "stars_and_tstats",
151 + "magnitude_statements_in_intro": true
152 + },
153 + "robustness": {
154 + "location": "main_text_section",
155 + "types": [
156 + "alternative_measure",
157 + "subsample",
158 + "alternative_specification",
159 + "falsification",
160 + "placebo"
161 + ]
162 + },
163 + "tables": {
164 + "convention_notes": "Regression tables report standard errors in parentheses, clustered by date, with stars marking 10/5/1% significance. Notes define variable construction and return windows and refer back to Table II for definitions.",
165 + "units_stated_in_notes": true,
166 + "standard_errors_description": "clustered"
167 + },
168 + "figures": {
169 + "uncertainty_shown": "unclear"
170 + },
171 + "conclusion": {
172 + "bounded_claims": true,
173 + "policy_implications": true,
174 + "open_questions": true,
175 + "restates_contribution": true
176 + },
177 + "hedging": {
178 + "level": "moderate",
179 + "typical_devices": [
180 + "consistent with",
181 + "suggestive evidence",
182 + "potentially",
183 + "likely",
184 + "we leave for future work"
185 + ]
186 + },
187 + "claim_evidence": [
188 + {
189 + "claim": "Return reaction to today's announcement falls as the prior-day salient surprise rises, controlling for own surprise",
190 + "evidence_type": "figure",
191 + "evidence_ref": "Figure 1; Table II"
192 + },
193 + {
194 + "claim": "The effect is a level shift rather than an interaction with own surprise",
195 + "evidence_type": "table",
196 + "evidence_ref": "Table III; Figure 3"
197 + },
198 + {
199 + "claim": "Distortions reverse over roughly 50 trading days",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table IV"
202 + },
203 + {
204 + "claim": "Long-short strategy on prior-day surprise earns about 15% per year among large firms",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table V; Figure 4"
207 + },
208 + {
209 + "claim": "Prior-day surprise does not predict next-day announcer surprises, against information transmission",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table VI"
212 + },
213 + {
214 + "claim": "Results are not driven by risk, trading frictions or strategic timing",
215 + "evidence_type": "table",
216 + "evidence_ref": "Tables VII, VIII"
217 + },
218 + {
219 + "claim": "Effect is concentrated at t-1 and holds within the day; heterogeneity exists by size and industry",
220 + "evidence_type": "table",
221 + "evidence_ref": "Tables X, XI, XII"
222 + }
223 + ],
224 + "structural_summary": "Opens with a classical epigraph and lab or popular examples of contrast effects, then sketches the prediction for earnings announcements. It previews the main figure and its magnitude and defends against information-transmission explanations through timing, reversals and a trading strategy. Supplementary timing and heterogeneity results follow, then contributions positioned against limited-attention and expectational-error literatures. The body proceeds through data, baseline results, a long section on alternative explanations, robustness, further exploration and a conclusion with broader implications.",
225 + "writing_profile_keywords": [
226 + "behavioral bias in markets",
227 + "lab-to-field",
228 + "alternative explanations section",
229 + "trading strategy",
230 + "earnings announcements",
231 + "epigraph opening"
232 + ],
233 + "confidence": 0.8,
234 + "notes": "Introduction paragraphs 0-6 are fragments of the epigraph and are treated as a single 'other' move. Paragraph 7 is truncated. No explicit roadmap is visible in the excerpts."
19 235 },
20 − "cost_usd": 0.07647,
21 − "request_id": "req_011CggpyixtzBwxJLpmxpdtR",
22 − "annotation": null,
23 236 "deterministic_checks": {
24 237 "roadmap_regex": false,
25 238 "contribution_regex_in_intro": true,
@@ -42,7 +255,6 @@
42 255 "6": "conclusion"
43 256 }
44 257 },
45 − "schema_valid": false,
46 − "code_version": "feb6c9e-dirty",
47 − "raw_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\":\"contribu"
258 + "schema_valid": true,
259 + "code_version": "87c73cd-dirty"
48 260 }
\ No newline at end of file
modified data/annotations/documents/jof_12695.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('notes_placeholder_removed' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.notes_placeholder_removed"
4 5 ],
5 6 "doc_id": "jof_12695",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:03:03+00:00",
11 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 12 "source_sha256": "72406e6c4d5127db79585517238a375fada0c728748d832063d41bb2274bbec4",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 20098,
@@ -141,8 +142,7 @@
141 142 "standard_errors_description": "none"
142 143 },
143 144 "figures": {
144 − "uncertainty_shown": "no",
145 − "notes_placeholder_removed": null
145 + "uncertainty_shown": "no"
146 146 },
147 147 "conclusion": {
148 148 "bounded_claims": true,
@@ -214,6 +214,6 @@
214 214 "4": "conclusion"
215 215 }
216 216 },
217 − "schema_valid": false,
218 − "code_version": "feb6c9e-dirty"
217 + "schema_valid": true,
218 + "code_version": "87c73cd-dirty"
219 219 }
\ No newline at end of file
modified data/annotations/documents/jof_12696.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('notes_placeholder_removed' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.notes_placeholder_removed"
4 5 ],
5 6 "doc_id": "jof_12696",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:02:39+00:00",
11 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 12 "source_sha256": "69290b0f8b85853098a638f914271cdd1d90690c05130b354115d030f73647a2",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 19268,
@@ -174,8 +175,7 @@
174 175 "standard_errors_description": "none"
175 176 },
176 177 "figures": {
177 − "uncertainty_shown": "no",
178 − "notes_placeholder_removed": ""
178 + "uncertainty_shown": "no"
179 179 },
180 180 "conclusion": {
181 181 "bounded_claims": false,
@@ -259,6 +259,6 @@
259 259 "8": "appendix"
260 260 }
261 261 },
262 − "schema_valid": false,
263 − "code_version": "feb6c9e-dirty"
262 + "schema_valid": true,
263 + "code_version": "87c73cd-dirty"
264 264 }
\ No newline at end of file
modified data/annotations/documents/jof_12700.json +6 −9
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('results_note_placeholder' was unexpected) @ abstract_moves"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped abstract_moves.results_note_placeholder"
4 5 ],
5 6 "doc_id": "jof_12700",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:03:32+00:00",
11 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 12 "source_sha256": "2376561d743450ac3bcfb4700b6448879ff8b756671a4d4df947378d217691ac",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 21647,
@@ -66,10 +67,6 @@
66 67 "contribution": {
67 68 "present": true,
68 69 "sentence_index": 3
69 − },
70 − "results_note_placeholder": {
71 − "present": false,
72 − "sentence_index": -1
73 70 }
74 71 },
75 72 "introduction": {
@@ -293,6 +290,6 @@
293 290 "15": "conclusion"
294 291 }
295 292 },
296 − "schema_valid": false,
297 − "code_version": "feb6c9e-dirty"
293 + "schema_valid": true,
294 + "code_version": "87c73cd-dirty"
298 295 }
\ No newline at end of file
modified data/annotations/documents/jof_12719.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('typical_devices_note' was unexpected) @ hedging"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped hedging.typical_devices_note"
4 5 ],
5 6 "doc_id": "jof_12719",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:04:23+00:00",
11 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 12 "source_sha256": "c2e062783a97b85465abe7f963477035699713f6a7b03961098569c8408ecd13",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 17584,
@@ -188,8 +189,7 @@
188 189 "'we present some evidence... further work is needed'",
189 190 "acknowledging what is not systematically examined",
190 191 "'helps explain' / 'offers an alternative explanation'"
191 − ],
192 − "typical_devices_note": null
192 + ]
193 193 },
194 194 "claim_evidence": [
195 195 {
@@ -259,6 +259,6 @@
259 259 "7": "appendix"
260 260 }
261 261 },
262 − "schema_valid": false,
263 − "code_version": "feb6c9e-dirty"
262 + "schema_valid": true,
263 + "code_version": "87c73cd-dirty"
264 264 }
\ No newline at end of file
modified data/annotations/documents/jof_12720.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('figure_note' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.figure_note"
4 5 ],
5 6 "doc_id": "jof_12720",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:04:27+00:00",
11 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 12 "source_sha256": "d409a0750d2309addcb8652230c03b2021d7e9e4a7f17db941dd7e9a86d2a839",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 17654,
@@ -161,8 +162,7 @@
161 162 "standard_errors_description": "unclear"
162 163 },
163 164 "figures": {
164 − "uncertainty_shown": "mixed",
165 − "figure_note": "none"
165 + "uncertainty_shown": "mixed"
166 166 },
167 167 "conclusion": {
168 168 "bounded_claims": true,
@@ -239,6 +239,6 @@
239 239 "4": "conclusion"
240 240 }
241 241 },
242 − "schema_valid": false,
243 − "code_version": "feb6c9e-dirty"
242 + "schema_valid": true,
243 + "code_version": "87c73cd-dirty"
244 244 }
\ No newline at end of file
modified data/annotations/documents/jof_12732.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('no_figures' was unexpected) @ figures"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.no_figures"
4 5 ],
5 6 "doc_id": "jof_12732",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:05:02+00:00",
11 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 12 "source_sha256": "0d4b1b03704ba9c76e8d8bdca98cd156f66cbf5d320faf72e8327dbcf6dc061b",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 21653,
@@ -194,8 +195,7 @@
194 195 "standard_errors_description": "robust"
195 196 },
196 197 "figures": {
197 − "uncertainty_shown": "no",
198 − "no_figures": null
198 + "uncertainty_shown": "no"
199 199 },
200 200 "conclusion": {
201 201 "bounded_claims": true,
@@ -287,6 +287,6 @@
287 287 "13": "appendix"
288 288 }
289 289 },
290 − "schema_valid": false,
291 − "code_version": "feb6c9e-dirty"
290 + "schema_valid": true,
291 + "code_version": "87c73cd-dirty"
292 292 }
\ No newline at end of file
added data/annotations/documents/jof_12734.json +251 −0
@@ -0,0 +1,251 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12734",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:13+00:00",
9 + "source_sha256": "e94f1ac2e8a9a8db875d86ddfe1e17ed29c2bbf94dafac6e28594aebdd0c7db1",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 16097,
12 + "usage": {
13 + "input": 5901,
14 + "output": 2185,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.068098,
19 + "request_id": "req_011CggqGABeYE9wTYTzhNLup",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Degree of expectation stickiness (lambda) in analyst forecasts, and how strongly profitability-anomaly returns vary with stickiness and with profit persistence",
27 + "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.",
28 + "key_assumptions": [
29 + "Analyst forecasts represent market participants' beliefs",
30 + "Profits are predictable from persistent public signals",
31 + "Sticky-expectation updating nests rational expectations"
32 + ],
33 + "threats_addressed": [
34 + "look-ahead bias in firm-level stickiness estimation",
35 + "mechanical correlation under rational-expectations null (simulations)",
36 + "past trend in profits as alternative explanation",
37 + "risk adjustment via factor models"
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": 3
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": 0
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": 1,
80 + "move": "motivation"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "related_literature_positioning"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "contribution_statement"
97 + }
98 + ],
99 + "contribution_paragraph_index": 2,
100 + "contribution_types": [
101 + "new_theory_or_model",
102 + "new_evidence",
103 + "new_data_or_measure"
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": "data"
120 + },
121 + {
122 + "idx": 3,
123 + "role": "results"
124 + },
125 + {
126 + "idx": 4,
127 + "role": "robustness"
128 + },
129 + {
130 + "idx": 5,
131 + "role": "conclusion"
132 + },
133 + {
134 + "idx": 6,
135 + "role": "appendix"
136 + }
137 + ],
138 + "results_presentation": {
139 + "economic_significance_discussed": true,
140 + "statistical_significance_language": "unclear",
141 + "magnitude_statements_in_intro": false
142 + },
143 + "robustness": {
144 + "location": "main_text_section",
145 + "types": [
146 + "placebo",
147 + "alternative_specification",
148 + "other"
149 + ]
150 + },
151 + "tables": {
152 + "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).",
153 + "units_stated_in_notes": true,
154 + "standard_errors_description": "unclear"
155 + },
156 + "figures": {
157 + "uncertainty_shown": "unclear"
158 + },
159 + "conclusion": {
160 + "bounded_claims": true,
161 + "policy_implications": false,
162 + "open_questions": false,
163 + "restates_contribution": true
164 + },
165 + "hedging": {
166 + "level": "moderate",
167 + "typical_devices": [
168 + "conditional assumptions (assuming analyst forecasts are representative)",
169 + "'in line with' interpretations",
170 + "'suggests'"
171 + ]
172 + },
173 + "claim_evidence": [
174 + {
175 + "claim": "The profitability anomaly carries large economic significance: high Sharpe ratio, no crash risk, high capacity",
176 + "evidence_type": "prior_literature",
177 + "evidence_ref": "Novy-Marx (2013, 2015); Lemperiere et al. (2017); Landier, Simon, Thesmar (2015)"
178 + },
179 + {
180 + "claim": "Analyst expectations are sticky: forecast errors are positively related to past forecast revisions",
181 + "evidence_type": "table",
182 + "evidence_ref": "Table III, Figure 1"
183 + },
184 + {
185 + "claim": "Analysts are too pessimistic about future profits of high-profit firms",
186 + "evidence_type": "table",
187 + "evidence_ref": "Table VI, Figure 2"
188 + },
189 + {
190 + "claim": "The profitability anomaly is stronger for stocks followed by stickier analysts",
191 + "evidence_type": "table",
192 + "evidence_ref": "Table VII"
193 + },
194 + {
195 + "claim": "The profitability anomaly is stronger for firms with more persistent profits",
196 + "evidence_type": "table",
197 + "evidence_ref": "Table VIII"
198 + },
199 + {
200 + "claim": "Double-sort results are not mechanical under the rational-expectations null",
201 + "evidence_type": "figure",
202 + "evidence_ref": "Figure 3"
203 + },
204 + {
205 + "claim": "Less experienced and busier analysts have stickier beliefs, consistent with limited attention",
206 + "evidence_type": "table",
207 + "evidence_ref": "Table V"
208 + },
209 + {
210 + "claim": "In the sticky-belief model, past profits forecast future returns",
211 + "evidence_type": "equation_or_proof",
212 + "evidence_ref": "Section I; Appendix proofs"
213 + }
214 + ],
215 + "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.",
216 + "writing_profile_keywords": [
217 + "proposition-driven testing",
218 + "behavioral asset pricing",
219 + "analyst forecasts",
220 + "portfolio sorts",
221 + "model-to-data mapping",
222 + "sticky expectations"
223 + ],
224 + "confidence": 0.8,
225 + "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."
226 + },
227 + "deterministic_checks": {
228 + "roadmap_regex": false,
229 + "contribution_regex_in_intro": false,
230 + "magnitude_in_abstract": false,
231 + "abstract_n_sentences": 4,
232 + "intro_n_paragraphs": 3,
233 + "intro_n_words": 624,
234 + "causal_words_in_abstract": 0,
235 + "hedge_words_in_abstract": 0,
236 + "stars_mentioned": false,
237 + "clustered_se_mentioned": false,
238 + "internet_appendix_mentioned": false,
239 + "section_roles_rule": {
240 + "0": "introduction",
241 + "1": "theory",
242 + "2": "data",
243 + "3": "theory",
244 + "4": "robustness",
245 + "5": "conclusion",
246 + "6": "appendix"
247 + }
248 + },
249 + "schema_valid": true,
250 + "code_version": "feb6c9e-dirty"
251 +}
\ No newline at end of file
added data/annotations/documents/jof_12743.json +268 −0
@@ -0,0 +1,268 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12743",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:44+00:00",
9 + "source_sha256": "b3bd207c989a7e383a1c96c2e78b6c328920f716aea7167c6f58f78cb21deb4b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23027,
12 + "usage": {
13 + "input": 7927,
14 + "output": 2100,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.074502,
19 + "request_id": "req_011CggqJTkdqUhgQRaTD2TLv",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "behavioral_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Marginal effect of investors' unrealized returns since purchase on announcement-window selling probability and abnormal volume/returns, conditional on earnings news",
27 + "strategy": "Cox proportional hazard models of fund-manager sales using Ancerno position-level data, controlling for earnings surprise; stock-level OLS regressions of abnormal volume and returns on capital gains overhang interacted with unexpected earnings, with portfolio sorts",
28 + "key_assumptions": [
29 + "Trades in short windows around earnings announcements are motivated by the announcements",
30 + "Proportional hazards assumption holds",
31 + "CGO proxies the representative investor's unrealized gain/loss position",
32 + "Controlling for earnings news isolates preference effects from belief shocks"
33 + ],
34 + "threats_addressed": [
35 + "Belief shocks confounding preference-driven selling (controls for earnings surprise)",
36 + "Violation of proportional hazards assumption",
37 + "Tax-motivated trading and turn-of-year effects",
38 + "January-performance explanation for CGO return predictability (Novy-Marx)",
39 + "Alternative CGO measure (Frazzini)"
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": 0
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": "tension_or_puzzle"
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_approach"
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_results"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "mechanism_or_theory_sketch"
107 + },
108 + {
109 + "paragraph_index": 5,
110 + "move": "contribution_statement"
111 + },
112 + {
113 + "paragraph_index": 5,
114 + "move": "related_literature_positioning"
115 + },
116 + {
117 + "paragraph_index": 6,
118 + "move": "contribution_statement"
119 + },
120 + {
121 + "paragraph_index": 6,
122 + "move": "related_literature_positioning"
123 + },
124 + {
125 + "paragraph_index": 7,
126 + "move": "roadmap"
127 + }
128 + ],
129 + "contribution_paragraph_index": 5,
130 + "contribution_types": [
131 + "new_evidence",
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": "conclusion"
161 + },
162 + {
163 + "idx": 6,
164 + "role": "appendix"
165 + }
166 + ],
167 + "results_presentation": {
168 + "economic_significance_discussed": true,
169 + "statistical_significance_language": "unclear",
170 + "magnitude_statements_in_intro": true
171 + },
172 + "robustness": {
173 + "location": "mixed",
174 + "types": [
175 + "alternative_measure",
176 + "subsample",
177 + "alternative_specification",
178 + "other"
179 + ]
180 + },
181 + "tables": {
182 + "convention_notes": "Notes define the sample and estimator (OLS), refer to the Appendix for variable definitions, state that continuous regressors are ranked into quarterly deciles and scaled, and report significance thresholds for correlations.",
183 + "units_stated_in_notes": true,
184 + "standard_errors_description": "unclear"
185 + },
186 + "figures": {
187 + "uncertainty_shown": "unclear"
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 + "'consistent with' framing",
199 + "'suggest' verbs",
200 + "'generally consistent'",
201 + "ceteris paribus qualifiers"
202 + ]
203 + },
204 + "claim_evidence": [
205 + {
206 + "claim": "Fund managers' marginal selling schedule approximates an inverse V-shape around zero unrealized returns, steeper in the loss domain",
207 + "evidence_type": "figure",
208 + "evidence_ref": "Figure 1, Figure 2"
209 + },
210 + {
211 + "claim": "Gain-position holders sell more and loss-position holders sell less as earnings news improves",
212 + "evidence_type": "figure",
213 + "evidence_ref": "Figure 3"
214 + },
215 + {
216 + "claim": "Hazard-of-sale for 22–26% winners is about 1.24 times that of 22–26% losers",
217 + "evidence_type": "figure",
218 + "evidence_ref": "Figure 2"
219 + },
220 + {
221 + "claim": "Stockholders' CGO moderates abnormal seller-initiated volume around announcements",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table IV"
224 + },
225 + {
226 + "claim": "CGO is associated with short-window return underreaction to earnings news",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table V, Table VI, Figure 4, Figure 5"
229 + }
230 + ],
231 + "structural_summary": "The introduction motivates the work with a puzzle: preference-based accounts of the disposition effect are questioned by evidence of V-shaped selling. It then previews investor-level hazard models and stock-level volume and return tests. The results are presented with a magnitude example and an emotion-based interpretation, followed by contribution claims positioned against prior literature and a roadmap. The body moves from data, to investor-level tests (motivation, method, proportional-hazards checks), to two stock-level test sections. A combined robustness-and-conclusion section summarizes untabulated and Internet Appendix checks and suggests directions for future research.",
232 + "writing_profile_keywords": [
233 + "disposition effect",
234 + "prospect theory",
235 + "hazard model",
236 + "earnings announcements",
237 + "multi-level evidence",
238 + "first-person singular",
239 + "consistent-with hedging"
240 + ],
241 + "confidence": 0.8,
242 + "notes": "Single-author paper written in the first person. Robustness checks are partly untabulated and partly placed in the Internet Appendix. Standard-error treatment is not visible in the excerpts."
243 + },
244 + "deterministic_checks": {
245 + "roadmap_regex": true,
246 + "contribution_regex_in_intro": true,
247 + "magnitude_in_abstract": false,
248 + "abstract_n_sentences": 5,
249 + "intro_n_paragraphs": 8,
250 + "intro_n_words": 1148,
251 + "causal_words_in_abstract": 0,
252 + "hedge_words_in_abstract": 2,
253 + "stars_mentioned": false,
254 + "clustered_se_mentioned": false,
255 + "internet_appendix_mentioned": false,
256 + "section_roles_rule": {
257 + "0": "introduction",
258 + "1": "data",
259 + "2": "other",
260 + "3": "other",
261 + "4": "other",
262 + "5": "conclusion",
263 + "6": "appendix"
264 + }
265 + },
266 + "schema_valid": true,
267 + "code_version": "87c73cd-dirty"
268 +}
\ No newline at end of file
added data/annotations/documents/jof_12744.json +287 −0
@@ -0,0 +1,287 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12744",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:37+00:00",
9 + "source_sha256": "b2d7e013a77f41f5d17444cce2cf5a874fcfcd76033548c1dcffe23045ba96f6",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 28363,
12 + "usage": {
13 + "input": 10054,
14 + "output": 2796,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.09693,
19 + "request_id": "req_011CggqHRA69YfTqZ2qWpPsC",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "market_microstructure",
24 + "primary_method": "calibration_simulation",
25 + "identification": {
26 + "estimand": "not applicable (theory paper; comparative statics of equilibrium prices, returns, welfare with respect to proportional trading fees)",
27 + "strategy": "not applicable; dynamic multiagent general-equilibrium model with incomplete markets solved numerically via a novel synchronizing algorithm, with illustrative calibrated examples",
28 + "key_assumptions": [
29 + "finite-horizon discrete-time exchange economy on a tree/lattice",
30 + "incomplete markets with unhedgeable endowments giving an every-period trading motive",
31 + "proportional trading fees on the stock, liquid bond",
32 + "time-additive utility in illustrations"
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": 0
50 + },
51 + "results": {
52 + "present": true,
53 + "sentence_index": 3
54 + },
55 + "magnitude": {
56 + "present": false,
57 + "sentence_index": -1
58 + },
59 + "contribution": {
60 + "present": true,
61 + "sentence_index": 1
62 + }
63 + },
64 + "introduction": {
65 + "moves": [
66 + {
67 + "paragraph_index": 0,
68 + "move": "motivation"
69 + },
70 + {
71 + "paragraph_index": 0,
72 + "move": "preview_of_approach"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "mechanism_or_theory_sketch"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 2,
84 + "move": "preview_of_results"
85 + },
86 + {
87 + "paragraph_index": 3,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 4,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "mechanism_or_theory_sketch"
97 + },
98 + {
99 + "paragraph_index": 5,
100 + "move": "contribution_statement"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "policy_or_broader_implication"
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": "roadmap"
117 + }
118 + ],
119 + "contribution_paragraph_index": 5,
120 + "contribution_types": [
121 + "new_theory_or_model",
122 + "new_method"
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": "theory"
135 + },
136 + {
137 + "idx": 2,
138 + "role": "theory"
139 + },
140 + {
141 + "idx": 3,
142 + "role": "theory"
143 + },
144 + {
145 + "idx": 4,
146 + "role": "results"
147 + },
148 + {
149 + "idx": 5,
150 + "role": "results"
151 + },
152 + {
153 + "idx": 6,
154 + "role": "conclusion"
155 + },
156 + {
157 + "idx": 7,
158 + "role": "appendix"
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": "unclear",
180 + "magnitude_statements_in_intro": false
181 + },
182 + "robustness": {
183 + "location": "within_results",
184 + "types": [
185 + "sensitivity_to_parameters",
186 + "other"
187 + ]
188 + },
189 + "tables": {
190 + "convention_notes": "Few tables; notes describe model quantities normalized by aggregate output, simulation path counts and data sources for calibration moments; no standard errors or significance markers.",
191 + "units_stated_in_notes": true,
192 + "standard_errors_description": "none"
193 + },
194 + "figures": {
195 + "uncertainty_shown": "no"
196 + },
197 + "conclusion": {
198 + "bounded_claims": true,
199 + "policy_implications": false,
200 + "open_questions": false,
201 + "restates_contribution": true
202 + },
203 + "hedging": {
204 + "level": "moderate",
205 + "typical_devices": [
206 + "framing numerical results as illustrative rather than calibrated to real data",
207 + "'in calibrated examples' qualifiers",
208 + "'so far' framing when comparing competing models",
209 + "explicit statement of model limitations in appendix"
210 + ]
211 + },
212 + "claim_evidence": [
213 + {
214 + "claim": "Trading fees raise both bond and stock prices via a precautionary-savings effect from higher individual consumption volatility",
215 + "evidence_type": "figure",
216 + "evidence_ref": "Figures 4 and 5"
217 + },
218 + {
219 + "claim": "Interest rate falls while equity premium, return volatility and Sharpe ratio rise with fees",
220 + "evidence_type": "figure",
221 + "evidence_ref": "Figure 7"
222 + },
223 + {
224 + "claim": "Expected liquidity and liquidity risk generate deviations from the classic CCAPM, with time variation driven mainly by expected liquidity",
225 + "evidence_type": "figure",
226 + "evidence_ref": "Figure 8"
227 + },
228 + {
229 + "claim": "Adding a third trader does not shrink deviations from frictionless equilibrium; many trades are bilateral",
230 + "evidence_type": "figure",
231 + "evidence_ref": "Figure 9; Table II"
232 + },
233 + {
234 + "claim": "Slow-moving capital under fees generates slow price reversal, similar to inattention models",
235 + "evidence_type": "figure",
236 + "evidence_ref": "Figure 10"
237 + },
238 + {
239 + "claim": "Analytical bounds and comparisons of prices with and without fees",
240 + "evidence_type": "equation_or_proof",
241 + "evidence_ref": "Section III; Appendix C (Proposition 2)"
242 + }
243 + ],
244 + "structural_summary": "Theory-driven paper. The introduction motivates trading fees as a financial-sector service and poses three questions: mechanisms, pricing of liquidity risk, and slow price reversal. It previews the model and its results paragraph by paragraph, claims a methodological contribution (a Walrasian equilibrium concept and a synchronizing algorithm), positions the work against earlier transaction-cost equilibrium papers and dual methods, and ends with a roadmap. The body builds a general model, solves a frictionless benchmark, adds fees with analytical price bounds and the algorithm, then reports numerical illustrations across fee levels. Three applications follow (a CCAPM decomposition, three traders, impulse responses). The conclusion restates the contributions, while appendices hold the setup, FOCs, proof and algorithm details.",
245 + "writing_profile_keywords": [
246 + "general equilibrium",
247 + "trading fees",
248 + "numerical algorithm",
249 + "illustrative calibration",
250 + "comparative statics across fee levels",
251 + "figure-heavy results",
252 + "liquidity premia",
253 + "methodological contribution"
254 + ],
255 + "confidence": 0.8,
256 + "notes": "Classified as theoretical with calibration/simulation as the primary method; the empirical-identification fields are not applicable. Robustness consists mainly of varying fee levels and a three-trader extension. The 'question' move is implicit rather than stated in the abstract."
257 + },
258 + "deterministic_checks": {
259 + "roadmap_regex": true,
260 + "contribution_regex_in_intro": true,
261 + "magnitude_in_abstract": false,
262 + "abstract_n_sentences": 16,
263 + "intro_n_paragraphs": 9,
264 + "intro_n_words": 1265,
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": "theory",
273 + "2": "theory",
274 + "3": "theory",
275 + "4": "results",
276 + "5": "other",
277 + "6": "conclusion",
278 + "7": "appendix",
279 + "8": "appendix",
280 + "9": "appendix",
281 + "10": "appendix",
282 + "11": "appendix"
283 + }
284 + },
285 + "schema_valid": true,
286 + "code_version": "87c73cd-dirty"
287 +}
\ No newline at end of file
added data/annotations/documents/jof_12745.json +262 −0
@@ -0,0 +1,262 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12745",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:29+00:00",
9 + "source_sha256": "e5585924277d16540b27d0cd93241c2b6b802715a768cf07c7eb31bb0a93b13a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20419,
12 + "usage": {
13 + "input": 6612,
14 + "output": 2195,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.071142,
19 + "request_id": "req_011CggqHH9dcjTCJ3Jp2ymKg",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Association between a model's share of captive (manufacturer) financing and its resale price depreciation, LTVs and production quantities; also the effect of a rise in captive share after an acquisition",
27 + "strategy": "Cross-sectional regressions of model-level depreciation on captive share. Depreciation is decomposed into new-price versus productivity-driven components to separate the commitment and information-asymmetry hypotheses. A natural experiment uses Volvo's 2007 acquisition of an Ingersoll Rand unit. Further tests use within-manufacturer regressions of quantities and time-series variation in loan-officer collateral tightening.",
28 + "key_assumptions": [
29 + "Information asymmetry about quality would show up in productivity-driven (cross-sectional age) depreciation rather than in new-price depreciation",
30 + "The acquisition shifted captive share for reasons unrelated to the machines' ex-ante quality",
31 + "Used machine prices measured at auction reflect market resale values"
32 + ],
33 + "threats_addressed": [
34 + "ex-ante machine quality / signaling alternative",
35 + "market power confound",
36 + "bank credit pull-back as an alternative explanation for machine choice"
37 + ],
38 + "causal_language_strength": 3,
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": 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": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "mechanism_or_theory_sketch"
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": "identification_defense"
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": "related_literature_positioning"
116 + },
117 + {
118 + "paragraph_index": 7,
119 + "move": "contribution_statement"
120 + },
121 + {
122 + "paragraph_index": 7,
123 + "move": "roadmap"
124 + }
125 + ],
126 + "contribution_paragraph_index": 7,
127 + "contribution_types": [
128 + "new_evidence",
129 + "new_theory_or_model"
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": "data"
146 + },
147 + {
148 + "idx": 3,
149 + "role": "results"
150 + },
151 + {
152 + "idx": 4,
153 + "role": "conclusion"
154 + }
155 + ],
156 + "results_presentation": {
157 + "economic_significance_discussed": true,
158 + "statistical_significance_language": "unclear",
159 + "magnitude_statements_in_intro": false
160 + },
161 + "robustness": {
162 + "location": "none_seen",
163 + "types": []
164 + },
165 + "tables": {
166 + "convention_notes": "Table notes describe the sample, the unit of analysis and how depreciation rates are estimated (model-level log price on log age). Variable definitions such as the down payment ratio are given in the notes. Excerpts are truncated, so the standard-error treatment is not visible.",
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": false,
177 + "restates_contribution": true
178 + },
179 + "hedging": {
180 + "level": "moderate",
181 + "typical_devices": [
182 + "'consistent with' framing",
183 + "'strongly suggestive evidence'",
184 + "'move closer to causal inference'",
185 + "'may be reasonable to think'",
186 + "explicit statement that the paper does not rule out other motives"
187 + ]
188 + },
189 + "claim_evidence": [
190 + {
191 + "claim": "Manufacturer-financed shares of new durable investment are large across equipment industries",
192 + "evidence_type": "figure",
193 + "evidence_ref": "Figure 1"
194 + },
195 + {
196 + "claim": "Models with more captive financing depreciate less in resale markets",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table I; Figure 2"
199 + },
200 + {
201 + "claim": "Captive finance is unrelated to productivity-driven depreciation, undermining the quality-signaling view",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table II"
204 + },
205 + {
206 + "claim": "Resale performance of Ingersoll Rand machines improved after the Volvo acquisition raised captive share",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table IV; Figure 4"
209 + },
210 + {
211 + "claim": "Equipment types with a history of captive backing show lower future production within a manufacturer",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table V"
214 + },
215 + {
216 + "claim": "Captive-backed models carry higher LTVs, even when financed by banks",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table VI"
219 + },
220 + {
221 + "claim": "Bank-financed buyers shift toward captive-backed models when collateral requirements tighten",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table VII; Figure 5"
224 + }
225 + ],
226 + "structural_summary": "The paper opens with stylized facts on the scale of manufacturer lending and asks why manufacturers finance their own sales. It proposes a Coase-commitment explanation and contrasts it with an information-asymmetry alternative, then separates the two by decomposing depreciation. A background and hypothesis section precedes a brief data section. A long results section moves from correlations to the alternative-hypothesis test, market power, an acquisition-based natural experiment, quantity effects, and pledgeability spillovers. A short discussion restates the findings and frames captive finance as a costly hedge.",
227 + "writing_profile_keywords": [
228 + "captive finance",
229 + "Coase conjecture",
230 + "durable goods",
231 + "hypothesis discrimination",
232 + "natural experiment",
233 + "pledgeability",
234 + "heavy equipment auction data",
235 + "competing hypotheses"
236 + ],
237 + "confidence": 0.8,
238 + "notes": "No explicit robustness section appears in the excerpts. An internet appendix is mentioned but its contents are not visible. Statistical significance conventions and standard-error treatment are not visible."
239 + },
240 + "deterministic_checks": {
241 + "roadmap_regex": true,
242 + "contribution_regex_in_intro": true,
243 + "magnitude_in_abstract": false,
244 + "abstract_n_sentences": 5,
245 + "intro_n_paragraphs": 8,
246 + "intro_n_words": 1334,
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": false,
252 + "section_roles_rule": {
253 + "0": "introduction",
254 + "1": "literature",
255 + "2": "data",
256 + "3": "results",
257 + "4": "conclusion"
258 + }
259 + },
260 + "schema_valid": true,
261 + "code_version": "87c73cd-dirty"
262 +}
\ No newline at end of file
added data/annotations/documents/jof_12746.json +249 −0
@@ -0,0 +1,249 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12746",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:46+00:00",
9 + "source_sha256": "3c7086c63268ccf590e670a90fa4077b282c62c6ba180aa41a207109a68a9d6a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 15969,
12 + "usage": {
13 + "input": 5907,
14 + "output": 2012,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.064662,
19 + "request_id": "req_011CggqJcWV9Jv5pNpJxSFhb",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "financial_econometrics",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "How well parameter-free earnings surprise measures (CE vs FOM) approximate an ideal bias-adjusted surprise and explain announcement and post-announcement returns",
27 + "strategy": "Analytical model that derives relative-efficiency bounds, followed by comparison of explanatory power in canonical event-study regressions of CAR/POSTCAR on CE, Rank(CE), FOM and indicators, including model selection",
28 + "key_assumptions": [
29 + "A subset of analyst forecasts is biased",
30 + "Market expectation is a filter of individual forecasts",
31 + "Abnormal returns reflect earnings surprise"
32 + ],
33 + "threats_addressed": [
34 + "outliers in forecasts",
35 + "stale forecasts / walk-down effects",
36 + "earnings persistence",
37 + "forecast dispersion",
38 + "international external validity"
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": true,
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": 3
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": "tension_or_puzzle"
82 + },
83 + {
84 + "paragraph_index": 2,
85 + "move": "preview_of_approach"
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": 2,
97 + "move": "policy_or_broader_implication"
98 + }
99 + ],
100 + "contribution_paragraph_index": 2,
101 + "contribution_types": [
102 + "new_data_or_measure",
103 + "new_theory_or_model",
104 + "new_evidence"
105 + ],
106 + "roadmap_present": false,
107 + "literature_positioning_location": "inside_introduction"
108 + },
109 + "section_roles": [
110 + {
111 + "idx": 0,
112 + "role": "introduction"
113 + },
114 + {
115 + "idx": 1,
116 + "role": "theory"
117 + },
118 + {
119 + "idx": 2,
120 + "role": "data"
121 + },
122 + {
123 + "idx": 3,
124 + "role": "results"
125 + },
126 + {
127 + "idx": 4,
128 + "role": "other"
129 + },
130 + {
131 + "idx": 5,
132 + "role": "other"
133 + },
134 + {
135 + "idx": 6,
136 + "role": "conclusion"
137 + },
138 + {
139 + "idx": 7,
140 + "role": "appendix"
141 + }
142 + ],
143 + "results_presentation": {
144 + "economic_significance_discussed": true,
145 + "statistical_significance_language": "stars_and_tstats",
146 + "magnitude_statements_in_intro": true
147 + },
148 + "robustness": {
149 + "location": "within_results",
150 + "types": [
151 + "alternative_measure",
152 + "subsample",
153 + "alternative_specification",
154 + "other"
155 + ]
156 + },
157 + "tables": {
158 + "convention_notes": "Table notes define the dependent variables (CAR over the three-day window, POSTCAR over trading days 2–126) and the surprise measures. Regressions include year dummies. Results are reported as coefficients with t-statistics and R2. The visible notes do not state the standard-error method. The first table reports numerical efficiency bounds from the model.",
159 + "units_stated_in_notes": true,
160 + "standard_errors_description": "unclear"
161 + },
162 + "figures": {
163 + "uncertainty_shown": "no"
164 + },
165 + "conclusion": {
166 + "bounded_claims": false,
167 + "policy_implications": false,
168 + "open_questions": false,
169 + "restates_contribution": true
170 + },
171 + "hedging": {
172 + "level": "low",
173 + "typical_devices": [
174 + "'potentially superior'",
175 + "'where bias is potentially large'",
176 + "conditional framing ('if a subset of forecasts can be biased')"
177 + ]
178 + },
179 + "claim_evidence": [
180 + {
181 + "claim": "CE is a poor parameter-free approximation of the ideal bias-adjusted surprise measure; FOM approximates it far better",
182 + "evidence_type": "equation_or_proof",
183 + "evidence_ref": "Section I; Appendix A"
184 + },
185 + {
186 + "claim": "Lower bound on FOM's relative efficiency versus the infeasible ideal can reach about 50% under plausible parameters",
187 + "evidence_type": "table",
188 + "evidence_ref": "Table I"
189 + },
190 + {
191 + "claim": "FOM explains announcement returns much better than CE; a 1-SD rise in FOM raises CAR by about 1.5%",
192 + "evidence_type": "table",
193 + "evidence_ref": "Table III"
194 + },
195 + {
196 + "claim": "FOM also outperforms in post-announcement drift regressions and in international stocks",
197 + "evidence_type": "table",
198 + "evidence_ref": "Tables IV, V"
199 + },
200 + {
201 + "claim": "Model selection favors FOM-based specifications",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table VI"
204 + },
205 + {
206 + "claim": "FOM beats parameter-dependent bias-adjusted CE in explaining CAR",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table X"
209 + }
210 + ],
211 + "structural_summary": "The paper opens with the standard use of the consensus error in event studies and argues that its validity rests on unbiased forecasts, which analyst bias undermines. It then proposes the fraction-of-misses measure and builds a model showing that the ideal surprise measure is a nonlinear filter downweighting large errors. From the model it derives efficiency bounds and comparative statics. Empirically, it compares CE and FOM in canonical CAR/POSTCAR regressions on I/B/E/S data, with international, model-selection, timing, subsample and parameter-dependent checks inside the results section. Proofs are in the appendix, and the conclusion restates the contribution and its broader applicability to other forecasts.",
212 + "writing_profile_keywords": [
213 + "measurement",
214 + "earnings surprise",
215 + "event study",
216 + "analyst bias",
217 + "theory-then-evidence",
218 + "parameter-free measure",
219 + "relative efficiency"
220 + ],
221 + "confidence": 0.8,
222 + "notes": "Sections idx 4 and 5 appear to be equation fragments mis-parsed as headings. The introduction excerpt is fragmented, with paragraphs 1–2 being parts of the FOM formula and its explanation. No explicit roadmap is visible."
223 + },
224 + "deterministic_checks": {
225 + "roadmap_regex": false,
226 + "contribution_regex_in_intro": false,
227 + "magnitude_in_abstract": false,
228 + "abstract_n_sentences": 7,
229 + "intro_n_paragraphs": 4,
230 + "intro_n_words": 510,
231 + "causal_words_in_abstract": 0,
232 + "hedge_words_in_abstract": 0,
233 + "stars_mentioned": false,
234 + "clustered_se_mentioned": false,
235 + "internet_appendix_mentioned": false,
236 + "section_roles_rule": {
237 + "0": "introduction",
238 + "1": "theory",
239 + "2": "data",
240 + "3": "results",
241 + "4": "other",
242 + "5": "other",
243 + "6": "conclusion",
244 + "7": "appendix"
245 + }
246 + },
247 + "schema_valid": true,
248 + "code_version": "87c73cd-dirty"
249 +}
\ No newline at end of file
modified data/annotations/documents/jof_12747.json +7 −7
@@ -1,13 +1,14 @@
1 1 {
2 − "validation_errors": [
3 − "Additional properties are not allowed ('typical_devices_note' was unexpected) @ hedging"
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped hedging.typical_devices_note"
4 5 ],
5 6 "doc_id": "jof_12747",
6 7 "ontology_version": "1.0",
7 8 "prompt_version": "1.0",
8 9 "model": "claude-opus-5-5",
9 10 "effort": "medium",
10 − "annotated_at": "2026-11-03T23:06:07+00:00",
11 + "annotated_at": "2026-11-03T23:26:27+00:00",
11 12 "source_sha256": "1003c0785087714d103c43bb64b6d58afe608257d5da747efb905113a59af3fe",
12 13 "parser": "pymupdf-jof-heuristics-1.3",
13 14 "payload_chars": 35706,
@@ -225,8 +226,7 @@
225 226 "'to the extent that'",
226 227 "explicit caveat that results need not imply investor segmentation",
227 228 "'suggest' for implications"
228 − ],
229 − "typical_devices_note": ""
229 + ]
230 230 },
231 231 "claim_evidence": [
232 232 {
@@ -300,6 +300,6 @@
300 300 "6": "conclusion"
301 301 }
302 302 },
303 − "schema_valid": false,
304 − "code_version": "feb6c9e-dirty"
303 + "schema_valid": true,
304 + "code_version": "87c73cd-dirty"
305 305 }
\ No newline at end of file
added data/annotations/documents/jof_12748.json +285 −0
@@ -0,0 +1,285 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12748",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:45+00:00",
9 + "source_sha256": "10fbce238bb52c53fd28899a1f57dbd850988a95b653bfd9e898aaa72d52d5c7",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 29101,
12 + "usage": {
13 + "input": 9665,
14 + "output": 2285,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.085154,
19 + "request_id": "req_011CggqJSWyBNF1hAkmG3TQX",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "calibration_simulation",
25 + "identification": {
26 + "estimand": "Return predictability (factor-model alphas) of the gap between structural model-implied equity value and market equity value, interpreted as mispricing of the default option",
27 + "strategy": "Implement a structural real-option equity valuation model with firm-specific inputs, sort stocks on model-to-market value, and test long-short alphas. Double-sorts by default-related characteristics, a counterfactual model with the default option shut down, earnings-announcement returns, Fama-MacBeth regressions, and an international out-of-sample test.",
28 + "key_assumptions": [
29 + "Model inputs (industry leverage refinancing, constant parameters) adequately capture fundamental value",
30 + "Factor models adequately control for risk so that alphas reflect mispricing",
31 + "Return spreads concentrated around information events indicate mispricing correction rather than risk"
32 + ],
33 + "threats_addressed": [
34 + "value premium in disguise (double-sorts on market-to-book and residual income value)",
35 + "alternative factor models (FF5, q-factor)",
36 + "model specification concerns (shutting down default option)",
37 + "out-of-sample validity (international markets)",
38 + "time variation and subperiods"
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": true,
62 + "sentence_index": 2
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": 1,
77 + "move": "literature_gap"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 2,
85 + "move": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 3,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 4,
93 + "move": "preview_of_results"
94 + },
95 + {
96 + "paragraph_index": 4,
97 + "move": "identification_defense"
98 + },
99 + {
100 + "paragraph_index": 5,
101 + "move": "robustness_mention"
102 + },
103 + {
104 + "paragraph_index": 6,
105 + "move": "mechanism_or_theory_sketch"
106 + },
107 + {
108 + "paragraph_index": 7,
109 + "move": "mechanism_or_theory_sketch"
110 + },
111 + {
112 + "paragraph_index": 8,
113 + "move": "related_literature_positioning"
114 + },
115 + {
116 + "paragraph_index": 9,
117 + "move": "roadmap"
118 + }
119 + ],
120 + "contribution_paragraph_index": 1,
121 + "contribution_types": [
122 + "new_evidence",
123 + "new_data_or_measure"
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": "theory"
136 + },
137 + {
138 + "idx": 2,
139 + "role": "theory"
140 + },
141 + {
142 + "idx": 3,
143 + "role": "data"
144 + },
145 + {
146 + "idx": 4,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 5,
151 + "role": "robustness"
152 + },
153 + {
154 + "idx": 6,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 7,
159 + "role": "conclusion"
160 + },
161 + {
162 + "idx": 8,
163 + "role": "appendix"
164 + },
165 + {
166 + "idx": 9,
167 + "role": "appendix"
168 + }
169 + ],
170 + "results_presentation": {
171 + "economic_significance_discussed": true,
172 + "statistical_significance_language": "stars_and_tstats",
173 + "magnitude_statements_in_intro": true
174 + },
175 + "robustness": {
176 + "location": "mixed",
177 + "types": [
178 + "alternative_specification",
179 + "subsample",
180 + "alternative_measure",
181 + "alternative_sample_period",
182 + "falsification"
183 + ]
184 + },
185 + "tables": {
186 + "convention_notes": "Notes define the sort procedure: decile 1 is most overvalued and decile 10 most undervalued, with value-weighted one-month holding. Later tables refer back to the Table II sort. Alphas are reported in percent per month with t-statistics. Factor models are named explicitly, and international tables are organized by country.",
187 + "units_stated_in_notes": true,
188 + "standard_errors_description": "unclear"
189 + },
190 + "figures": {
191 + "uncertainty_shown": "no_figures"
192 + },
193 + "conclusion": {
194 + "bounded_claims": true,
195 + "policy_implications": false,
196 + "open_questions": false,
197 + "restates_contribution": true
198 + },
199 + "hedging": {
200 + "level": "moderate",
201 + "typical_devices": [
202 + "'at least in part'",
203 + "'suggests'",
204 + "'we conjecture'",
205 + "explicit disclaimers of what is not claimed (e.g., not resolving the distress puzzle)"
206 + ]
207 + },
208 + "claim_evidence": [
209 + {
210 + "claim": "Long-short strategy on model-implied relative value earns roughly 11% annual four-factor alpha in U.S. stocks",
211 + "evidence_type": "table",
212 + "evidence_ref": "Table III"
213 + },
214 + {
215 + "claim": "Model performance is stronger among distressed and highly volatile stocks",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table VII"
218 + },
219 + {
220 + "claim": "Removing the default option sharply weakens the alphas, implicating the option as the driver",
221 + "evidence_type": "table",
222 + "evidence_ref": "Table VIII"
223 + },
224 + {
225 + "claim": "Return spread is larger on earnings announcement days, consistent with mispricing correction",
226 + "evidence_type": "table",
227 + "evidence_ref": "Table X"
228 + },
229 + {
230 + "claim": "Results are not the value premium in disguise",
231 + "evidence_type": "table",
232 + "evidence_ref": "Table VI"
233 + },
234 + {
235 + "claim": "Model value predicts returns in several of nine developed markets",
236 + "evidence_type": "table",
237 + "evidence_ref": "Tables XV-XVI"
238 + },
239 + {
240 + "claim": "Investors may lack the skills to value options correctly",
241 + "evidence_type": "prior_literature",
242 + "evidence_ref": "Poteshman and Serbin (2003); Benartzi and Thaler (2001); Hirshleifer and Teoh (2003)"
243 + }
244 + ],
245 + "structural_summary": "The paper motivates equity misvaluation and proposes an unrecognized source: investors fail to price the default option. It builds a structural real-option valuation model and applies it monthly to the full cross section. It then shows portfolio-sort alphas, rules out a value-premium explanation, and ties performance to the default option via characteristic double-sorts and a shut-down counterfactual. Mispricing-consistent tests (earnings announcements, transparency, time variation) and an international out-of-sample test follow. The paper closes with a brief restatement of findings.",
246 + "writing_profile_keywords": [
247 + "structural valuation model",
248 + "portfolio sorts",
249 + "factor alphas",
250 + "mispricing",
251 + "default option",
252 + "out-of-sample international",
253 + "counterfactual model"
254 + ],
255 + "confidence": 0.8,
256 + "notes": "Primary method is labeled calibration_simulation because the paper implements a calibrated structural model whose outputs feed portfolio-sort tests; it could arguably be classified as reduced_form_regression or other. Section II (Model Descriptives) is labeled data/descriptive even though its rule role was theory. Standard-error treatment is not visible in the excerpts."
257 + },
258 + "deterministic_checks": {
259 + "roadmap_regex": true,
260 + "contribution_regex_in_intro": true,
261 + "magnitude_in_abstract": true,
262 + "abstract_n_sentences": 6,
263 + "intro_n_paragraphs": 10,
264 + "intro_n_words": 1622,
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": "theory",
273 + "2": "other",
274 + "3": "theory",
275 + "4": "theory",
276 + "5": "robustness",
277 + "6": "results",
278 + "7": "conclusion",
279 + "8": "appendix",
280 + "9": "appendix"
281 + }
282 + },
283 + "schema_valid": true,
284 + "code_version": "87c73cd-dirty"
285 +}
\ No newline at end of file
added data/annotations/documents/jof_12750.json +278 −0
@@ -0,0 +1,278 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [],
4 + "doc_id": "jof_12750",
5 + "ontology_version": "1.0",
6 + "prompt_version": "1.0",
7 + "model": "claude-opus-5-5",
8 + "effort": "medium",
9 + "annotated_at": "2026-11-03T23:26:44+00:00",
10 + "source_sha256": "b95c6618b46267c00f197301c8a90f3cc0ae2b652e2004cde9042bb5fff76431",
11 + "parser": "pymupdf-jof-heuristics-1.3",
12 + "payload_chars": 21129,
13 + "usage": {
14 + "input": 7529,
15 + "output": 2048,
16 + "cache_read": 3968,
17 + "cache_write": 0
18 + },
19 + "cost_usd": 0.07187,
20 + "request_id": "req_011CggrpukGjUVLF7kFJooeJ",
21 + "annotation": {
22 + "genre": "mixed_theory_empirical",
23 + "subfield": "asset_pricing",
24 + "subfield_secondary": "financial_intermediation_banking",
25 + "primary_method": "reduced_form_regression",
26 + "identification": {
27 + "estimand": "Association between quarterly changes in aggregate pension underfunding ratio and changes in 30-year swap spreads",
28 + "strategy": "Time-series OLS regressions with controls, asymmetric underfunded/overfunded terms, and 2SLS using stock market excess returns as instrument for underfunding; cross-maturity and cross-country comparisons",
29 + "key_assumptions": [
30 + "Stock returns affect funding status but are unrelated to swap spreads except through underfunding",
31 + "Controls capture other known swap spread drivers"
32 + ],
33 + "threats_addressed": [
34 + "Yield curve level jointly driving swap spreads and underfunding",
35 + "Omitted variables such as sovereign credit risk and corporate issuance",
36 + "Time-period dependence",
37 + "Measurement of underfunding"
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": true,
49 + "sentence_index": 1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 2
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 4
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": "motivation"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "preview_of_approach"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "tension_or_puzzle"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "identification_defense"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "robustness_mention"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "preview_of_results"
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": 5,
108 + "move": "contribution_statement"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "roadmap"
113 + }
114 + ],
115 + "contribution_paragraph_index": 5,
116 + "contribution_types": [
117 + "new_evidence",
118 + "new_data_or_measure",
119 + "new_theory_or_model",
120 + "resolution_of_puzzle"
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": "results"
133 + },
134 + {
135 + "idx": 2,
136 + "role": "other"
137 + },
138 + {
139 + "idx": 3,
140 + "role": "theory"
141 + },
142 + {
143 + "idx": 4,
144 + "role": "results"
145 + },
146 + {
147 + "idx": 5,
148 + "role": "results"
149 + },
150 + {
151 + "idx": 6,
152 + "role": "conclusion"
153 + },
154 + {
155 + "idx": 7,
156 + "role": "appendix"
157 + },
158 + {
159 + "idx": 8,
160 + "role": "appendix"
161 + }
162 + ],
163 + "results_presentation": {
164 + "economic_significance_discussed": false,
165 + "statistical_significance_language": "unclear",
166 + "magnitude_statements_in_intro": false
167 + },
168 + "robustness": {
169 + "location": "main_text_section",
170 + "types": [
171 + "alternative_specification",
172 + "alternative_measure",
173 + "alternative_sample_period",
174 + "subsample",
175 + "falsification"
176 + ]
177 + },
178 + "tables": {
179 + "convention_notes": "Table notes define variables in quarterly changes, reference equations for constructed measures, and describe panels and model columns; one illustrative cash-flow table and one balance sheet table with stated units.",
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 + "possible reasons",
196 + "can drive",
197 + "helps explain",
198 + "support the view",
199 + "fairly consistent"
200 + ]
201 + },
202 + "claim_evidence": [
203 + {
204 + "claim": "30-year U.S. swap spread turned negative after Lehman default and stayed negative",
205 + "evidence_type": "figure",
206 + "evidence_ref": "Figure 1"
207 + },
208 + {
209 + "claim": "Underfunded pension plans optimally receive fixed in long-dated swaps; dealer constraints imply negative spreads",
210 + "evidence_type": "equation_or_proof",
211 + "evidence_ref": "Section III; Appendix: Model and Proofs"
212 + },
213 + {
214 + "claim": "Changes in underfunding ratio explain changes in 30-year swap spreads, mainly when plans are underfunded",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table IV; Figure 2"
217 + },
218 + {
219 + "claim": "Funding status does not explain 2- to 10-year swap spreads",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table V"
222 + },
223 + {
224 + "claim": "Results hold in 2SLS using stock returns as instrument",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table VI"
227 + },
228 + {
229 + "claim": "Link robust to extra controls, period dummies and alternative measure; strongest right after Lehman",
230 + "evidence_type": "table",
231 + "evidence_ref": "Table VII"
232 + },
233 + {
234 + "claim": "Similar underfunding-swap spread link in Japan and the Netherlands",
235 + "evidence_type": "table",
236 + "evidence_ref": "Tables VIII-IX"
237 + }
238 + ],
239 + "structural_summary": "Opens with a persistent pricing anomaly that contradicts the prior credit-risk and convenience-yield view of swap spreads, then proposes a demand-plus-frictions hypothesis. Stylized facts come first, followed by institutional background on pension duration needs and swap supply. An illustrative model, with proofs in the appendix, formalizes the mechanism. U.S. time-series regressions follow, covering regimes, OLS, 2SLS and robustness checks, with a maturity-based falsification. International extensions to Japan and the Netherlands come next, and a brief conclusion restates the contribution.",
240 + "writing_profile_keywords": [
241 + "anomaly-driven",
242 + "demand-based explanation",
243 + "limits to arbitrage",
244 + "illustrative model",
245 + "instrumental variables",
246 + "cross-country extension",
247 + "institutional detail"
248 + ],
249 + "confidence": 0.8,
250 + "notes": "Standard error type and significance notation not visible in table notes excerpts. Section II is institutional background."
251 + },
252 + "deterministic_checks": {
253 + "roadmap_regex": true,
254 + "contribution_regex_in_intro": true,
255 + "magnitude_in_abstract": false,
256 + "abstract_n_sentences": 7,
257 + "intro_n_paragraphs": 7,
258 + "intro_n_words": 1096,
259 + "causal_words_in_abstract": 0,
260 + "hedge_words_in_abstract": 0,
261 + "stars_mentioned": false,
262 + "clustered_se_mentioned": false,
263 + "internet_appendix_mentioned": false,
264 + "section_roles_rule": {
265 + "0": "introduction",
266 + "1": "results",
267 + "2": "other",
268 + "3": "theory",
269 + "4": "other",
270 + "5": "results",
271 + "6": "conclusion",
272 + "7": "appendix",
273 + "8": "appendix"
274 + }
275 + },
276 + "schema_valid": true,
277 + "code_version": "87c73cd-dirty"
278 +}
\ No newline at end of file
added data/annotations/documents/jof_12752.json +238 −0
@@ -0,0 +1,238 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12752",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:03+00:00",
9 + "source_sha256": "4c0ef0337db02c3988e9a5d3f5274cefcf5524ffd56c6d44b53d36583e4530c2",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 15617,
12 + "usage": {
13 + "input": 5360,
14 + "output": 2038,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.062994,
19 + "request_id": "req_011CggqKr2cigGf4YaUW4fBG",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "asset_pricing (option valuation); executive compensation",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Conditional daily rate at which employees voluntarily exercise vested in-the-money options, as a function of option, stock, firm and holder characteristics, and the implied option cost to the firm.",
27 + "strategy": "Reduced-form fractional-logistic exercise function (hazard-like) estimated by GMM that is robust to heteroskedasticity and within-employee correlation. The estimated function is then fed into simulation-based valuation.",
28 + "key_assumptions": [
29 + "The estimated exercise function describes future exercise behavior well enough to serve as a valuation input.",
30 + "Covariates capture the relevant drivers of exercise.",
31 + "Employee termination and cancellation hazards are modeled separately."
32 + ],
33 + "threats_addressed": [
34 + "Correlation across exercises by the same employee",
35 + "Heteroskedasticity",
36 + "Omitted-variable bias from using only standard exercise factors (shown via specifications 1-2)"
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": true,
60 + "sentence_index": 2
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": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "literature_gap"
80 + },
81 + {
82 + "paragraph_index": 0,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 0,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 1,
91 + "move": "roadmap"
92 + }
93 + ],
94 + "contribution_paragraph_index": 0,
95 + "contribution_types": [
96 + "new_method",
97 + "new_evidence",
98 + "new_data_or_measure"
99 + ],
100 + "roadmap_present": true,
101 + "literature_positioning_location": "separate_section"
102 + },
103 + "section_roles": [
104 + {
105 + "idx": 0,
106 + "role": "introduction"
107 + },
108 + {
109 + "idx": 1,
110 + "role": "literature"
111 + },
112 + {
113 + "idx": 2,
114 + "role": "empirical_design"
115 + },
116 + {
117 + "idx": 3,
118 + "role": "data"
119 + },
120 + {
121 + "idx": 4,
122 + "role": "results"
123 + },
124 + {
125 + "idx": 5,
126 + "role": "results"
127 + },
128 + {
129 + "idx": 6,
130 + "role": "conclusion"
131 + },
132 + {
133 + "idx": 7,
134 + "role": "appendix"
135 + }
136 + ],
137 + "results_presentation": {
138 + "economic_significance_discussed": true,
139 + "statistical_significance_language": "unclear",
140 + "magnitude_statements_in_intro": false
141 + },
142 + "robustness": {
143 + "location": "none_seen",
144 + "types": []
145 + },
146 + "tables": {
147 + "convention_notes": "Table notes define the dependent variable and specifications, and report coefficients with standard errors in parentheses. One table translates coefficients into percentage changes in the exercise rate for a one-SD change in each factor. Valuation tables state their base-case parameters (strike 100, 5% rate, 10% termination, SIC 3). Summary tables are grouped by one-digit SIC.",
148 + "units_stated_in_notes": true,
149 + "standard_errors_description": "unclear"
150 + },
151 + "figures": {
152 + "uncertainty_shown": "no_figures"
153 + },
154 + "conclusion": {
155 + "bounded_claims": false,
156 + "policy_implications": false,
157 + "open_questions": false,
158 + "restates_contribution": true
159 + },
160 + "hedging": {
161 + "level": "low",
162 + "typical_devices": [
163 + "ranges of magnitudes (e.g., 11% to 16%)",
164 + "'typically' qualifiers",
165 + "assumption-conditioned valuations"
166 + ]
167 + },
168 + "claim_evidence": [
169 + {
170 + "claim": "Moving from annual or cliff vesting to monthly vesting lowers option value by roughly 11-16%.",
171 + "evidence_type": "table",
172 + "evidence_ref": "Table VIII"
173 + },
174 + {
175 + "claim": "Men exercise faster, lowering option value by 2-4%.",
176 + "evidence_type": "table",
177 + "evidence_ref": "Tables IV, VIII"
178 + },
179 + {
180 + "claim": "Top-ranked employees exercise more slowly, raising option value by 2-7%.",
181 + "evidence_type": "table",
182 + "evidence_ref": "Tables IV, VI, VIII"
183 + },
184 + {
185 + "claim": "Black-Scholes approximations used in practice produce valuation errors, and the new analytic approximation is more accurate.",
186 + "evidence_type": "table",
187 + "evidence_ref": "Table IX"
188 + },
189 + {
190 + "claim": "Omitting portfolio and additional factors distorts exercise estimates.",
191 + "evidence_type": "table",
192 + "evidence_ref": "Table IV specs (1)-(2)"
193 + },
194 + {
195 + "claim": "ESOs are a major share of CEO pay and their valuation is controversial in practice.",
196 + "evidence_type": "prior_literature",
197 + "evidence_ref": "Murphy (2013); Meridian (2018); ISS/IBM example"
198 + }
199 + ],
200 + "structural_summary": "The introduction motivates the topic with the economic importance of ESOs and a practical valuation controversy. It argues that standard theory and Black-Scholes modifications are inadequate, then previews a GMM fractional-logistic exercise model applied to proprietary data. A separate literature section contrasts utility-based theoretical models with empirical work. Next come the methodology, a long data section, estimation results with rank interactions, and a valuation section that quantifies cost effects, critiques practitioner approximations and proposes a new analytic approximation. The conclusion restates the method and findings.",
201 + "writing_profile_keywords": [
202 + "proprietary data",
203 + "reduced-form exercise function",
204 + "option valuation",
205 + "fractional logit GMM",
206 + "economic magnitudes in valuation terms",
207 + "practitioner relevance",
208 + "no figures"
209 + ],
210 + "confidence": 0.75,
211 + "notes": "The primary method is borderline between structural_estimation and reduced_form_regression: the exercise model is explicitly reduced-form, but it is used as an input to simulation-based valuation. Section IV's role_rule is 'empirical_design', but its content is estimation results, so it is labeled results. The excerpts do not show how standard errors are clustered beyond 'robust to correlation'. Robustness checks are not visible."
212 + },
213 + "deterministic_checks": {
214 + "roadmap_regex": true,
215 + "contribution_regex_in_intro": false,
216 + "magnitude_in_abstract": true,
217 + "abstract_n_sentences": 5,
218 + "intro_n_paragraphs": 2,
219 + "intro_n_words": 328,
220 + "causal_words_in_abstract": 0,
221 + "hedge_words_in_abstract": 0,
222 + "stars_mentioned": false,
223 + "clustered_se_mentioned": false,
224 + "internet_appendix_mentioned": false,
225 + "section_roles_rule": {
226 + "0": "introduction",
227 + "1": "literature",
228 + "2": "empirical_design",
229 + "3": "data",
230 + "4": "empirical_design",
231 + "5": "other",
232 + "6": "conclusion",
233 + "7": "appendix"
234 + }
235 + },
236 + "schema_valid": true,
237 + "code_version": "87c73cd-dirty"
238 +}
\ No newline at end of file
added data/annotations/documents/jof_12753.json +253 −0
@@ -0,0 +1,253 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12753",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:27+00:00",
9 + "source_sha256": "aaf66a721758f107ee22a1b60f714fcf8f67191cc1293dc98347ff3e42b88c9a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23131,
12 + "usage": {
13 + "input": 7685,
14 + "output": 2124,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.074014,
19 + "request_id": "req_011CggqH7KnVn4Q3EjH2b4Gu",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "economic history / banking regulation",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of state deposit insurance on insured banks' deposits, loans, cash and capital ratios, and the sensitivity of deposit flows to bank risk (market discipline)",
27 + "strategy": "Within-state comparison of insured state banks with uninsured national banks, plus uninsured banks in adjacent non-insurance states, before and after the laws passed. Uses bank fixed effects in a panel, a border closest-neighbor-pair sample, and cross-county variation in the WWI agricultural price boom.",
28 + "key_assumptions": [
29 + "Insured and uninsured banks in the same and neighboring states would have followed parallel trends absent insurance",
30 + "Adoption of insurance was not driven by idiosyncratic banking shocks (states where it may have been are excluded)",
31 + "Adjacent non-insurance states form a valid comparison group"
32 + ],
33 + "threats_addressed": [
34 + "Omitted factors coinciding with deposit insurance adoption, the weakness of cross-country studies",
35 + "Pre-existing trends, tested with state-level pretrend regressions for 1900-1908",
36 + "Geographic heterogeneity, addressed with border closest-neighbor pairs",
37 + "Endogenous adoption, addressed by excluding Mississippi and Washington",
38 + "Voluntary versus mandatory insurance differences"
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": 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": "literature_gap"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "contribution_statement"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "mechanism_or_theory_sketch"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 3,
101 + "move": "roadmap"
102 + }
103 + ],
104 + "contribution_paragraph_index": 1,
105 + "contribution_types": [
106 + "new_evidence",
107 + "new_identification",
108 + "new_data_or_measure"
109 + ],
110 + "roadmap_present": true,
111 + "literature_positioning_location": "inside_introduction"
112 + },
113 + "section_roles": [
114 + {
115 + "idx": 0,
116 + "role": "introduction"
117 + },
118 + {
119 + "idx": 1,
120 + "role": "other"
121 + },
122 + {
123 + "idx": 2,
124 + "role": "data"
125 + },
126 + {
127 + "idx": 3,
128 + "role": "results"
129 + },
130 + {
131 + "idx": 4,
132 + "role": "results"
133 + },
134 + {
135 + "idx": 5,
136 + "role": "results"
137 + },
138 + {
139 + "idx": 6,
140 + "role": "conclusion"
141 + }
142 + ],
143 + "results_presentation": {
144 + "economic_significance_discussed": true,
145 + "statistical_significance_language": "unclear",
146 + "magnitude_statements_in_intro": false
147 + },
148 + "robustness": {
149 + "location": "within_results",
150 + "types": [
151 + "subsample",
152 + "alternative_specification"
153 + ]
154 + },
155 + "tables": {
156 + "convention_notes": "Notes define the dependent variable (given in the column titles) and the DI and DI Area indicators, and state the sample states and data sources. Most tables report linear regressions at the state or bank level.",
157 + "units_stated_in_notes": false,
158 + "standard_errors_description": "unclear"
159 + },
160 + "figures": {
161 + "uncertainty_shown": "unclear"
162 + },
163 + "conclusion": {
164 + "bounded_claims": true,
165 + "policy_implications": true,
166 + "open_questions": false,
167 + "restates_contribution": true
168 + },
169 + "hedging": {
170 + "level": "moderate",
171 + "typical_devices": [
172 + "seemingly",
173 + "apparently",
174 + "presumably",
175 + "it is conceivable that",
176 + "framing results as a conservative measure"
177 + ]
178 + },
179 + "claim_evidence": [
180 + {
181 + "claim": "Deposit insurance raised insured banks' deposits and loans and lowered their cash and capital ratios",
182 + "evidence_type": "table",
183 + "evidence_ref": "Tables V, VI"
184 + },
185 + {
186 + "claim": "State-level aggregates show deposits shifting toward insured state banks after adoption",
187 + "evidence_type": "figure",
188 + "evidence_ref": "Figures 2, 4, 6; Table IV"
189 + },
190 + {
191 + "claim": "No differential pretrends existed before deposit insurance",
192 + "evidence_type": "table",
193 + "evidence_ref": "Table III"
194 + },
195 + {
196 + "claim": "Insurance removed market discipline: deposit growth at insured banks was less sensitive to risk measures",
197 + "evidence_type": "table",
198 + "evidence_ref": "Tables VIII, IX"
199 + },
200 + {
201 + "claim": "Results hold in a closest-neighbor border sample",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table VII"
204 + },
205 + {
206 + "claim": "After the price collapse, banks fled the insurance systems by converting to national charters",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table X"
209 + },
210 + {
211 + "claim": "The WWI price boom explains one- to two-thirds of the deposit insurance effects",
212 + "evidence_type": "none_stated",
213 + "evidence_ref": "Section III.D.2 (stated in conclusion)"
214 + }
215 + ],
216 + "structural_summary": "The introduction sets broad policy support for deposit insurance against empirical evidence of moral hazard. It argues that cross-country studies are confounded and positions within-state historical variation as a cleaner test. It previews findings on risk-taking, loss of market discipline, and the WWI agricultural boom and bust, then gives a roadmap. The body covers institutional history with aggregate trends, then data, then bank-level difference-style regressions with pretrend, border-sample and voluntary-versus-mandatory extensions. A deposit-flow market discipline analysis follows, and finally the post-1920 collapse. The conclusion restates the three findings, draws methodological lessons on time-varying risk-taking opportunities, and offers policy and political-economy implications.",
217 + "writing_profile_keywords": [
218 + "economic history",
219 + "natural experiment",
220 + "moral hazard",
221 + "market discipline",
222 + "bank-level panel",
223 + "policy implication",
224 + "state deposit insurance"
225 + ],
226 + "confidence": 0.8,
227 + "notes": "No standard error or significance details are visible in the excerpts. Theory is in the Internet Appendix. The strategy is DiD-like, combining insured state banks with national banks and neighboring states."
228 + },
229 + "deterministic_checks": {
230 + "roadmap_regex": true,
231 + "contribution_regex_in_intro": true,
232 + "magnitude_in_abstract": false,
233 + "abstract_n_sentences": 7,
234 + "intro_n_paragraphs": 4,
235 + "intro_n_words": 570,
236 + "causal_words_in_abstract": 1,
237 + "hedge_words_in_abstract": 0,
238 + "stars_mentioned": false,
239 + "clustered_se_mentioned": false,
240 + "internet_appendix_mentioned": true,
241 + "section_roles_rule": {
242 + "0": "introduction",
243 + "1": "other",
244 + "2": "data",
245 + "3": "other",
246 + "4": "other",
247 + "5": "other",
248 + "6": "conclusion"
249 + }
250 + },
251 + "schema_valid": true,
252 + "code_version": "87c73cd-dirty"
253 +}
\ No newline at end of file
added data/annotations/documents/jof_12754.json +243 −0
@@ -0,0 +1,243 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12754",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:06:55+00:00",
9 + "source_sha256": "ed2c10f5a7f6092f564d884c244861608398cca952a77b1ab5497076c49d738d",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 18483,
12 + "usage": {
13 + "input": 6179,
14 + "output": 2018,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06587,
19 + "request_id": "req_011CggqKAV5xYLQsCH9gaujz",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "behavioral_finance",
23 + "subfield_secondary": "asset_pricing; institutional investors",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Association between changes in investor sentiment metrics and cross-sectional institutional versus individual demand shocks for volatile versus safe stocks.",
27 + "strategy": "Time-series correlations between orthogonalized sentiment changes and 13(f)-inferred institutional demand shocks across volatility deciles. These are supplemented by panel regressions, decomposition of demand into flow-induced versus manager-driven components, and sorts by client type and institution type.",
28 + "key_assumptions": [
29 + "The complement of 13(f) institutional demand captures individual investors' demand.",
30 + "Sentiment metrics such as the Baker–Wurgler (BW) index proxy for sentiment.",
31 + "Stock volatility proxies speculative versus safe stocks."
32 + ],
33 + "threats_addressed": [
34 + "Mismeasurement of retail demand via the inverse of 13(f) demand, checked against Odean retail broker data.",
35 + "Retail flows driving institutional trades.",
36 + "Changes in risk and alternative risk metrics.",
37 + "Choice of sentiment metric, using 17 alternatives."
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": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_results"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "preview_of_approach"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "identification_defense"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "policy_or_broader_implication"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "policy_or_broader_implication"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "related_literature_positioning"
105 + }
106 + ],
107 + "contribution_paragraph_index": 0,
108 + "contribution_types": [
109 + "new_evidence"
110 + ],
111 + "roadmap_present": false,
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": "robustness"
130 + },
131 + {
132 + "idx": 4,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 5,
137 + "role": "conclusion"
138 + }
139 + ],
140 + "results_presentation": {
141 + "economic_significance_discussed": true,
142 + "statistical_significance_language": "p_values",
143 + "magnitude_statements_in_intro": false
144 + },
145 + "robustness": {
146 + "location": "mixed",
147 + "types": [
148 + "alternative_measure",
149 + "alternative_specification",
150 + "subsample"
151 + ]
152 + },
153 + "tables": {
154 + "convention_notes": "Notes state the sample period (1980–2010 quarterly), how variables are constructed (volatility from the prior 12 months of returns), and the panel structure. Results are reported as correlations with p-values and as panel regression coefficients.",
155 + "units_stated_in_notes": true,
156 + "standard_errors_description": "unclear"
157 + },
158 + "figures": {
159 + "uncertainty_shown": "no_figures"
160 + },
161 + "conclusion": {
162 + "bounded_claims": true,
163 + "policy_implications": false,
164 + "open_questions": true,
165 + "restates_contribution": true
166 + },
167 + "hedging": {
168 + "level": "moderate",
169 + "typical_devices": [
170 + "conditional framing ('if these metrics capture sentiment, then...')",
171 + "explicit correlation-not-causation caveat",
172 + "'largely consistent' qualifiers",
173 + "'some evidence'"
174 + ]
175 + },
176 + "claim_evidence": [
177 + {
178 + "claim": "When sentiment rises, institutions buy volatile stocks from individuals and sell safe stocks to them.",
179 + "evidence_type": "table",
180 + "evidence_ref": "Table I, Table II"
181 + },
182 + {
183 + "claim": "The result is not explained by changes in risk or by the choice of risk metric.",
184 + "evidence_type": "table",
185 + "evidence_ref": "Table III"
186 + },
187 + {
188 + "claim": "None of 17 sentiment metrics captures individual demand shocks; 10 relate to institutional demand, and only these predict returns.",
189 + "evidence_type": "table",
190 + "evidence_ref": "Table IV"
191 + },
192 + {
193 + "claim": "Underlying investor flows do not drive the aggregate relation between institutional demand and sentiment.",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table VI"
196 + },
197 + {
198 + "claim": "Institutional investment styles such as risk management and momentum explain much of institutional sentiment trading.",
199 + "evidence_type": "table",
200 + "evidence_ref": "Tables VII, VIII"
201 + },
202 + {
203 + "claim": "Institutions with retail versus institutional client bases differ in sentiment trading.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table V"
206 + }
207 + ],
208 + "structural_summary": "The introduction opens by challenging the long-standing assumption that individuals are sentiment traders, then states the counterclaim and previews results. It follows with tests ruling out retail-driven explanations and closes on implications for market efficiency and for the prior sentiment literature. The data section is followed by core evidence on whose demand the BW metric captures, then a robustness section on 17 alternative metrics. A long mechanism section examines retail measurement, client base, flows, stock-level panels, institution types, and individual institutions. The conclusion restates findings with an explicit causation caveat.",
209 + "writing_profile_keywords": [
210 + "conventional-wisdom reversal",
211 + "13(f) holdings",
212 + "investor sentiment",
213 + "mechanism exploration",
214 + "correlation evidence",
215 + "reinterpretation of literature"
216 + ],
217 + "confidence": 0.8,
218 + "notes": "No figures or equations. A roadmap is not visible in the truncated introduction. Significance appears to be reported via p-values in correlation tables; regression tables may also report t-statistics, which is not visible in the excerpts."
219 + },
220 + "deterministic_checks": {
221 + "roadmap_regex": false,
222 + "contribution_regex_in_intro": true,
223 + "magnitude_in_abstract": false,
224 + "abstract_n_sentences": 3,
225 + "intro_n_paragraphs": 5,
226 + "intro_n_words": 910,
227 + "causal_words_in_abstract": 0,
228 + "hedge_words_in_abstract": 0,
229 + "stars_mentioned": false,
230 + "clustered_se_mentioned": false,
231 + "internet_appendix_mentioned": false,
232 + "section_roles_rule": {
233 + "0": "introduction",
234 + "1": "data",
235 + "2": "other",
236 + "3": "robustness",
237 + "4": "other",
238 + "5": "conclusion"
239 + }
240 + },
241 + "schema_valid": true,
242 + "code_version": "87c73cd-dirty"
243 +}
\ No newline at end of file
added data/annotations/documents/jof_12755.json +312 −0
@@ -0,0 +1,312 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12755",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:22+00:00",
9 + "source_sha256": "88beecf8b426e07c50fc96962fcd2514c1528ee0d203d4ee78f278aa8010b390",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24491,
12 + "usage": {
13 + "input": 8390,
14 + "output": 2283,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.080014,
19 + "request_id": "req_011CggqM7zakNpsuj1KdsaEz",
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": "Cross-sectional relation between a country's trade network centrality and its unconditional interest rate differentials, currency risk premia, and consumption growth betas.",
27 + "strategy": "Panel regressions with year fixed effects on lagged standardized centrality, size and trade-to-GDP; currency portfolio sorts on centrality; time-series factor regressions of carry factors on the PMC factor; all guided by a general equilibrium network model.",
28 + "key_assumptions": [
29 + "Centrality is an unconditional, slowly changing country characteristic.",
30 + "Lagged centrality is predetermined relative to returns.",
31 + "The model's mechanism, common consumption exposure, maps to the empirical consumption betas."
32 + ],
33 + "threats_addressed": [
34 + "Confounding by country size (GDP share) and trade openness",
35 + "Structural changes in the trade network such as euro introduction and country entry and exit",
36 + "Pegged currencies",
37 + "CIP deviations after 2007",
38 + "Developed-country subset differences"
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": 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": 0,
77 + "move": "tension_or_puzzle"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "literature_gap"
82 + },
83 + {
84 + "paragraph_index": 0,
85 + "move": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "preview_of_approach"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "mechanism_or_theory_sketch"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 2,
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": "preview_of_results"
110 + },
111 + {
112 + "paragraph_index": 5,
113 + "move": "contribution_statement"
114 + },
115 + {
116 + "paragraph_index": 6,
117 + "move": "related_literature_positioning"
118 + },
119 + {
120 + "paragraph_index": 7,
121 + "move": "roadmap"
122 + }
123 + ],
124 + "contribution_paragraph_index": 5,
125 + "contribution_types": [
126 + "new_evidence",
127 + "new_theory_or_model",
128 + "new_data_or_measure"
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": "conclusion"
149 + },
150 + {
151 + "idx": 4,
152 + "role": "appendix"
153 + },
154 + {
155 + "idx": 5,
156 + "role": "appendix"
157 + },
158 + {
159 + "idx": 6,
160 + "role": "appendix"
161 + },
162 + {
163 + "idx": 7,
164 + "role": "robustness"
165 + },
166 + {
167 + "idx": 8,
168 + "role": "robustness"
169 + },
170 + {
171 + "idx": 9,
172 + "role": "robustness"
173 + },
174 + {
175 + "idx": 10,
176 + "role": "robustness"
177 + },
178 + {
179 + "idx": 11,
180 + "role": "robustness"
181 + },
182 + {
183 + "idx": 12,
184 + "role": "robustness"
185 + },
186 + {
187 + "idx": 13,
188 + "role": "robustness"
189 + },
190 + {
191 + "idx": 14,
192 + "role": "robustness"
193 + },
194 + {
195 + "idx": 15,
196 + "role": "appendix"
197 + }
198 + ],
199 + "results_presentation": {
200 + "economic_significance_discussed": true,
201 + "statistical_significance_language": "unclear",
202 + "magnitude_statements_in_intro": true
203 + },
204 + "robustness": {
205 + "location": "appendix",
206 + "types": [
207 + "alternative_measure",
208 + "subsample",
209 + "alternative_sample_period",
210 + "alternative_specification"
211 + ]
212 + },
213 + "tables": {
214 + "convention_notes": "Notes define variables, sorting procedures and fixed effects. Regressors are standardized and returns are annualized. Factor construction is described as long/short portfolios. Standard error method is not visible in the excerpts.",
215 + "units_stated_in_notes": true,
216 + "standard_errors_description": "unclear"
217 + },
218 + "figures": {
219 + "uncertainty_shown": "unclear"
220 + },
221 + "conclusion": {
222 + "bounded_claims": false,
223 + "policy_implications": false,
224 + "open_questions": false,
225 + "restates_contribution": true
226 + },
227 + "hedging": {
228 + "level": "low",
229 + "typical_devices": [
230 + "consistent with the model's implications",
231 + "suggest",
232 + "helps explain"
233 + ]
234 + },
235 + "claim_evidence": [
236 + {
237 + "claim": "Over 1988-2016, a peripheral-minus-central currency portfolio earns a Sharpe ratio similar to the unconditional carry trade.",
238 + "evidence_type": "table",
239 + "evidence_ref": "Table III, Table IV"
240 + },
241 + {
242 + "claim": "Interest rates and currency risk premia decline with trade network centrality, controlling for size and trade openness.",
243 + "evidence_type": "table",
244 + "evidence_ref": "Table I, Figure 2"
245 + },
246 + {
247 + "claim": "Central countries' consumption growth covaries more with world consumption growth.",
248 + "evidence_type": "table",
249 + "evidence_ref": "Table II, Figure 4"
250 + },
251 + {
252 + "claim": "The unconditional carry factor loads nearly one-for-one on PMC and leaves no unexplained excess return.",
253 + "evidence_type": "table",
254 + "evidence_ref": "Table V, Table VI"
255 + },
256 + {
257 + "claim": "In the model, central countries have higher exposure to global consumption shocks, hence lower interest rates and risk premia.",
258 + "evidence_type": "equation_or_proof",
259 + "evidence_ref": "Propositions 1-2, Appendix A.4-A.5"
260 + },
261 + {
262 + "claim": "Results hold without the euro aggregate, without pegs, for developed countries only, and for a sample ending in 2007.",
263 + "evidence_type": "table",
264 + "evidence_ref": "Appendix Tables B.V-B.IX"
265 + }
266 + ],
267 + "structural_summary": "The introduction opens with the carry trade Sharpe ratio puzzle, then motivates trade network centrality with a network figure and a scatter of returns against centrality. It previews a multicountry network-production model and three empirical tests: regressions, portfolio sorts, and a factor comparison showing PMC subsumes the unconditional carry factor. Contributions to the networks, exchange rate, and risk-sharing literatures are positioned within the introduction. The model section derives two propositions; the data section builds the centrality measure and tests both propositions. A short conclusion restates the findings. Appendices hold the proofs and a battery of robustness portfolio sorts.",
268 + "writing_profile_keywords": [
269 + "model-then-test",
270 + "carry trade",
271 + "portfolio sorts",
272 + "network centrality",
273 + "first-person singular",
274 + "mechanism validation"
275 + ],
276 + "confidence": 0.8,
277 + "notes": "Single-author JF paper. The empirical section is labeled data but contains the main results. Standard error and significance conventions are not visible in the excerpts."
278 + },
279 + "deterministic_checks": {
280 + "roadmap_regex": true,
281 + "contribution_regex_in_intro": true,
282 + "magnitude_in_abstract": false,
283 + "abstract_n_sentences": 5,
284 + "intro_n_paragraphs": 8,
285 + "intro_n_words": 1165,
286 + "causal_words_in_abstract": 1,
287 + "hedge_words_in_abstract": 0,
288 + "stars_mentioned": false,
289 + "clustered_se_mentioned": false,
290 + "internet_appendix_mentioned": false,
291 + "section_roles_rule": {
292 + "0": "introduction",
293 + "1": "theory",
294 + "2": "data",
295 + "3": "conclusion",
296 + "4": "appendix",
297 + "5": "appendix",
298 + "6": "appendix",
299 + "7": "appendix",
300 + "8": "appendix",
301 + "9": "appendix",
302 + "10": "appendix",
303 + "11": "appendix",
304 + "12": "appendix",
305 + "13": "appendix",
306 + "14": "appendix",
307 + "15": "appendix"
308 + }
309 + },
310 + "schema_valid": true,
311 + "code_version": "87c73cd-dirty"
312 +}
\ No newline at end of file
added data/annotations/documents/jof_12756.json +238 −0
@@ -0,0 +1,238 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12756",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:20+00:00",
9 + "source_sha256": "fe661656f0bccf6917825d8844475fd548c8858f7686a31c616748fba4e687f4",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 11381,
12 + "usage": {
13 + "input": 4106,
14 + "output": 2370,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.064618,
19 + "request_id": "req_011CggqLwaX1ACqpRRCHpaDU",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "entrepreneurial_finance_private_equity",
23 + "subfield_secondary": "career concerns / delegated investment; IPO market information frictions",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; the model compares a profit-motivated VC equilibrium with a reputation-motivated VC equilibrium",
28 + "key_assumptions": [
29 + "VCs care about reputation beyond profits because fixed fees depend on capital flows",
30 + "VC skill is the ability to identify good firms (being informed)",
31 + "Asymmetric information between VC-backed firms and IPO bidders"
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": 0
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": "related_literature_positioning"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "preview_of_approach"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "preview_of_results"
80 + },
81 + {
82 + "paragraph_index": 0,
83 + "move": "mechanism_or_theory_sketch"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "other"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "contribution_statement"
92 + },
93 + {
94 + "paragraph_index": 2,
95 + "move": "related_literature_positioning"
96 + }
97 + ],
98 + "contribution_paragraph_index": 2,
99 + "contribution_types": [
100 + "new_theory_or_model"
101 + ],
102 + "roadmap_present": false,
103 + "literature_positioning_location": "inside_introduction"
104 + },
105 + "section_roles": [
106 + {
107 + "idx": 0,
108 + "role": "introduction"
109 + },
110 + {
111 + "idx": 1,
112 + "role": "theory"
113 + },
114 + {
115 + "idx": 2,
116 + "role": "theory"
117 + },
118 + {
119 + "idx": 3,
120 + "role": "results"
121 + },
122 + {
123 + "idx": 4,
124 + "role": "robustness"
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": "main_text_section",
142 + "types": [
143 + "alternative_specification",
144 + "sensitivity_to_parameters"
145 + ]
146 + },
147 + "tables": {
148 + "convention_notes": "No tables. One figure shows the timeline of moves. Results are stated as lemmas, propositions and corollaries, with all proofs in Appendix A.",
149 + "units_stated_in_notes": false,
150 + "standard_errors_description": "none"
151 + },
152 + "figures": {
153 + "uncertainty_shown": "no"
154 + },
155 + "conclusion": {
156 + "bounded_claims": true,
157 + "policy_implications": false,
158 + "open_questions": false,
159 + "restates_contribution": true
160 + },
161 + "hedging": {
162 + "level": "moderate",
163 + "typical_devices": [
164 + "modal 'can' for effects",
165 + "'arguably' when justifying assumptions",
166 + "conditional welfare statements (depends on abundance or profitability of good firms)"
167 + ]
168 + },
169 + "claim_evidence": [
170 + {
171 + "claim": "Uninformed reputation-motivated VCs are biased against backing firms so as to avoid revealing their lack of information",
172 + "evidence_type": "equation_or_proof",
173 + "evidence_ref": "Section II.C; Appendix A propositions"
174 + },
175 + {
176 + "claim": "Reputation motivation creates a certification effect that can prevent IPO market breakdowns",
177 + "evidence_type": "equation_or_proof",
178 + "evidence_ref": "Section III.A; Appendix A"
179 + },
180 + {
181 + "claim": "Reputation motivation lowers welfare when good firms are abundant or profitable",
182 + "evidence_type": "equation_or_proof",
183 + "evidence_ref": "Section III.B; Appendix A corollaries"
184 + },
185 + {
186 + "claim": "The VC's equilibrium profits can exceed those of a purely profit-maximizing VC",
187 + "evidence_type": "equation_or_proof",
188 + "evidence_ref": "Section III"
189 + },
190 + {
191 + "claim": "VC reputation motivation stems from fixed fees on attracted capital, and picking good firms is central to VC skill",
192 + "evidence_type": "prior_literature",
193 + "evidence_ref": "Introduction subsection A; Section I.D; Nelson et al. (2004), Gompers et al. (2006)"
194 + },
195 + {
196 + "claim": "Results survive extensions such as assets in place, IPO discount, nonlinear flow-performance and partially informed VCs",
197 + "evidence_type": "equation_or_proof",
198 + "evidence_ref": "Section IV"
199 + }
200 + ],
201 + "structural_summary": "Theory paper. The introduction contrasts the known harms of career concerns in secondary markets with the unexplored primary market, poses the question for VCs, and previews a counterintuitive result and its mechanism. It then discusses the realism of the assumptions and empirical proxies, and positions the paper against the career-concerns literature. The body sets up the model, characterizes the equilibria under profit and reputation motivation, compares their costs and benefits, and runs extensions as robustness checks. A short conclusion restates the main finding, and the proofs are collected in an appendix.",
202 + "writing_profile_keywords": [
203 + "career concerns",
204 + "certification effect",
205 + "counterintuitive result",
206 + "equilibrium comparison",
207 + "proofs in appendix",
208 + "extensions as robustness",
209 + "empirical proxies discussion"
210 + ],
211 + "confidence": 0.85,
212 + "notes": "The final abstract sentence fragment is truncated introduction text and was ignored. No roadmap paragraph is visible in the excerpts."
213 + },
214 + "deterministic_checks": {
215 + "roadmap_regex": false,
216 + "contribution_regex_in_intro": true,
217 + "magnitude_in_abstract": false,
218 + "abstract_n_sentences": 5,
219 + "intro_n_paragraphs": 3,
220 + "intro_n_words": 700,
221 + "causal_words_in_abstract": 0,
222 + "hedge_words_in_abstract": 1,
223 + "stars_mentioned": false,
224 + "clustered_se_mentioned": false,
225 + "internet_appendix_mentioned": false,
226 + "section_roles_rule": {
227 + "0": "introduction",
228 + "1": "theory",
229 + "2": "theory",
230 + "3": "other",
231 + "4": "robustness",
232 + "5": "conclusion",
233 + "6": "appendix"
234 + }
235 + },
236 + "schema_valid": true,
237 + "code_version": "87c73cd-dirty"
238 +}
\ No newline at end of file
added data/annotations/documents/jof_12757.json +243 −0
@@ -0,0 +1,243 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12757",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:01+00:00",
9 + "source_sha256": "246ac7ac6e4b23de358a71c3c87c8386634a8fd3430da867a252f1ef020bde18",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 17548,
12 + "usage": {
13 + "input": 6182,
14 + "output": 1880,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.063122,
19 + "request_id": "req_011CggqKoAxwiQXKFBm5JRfa",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; analytical characterization of equilibrium in a continuous-time model",
28 + "key_assumptions": [
29 + "fixed exogenous riskless rate",
30 + "hedgers' asset demand independent of wealth",
31 + "arbitrageurs have CRRA utility",
32 + "no asymmetric information or leverage constraints"
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": 0
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": 0,
72 + "move": "preview_of_approach"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "contribution_statement"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "related_literature_positioning"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "preview_of_results"
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 + "contribution_paragraph_index": 0,
108 + "contribution_types": [
109 + "new_theory_or_model"
110 + ],
111 + "roadmap_present": false,
112 + "literature_positioning_location": "inside_introduction"
113 + },
114 + "section_roles": [
115 + {
116 + "idx": 0,
117 + "role": "introduction"
118 + },
119 + {
120 + "idx": 1,
121 + "role": "theory"
122 + },
123 + {
124 + "idx": 2,
125 + "role": "theory"
126 + },
127 + {
128 + "idx": 3,
129 + "role": "theory"
130 + },
131 + {
132 + "idx": 4,
133 + "role": "theory"
134 + },
135 + {
136 + "idx": 5,
137 + "role": "theory"
138 + },
139 + {
140 + "idx": 6,
141 + "role": "conclusion"
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": "none_seen",
151 + "types": []
152 + },
153 + "tables": {
154 + "convention_notes": "No tables; results are delivered as propositions and numerical illustrations in figures with parameter values stated in the captions.",
155 + "units_stated_in_notes": false,
156 + "standard_errors_description": "none"
157 + },
158 + "figures": {
159 + "uncertainty_shown": "no"
160 + },
161 + "conclusion": {
162 + "bounded_claims": true,
163 + "policy_implications": false,
164 + "open_questions": true,
165 + "restates_contribution": true
166 + },
167 + "hedging": {
168 + "level": "moderate",
169 + "typical_devices": [
170 + "modal 'can' for possibility results",
171 + "conditioning on parameter ranges (e.g., CRRA below one)",
172 + "explicit statement of simplifying assumptions and their scope"
173 + ]
174 + },
175 + "claim_evidence": [
176 + {
177 + "claim": "Arbitrageur risk aversion equals static CRRA plus a forward-looking intertemporal hedging component",
178 + "evidence_type": "equation_or_proof",
179 + "evidence_ref": "Section II.C.2"
180 + },
181 + {
182 + "claim": "More risk-averse hedgers can make arbitrageurs with CRRA below one endogenously more risk-averse",
183 + "evidence_type": "figure",
184 + "evidence_ref": "Figure 1; Section II.C.2"
185 + },
186 + {
187 + "claim": "Risk premia can rise with arbitrageur wealth at low wealth even as Sharpe ratios fall, due to endogenous risk",
188 + "evidence_type": "figure",
189 + "evidence_ref": "Figure 2; Section III.B"
190 + },
191 + {
192 + "claim": "Arbitrageur wealth is the single priced factor, and aggregate illiquidity captures it",
193 + "evidence_type": "equation_or_proof",
194 + "evidence_ref": "Section IV"
195 + },
196 + {
197 + "claim": "Zero-supply assumption is without loss of generality with long-lived hedgers",
198 + "evidence_type": "equation_or_proof",
199 + "evidence_ref": "Proposition 12; Section V"
200 + },
201 + {
202 + "claim": "Aggregate liquidity is empirically time-varying and carries a positive premium",
203 + "evidence_type": "prior_literature",
204 + "evidence_ref": "Introduction paragraph 3, footnote 2"
205 + }
206 + ],
207 + "structural_summary": "A purely theoretical asset-pricing paper. The introduction motivates the work with the role of intermediary capital in liquidity provision. It states a minimal-friction modeling choice and previews analytical results on dynamic risk aversion, endogenous risk, and the priced arbitrageur-wealth factor proxied by aggregate illiquidity. It closes by positioning against margin-constraint and log-utility arbitrage models. The body moves from model setup to equilibrium with short-maturity assets, then long-maturity assets, liquidity risk, and a positive-supply extension. The conclusion restates the contributions and proposes stochastic hedger supply as an extension.",
208 + "writing_profile_keywords": [
209 + "continuous-time model",
210 + "analytical propositions",
211 + "endogenous risk",
212 + "intermediary asset pricing",
213 + "numerical illustrations",
214 + "literature differentiation"
215 + ],
216 + "confidence": 0.85,
217 + "notes": "No empirical analysis. Figures are numerical illustrations of the model, not data. Literature positioning occupies the latter introduction paragraphs and reads like a related-literature subsection, but the excerpts do not show it as separate."
218 + },
219 + "deterministic_checks": {
220 + "roadmap_regex": false,
221 + "contribution_regex_in_intro": false,
222 + "magnitude_in_abstract": false,
223 + "abstract_n_sentences": 6,
224 + "intro_n_paragraphs": 6,
225 + "intro_n_words": 1204,
226 + "causal_words_in_abstract": 0,
227 + "hedge_words_in_abstract": 0,
228 + "stars_mentioned": false,
229 + "clustered_se_mentioned": false,
230 + "internet_appendix_mentioned": false,
231 + "section_roles_rule": {
232 + "0": "introduction",
233 + "1": "theory",
234 + "2": "theory",
235 + "3": "theory",
236 + "4": "other",
237 + "5": "other",
238 + "6": "conclusion"
239 + }
240 + },
241 + "schema_valid": true,
242 + "code_version": "87c73cd-dirty"
243 +}
\ No newline at end of file
added data/annotations/documents/jof_12758.json +259 −0
@@ -0,0 +1,259 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12758",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:31+00:00",
9 + "source_sha256": "8ae0853b4885957d3b2a9704e7cecb575225be9f726338b798d1c653efac5dfa",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 18158,
12 + "usage": {
13 + "input": 6351,
14 + "output": 2125,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.068698,
19 + "request_id": "req_011CggqMveUkwtMKV3FMVLW2",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "corporate_finance (capital structure / structural credit risk models)",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Cross-sectional pricing of unlevered market beta, size, BTM and volatility in leverage-adjusted (unlevered) equity returns, compared with levered stock returns",
27 + "strategy": "Unlever stock returns in two ways: data-driven scaling by (1 - leverage), and a parametric Merton or Leland-Toft structural model. Then run portfolio sorts and Fama-MacBeth regressions on the unlevered returns.",
28 + "key_assumptions": [
29 + "Linear asset pricing models allow levered risk to be decomposed into asset risk plus leverage-related risk",
30 + "The structural credit model (Merton/Leland-Toft) correctly maps equity returns to asset returns",
31 + "Book liabilities and assumed debt maturity proxy adequately for the debt structure"
32 + ],
33 + "threats_addressed": [
34 + "Model dependence of the unlevering (checked with Leland-Toft and the scaling approach)",
35 + "Debt maturity and face-value measurement choices",
36 + "Portfolio weighting choices",
37 + "Nonlinearity of leverage that regression controls fail to capture (heteroskedasticity)"
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": 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": "literature_gap"
81 + },
82 + {
83 + "paragraph_index": 0,
84 + "move": "contribution_statement"
85 + },
86 + {
87 + "paragraph_index": 0,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 1,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 1,
96 + "move": "robustness_mention"
97 + }
98 + ],
99 + "contribution_paragraph_index": 0,
100 + "contribution_types": [
101 + "new_evidence",
102 + "new_theory_or_model",
103 + "resolution_of_puzzle"
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": "data"
120 + },
121 + {
122 + "idx": 3,
123 + "role": "results"
124 + },
125 + {
126 + "idx": 4,
127 + "role": "results"
128 + },
129 + {
130 + "idx": 5,
131 + "role": "robustness"
132 + },
133 + {
134 + "idx": 6,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 7,
139 + "role": "conclusion"
140 + },
141 + {
142 + "idx": 8,
143 + "role": "appendix"
144 + }
145 + ],
146 + "results_presentation": {
147 + "economic_significance_discussed": true,
148 + "statistical_significance_language": "stars_and_tstats",
149 + "magnitude_statements_in_intro": false
150 + },
151 + "robustness": {
152 + "location": "main_text_section",
153 + "types": [
154 + "alternative_specification",
155 + "alternative_measure",
156 + "sensitivity_to_parameters",
157 + "other"
158 + ]
159 + },
160 + "tables": {
161 + "convention_notes": "Panels report value-weighted returns for 25 double-sorted or 5 single-sorted portfolios, followed by Fama-MacBeth regression panels with t-statistics in parentheses. Notes define the leverage measure and its timing, and give the simulation parameters.",
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": false,
171 + "open_questions": false,
172 + "restates_contribution": true
173 + },
174 + "hedging": {
175 + "level": "moderate",
176 + "typical_devices": [
177 + "'in part' qualifiers",
178 + "'suggest'",
179 + "'virtually disappear' / 'weakened'",
180 + "'consistent with theoretical predictions'"
181 + ]
182 + },
183 + "claim_evidence": [
184 + {
185 + "claim": "Unlevered market beta is more strongly priced than levered beta, even after controlling for size and BTM",
186 + "evidence_type": "table",
187 + "evidence_ref": "Tables I, III, IV; Figure 2"
188 + },
189 + {
190 + "claim": "The value premium virtually disappears and the size effect weakens for unlevered returns",
191 + "evidence_type": "table",
192 + "evidence_ref": "Tables III, IV"
193 + },
194 + {
195 + "claim": "Leverage induces heteroskedasticity in stock returns that unlevering removes",
196 + "evidence_type": "figure",
197 + "evidence_ref": "Figure 1, Figure 3"
198 + },
199 + {
200 + "claim": "Controlling for leverage in regressions does not adequately address its nonlinear effect",
201 + "evidence_type": "table",
202 + "evidence_ref": "Table V"
203 + },
204 + {
205 + "claim": "The volatility puzzle holds for levered returns but not for unlevered returns",
206 + "evidence_type": "table",
207 + "evidence_ref": "Table VIII"
208 + },
209 + {
210 + "claim": "Results are robust to debt maturity, face value of debt, the Leland-Toft model and weighting choices",
211 + "evidence_type": "table",
212 + "evidence_ref": "Tables VI, VII"
213 + },
214 + {
215 + "claim": "Derivative of expected levered returns with respect to levered volatility is negative",
216 + "evidence_type": "equation_or_proof",
217 + "evidence_ref": "Appendix, Equation (3)"
218 + }
219 + ],
220 + "structural_summary": "The paper opens with a question and motivates it with well-known anomalies, then positions itself as the first to test hypotheses on unlevered returns. A theory section derives predictions for asset beta, size, BTM and volatility. After the data section, the empirical work unlevers returns twice: first by data-driven leverage scaling, then with the Merton model, and shows why leverage controls in regressions are inadequate. A dedicated robustness section varies the model and measurement choices, and a separate section treats volatility. The conclusion restates the contribution, and an appendix contains a proof.",
221 + "writing_profile_keywords": [
222 + "unlevered returns",
223 + "anomalies",
224 + "structural credit models",
225 + "Fama-MacBeth",
226 + "portfolio sorts",
227 + "theory-derived hypotheses",
228 + "parsimonious factor model"
229 + ],
230 + "confidence": 0.8,
231 + "notes": "The introduction excerpt is truncated, so a roadmap may exist but is not visible. The standard-error type is not stated in the visible excerpts."
232 + },
233 + "deterministic_checks": {
234 + "roadmap_regex": false,
235 + "contribution_regex_in_intro": false,
236 + "magnitude_in_abstract": false,
237 + "abstract_n_sentences": 5,
238 + "intro_n_paragraphs": 2,
239 + "intro_n_words": 446,
240 + "causal_words_in_abstract": 1,
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": "theory",
248 + "2": "data",
249 + "3": "other",
250 + "4": "theory",
251 + "5": "robustness",
252 + "6": "other",
253 + "7": "conclusion",
254 + "8": "appendix"
255 + }
256 + },
257 + "schema_valid": true,
258 + "code_version": "87c73cd-dirty"
259 +}
\ No newline at end of file
added data/annotations/documents/jof_12759.json +279 −0
@@ -0,0 +1,279 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12759",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:03+00:00",
9 + "source_sha256": "726c51092b92ad4405d3e90a123f4dac31f6b84c1dcb12552dee3fea64e57f5b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 28533,
12 + "usage": {
13 + "input": 9380,
14 + "output": 2241,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.083134,
19 + "request_id": "req_011CggqKjnp2Bz2RepMqRRRb",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "institutional investors / high-frequency trading",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Association between HFT signed net flow and the presence, duration and information content of institutional meta orders, plus investor execution costs and HFT revenue",
27 + "strategy": "Proprietary institutional child-order data linked to HFT trades identified by name; descriptive event-time patterns, a matched placebo sample with similar market conditions, and panel regressions with controls",
28 + "key_assumptions": [
29 + "Placebo periods matched on volume, returns, volatility and order imbalance capture common market conditions",
30 + "Permanent price impact proxies the information content of orders",
31 + "NASDAQ OMX flow is representative because offsetting positions in other venues are unlikely"
32 + ],
33 + "threats_addressed": [
34 + "Common response to market conditions, addressed with the matched placebo sample",
35 + "Offsetting positions in other markets or correlated securities",
36 + "Volume-clock vs wall-clock time",
37 + "Market share of the venue",
38 + "HFT positioning before the first child trade"
39 + ],
40 + "causal_language_strength": 2,
41 + "uses_causal_claim": false
42 + },
43 + "abstract_moves": {
44 + "question": {
45 + "present": true,
46 + "sentence_index": 1
47 + },
48 + "gap": {
49 + "present": false,
50 + "sentence_index": -1
51 + },
52 + "method": {
53 + "present": false,
54 + "sentence_index": -1
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": false,
66 + "sentence_index": -1
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 1,
77 + "move": "literature_gap"
78 + },
79 + {
80 + "paragraph_index": 2,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 3,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 4,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 4,
93 + "move": "identification_defense"
94 + },
95 + {
96 + "paragraph_index": 5,
97 + "move": "identification_defense"
98 + },
99 + {
100 + "paragraph_index": 5,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 19,
105 + "move": "related_literature_positioning"
106 + },
107 + {
108 + "paragraph_index": 20,
109 + "move": "contribution_statement"
110 + },
111 + {
112 + "paragraph_index": 21,
113 + "move": "related_literature_positioning"
114 + },
115 + {
116 + "paragraph_index": 22,
117 + "move": "contribution_statement"
118 + },
119 + {
120 + "paragraph_index": 23,
121 + "move": "other"
122 + }
123 + ],
124 + "contribution_paragraph_index": 20,
125 + "contribution_types": [
126 + "new_evidence",
127 + "new_data_or_measure"
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": "theory"
140 + },
141 + {
142 + "idx": 2,
143 + "role": "data"
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": "robustness"
160 + },
161 + {
162 + "idx": 7,
163 + "role": "conclusion"
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": "internet_appendix",
173 + "types": [
174 + "alternative_measure",
175 + "placebo",
176 + "other"
177 + ]
178 + },
179 + "tables": {
180 + "convention_notes": "Table notes define the sample and observation unit (stock-period meta orders), how variables are constructed and signed (sell orders converted to pseudo-buys), and their standardization and units (dollars, % of shares outstanding). Subsample splits are by order size and duration; odd hours are reported for brevity. Standard-error treatment is not visible in the excerpts.",
181 + "units_stated_in_notes": true,
182 + "standard_errors_description": "unclear"
183 + },
184 + "figures": {
185 + "uncertainty_shown": "unclear"
186 + },
187 + "conclusion": {
188 + "bounded_claims": true,
189 + "policy_implications": true,
190 + "open_questions": false,
191 + "restates_contribution": true
192 + },
193 + "hedging": {
194 + "level": "moderate",
195 + "typical_devices": [
196 + "appear to / seem to",
197 + "consistent with",
198 + "explicit 'Some Words of Caution' paragraph disclaiming causality",
199 + "we believe / we suspect",
200 + "suggests"
201 + ]
202 + },
203 + "claim_evidence": [
204 + {
205 + "claim": "HFTs initially trade against institutional orders, then trade with long-lasting orders after about seven hours, building speculative positions",
206 + "evidence_type": "figure",
207 + "evidence_ref": "Figure 2, Table III"
208 + },
209 + {
210 + "claim": "With-wind trading is absent in matched placebo periods, linking it to the presence of institutional orders",
211 + "evidence_type": "figure",
212 + "evidence_ref": "Figure 3, Table II"
213 + },
214 + {
215 + "claim": "Institutional orders are mostly information-motivated; institutions trade few names per day",
216 + "evidence_type": "figure",
217 + "evidence_ref": "Figure 4, Figure 5, Table V"
218 + },
219 + {
220 + "claim": "Half of investors' private signal is revealed in prices after roughly three hours",
221 + "evidence_type": "figure",
222 + "evidence_ref": "Figure 6"
223 + },
224 + {
225 + "claim": "HFTs trade with-wind most strongly on the most informed long-lasting orders",
226 + "evidence_type": "figure",
227 + "evidence_ref": "Figure 7"
228 + },
229 + {
230 + "claim": "Investors cut their volume participation by about 50% after HFTs turn with-wind",
231 + "evidence_type": "table",
232 + "evidence_ref": "Table VI"
233 + },
234 + {
235 + "claim": "A one-SD rise in HFT with-wind flow is associated with $4,480 higher execution cost and $850 higher HFT gross profit",
236 + "evidence_type": "table",
237 + "evidence_ref": "Table VII, Figure 8"
238 + }
239 + ],
240 + "structural_summary": "The paper opens with HFT background and a gap: little is known about HFT interaction with institutional investors. It sets up three theoretical benchmarks (market-making, predatory trading, back-running) and then describes the linked proprietary and public data. The core results show against-wind then with-wind HFT flow, validated with a placebo match. Information content of orders is used to discriminate between predatory trading and back-running. A final section derives four predictions from an economic narrative and tests each one. Robustness checks are placed in the internet appendix. The introduction contains explicit literature differentiation and a caution paragraph disclaiming causality. The conclusion reframes HFTs as high-tech traders and calls for more granular regulatory data.",
241 + "writing_profile_keywords": [
242 + "theory-benchmark comparison",
243 + "proprietary data",
244 + "placebo matching",
245 + "hypothesis discrimination",
246 + "explicit caveats",
247 + "bulleted findings",
248 + "economic narrative",
249 + "policy discussion"
250 + ],
251 + "confidence": 0.82,
252 + "notes": "Paragraphs 6-18 of the introduction are fragments of a bullet list (findings). The excerpts do not show standard errors or how statistical significance is reported."
253 + },
254 + "deterministic_checks": {
255 + "roadmap_regex": false,
256 + "contribution_regex_in_intro": true,
257 + "magnitude_in_abstract": false,
258 + "abstract_n_sentences": 6,
259 + "intro_n_paragraphs": 24,
260 + "intro_n_words": 1757,
261 + "causal_words_in_abstract": 0,
262 + "hedge_words_in_abstract": 1,
263 + "stars_mentioned": false,
264 + "clustered_se_mentioned": false,
265 + "internet_appendix_mentioned": true,
266 + "section_roles_rule": {
267 + "0": "introduction",
268 + "1": "theory",
269 + "2": "other",
270 + "3": "other",
271 + "4": "other",
272 + "5": "other",
273 + "6": "robustness",
274 + "7": "conclusion"
275 + }
276 + },
277 + "schema_valid": true,
278 + "code_version": "87c73cd-dirty"
279 +}
\ No newline at end of file
added data/annotations/documents/jof_12760.json +264 −0
@@ -0,0 +1,264 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12760",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:38+00:00",
9 + "source_sha256": "5c698f26f1599243f8ec22f494096ac1294d780501a09915db896903d26cf695",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 14881,
12 + "usage": {
13 + "input": 4963,
14 + "output": 2172,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.064086,
19 + "request_id": "req_011CggqNPAg7Jb6TyY4G9qb8",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "macro_finance",
23 + "subfield_secondary": "household_finance",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; analytical characterization of equilibrium in a two-date labor-search model with limited-liability household debt",
28 + "key_assumptions": [
29 + "households protected by limited liability",
30 + "Diamond-Mortensen-Pissarides random matching with bilateral wage bargaining",
31 + "square-root matching probabilities",
32 + "perfectly competitive banks",
33 + "repayment schedules nondecreasing in wages"
34 + ],
35 + "threats_addressed": [
36 + "relaxing equal bargaining power",
37 + "alternative matching function",
38 + "imperfectly competitive banks",
39 + "nonincreasing financial contracts"
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": 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": "preview_of_approach"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "mechanism_or_theory_sketch"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 3,
94 + "move": "preview_of_approach"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "policy_or_broader_implication"
99 + },
100 + {
101 + "paragraph_index": 5,
102 + "move": "other"
103 + },
104 + {
105 + "paragraph_index": 6,
106 + "move": "related_literature_positioning"
107 + },
108 + {
109 + "paragraph_index": 7,
110 + "move": "related_literature_positioning"
111 + },
112 + {
113 + "paragraph_index": 8,
114 + "move": "roadmap"
115 + }
116 + ],
117 + "contribution_paragraph_index": 6,
118 + "contribution_types": [
119 + "new_theory_or_model"
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": "theory"
132 + },
133 + {
134 + "idx": 2,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 3,
139 + "role": "other"
140 + },
141 + {
142 + "idx": 4,
143 + "role": "robustness"
144 + },
145 + {
146 + "idx": 5,
147 + "role": "conclusion"
148 + },
149 + {
150 + "idx": 6,
151 + "role": "appendix"
152 + }
153 + ],
154 + "results_presentation": {
155 + "economic_significance_discussed": false,
156 + "statistical_significance_language": "unclear",
157 + "magnitude_statements_in_intro": false
158 + },
159 + "robustness": {
160 + "location": "main_text_section",
161 + "types": [
162 + "sensitivity_to_parameters",
163 + "alternative_specification"
164 + ]
165 + },
166 + "tables": {
167 + "convention_notes": "No tables; results are delivered as propositions and equations, with proofs collected in the Appendix.",
168 + "units_stated_in_notes": false,
169 + "standard_errors_description": "none"
170 + },
171 + "figures": {
172 + "uncertainty_shown": "no_figures"
173 + },
174 + "conclusion": {
175 + "bounded_claims": false,
176 + "policy_implications": true,
177 + "open_questions": false,
178 + "restates_contribution": true
179 + },
180 + "hedging": {
181 + "level": "moderate",
182 + "typical_devices": [
183 + "'our model is stylized, but may'",
184 + "'our model suggests'",
185 + "'may contribute' / 'may exacerbate' in extension previews",
186 + "'could be a bad thing'"
187 + ]
188 + },
189 + "claim_evidence": [
190 + {
191 + "claim": "Limited-liability household debt creates a debt overhang that makes indebted households demand high wages to work",
192 + "evidence_type": "equation_or_proof",
193 + "evidence_ref": "Section II.A-B; proofs in Appendix"
194 + },
195 + {
196 + "claim": "Debt is the optimal financing contract within the class of repayment schedules increasing in wages",
197 + "evidence_type": "equation_or_proof",
198 + "evidence_ref": "Section II.A; Section IV.D"
199 + },
200 + {
201 + "claim": "Firms respond with high wages and few vacancies, so high household debt raises unemployment (vacancy posting effect)",
202 + "evidence_type": "equation_or_proof",
203 + "evidence_ref": "Section II.B-C"
204 + },
205 + {
206 + "claim": "Equilibrium household debt is inefficiently high because of a household debt externality; debt caps can mitigate it",
207 + "evidence_type": "equation_or_proof",
208 + "evidence_ref": "Section II.D"
209 + },
210 + {
211 + "claim": "Lower default penalties can worsen the vacancy posting effect",
212 + "evidence_type": "equation_or_proof",
213 + "evidence_ref": "Section III.C"
214 + },
215 + {
216 + "claim": "Household debt reducing labor supply is consistent with recent empirical studies",
217 + "evidence_type": "prior_literature",
218 + "evidence_ref": "Bernstein (2018); Herkenhoff (2012); Dobbie and Goldsmith-Pinkham (2015)"
219 + },
220 + {
221 + "claim": "The model matches macro stylized facts on household leverage, wage rigidity and collateral shocks",
222 + "evidence_type": "prior_literature",
223 + "evidence_ref": "Mian and Sufi (2010, 2014b); Bewley (1999); Daly and Hobijn (2015)"
224 + }
225 + ],
226 + "structural_summary": "A theory paper. The introduction opens with the prevalence of personal bankruptcy and limited liability, then poses explicit research questions. It previews a two-date search model and its headline results (debt overhang, optimal debt contract, vacancy posting effect, externality). It then sketches the extensions, draws policy implications, links predictions to existing empirical findings and stylized facts, positions the paper against the aggregate-demand and debt-overhang literatures, and closes with a roadmap. The body proceeds in order: model setup with assumptions, analytical results (including the constrained-efficient benchmark and multiple equilibria), reduced-form extensions, and robustness to relaxed modeling assumptions. Proofs are in the Appendix. The conclusion restates the mechanism and gives a regulatory and monetary-policy implication.",
227 + "writing_profile_keywords": [
228 + "labor search model",
229 + "household debt overhang",
230 + "limited liability",
231 + "externality",
232 + "macroprudential policy",
233 + "propositions with appendix proofs",
234 + "corporate finance analogy",
235 + "labeled intro paragraphs"
236 + ],
237 + "confidence": 0.85,
238 + "notes": "No tables or figures. The intro uses inline labels (Model preview, Results preview, Policy, Empirical content, Related literature, Layout). The Extensions section was coded 'other' because it adds new results rather than serving as robustness. Causal language about the model's mechanism is strong.\n"
239 + },
240 + "deterministic_checks": {
241 + "roadmap_regex": false,
242 + "contribution_regex_in_intro": true,
243 + "magnitude_in_abstract": false,
244 + "abstract_n_sentences": 7,
245 + "intro_n_paragraphs": 9,
246 + "intro_n_words": 1101,
247 + "causal_words_in_abstract": 1,
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": "theory",
255 + "2": "results",
256 + "3": "robustness",
257 + "4": "robustness",
258 + "5": "conclusion",
259 + "6": "appendix"
260 + }
261 + },
262 + "schema_valid": true,
263 + "code_version": "87c73cd-dirty"
264 +}
\ No newline at end of file
added data/annotations/documents/jof_12761.json +320 −0
@@ -0,0 +1,320 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [],
4 + "doc_id": "jof_12761",
5 + "ontology_version": "1.0",
6 + "prompt_version": "1.0",
7 + "model": "claude-opus-5-5",
8 + "effort": "medium",
9 + "annotated_at": "2026-11-03T23:26:47+00:00",
10 + "source_sha256": "538eafef0d3194164dce82369af7a6f77264ca6b288db5d4bacc1e20126e9c6d",
11 + "parser": "pymupdf-jof-heuristics-1.3",
12 + "payload_chars": 31495,
13 + "usage": {
14 + "input": 11283,
15 + "output": 2375,
16 + "cache_read": 3968,
17 + "cache_write": 0
18 + },
19 + "cost_usd": 0.093426,
20 + "request_id": "req_011CggrpyP4Ddma4wQ4m3gJf",
21 + "annotation": {
22 + "genre": "theoretical",
23 + "subfield": "corporate_finance",
24 + "subfield_secondary": "asset_pricing; dynamic contracting; liquidity and risk management",
25 + "primary_method": "theoretical_model_analytical",
26 + "identification": {
27 + "estimand": "not applicable (theoretical model)",
28 + "strategy": "not applicable; continuous-time dynamic optimization with limited commitment (inalienable human capital), solved via an HJB/ODE and shown equivalent to a dual optimal contract, complemented by numerical calibration",
29 + "key_assumptions": [
30 + "entrepreneur cannot irrevocably commit human capital and can leave to recontract elsewhere",
31 + "otherwise frictionless, competitive and complete financial markets",
32 + "risk-averse entrepreneur and risk-averse diversified investors",
33 + "q-theory adjustment costs on investment",
34 + "observable, verifiable output and no private saving"
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": false,
51 + "sentence_index": -1
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": 2
64 + }
65 + },
66 + "introduction": {
67 + "moves": [
68 + {
69 + "paragraph_index": 0,
70 + "move": "motivation"
71 + },
72 + {
73 + "paragraph_index": 0,
74 + "move": "contribution_statement"
75 + },
76 + {
77 + "paragraph_index": 0,
78 + "move": "preview_of_results"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "preview_of_results"
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": "related_literature_positioning"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "contribution_statement"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "related_literature_positioning"
103 + },
104 + {
105 + "paragraph_index": 5,
106 + "move": "mechanism_or_theory_sketch"
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": 7,
118 + "move": "related_literature_positioning"
119 + },
120 + {
121 + "paragraph_index": 8,
122 + "move": "related_literature_positioning"
123 + },
124 + {
125 + "paragraph_index": 9,
126 + "move": "related_literature_positioning"
127 + },
128 + {
129 + "paragraph_index": 10,
130 + "move": "related_literature_positioning"
131 + },
132 + {
133 + "paragraph_index": 11,
134 + "move": "related_literature_positioning"
135 + },
136 + {
137 + "paragraph_index": 12,
138 + "move": "related_literature_positioning"
139 + },
140 + {
141 + "paragraph_index": 13,
142 + "move": "roadmap"
143 + }
144 + ],
145 + "contribution_paragraph_index": 0,
146 + "contribution_types": [
147 + "new_theory_or_model"
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": "theory"
160 + },
161 + {
162 + "idx": 2,
163 + "role": "theory"
164 + },
165 + {
166 + "idx": 3,
167 + "role": "theory"
168 + },
169 + {
170 + "idx": 4,
171 + "role": "theory"
172 + },
173 + {
174 + "idx": 5,
175 + "role": "results"
176 + },
177 + {
178 + "idx": 6,
179 + "role": "theory"
180 + },
181 + {
182 + "idx": 7,
183 + "role": "theory"
184 + },
185 + {
186 + "idx": 8,
187 + "role": "theory"
188 + },
189 + {
190 + "idx": 9,
191 + "role": "conclusion"
192 + },
193 + {
194 + "idx": 10,
195 + "role": "appendix"
196 + },
197 + {
198 + "idx": 11,
199 + "role": "appendix"
200 + },
201 + {
202 + "idx": 12,
203 + "role": "appendix"
204 + },
205 + {
206 + "idx": 13,
207 + "role": "appendix"
208 + }
209 + ],
210 + "results_presentation": {
211 + "economic_significance_discussed": true,
212 + "statistical_significance_language": "unclear",
213 + "magnitude_statements_in_intro": false
214 + },
215 + "robustness": {
216 + "location": "main_text_section",
217 + "types": [
218 + "sensitivity_to_parameters",
219 + "other"
220 + ]
221 + },
222 + "tables": {
223 + "convention_notes": "Few tables. Two tables contrast the primal (liquidity-based) problem with the dual (contracting) problem side by side; one lists annualized baseline parameter values with symbols. There are no regression tables.",
224 + "units_stated_in_notes": true,
225 + "standard_errors_description": "none"
226 + },
227 + "figures": {
228 + "uncertainty_shown": "no"
229 + },
230 + "conclusion": {
231 + "bounded_claims": true,
232 + "policy_implications": false,
233 + "open_questions": true,
234 + "restates_contribution": true
235 + },
236 + "hedging": {
237 + "level": "low",
238 + "typical_devices": [
239 + "'in general' qualifiers on theoretical results",
240 + "consistent-with references to prior empirical findings",
241 + "acknowledging strong assumptions in the conclusion"
242 + ]
243 + },
244 + "claim_evidence": [
245 + {
246 + "claim": "Inalienable human capital creates an endogenous debt capacity and makes liquidity and risk management valuable even with frictionless markets",
247 + "evidence_type": "equation_or_proof",
248 + "evidence_ref": "Sections III-IV, Appendix A-B"
249 + },
250 + {
251 + "claim": "Entrepreneur's net worth is overexposed to idiosyncratic risk and underexposed to systematic risk relative to first-best",
252 + "evidence_type": "figure",
253 + "evidence_ref": "Figures 2 and 4"
254 + },
255 + {
256 + "claim": "The firm underinvests and undercompensates the entrepreneur, more so near the debt limit",
257 + "evidence_type": "figure",
258 + "evidence_ref": "Figures 5 and 6"
259 + },
260 + {
261 + "claim": "The liquidity-based optimization problem is equivalent to a dual optimal long-term contract",
262 + "evidence_type": "table",
263 + "evidence_ref": "Tables I-II, Section IV.F"
264 + },
265 + {
266 + "claim": "Less constrained firms investing more in the market portfolio is consistent with observed corporate holdings of risky assets",
267 + "evidence_type": "prior_literature",
268 + "evidence_ref": "Duchin et al. (2017)"
269 + },
270 + {
271 + "claim": "Debt optimality in Hart and Moore is not robust once the model is generalized",
272 + "evidence_type": "figure",
273 + "evidence_ref": "Section VII, Figure 11"
274 + }
275 + ],
276 + "structural_summary": "Theoretical corporate finance paper. The introduction motivates the problem with key-man risk, states the new theory of liquidity and risk management, and previews the policy distortions. It then gives extended positioning against Hart-Moore, Ai-Li, Rampini-Viswanathan and the limited-commitment and compensation literatures, and ends with a roadmap. The body sets up the model, derives the first-best and the constrained solution, and shows equivalence to a dual contracting problem. A calibrated quantitative section then presents policy functions in figures. Extensions cover persistent shocks, the deterministic Hart-Moore case and two-sided commitment. The conclusion restates the mechanism and suggests future work. Proofs are in appendices.",
277 + "writing_profile_keywords": [
278 + "continuous-time dynamic contracting",
279 + "limited commitment",
280 + "primal-dual equivalence",
281 + "calibrated numerical illustration",
282 + "extensive literature positioning",
283 + "extensions sections",
284 + "policy function figures"
285 + ],
286 + "confidence": 0.85,
287 + "notes": "The quantitative section is a calibration-based illustration of the model rather than an empirical analysis. Empirical links appear only through consistency with prior literature."
288 + },
289 + "deterministic_checks": {
290 + "roadmap_regex": true,
291 + "contribution_regex_in_intro": true,
292 + "magnitude_in_abstract": false,
293 + "abstract_n_sentences": 7,
294 + "intro_n_paragraphs": 14,
295 + "intro_n_words": 1815,
296 + "causal_words_in_abstract": 0,
297 + "hedge_words_in_abstract": 0,
298 + "stars_mentioned": false,
299 + "clustered_se_mentioned": false,
300 + "internet_appendix_mentioned": false,
301 + "section_roles_rule": {
302 + "0": "introduction",
303 + "1": "theory",
304 + "2": "other",
305 + "3": "other",
306 + "4": "other",
307 + "5": "other",
308 + "6": "other",
309 + "7": "other",
310 + "8": "other",
311 + "9": "conclusion",
312 + "10": "appendix",
313 + "11": "appendix",
314 + "12": "appendix",
315 + "13": "appendix"
316 + }
317 + },
318 + "schema_valid": true,
319 + "code_version": "87c73cd-dirty"
320 +}
\ No newline at end of file
added data/annotations/documents/jof_12762.json +218 −0
@@ -0,0 +1,218 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped claim_evidence[0].equation_or_proof"
5 + ],
6 + "doc_id": "jof_12762",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:28+00:00",
12 + "source_sha256": "7438749b916f110b3d338ba03448fa009f27721aa6ec6a8881722d99f4f7ba56",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 8571,
15 + "usage": {
16 + "input": 3142,
17 + "output": 1720,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.047762,
22 + "request_id": "req_011CggqP8hJc42PebUm37ZTR",
23 + "annotation": {
24 + "genre": "replication_or_correction",
25 + "subfield": "financial_intermediation_banking",
26 + "subfield_secondary": "asset_pricing (contingent capital pricing, equilibrium uniqueness)",
27 + "primary_method": "theoretical_model_analytical",
28 + "identification": {
29 + "estimand": "not applicable (theoretical correction)",
30 + "strategy": "not applicable; analytical counterexample within the original continuous-time framework",
31 + "key_assumptions": [
32 + "Model consistent with SW's continuous-time framework",
33 + "Conversion triggered by the stock price",
34 + "Stochastic continuous-time setting"
35 + ],
36 + "threats_addressed": [],
37 + "causal_language_strength": 1,
38 + "uses_causal_claim": false
39 + },
40 + "abstract_moves": {
41 + "question": {
42 + "present": false,
43 + "sentence_index": -1
44 + },
45 + "gap": {
46 + "present": true,
47 + "sentence_index": 0
48 + },
49 + "method": {
50 + "present": true,
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": 0,
78 + "move": "preview_of_results"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "related_literature_positioning"
83 + },
84 + {
85 + "paragraph_index": 0,
86 + "move": "contribution_statement"
87 + }
88 + ],
89 + "contribution_paragraph_index": 0,
90 + "contribution_types": [
91 + "replication_or_correction",
92 + "new_theory_or_model"
93 + ],
94 + "roadmap_present": false,
95 + "literature_positioning_location": "inside_introduction"
96 + },
97 + "section_roles": [
98 + {
99 + "idx": 0,
100 + "role": "introduction"
101 + },
102 + {
103 + "idx": 1,
104 + "role": "theory"
105 + },
106 + {
107 + "idx": 2,
108 + "role": "theory"
109 + },
110 + {
111 + "idx": 3,
112 + "role": "theory"
113 + },
114 + {
115 + "idx": 4,
116 + "role": "conclusion"
117 + },
118 + {
119 + "idx": 5,
120 + "role": "appendix"
121 + }
122 + ],
123 + "results_presentation": {
124 + "economic_significance_discussed": false,
125 + "statistical_significance_language": "unclear",
126 + "magnitude_statements_in_intro": false
127 + },
128 + "robustness": {
129 + "location": "none_seen",
130 + "types": []
131 + },
132 + "tables": {
133 + "convention_notes": "No tables. Results are stated as lemmas, propositions, a theorem and a corollary, with proofs in the appendix.",
134 + "units_stated_in_notes": false,
135 + "standard_errors_description": "none"
136 + },
137 + "figures": {
138 + "uncertainty_shown": "no_figures"
139 + },
140 + "conclusion": {
141 + "bounded_claims": true,
142 + "policy_implications": true,
143 + "open_questions": false,
144 + "restates_contribution": true
145 + },
146 + "hedging": {
147 + "level": "low",
148 + "typical_devices": [
149 + "modal 'can exist' for uniqueness under broader terms",
150 + "explicit scope conditions (e.g., when conversion terms penalize shareholders)",
151 + "'not necessarily' when stating design implications"
152 + ]
153 + },
154 + "claim_evidence": [
155 + {
156 + "claim": "SW's Theorem 1 is invalid because its pricing restriction is needed only at conversion, not before",
157 + "evidence_type": "equation_or_proof",
158 + "evidence_ref": "Section II; Appendix proofs"
159 + },
160 + {
161 + "claim": "A parametric model consistent with SW yields closed-form stock and CC prices with a unique equilibrium under value-transferring terms",
162 + "evidence_type": "equation_or_proof",
163 + "evidence_ref": "Section I, Corollary 1; Appendix E"
164 + },
165 + {
166 + "claim": "Uniqueness requires that conversion occur when the post-conversion stock price equals the trigger",
167 + "evidence_type": "equation_or_proof",
168 + "evidence_ref": "Proposition 1; Appendix B"
169 + },
170 + {
171 + "claim": "Multiple equilibria can arise in static or deterministic settings but not in the continuous-time stochastic one, even with sunspots",
172 + "evidence_type": "equation_or_proof",
173 + "evidence_ref": "Section III"
174 + },
175 + {
176 + "claim": "Concurrent work also finds uniqueness when conversion transfers value to CC investors",
177 + "evidence_type": "prior_literature",
178 + "evidence_ref": "Glasserman and Nouri (2016)"
179 + }
180 + ],
181 + "structural_summary": "A short corrigendum. The abstract and a single introductory paragraph flag an error in a prior theorem, state the corrected result, and distinguish the paper from concurrent work. Section I builds a counterexample model with closed-form prices. Section II pinpoints the faulty step in the original proof. Section III reconciles the findings with static, deterministic and sunspot settings where multiple equilibria can occur. The conclusion restates the correction and draws an implication for contingent-capital design. Formal proofs are collected in an appendix.",
182 + "writing_profile_keywords": [
183 + "corrigendum",
184 + "counterexample",
185 + "closed-form solutions",
186 + "continuous-time model",
187 + "equilibrium uniqueness",
188 + "contingent capital",
189 + "proof in appendix",
190 + "concise"
191 + ],
192 + "confidence": 0.85,
193 + "notes": "The abstract's final sentence appears to be an extraction artifact that duplicates the opening of the introduction. The `claim_evidence` array contains a malformed key (`equation_or_proof: null`) in its first entry, which is not allowed by the schema. Roles for Section I were set to theory although the role_rule labeled it 'other'."
194 + },
195 + "deterministic_checks": {
196 + "roadmap_regex": false,
197 + "contribution_regex_in_intro": false,
198 + "magnitude_in_abstract": false,
199 + "abstract_n_sentences": 5,
200 + "intro_n_paragraphs": 1,
201 + "intro_n_words": 241,
202 + "causal_words_in_abstract": 0,
203 + "hedge_words_in_abstract": 1,
204 + "stars_mentioned": false,
205 + "clustered_se_mentioned": false,
206 + "internet_appendix_mentioned": false,
207 + "section_roles_rule": {
208 + "0": "introduction",
209 + "1": "other",
210 + "2": "theory",
211 + "3": "theory",
212 + "4": "conclusion",
213 + "5": "appendix"
214 + }
215 + },
216 + "schema_valid": true,
217 + "code_version": "87c73cd-dirty"
218 +}
\ No newline at end of file
added data/annotations/documents/jof_12763.json +273 −0
@@ -0,0 +1,273 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12763",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:07:14+00:00",
9 + "source_sha256": "f2a068f4774f1dd01c3a71b5d3255901c8d1e807c3fec7a9c012888f81d6950c",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 26108,
12 + "usage": {
13 + "input": 8663,
14 + "output": 2230,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.080046,
19 + "request_id": "req_011CggqLYrHGW6qRXytiHu9C",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "How much broker incentives (fees) and consumer search costs distort retail purchases toward dominated reverse convertible bonds, and the resulting losses to consumers",
27 + "strategy": "Within-set comparisons of simultaneously issued, otherwise-identical dominated and superior bonds using set fixed effects, plus maximum-likelihood estimation of a broker-intermediated search model with heterogeneous consumer sophistication and counterfactual search regimes",
28 + "key_assumptions": [
29 + "Reverse convertibles are fully described by a few observable features, so dominance is unambiguous",
30 + "Disclosed broker fees do not reflect differences in advice or bundled services",
31 + "Consumers use the same search strategies across markets (the reservation-utility distribution is constant)",
32 + "Consumers' beliefs about offered utilities match the empirical density"
33 + ],
34 + "threats_addressed": [
35 + "Unobserved product characteristics and preference heterogeneity that could rationalize dispersion",
36 + "Endogeneity of fees with respect to unobserved product quality"
37 + ],
38 + "causal_language_strength": 3,
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": 3
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": 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": "policy_or_broader_implication"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "literature_gap"
80 + },
81 + {
82 + "paragraph_index": 2,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 3,
87 + "move": "identification_defense"
88 + },
89 + {
90 + "paragraph_index": 4,
91 + "move": "identification_defense"
92 + },
93 + {
94 + "paragraph_index": 5,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 6,
99 + "move": "mechanism_or_theory_sketch"
100 + },
101 + {
102 + "paragraph_index": 6,
103 + "move": "preview_of_approach"
104 + },
105 + {
106 + "paragraph_index": 7,
107 + "move": "related_literature_positioning"
108 + },
109 + {
110 + "paragraph_index": 7,
111 + "move": "contribution_statement"
112 + },
113 + {
114 + "paragraph_index": 8,
115 + "move": "related_literature_positioning"
116 + },
117 + {
118 + "paragraph_index": 9,
119 + "move": "related_literature_positioning"
120 + },
121 + {
122 + "paragraph_index": 9,
123 + "move": "roadmap"
124 + }
125 + ],
126 + "contribution_paragraph_index": 7,
127 + "contribution_types": [
128 + "new_data_or_measure",
129 + "new_evidence",
130 + "new_theory_or_model"
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": "data"
143 + },
144 + {
145 + "idx": 2,
146 + "role": "results"
147 + },
148 + {
149 + "idx": 3,
150 + "role": "theory"
151 + },
152 + {
153 + "idx": 4,
154 + "role": "results"
155 + },
156 + {
157 + "idx": 5,
158 + "role": "other"
159 + },
160 + {
161 + "idx": 6,
162 + "role": "conclusion"
163 + },
164 + {
165 + "idx": 7,
166 + "role": "appendix"
167 + }
168 + ],
169 + "results_presentation": {
170 + "economic_significance_discussed": true,
171 + "statistical_significance_language": "stars_and_tstats",
172 + "magnitude_statements_in_intro": true
173 + },
174 + "robustness": {
175 + "location": "none_seen",
176 + "types": []
177 + },
178 + "tables": {
179 + "convention_notes": "Notes give the sample (SEC-registered one-year reverse convertibles, 2008-2012, largest 1% of issues dropped), the equation being estimated, and the fixed effects used (issuer, underlying equity, month, dominated/superior set). Stars mark 10/5/1% significance. Standard errors for the structural model come from the observed Fisher information matrix. Counterfactual tables report changes in expected returns.",
180 + "units_stated_in_notes": true,
181 + "standard_errors_description": "unclear"
182 + },
183 + "figures": {
184 + "uncertainty_shown": "unclear"
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 + "evidence consistent with",
196 + "the evidence suggests",
197 + "I argue",
198 + "likely not unique",
199 + "may not be acting"
200 + ]
201 + },
202 + "claim_evidence": [
203 + {
204 + "claim": "Implied market values of reverse convertibles are widely dispersed (SD above 2.4% of notional).",
205 + "evidence_type": "figure",
206 + "evidence_ref": "Figure 3"
207 + },
208 + {
209 + "claim": "When dominated and superior versions coexist, consumers buy more of the dominated one.",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table II, Figure 4"
212 + },
213 + {
214 + "claim": "Products with higher broker fees have lower payoffs to consumers.",
215 + "evidence_type": "table",
216 + "evidence_ref": "Table III, Figure 5"
217 + },
218 + {
219 + "claim": "Brokers earn roughly twice the fee for selling dominated bonds.",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table III"
222 + },
223 + {
224 + "claim": "Estimated search costs are a key friction, and brokers price discriminate by consumer sophistication.",
225 + "evidence_type": "table",
226 + "evidence_ref": "Tables IV, V; Figure 7"
227 + },
228 + {
229 + "claim": "Consumer returns would change under random or optimally directed search compared with broker-directed search.",
230 + "evidence_type": "table",
231 + "evidence_ref": "Table VI"
232 + }
233 + ],
234 + "structural_summary": "The introduction is motivated by price dispersion in financial products and policy concern over brokers not held to a fiduciary standard. It names data challenges in prior work and presents a novel reverse convertible dataset as a clean setting where dominance is unambiguous and broker fees are disclosed. It then previews three stylized facts and a broker-intermediated search model. Literature positioning and a roadmap close the introduction. The body moves from institutional background and data, to reduced-form evidence linking issue size and fees to product quality, to model setup and equilibrium, to maximum-likelihood estimation, and finally to search counterfactuals. The conclusion restates the findings and argues they likely generalize beyond this market.",
235 + "writing_profile_keywords": [
236 + "clean setting",
237 + "dominated products",
238 + "conflict of interest",
239 + "structural search model",
240 + "stylized facts",
241 + "counterfactuals",
242 + "novel data",
243 + "first-person singular"
244 + ],
245 + "confidence": 0.8,
246 + "notes": "The section labeled 'literature' by rule is actually institutional background and data. Robustness checks are not visible in the excerpts. Policy discussion appears in the introduction (the Obama fiduciary quote) but not explicitly in the conclusion excerpt."
247 + },
248 + "deterministic_checks": {
249 + "roadmap_regex": true,
250 + "contribution_regex_in_intro": true,
251 + "magnitude_in_abstract": false,
252 + "abstract_n_sentences": 5,
253 + "intro_n_paragraphs": 10,
254 + "intro_n_words": 1608,
255 + "causal_words_in_abstract": 0,
256 + "hedge_words_in_abstract": 1,
257 + "stars_mentioned": true,
258 + "clustered_se_mentioned": false,
259 + "internet_appendix_mentioned": false,
260 + "section_roles_rule": {
261 + "0": "introduction",
262 + "1": "literature",
263 + "2": "other",
264 + "3": "theory",
265 + "4": "theory",
266 + "5": "other",
267 + "6": "conclusion",
268 + "7": "appendix"
269 + }
270 + },
271 + "schema_valid": true,
272 + "code_version": "87c73cd-dirty"
273 +}
\ No newline at end of file
added data/annotations/documents/jof_12764.json +223 −0
@@ -0,0 +1,223 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.figures_note_placeholder_removed"
5 + ],
6 + "doc_id": "jof_12764",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:28+00:00",
12 + "source_sha256": "188e7ca2a35e22875fc246dfb34d81351980f7bfe852121cbd3d2f936eead2c4",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 12786,
15 + "usage": {
16 + "input": 4806,
17 + "output": 1656,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.053138,
22 + "request_id": "req_011CggqNW87nM8jy5NKVkRHS",
23 + "annotation": {
24 + "genre": "theoretical",
25 + "subfield": "macro_finance",
26 + "subfield_secondary": "information economics / feedback between financial markets and real economy",
27 + "primary_method": "theoretical_model_analytical",
28 + "identification": {
29 + "estimand": "not applicable (theoretical model)",
30 + "strategy": "not applicable; analytical general equilibrium model with numerical illustrations",
31 + "key_assumptions": [
32 + "mutual learning between firms and financial markets",
33 + "monopolistic competition with constant markup",
34 + "CARA speculators",
35 + "OLG dynamic structure"
36 + ],
37 + "threats_addressed": [],
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": 1
57 + },
58 + "magnitude": {
59 + "present": false,
60 + "sentence_index": -1
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": 0,
79 + "move": "preview_of_approach"
80 + },
81 + {
82 + "paragraph_index": 0,
83 + "move": "mechanism_or_theory_sketch"
84 + }
85 + ],
86 + "contribution_paragraph_index": 0,
87 + "contribution_types": [
88 + "new_theory_or_model"
89 + ],
90 + "roadmap_present": false,
91 + "literature_positioning_location": "inside_introduction"
92 + },
93 + "section_roles": [
94 + {
95 + "idx": 0,
96 + "role": "introduction"
97 + },
98 + {
99 + "idx": 1,
100 + "role": "theory"
101 + },
102 + {
103 + "idx": 2,
104 + "role": "theory"
105 + },
106 + {
107 + "idx": 3,
108 + "role": "theory"
109 + },
110 + {
111 + "idx": 4,
112 + "role": "theory"
113 + },
114 + {
115 + "idx": 5,
116 + "role": "other"
117 + },
118 + {
119 + "idx": 6,
120 + "role": "conclusion"
121 + },
122 + {
123 + "idx": 7,
124 + "role": "appendix"
125 + },
126 + {
127 + "idx": 8,
128 + "role": "appendix"
129 + }
130 + ],
131 + "results_presentation": {
132 + "economic_significance_discussed": false,
133 + "statistical_significance_language": "unclear",
134 + "magnitude_statements_in_intro": false
135 + },
136 + "robustness": {
137 + "location": "none_seen",
138 + "types": []
139 + },
140 + "tables": {
141 + "convention_notes": "Two tables: one lists the parameter values used for the numerical illustration, with descriptions and sources; the other compares good versus bad steady-state equilibrium outcomes (GDP, investment, TFP, information precisions). No inferential statistics.",
142 + "units_stated_in_notes": false,
143 + "standard_errors_description": "none"
144 + },
145 + "figures": {
146 + "uncertainty_shown": "no"
147 + },
148 + "conclusion": {
149 + "bounded_claims": true,
150 + "policy_implications": false,
151 + "open_questions": true,
152 + "restates_contribution": true
153 + },
154 + "hedging": {
155 + "level": "low",
156 + "typical_devices": [
157 + "'possibly' for multiplicity of steady states",
158 + "acknowledging the model is too stylized to calibrate",
159 + "noting modeling assumptions such as constant markup"
160 + ]
161 + },
162 + "claim_evidence": [
163 + {
164 + "claim": "Uncertainty in financial and real sectors spiked jointly in the Great Recession, motivating the question",
165 + "evidence_type": "prior_literature",
166 + "evidence_ref": "VIX; Jurado, Ludvigson, and Ng (2015); Bloom et al. (2012)"
167 + },
168 + {
169 + "claim": "Mutual learning creates strategic complementarity in information production between firms and speculators",
170 + "evidence_type": "equation_or_proof",
171 + "evidence_ref": "Sections I–II; Appendix B"
172 + },
173 + {
174 + "claim": "General equilibrium model yields amplification, co-movement of uncertainties, contagion and cross-sectional implications",
175 + "evidence_type": "equation_or_proof",
176 + "evidence_ref": "Section III.C; Figures 2–4"
177 + },
178 + {
179 + "claim": "Dynamic OLG model can have two steady states (uncertainty traps) and a two-stage crisis",
180 + "evidence_type": "figure",
181 + "evidence_ref": "Figures 5, 6, 8; Table II"
182 + }
183 + ],
184 + "structural_summary": "The paper opens with stylized facts on joint uncertainty spikes and poses them as research questions. It then builds the theory incrementally:\n- a one-firm baseline with exogenous information establishing two-way learning;\n- endogenous information acquisition generating strategic complementarity;\n- a macro GE extension that delivers four implications;\n- a dynamic OLG extension that delivers uncertainty traps and transitional crisis dynamics.\nNumerical illustrations with assigned parameters follow, then a conclusion restating the mechanism and noting extensions. Proofs are in the appendix.",
185 + "writing_profile_keywords": [
186 + "theory-building",
187 + "incremental model extension",
188 + "general equilibrium",
189 + "multiple equilibria",
190 + "stylized facts motivation",
191 + "numerical illustration",
192 + "proofs in appendix"
193 + ],
194 + "confidence": 0.8,
195 + "notes": "The introduction is captured as a single paragraph, so moves beyond the motivation and approach preview are not visible. The numerical section explicitly disclaims calibration."
196 + },
197 + "deterministic_checks": {
198 + "roadmap_regex": false,
199 + "contribution_regex_in_intro": false,
200 + "magnitude_in_abstract": false,
201 + "abstract_n_sentences": 4,
202 + "intro_n_paragraphs": 1,
203 + "intro_n_words": 216,
204 + "causal_words_in_abstract": 0,
205 + "hedge_words_in_abstract": 0,
206 + "stars_mentioned": false,
207 + "clustered_se_mentioned": false,
208 + "internet_appendix_mentioned": false,
209 + "section_roles_rule": {
210 + "0": "introduction",
211 + "1": "theory",
212 + "2": "theory",
213 + "3": "theory",
214 + "4": "theory",
215 + "5": "other",
216 + "6": "conclusion",
217 + "7": "appendix",
218 + "8": "appendix"
219 + }
220 + },
221 + "schema_valid": true,
222 + "code_version": "87c73cd-dirty"
223 +}
\ No newline at end of file
added data/annotations/documents/jof_12765.json +259 −0
@@ -0,0 +1,259 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12765",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:17+00:00",
9 + "source_sha256": "7bf1a8e2550a0c6bf66185f46f6a462ac2bc722c2cdfd1d8ee9af3dfbd9766bb",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21279,
12 + "usage": {
13 + "input": 7423,
14 + "output": 2087,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.072226,
19 + "request_id": "req_011CggqRLX1KiMN2ygSHLMHR",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "corporate_finance (structural credit risk)",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Firm-level asset variance risk premium and jump risk premia, plus their contribution to credit spreads",
27 + "strategy": "Firm-by-firm simulated maximum likelihood estimation of a structural SVJ credit model on CDS term structures and option-implied equity volatility, where two observables are assumed accurately priced and the rest observed with error; nested-model likelihood ratio tests; calibration to representative Baa and Aa firms",
28 + "key_assumptions": [
29 + "Structural first-passage model with exogenous default boundary in the baseline",
30 + "Variance risk premium is affine (proportional) in asset variance",
31 + "Two observables are measured without error and the others with error",
32 + "Heston-type asset variance dynamics"
33 + ],
34 + "threats_addressed": [
35 + "Model misspecification via an extended affine risk premium alternative",
36 + "Short-maturity fit, addressed by adding jumps",
37 + "Endogenous default and optimal leverage, examined as extensions"
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": 0
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": 3
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": 3,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "contribution_statement"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 5,
100 + "move": "related_literature_positioning"
101 + },
102 + {
103 + "paragraph_index": 6,
104 + "move": "mechanism_or_theory_sketch"
105 + },
106 + {
107 + "paragraph_index": 6,
108 + "move": "roadmap"
109 + }
110 + ],
111 + "contribution_paragraph_index": 1,
112 + "contribution_types": [
113 + "new_theory_or_model",
114 + "new_evidence",
115 + "resolution_of_puzzle"
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": "theory"
128 + },
129 + {
130 + "idx": 2,
131 + "role": "theory"
132 + },
133 + {
134 + "idx": 3,
135 + "role": "robustness"
136 + },
137 + {
138 + "idx": 4,
139 + "role": "theory"
140 + },
141 + {
142 + "idx": 5,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 6,
147 + "role": "robustness"
148 + },
149 + {
150 + "idx": 7,
151 + "role": "conclusion"
152 + }
153 + ],
154 + "results_presentation": {
155 + "economic_significance_discussed": true,
156 + "statistical_significance_language": "unclear",
157 + "magnitude_statements_in_intro": true
158 + },
159 + "robustness": {
160 + "location": "mixed",
161 + "types": [
162 + "alternative_specification",
163 + "sensitivity_to_parameters",
164 + "other"
165 + ]
166 + },
167 + "tables": {
168 + "convention_notes": "Calibration tables state the calibration targets (Sharpe ratio, default probability, recovery, rates) in their notes and compare risk-premium compositions. Estimation tables report cross-sectional quantiles of parameter estimates and RMSE pricing errors in basis points.",
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": false,
178 + "open_questions": false,
179 + "restates_contribution": true
180 + },
181 + "hedging": {
182 + "level": "moderate",
183 + "typical_devices": [
184 + "suggestive of",
185 + "possible link",
186 + "economically reasonable",
187 + "to the best of our knowledge"
188 + ]
189 + },
190 + "claim_evidence": [
191 + {
192 + "claim": "A reasonable asset variance risk premium lets the calibrated model match medium- and long-maturity spreads for Baa and Aa firms",
193 + "evidence_type": "table",
194 + "evidence_ref": "Tables I-II"
195 + },
196 + {
197 + "claim": "Jump risk premia help fit short-term spreads while the variance risk premium stays economically important",
198 + "evidence_type": "table",
199 + "evidence_ref": "Table V"
200 + },
201 + {
202 + "claim": "Firm-level asset variance risk premia are negative and significant for most firms; mean implied equity variance risk premium is about -16.7%",
203 + "evidence_type": "table",
204 + "evidence_ref": "Tables VII, IX"
205 + },
206 + {
207 + "claim": "The SVJ model dominates the nested SV and constant-volatility models in pricing performance",
208 + "evidence_type": "figure",
209 + "evidence_ref": "Figure 3; Table VIII"
210 + },
211 + {
212 + "claim": "The variance risk premium accounts for about 35% of five-year CDS spreads on average",
213 + "evidence_type": "figure",
214 + "evidence_ref": "Figure 4"
215 + },
216 + {
217 + "claim": "Debt Sharpe ratios exceed equity Sharpe ratios for levered firms, linking the credit spread and distress puzzles",
218 + "evidence_type": "figure",
219 + "evidence_ref": "Figure 2"
220 + }
221 + ],
222 + "structural_summary": "The paper opens by framing the credit spread puzzle and building on the insight that risk premia matter more than default probabilities. It then extends the Leland model with priced stochastic asset volatility and derives semi-closed-form solutions. Calibrations to representative Baa and Aa firms show the puzzle is resolved at medium and long maturities. An alternative risk-premium specification serves as a check, and jumps are added to fit short maturities. The SVJ model is estimated firm by firm with simulated maximum likelihood on CDS and option data, and spreads are decomposed into loss and premium components. Extensions to endogenous default and optimal leverage follow, and the conclusion recaps the findings.",
223 + "writing_profile_keywords": [
224 + "structural credit model",
225 + "puzzle resolution",
226 + "calibration then estimation",
227 + "nested model comparison",
228 + "variance risk premium",
229 + "firm-level estimation"
230 + ],
231 + "confidence": 0.8,
232 + "notes": "How statistical significance is reported is not visible in the excerpts. The estimation section includes data and methodology subsections; it is labeled as results."
233 + },
234 + "deterministic_checks": {
235 + "roadmap_regex": true,
236 + "contribution_regex_in_intro": true,
237 + "magnitude_in_abstract": false,
238 + "abstract_n_sentences": 6,
239 + "intro_n_paragraphs": 7,
240 + "intro_n_words": 1034,
241 + "causal_words_in_abstract": 0,
242 + "hedge_words_in_abstract": 0,
243 + "stars_mentioned": false,
244 + "clustered_se_mentioned": false,
245 + "internet_appendix_mentioned": true,
246 + "section_roles_rule": {
247 + "0": "introduction",
248 + "1": "theory",
249 + "2": "theory",
250 + "3": "robustness",
251 + "4": "theory",
252 + "5": "theory",
253 + "6": "robustness",
254 + "7": "conclusion"
255 + }
256 + },
257 + "schema_valid": true,
258 + "code_version": "87c73cd-dirty"
259 +}
\ No newline at end of file
added data/annotations/documents/jof_12766.json +256 −0
@@ -0,0 +1,256 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12766",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:13+00:00",
9 + "source_sha256": "2f9e655c4aaab4f12e6619b0ef5afd6ecc1a4237f6323e16f12f4b0f9961073f",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 18577,
12 + "usage": {
13 + "input": 6410,
14 + "output": 2189,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.070214,
19 + "request_id": "req_011CggqQxUx2n31r4CQYqMnf",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance; labor and technology adoption",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Cross-sectional association between a firm's within-industry routine-task labor share (RShare) and its systematic risk, expected returns, and the response of investment and employment to aggregate shocks.",
27 + "strategy": "Within-industry portfolio sorts on RShare (single and double sorts), unconditional and conditional CAPM betas, cash-flow and GDP betas, and panel regressions with controls. Mechanism tests interact RShare with aggregate shocks (GDP growth, recessions) for machine investment and routine employment.",
28 + "key_assumptions": [
29 + "RShare proxies for the presence of labor-technology switching options",
30 + "Within-industry comparisons absorb industry-level confounds",
31 + "Betas and future returns proxy for systematic risk and expected returns"
32 + ],
33 + "threats_addressed": [
34 + "operating leverage as an alternative explanation",
35 + "known firm characteristics that predict risk (via double sorts and panel controls)",
36 + "asymmetry between recessions and expansions (negative versus positive shocks)"
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": 4
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": "preview_of_approach"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "mechanism_or_theory_sketch"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_results"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "identification_defense"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "contribution_statement"
96 + },
97 + {
98 + "paragraph_index": 3,
99 + "move": "related_literature_positioning"
100 + },
101 + {
102 + "paragraph_index": 4,
103 + "move": "roadmap"
104 + }
105 + ],
106 + "contribution_paragraph_index": 3,
107 + "contribution_types": [
108 + "new_theory_or_model",
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": "data"
127 + },
128 + {
129 + "idx": 3,
130 + "role": "results"
131 + },
132 + {
133 + "idx": 4,
134 + "role": "conclusion"
135 + },
136 + {
137 + "idx": 5,
138 + "role": "appendix"
139 + },
140 + {
141 + "idx": 6,
142 + "role": "appendix"
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 + "alternative_measure",
154 + "alternative_specification",
155 + "subsample"
156 + ]
157 + },
158 + "tables": {
159 + "convention_notes": "Notes define sorting procedures (end-of-June, within Fama-French 17 industries, value-weighted quintiles), define variables, and state sample periods. Return units are given in some notes (e.g., monthly excess returns, annual returns in percentages). Standard-error treatment is not visible in the excerpts.",
160 + "units_stated_in_notes": true,
161 + "standard_errors_description": "unclear"
162 + },
163 + "figures": {
164 + "uncertainty_shown": "unclear"
165 + },
166 + "conclusion": {
167 + "bounded_claims": false,
168 + "policy_implications": false,
169 + "open_questions": false,
170 + "restates_contribution": true
171 + },
172 + "hedging": {
173 + "level": "moderate",
174 + "typical_devices": [
175 + "'it seems that'",
176 + "'suggesting that'",
177 + "'do seem to be able to'",
178 + "'to the best of my knowledge'",
179 + "'likely to interrupt'"
180 + ]
181 + },
182 + "claim_evidence": [
183 + {
184 + "claim": "Betas of high-RShare portfolios are over 20% lower than those of low-RShare portfolios, with a high-minus-low return spread of about -3.9% per year.",
185 + "evidence_type": "table",
186 + "evidence_ref": "Tables III, IV"
187 + },
188 + {
189 + "claim": "High-RShare firms have lower exposure to cash-flow news and to GDP growth.",
190 + "evidence_type": "table",
191 + "evidence_ref": "Table V"
192 + },
193 + {
194 + "claim": "RShare predicts lower conditional betas and returns after controlling for firm characteristics.",
195 + "evidence_type": "table",
196 + "evidence_ref": "Tables VI, VII"
197 + },
198 + {
199 + "claim": "In downturns, high-RShare firms invest more in machines and cut more routine-task labor than industry peers.",
200 + "evidence_type": "table",
201 + "evidence_ref": "Tables VIII, IX; Figure 2"
202 + },
203 + {
204 + "claim": "High-RShare firms cut costs more and lose less value during recessions, but not during expansions.",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table X"
207 + },
208 + {
209 + "claim": "The switching option is exercised when productivity falls below a threshold, which lowers firm risk.",
210 + "evidence_type": "equation_or_proof",
211 + "evidence_ref": "Section I; Appendix A"
212 + },
213 + {
214 + "claim": "Routine-task job loss is concentrated in recessions.",
215 + "evidence_type": "prior_literature",
216 + "evidence_ref": "Jaimovich and Siu (2014); Figure 1"
217 + }
218 + ],
219 + "structural_summary": "The paper opens with labor-economics evidence that routine jobs are automated mainly in recessions. It frames labor-technology substitution as a state-contingent investment option distinct from growth options. A simple real-options switching model then predicts lower risk for firms with routine-task labor. Next, an establishment-level measure of routine-labor share is built from occupational data. The empirical section proceeds in three steps: characteristics, then asset-pricing tests via sorts, betas and panel regressions, then mechanism tests on investment, employment and performance around aggregate shocks. It also separates the mechanism from an operating-leverage explanation. The conclusion restates the insight, and proofs and data details are placed in appendices.",
220 + "writing_profile_keywords": [
221 + "model-guided empirics",
222 + "real options",
223 + "portfolio sorts",
224 + "mechanism tests",
225 + "new measure",
226 + "first-person singular",
227 + "cross-sectional asset pricing"
228 + ],
229 + "confidence": 0.8,
230 + "notes": "Single-author paper. The significance reporting format and standard-error treatment are not visible in the excerpts. The causal claims concern the theoretical mechanism; the empirical tests are framed as consistent with the model."
231 + },
232 + "deterministic_checks": {
233 + "roadmap_regex": true,
234 + "contribution_regex_in_intro": true,
235 + "magnitude_in_abstract": false,
236 + "abstract_n_sentences": 5,
237 + "intro_n_paragraphs": 5,
238 + "intro_n_words": 744,
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": "theory",
247 + "2": "other",
248 + "3": "results",
249 + "4": "conclusion",
250 + "5": "appendix",
251 + "6": "appendix"
252 + }
253 + },
254 + "schema_valid": true,
255 + "code_version": "87c73cd-dirty"
256 +}
\ No newline at end of file
added data/annotations/documents/jof_12768.json +232 −0
@@ -0,0 +1,232 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12768",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:30+00:00",
9 + "source_sha256": "1d24c8f974b29cb6f2b1d51ef4a7478f16e44a772cee214e4bc82339b053cc3b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21201,
12 + "usage": {
13 + "input": 6629,
14 + "output": 1881,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06493,
19 + "request_id": "req_011CggqSHKnsFwpsf8BVeY35",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "market_microstructure; experimental finance",
24 + "primary_method": "survey_or_experiment",
25 + "identification": {
26 + "estimand": "Treatment effects of exogenous information-network structure (none, complete, circle, core-periphery) on signal acquisition, volume, liquidity, information aggregation, price informativeness and profits",
27 + "strategy": "Between-session laboratory treatments that randomly vary the information network, analyzed with ordered probit and OLS regressions on treatment dummies, using standard errors clustered at subject level",
28 + "key_assumptions": [
29 + "Network structure is exogenous and assigned by design",
30 + "Information sharing is automatic, perfect and nonstrategic",
31 + "Subjects in different sessions are comparable"
32 + ],
33 + "threats_addressed": [
34 + "Endogenous network formation (removed by imposing exogenous networks)",
35 + "Within-subject dependence (handled by clustering)"
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": 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": "literature_gap"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "preview_of_approach"
79 + },
80 + {
81 + "paragraph_index": 2,
82 + "move": "mechanism_or_theory_sketch"
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": "preview_of_results"
95 + },
96 + {
97 + "paragraph_index": 4,
98 + "move": "tension_or_puzzle"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "related_literature_positioning"
103 + }
104 + ],
105 + "contribution_paragraph_index": 1,
106 + "contribution_types": [
107 + "new_evidence"
108 + ],
109 + "roadmap_present": false,
110 + "literature_positioning_location": "inside_introduction"
111 + },
112 + "section_roles": [
113 + {
114 + "idx": 0,
115 + "role": "introduction"
116 + },
117 + {
118 + "idx": 1,
119 + "role": "empirical_design"
120 + },
121 + {
122 + "idx": 2,
123 + "role": "results"
124 + },
125 + {
126 + "idx": 3,
127 + "role": "other"
128 + },
129 + {
130 + "idx": 4,
131 + "role": "conclusion"
132 + }
133 + ],
134 + "results_presentation": {
135 + "economic_significance_discussed": true,
136 + "statistical_significance_language": "stars_and_tstats",
137 + "magnitude_statements_in_intro": true
138 + },
139 + "robustness": {
140 + "location": "none_seen",
141 + "types": []
142 + },
143 + "tables": {
144 + "convention_notes": "Tables pair period-average descriptive statistics, reported with standard deviations, with regression tables. The regressions use treatment dummies against a nonnetworked baseline and include a period control. Standard errors are clustered by subject and reported in parentheses, with stars marking significance. Notes define each dependent variable precisely.",
145 + "units_stated_in_notes": true,
146 + "standard_errors_description": "clustered"
147 + },
148 + "figures": {
149 + "uncertainty_shown": "mixed"
150 + },
151 + "conclusion": {
152 + "bounded_claims": true,
153 + "policy_implications": false,
154 + "open_questions": true,
155 + "restates_contribution": true
156 + },
157 + "hedging": {
158 + "level": "moderate",
159 + "typical_devices": [
160 + "on average",
161 + "typically",
162 + "explicit design simplifications acknowledged",
163 + "might not be as strong"
164 + ]
165 + },
166 + "claim_evidence": [
167 + {
168 + "claim": "Signal acquisition probability and quantity fall as a trader's number of network connections rises.",
169 + "evidence_type": "table",
170 + "evidence_ref": "Tables II, III, IV; Figure 2"
171 + },
172 + {
173 + "claim": "Information sharing raises trading volume and improves liquidity.",
174 + "evidence_type": "table",
175 + "evidence_ref": "Table V"
176 + },
177 + {
178 + "claim": "Prices aggregate more of the available information with sharing, and aggregation rises with network density.",
179 + "evidence_type": "table",
180 + "evidence_ref": "Table VI; Figures 3-4"
181 + },
182 + {
183 + "claim": "Price informativeness about the true state does not improve with social communication.",
184 + "evidence_type": "table",
185 + "evidence_ref": "Table IX; Figures 5-6"
186 + },
187 + {
188 + "claim": "Net earnings are higher under sharing because of savings on signal costs.",
189 + "evidence_type": "table",
190 + "evidence_ref": "Tables X, XIII"
191 + },
192 + {
193 + "claim": "Prices support semi-strong but not strong-form efficiency.",
194 + "evidence_type": "prior_literature",
195 + "evidence_ref": "Discussion; Page and Siemroth (2018)"
196 + }
197 + ],
198 + "structural_summary": "The introduction opens with the rational-expectations literature on costly information and flags the absence of social networks as an open question. It then describes the exogenous-network setup and frames a trade-off between richer information sets and free-riding. Results are previewed sequentially: crowding out of acquisition, then liquidity, aggregation, informativeness and earnings, with a counter-to-conventional-wisdom point on forecasting quality. A design section presents the lab procedures and the four network treatments. A long results section is organized by outcome, using regressions on treatment dummies. A discussion section links the findings to market efficiency, and a short conclusion restates the findings and lists extensions.",
199 + "writing_profile_keywords": [
200 + "laboratory experiment",
201 + "information networks",
202 + "free-riding",
203 + "treatment comparisons",
204 + "price efficiency",
205 + "outcome-by-outcome results"
206 + ],
207 + "confidence": 0.82,
208 + "notes": "Experimental asset-market paper with no equations shown. No explicit robustness section is visible in the excerpts. An Internet Appendix exists, but its contents are not shown."
209 + },
210 + "deterministic_checks": {
211 + "roadmap_regex": false,
212 + "contribution_regex_in_intro": false,
213 + "magnitude_in_abstract": false,
214 + "abstract_n_sentences": 4,
215 + "intro_n_paragraphs": 5,
216 + "intro_n_words": 792,
217 + "causal_words_in_abstract": 0,
218 + "hedge_words_in_abstract": 0,
219 + "stars_mentioned": false,
220 + "clustered_se_mentioned": true,
221 + "internet_appendix_mentioned": false,
222 + "section_roles_rule": {
223 + "0": "introduction",
224 + "1": "other",
225 + "2": "results",
226 + "3": "conclusion",
227 + "4": "conclusion"
228 + }
229 + },
230 + "schema_valid": true,
231 + "code_version": "87c73cd-dirty"
232 +}
\ No newline at end of file
added data/annotations/documents/jof_12769.json +266 −0
@@ -0,0 +1,266 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped abstract_moves.preview"
5 + ],
6 + "doc_id": "jof_12769",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:28+00:00",
12 + "source_sha256": "47249e70cbbd487c239415949c55cc9790d649e003c8025e78a50f338b232fe5",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 24050,
15 + "usage": {
16 + "input": 8092,
17 + "output": 2179,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.076742,
22 + "request_id": "req_011CggqPYdiPkQ99kXXDFk76",
23 + "annotation": {
24 + "genre": "empirical",
25 + "subfield": "market_microstructure",
26 + "subfield_secondary": "high-frequency trading and price discovery",
27 + "primary_method": "other",
28 + "identification": {
29 + "estimand": "Contribution of market vs. limit orders by HFTs and non-HFTs to efficient price variance (price discovery), and how this varies with volatility",
30 + "strategy": "Event-time VAR (Hasbrouck-style, extended by Fleming-Mizrach-Nguyen) on regulatory message data with participant identifiers; impulse responses, variance decompositions, information shares; pooled OLS of daily shares on volatility",
31 + "key_assumptions": [
32 + "Causal ordering in VAR given by event time sequence",
33 + "HFT classification of participant IDs is accurate",
34 + "Small cross-exchange reporting time differences do not drive results"
35 + ],
36 + "threats_addressed": [
37 + "Cross-exchange timestamp differences (one-second calendar-time information shares)",
38 + "Public information explaining HFT price impact",
39 + "Front-running of non-HFT orders"
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": 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": "preview_of_approach"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "literature_gap"
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": "tension_or_puzzle"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "related_literature_positioning"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "mechanism_or_theory_sketch"
103 + },
104 + {
105 + "paragraph_index": 5,
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": "policy_or_broader_implication"
115 + },
116 + {
117 + "paragraph_index": 8,
118 + "move": "related_literature_positioning"
119 + },
120 + {
121 + "paragraph_index": 9,
122 + "move": "preview_of_results"
123 + },
124 + {
125 + "paragraph_index": 10,
126 + "move": "roadmap"
127 + }
128 + ],
129 + "contribution_paragraph_index": 1,
130 + "contribution_types": [
131 + "new_evidence"
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": "results"
148 + },
149 + {
150 + "idx": 3,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 4,
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": "within_results",
165 + "types": [
166 + "alternative_specification",
167 + "alternative_measure"
168 + ]
169 + },
170 + "tables": {
171 + "convention_notes": "Table notes repeat sample (15 non-cross-listed TSX 60 stocks, Oct 2012–Jun 2013), report stock-day averages of VAR IRFs, variance decompositions and OLS coefficients, and define variables. Price impacts are given in basis points and market cap in billions of CAD.",
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": true,
180 + "policy_implications": true,
181 + "open_questions": true,
182 + "restates_contribution": true
183 + },
184 + "hedging": {
185 + "level": "moderate",
186 + "typical_devices": [
187 + "consistent with",
188 + "suggests",
189 + "lend some support",
190 + "does not imply",
191 + "seemingly at odds"
192 + ]
193 + },
194 + "claim_evidence": [
195 + {
196 + "claim": "HFT limit orders contribute over twice as much to price discovery as HFT market orders; for non-HFTs, market orders contribute more than limit orders",
197 + "evidence_type": "table",
198 + "evidence_ref": "Tables IV, VII"
199 + },
200 + {
201 + "claim": "More aggressive orders have higher individual price impact, but limit orders dominate price discovery because they are far more numerous",
202 + "evidence_type": "table",
203 + "evidence_ref": "Tables III, IV, VI"
204 + },
205 + {
206 + "claim": "HFT limit orders account for about 30% of price discovery vs about 15% for non-HFT limit orders",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table VII"
209 + },
210 + {
211 + "claim": "Limit order submissions and their price discovery share fall with volatility, driven by HFTs",
212 + "evidence_type": "table",
213 + "evidence_ref": "Tables VIII, IX"
214 + },
215 + {
216 + "claim": "HFTs dominate price discovery via limit orders on each exchange, even at one-second sampling",
217 + "evidence_type": "table",
218 + "evidence_ref": "Tables X, XI"
219 + },
220 + {
221 + "claim": "HFTs react more than non-HFTs to events on other exchanges, consistent with cross-market information integration",
222 + "evidence_type": "table",
223 + "evidence_ref": "Tables XII, XIII"
224 + },
225 + {
226 + "claim": "Positive limit order price impact is reconciled with prior negative impact of HFT liquidity-supplying trades through nonexecution",
227 + "evidence_type": "prior_literature",
228 + "evidence_ref": "Brogaard, Hendershott, and Riordan (2014)"
229 + }
230 + ],
231 + "structural_summary": "The introduction contrasts the traditional trades-versus-quotes view with modern limit order books and proposes a VAR decomposition of price discovery by order type and trader type. It previews the headline results and reconciles them with prior trade-based evidence. It then links the findings to GPR and Hoffmann models, using volatility to discriminate between them, and discusses stability, adverse selection and front-running concerns. A data section is followed by two results sections, market-wide and then within/across exchanges, all presented through tables. The conclusion restates the findings, discusses regulatory trade-offs, and flags future research.",
232 + "writing_profile_keywords": [
233 + "market microstructure",
234 + "high-frequency traders",
235 + "limit orders",
236 + "VAR price discovery",
237 + "regulatory message data",
238 + "theory-consistent interpretation",
239 + "table-heavy"
240 + ],
241 + "confidence": 0.8,
242 + "notes": "The primary method is a Hasbrouck-style VAR, coded as 'other'. Standard error and significance conventions are not visible in the excerpts. The excerpts contain no figures and no equations."
243 + },
244 + "deterministic_checks": {
245 + "roadmap_regex": true,
246 + "contribution_regex_in_intro": true,
247 + "magnitude_in_abstract": false,
248 + "abstract_n_sentences": 6,
249 + "intro_n_paragraphs": 11,
250 + "intro_n_words": 1643,
251 + "causal_words_in_abstract": 0,
252 + "hedge_words_in_abstract": 1,
253 + "stars_mentioned": false,
254 + "clustered_se_mentioned": false,
255 + "internet_appendix_mentioned": false,
256 + "section_roles_rule": {
257 + "0": "introduction",
258 + "1": "data",
259 + "2": "other",
260 + "3": "other",
261 + "4": "conclusion"
262 + }
263 + },
264 + "schema_valid": true,
265 + "code_version": "87c73cd-dirty"
266 +}
\ No newline at end of file
added data/annotations/documents/jof_12770.json +278 −0
@@ -0,0 +1,278 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12770",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:32+00:00",
9 + "source_sha256": "5f8166298a5c8b07d6d251b6f99a2f317bc1249a81f7ed3f2815adb50f79bfca",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22979,
12 + "usage": {
13 + "input": 8199,
14 + "output": 1961,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07281,
19 + "request_id": "req_011CggqSZshvXubj6wdBKhxF",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "executive compensation / dynamic contracting",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; continuous-time principal-agent model solved via stochastic maximum principle and HJB, supplemented by numerical solutions",
28 + "key_assumptions": [
29 + "CARA risk-averse CEO with private savings",
30 + "hidden effort and manipulation with persistent effect on performance",
31 + "finite tenure T with postretirement window tau"
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": 0
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": 5
61 + }
62 + },
63 + "introduction": {
64 + "moves": [
65 + {
66 + "paragraph_index": 0,
67 + "move": "motivation"
68 + },
69 + {
70 + "paragraph_index": 1,
71 + "move": "literature_gap"
72 + },
73 + {
74 + "paragraph_index": 1,
75 + "move": "preview_of_approach"
76 + },
77 + {
78 + "paragraph_index": 2,
79 + "move": "preview_of_results"
80 + },
81 + {
82 + "paragraph_index": 2,
83 + "move": "mechanism_or_theory_sketch"
84 + },
85 + {
86 + "paragraph_index": 3,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "contribution_statement"
92 + },
93 + {
94 + "paragraph_index": 4,
95 + "move": "preview_of_results"
96 + },
97 + {
98 + "paragraph_index": 5,
99 + "move": "related_literature_positioning"
100 + },
101 + {
102 + "paragraph_index": 6,
103 + "move": "related_literature_positioning"
104 + },
105 + {
106 + "paragraph_index": 7,
107 + "move": "roadmap"
108 + }
109 + ],
110 + "contribution_paragraph_index": 3,
111 + "contribution_types": [
112 + "new_theory_or_model"
113 + ],
114 + "roadmap_present": true,
115 + "literature_positioning_location": "inside_introduction"
116 + },
117 + "section_roles": [
118 + {
119 + "idx": 0,
120 + "role": "introduction"
121 + },
122 + {
123 + "idx": 1,
124 + "role": "theory"
125 + },
126 + {
127 + "idx": 2,
128 + "role": "theory"
129 + },
130 + {
131 + "idx": 3,
132 + "role": "theory"
133 + },
134 + {
135 + "idx": 4,
136 + "role": "theory"
137 + },
138 + {
139 + "idx": 5,
140 + "role": "theory"
141 + },
142 + {
143 + "idx": 6,
144 + "role": "theory"
145 + },
146 + {
147 + "idx": 7,
148 + "role": "theory"
149 + },
150 + {
151 + "idx": 8,
152 + "role": "other"
153 + },
154 + {
155 + "idx": 9,
156 + "role": "conclusion"
157 + },
158 + {
159 + "idx": 10,
160 + "role": "appendix"
161 + },
162 + {
163 + "idx": 11,
164 + "role": "appendix"
165 + },
166 + {
167 + "idx": 12,
168 + "role": "appendix"
169 + }
170 + ],
171 + "results_presentation": {
172 + "economic_significance_discussed": false,
173 + "statistical_significance_language": "unclear",
174 + "magnitude_statements_in_intro": false
175 + },
176 + "robustness": {
177 + "location": "none_seen",
178 + "types": []
179 + },
180 + "tables": {
181 + "convention_notes": "No tables. Figures are numerical solutions, and each caption lists the parameter values used.",
182 + "units_stated_in_notes": false,
183 + "standard_errors_description": "none"
184 + },
185 + "figures": {
186 + "uncertainty_shown": "no"
187 + },
188 + "conclusion": {
189 + "bounded_claims": true,
190 + "policy_implications": false,
191 + "open_questions": false,
192 + "restates_contribution": true
193 + },
194 + "hedging": {
195 + "level": "moderate",
196 + "typical_devices": [
197 + "'may' for nonmonotonic effort",
198 + "'generally' qualifying results",
199 + "'might not' regarding firms' choices",
200 + "noting limits of postretirement compensation"
201 + ]
202 + },
203 + "claim_evidence": [
204 + {
205 + "claim": "Optimal contracts defer compensation, and vesting accelerates near the end of tenure.",
206 + "evidence_type": "equation_or_proof",
207 + "evidence_ref": "Sections V-VI, Appendix C"
208 + },
209 + {
210 + "claim": "Manipulation escalates over the CEO's tenure, producing an endogenous horizon problem.",
211 + "evidence_type": "figure",
212 + "evidence_ref": "Figure 3"
213 + },
214 + {
215 + "claim": "Without manipulation, vesting is deterministic; with manipulation, vesting is performance-based.",
216 + "evidence_type": "equation_or_proof",
217 + "evidence_ref": "Sections IV, VI"
218 + },
219 + {
220 + "claim": "Short-termism is prevalent: most surveyed CEOs would sacrifice long-term value to meet expectations.",
221 + "evidence_type": "prior_literature",
222 + "evidence_ref": "Graham, Harvey, and Rajgopal (2005)"
223 + },
224 + {
225 + "claim": "Under deterministic contracts, long-term incentives can be reduced only by distorting effort.",
226 + "evidence_type": "figure",
227 + "evidence_ref": "Section VII, Figure 4"
228 + },
229 + {
230 + "claim": "Incentive compatibility conditions are necessary and sufficient.",
231 + "evidence_type": "equation_or_proof",
232 + "evidence_ref": "Proposition 1, Appendices A-B"
233 + }
234 + ],
235 + "structural_summary": "The introduction motivates the paper with survey and empirical evidence of CEO short-termism. It then identifies a gap: prior models either take incentives as given or study contracts that rule out manipulation entirely. The paper builds a continuous-time dynamic agency model with hidden effort and manipulation. The analysis proceeds in steps: the agent's incentive-compatibility conditions, the principal's control problem, an infinite-tenure benchmark, finite tenure solved by backward induction, the general optimal contract, and a deterministic-contract subclass. Numerical examples illustrate the results and link them to empirical implications. The conclusion restates the findings, and proofs are placed in the appendices.",
236 + "writing_profile_keywords": [
237 + "dynamic contracting",
238 + "continuous time",
239 + "CEO horizon",
240 + "short-termism",
241 + "vesting",
242 + "numerical illustration",
243 + "proofs in appendix"
244 + ],
245 + "confidence": 0.85,
246 + "notes": "Theory paper. Section VIII presents numerical comparative statics and relates them to existing evidence; it contains no estimation of its own. No tables are present."
247 + },
248 + "deterministic_checks": {
249 + "roadmap_regex": true,
250 + "contribution_regex_in_intro": false,
251 + "magnitude_in_abstract": false,
252 + "abstract_n_sentences": 6,
253 + "intro_n_paragraphs": 8,
254 + "intro_n_words": 1265,
255 + "causal_words_in_abstract": 0,
256 + "hedge_words_in_abstract": 0,
257 + "stars_mentioned": false,
258 + "clustered_se_mentioned": false,
259 + "internet_appendix_mentioned": false,
260 + "section_roles_rule": {
261 + "0": "introduction",
262 + "1": "theory",
263 + "2": "other",
264 + "3": "other",
265 + "4": "other",
266 + "5": "other",
267 + "6": "other",
268 + "7": "other",
269 + "8": "theory",
270 + "9": "conclusion",
271 + "10": "appendix",
272 + "11": "appendix",
273 + "12": "appendix"
274 + }
275 + },
276 + "schema_valid": true,
277 + "code_version": "87c73cd-dirty"
278 +}
\ No newline at end of file
added data/annotations/documents/jof_12771.json +400 −0
@@ -0,0 +1,400 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12771",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:00+00:00",
9 + "source_sha256": "bcecd3acb83f5a65eb917a96fa8b4e4ffc11fd738e41c9290c0a7512d9f4f08a",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 28749,
12 + "usage": {
13 + "input": 10062,
14 + "output": 2815,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.097342,
19 + "request_id": "req_011CggqPjU63RJeaRiqxi2n4",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance (misallocation, real effects of mispricing, lumpy investment)",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Aggregate real efficiency (value) losses from investment distortions attributable to cross-sectional alpha wedges, measured as a counterfactual gain from eliminating alphas",
27 + "strategy": "Estimate a dynamic lumpy-investment model by GMM on cross-sectional moments, with subjective-belief wedges disciplined by empirical alphas, then compare it with a counterfactual economy without alphas",
28 + "key_assumptions": [
29 + "Alphas from a postulated benchmark asset pricing model measure informational inefficiencies (joint-hypothesis caveat)",
30 + "Firms maximize market value under prevailing distorted subjective beliefs",
31 + "Model structure (stochastic lumpy adjustment, alpha process dynamics) is correctly specified",
32 + "Macroeconomic parameters calibrated from the literature"
33 + ],
34 + "threats_addressed": [
35 + "joint-hypothesis problem / alternative view that alphas reflect omitted risk factors",
36 + "general equilibrium effects",
37 + "independence and persistence of the alpha process",
38 + "role of adjustment frictions",
39 + "cash-raising channel for the asset growth–alpha relation"
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": 3
68 + }
69 + },
70 + "introduction": {
71 + "moves": [
72 + {
73 + "paragraph_index": 0,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "tension_or_puzzle"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "preview_of_results"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "related_literature_positioning"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "preview_of_approach"
91 + },
92 + {
93 + "paragraph_index": 1,
94 + "move": "identification_defense"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "motivation"
99 + },
100 + {
101 + "paragraph_index": 2,
102 + "move": "robustness_mention"
103 + },
104 + {
105 + "paragraph_index": 3,
106 + "move": "preview_of_approach"
107 + },
108 + {
109 + "paragraph_index": 3,
110 + "move": "preview_of_results"
111 + },
112 + {
113 + "paragraph_index": 4,
114 + "move": "contribution_statement"
115 + },
116 + {
117 + "paragraph_index": 4,
118 + "move": "related_literature_positioning"
119 + },
120 + {
121 + "paragraph_index": 5,
122 + "move": "mechanism_or_theory_sketch"
123 + },
124 + {
125 + "paragraph_index": 5,
126 + "move": "related_literature_positioning"
127 + },
128 + {
129 + "paragraph_index": 6,
130 + "move": "roadmap"
131 + }
132 + ],
133 + "contribution_paragraph_index": 4,
134 + "contribution_types": [
135 + "new_theory_or_model",
136 + "new_method",
137 + "new_evidence"
138 + ],
139 + "roadmap_present": true,
140 + "literature_positioning_location": "inside_introduction"
141 + },
142 + "section_roles": [
143 + {
144 + "idx": 0,
145 + "role": "introduction"
146 + },
147 + {
148 + "idx": 1,
149 + "role": "theory"
150 + },
151 + {
152 + "idx": 2,
153 + "role": "results"
154 + },
155 + {
156 + "idx": 3,
157 + "role": "theory"
158 + },
159 + {
160 + "idx": 4,
161 + "role": "empirical_design"
162 + },
163 + {
164 + "idx": 5,
165 + "role": "results"
166 + },
167 + {
168 + "idx": 6,
169 + "role": "results"
170 + },
171 + {
172 + "idx": 7,
173 + "role": "other"
174 + },
175 + {
176 + "idx": 8,
177 + "role": "robustness"
178 + },
179 + {
180 + "idx": 9,
181 + "role": "other"
182 + },
183 + {
184 + "idx": 10,
185 + "role": "conclusion"
186 + },
187 + {
188 + "idx": 11,
189 + "role": "appendix"
190 + },
191 + {
192 + "idx": 12,
193 + "role": "appendix"
194 + },
195 + {
196 + "idx": 13,
197 + "role": "appendix"
198 + },
199 + {
200 + "idx": 14,
201 + "role": "appendix"
202 + },
203 + {
204 + "idx": 15,
205 + "role": "appendix"
206 + },
207 + {
208 + "idx": 16,
209 + "role": "appendix"
210 + },
211 + {
212 + "idx": 17,
213 + "role": "appendix"
214 + },
215 + {
216 + "idx": 18,
217 + "role": "appendix"
218 + },
219 + {
220 + "idx": 19,
221 + "role": "appendix"
222 + },
223 + {
224 + "idx": 20,
225 + "role": "appendix"
226 + },
227 + {
228 + "idx": 21,
229 + "role": "appendix"
230 + },
231 + {
232 + "idx": 22,
233 + "role": "appendix"
234 + },
235 + {
236 + "idx": 23,
237 + "role": "appendix"
238 + },
239 + {
240 + "idx": 24,
241 + "role": "appendix"
242 + },
243 + {
244 + "idx": 25,
245 + "role": "appendix"
246 + },
247 + {
248 + "idx": 26,
249 + "role": "appendix"
250 + },
251 + {
252 + "idx": 27,
253 + "role": "appendix"
254 + },
255 + {
256 + "idx": 28,
257 + "role": "appendix"
258 + }
259 + ],
260 + "results_presentation": {
261 + "economic_significance_discussed": true,
262 + "statistical_significance_language": "confidence_intervals",
263 + "magnitude_statements_in_intro": true
264 + },
265 + "robustness": {
266 + "location": "mixed",
267 + "types": [
268 + "sensitivity_to_parameters",
269 + "alternative_specification",
270 + "other"
271 + ]
272 + },
273 + "tables": {
274 + "convention_notes": "Only two tables are shown. The first reports decile portfolio returns and CAPM alphas for the value and investment anomalies. The second lists calibrated macro parameters and GMM-estimated firm parameters, with normalization notes. Most quantitative results are delivered through figures.",
275 + "units_stated_in_notes": true,
276 + "standard_errors_description": "unclear"
277 + },
278 + "figures": {
279 + "uncertainty_shown": "mixed"
280 + },
281 + "conclusion": {
282 + "bounded_claims": true,
283 + "policy_implications": true,
284 + "open_questions": true,
285 + "restates_contribution": true
286 + },
287 + "hedging": {
288 + "level": "moderate",
289 + "typical_devices": [
290 + "'can'/'may' modal verbs",
291 + "'raises the possibility'",
292 + "conditional framing ('if alphas reflect inefficiencies', 'conditional on the view')",
293 + "'potential' gains",
294 + "'plausibly'",
295 + "explicit disclaimer of being a first step rather than the final word"
296 + ]
297 + },
298 + "claim_evidence": [
299 + {
300 + "claim": "Value and investment anomalies replicate over 1975–2014 consistent with prior literature",
301 + "evidence_type": "table",
302 + "evidence_ref": "Table I"
303 + },
304 + {
305 + "claim": "Estimated model matches targeted moments on size, Tobin's q, value premium and investment",
306 + "evidence_type": "figure",
307 + "evidence_ref": "Figure 1"
308 + },
309 + {
310 + "claim": "Model replicates untargeted investment-alpha relation",
311 + "evidence_type": "figure",
312 + "evidence_ref": "Figure 5"
313 + },
314 + {
315 + "claim": "Aggregate efficiency gains from eliminating alphas can be substantial and rise with the value spread",
316 + "evidence_type": "figure",
317 + "evidence_ref": "Figure 3"
318 + },
319 + {
320 + "claim": "Gains are concentrated among firms with particular Tobin's q levels",
321 + "evidence_type": "figure",
322 + "evidence_ref": "Figure 4"
323 + },
324 + {
325 + "claim": "Alpha persistence strongly affects aggregate gains, so alpha magnitude alone is a poor indicator",
326 + "evidence_type": "figure",
327 + "evidence_ref": "Figure 6"
328 + },
329 + {
330 + "claim": "Cash-raising channel does not explain the asset growth–alpha relation",
331 + "evidence_type": "none_stated",
332 + "evidence_ref": "Internet Appendix (referenced in intro)"
333 + },
334 + {
335 + "claim": "Alphas alone are uninformative about real distortions",
336 + "evidence_type": "equation_or_proof",
337 + "evidence_ref": "Section I one-period example"
338 + }
339 + ],
340 + "structural_summary": "The paper opens with the anomalies literature and asks how large the real losses are if alphas reflect inefficiency. It positions itself against misallocation work in macro and acknowledges the joint-hypothesis issue. A one-period example motivates the argument, followed by reduced-form replication of the value and investment anomalies. The core is a continuous-time lumpy-investment model with closed-form distributions, estimated by GMM. Counterfactuals then remove the alpha wedges to quantify efficiency gains. Later sections give an efficient-markets reinterpretation, robustness discussions of modeling assumptions, and implications for ranking anomalies. The conclusion is hedged and lists extensions.",
341 + "writing_profile_keywords": [
342 + "structural model",
343 + "counterfactual",
344 + "GMM",
345 + "lumpy investment",
346 + "joint-hypothesis caveat",
347 + "misallocation",
348 + "figure-heavy results",
349 + "hedged quantification"
350 + ],
351 + "confidence": 0.8,
352 + "notes": "Section idx 6 ('prd') is a parsing artifact that holds the continuation of the counterfactual analysis. Several appendix headings are equation fragments. The roadmap refers to 'Section X' for the conclusion, but the conclusion is numbered IX. Model-fit figures show 95% bounds, while the counterfactual figures do not appear to show uncertainty."
353 + },
354 + "deterministic_checks": {
355 + "roadmap_regex": true,
356 + "contribution_regex_in_intro": true,
357 + "magnitude_in_abstract": false,
358 + "abstract_n_sentences": 4,
359 + "intro_n_paragraphs": 7,
360 + "intro_n_words": 1140,
361 + "causal_words_in_abstract": 1,
362 + "hedge_words_in_abstract": 1,
363 + "stars_mentioned": false,
364 + "clustered_se_mentioned": false,
365 + "internet_appendix_mentioned": true,
366 + "section_roles_rule": {
367 + "0": "introduction",
368 + "1": "other",
369 + "2": "results",
370 + "3": "theory",
371 + "4": "theory",
372 + "5": "other",
373 + "6": "other",
374 + "7": "theory",
375 + "8": "robustness",
376 + "9": "other",
377 + "10": "conclusion",
378 + "11": "appendix",
379 + "12": "appendix",
380 + "13": "appendix",
381 + "14": "appendix",
382 + "15": "appendix",
383 + "16": "appendix",
384 + "17": "appendix",
385 + "18": "appendix",
386 + "19": "appendix",
387 + "20": "appendix",
388 + "21": "appendix",
389 + "22": "appendix",
390 + "23": "appendix",
391 + "24": "appendix",
392 + "25": "appendix",
393 + "26": "appendix",
394 + "27": "appendix",
395 + "28": "appendix"
396 + }
397 + },
398 + "schema_valid": true,
399 + "code_version": "87c73cd-dirty"
400 +}
\ No newline at end of file
added data/annotations/documents/jof_12772.json +246 −0
@@ -0,0 +1,246 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12772",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:06+00:00",
9 + "source_sha256": "efec61440f4f40e9e99f13299f92dc2600b9a4b5c865566d90210a3e7468452c",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 17403,
12 + "usage": {
13 + "input": 6192,
14 + "output": 2135,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.068262,
19 + "request_id": "req_011CggqQTyCQqugpGngRYxWr",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Cross-sectional risk prices (lambda) on exposures (betas) to multi-quarter capital share growth across portfolio cross-sections",
27 + "strategy": "Two-pass Fama-MacBeth expected return-beta regressions using long-horizon (H-quarter) capital share betas, with horse races against Fama-French and intermediary factors; household-level income-share proxy as an alternative SDF",
28 + "key_assumptions": [
29 + "Linearized SDF in capital share growth (and consumption) is an adequate representation",
30 + "Long-horizon betas capture the relevant risk exposure",
31 + "Mimicking income-share factor based on capital share proxies top-wealth consumption growth"
32 + ],
33 + "threats_addressed": [
34 + "Explanatory power could reflect aggregate consumption risk rather than capital share risk",
35 + "Overlap with existing pricing factors (SMB, HML, intermediary factors)",
36 + "Estimation uncertainty in two-pass regressions, handled via bootstrapped confidence intervals"
37 + ],
38 + "causal_language_strength": 2,
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": false,
52 + "sentence_index": -1
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": "preview_of_results"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "tension_or_puzzle"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "mechanism_or_theory_sketch"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "preview_of_approach"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "related_literature_positioning"
96 + },
97 + {
98 + "paragraph_index": 4,
99 + "move": "related_literature_positioning"
100 + },
101 + {
102 + "paragraph_index": 5,
103 + "move": "other"
104 + },
105 + {
106 + "paragraph_index": 6,
107 + "move": "roadmap"
108 + }
109 + ],
110 + "contribution_paragraph_index": 2,
111 + "contribution_types": [
112 + "new_evidence"
113 + ],
114 + "roadmap_present": true,
115 + "literature_positioning_location": "inside_introduction"
116 + },
117 + "section_roles": [
118 + {
119 + "idx": 0,
120 + "role": "introduction"
121 + },
122 + {
123 + "idx": 1,
124 + "role": "data"
125 + },
126 + {
127 + "idx": 2,
128 + "role": "empirical_design"
129 + },
130 + {
131 + "idx": 3,
132 + "role": "results"
133 + },
134 + {
135 + "idx": 4,
136 + "role": "conclusion"
137 + }
138 + ],
139 + "results_presentation": {
140 + "economic_significance_discussed": true,
141 + "statistical_significance_language": "mixed",
142 + "magnitude_statements_in_intro": false
143 + },
144 + "robustness": {
145 + "location": "within_results",
146 + "types": [
147 + "alternative_specification",
148 + "alternative_measure",
149 + "subsample"
150 + ]
151 + },
152 + "tables": {
153 + "convention_notes": "Risk-price tables share boilerplate notes: estimates scaled by 100, bootstrapped 95% confidence intervals in square brackets, sample period stated, and the cross-sectional regression equation reproduced. Fit is reported via adjusted R2, RMSE/RMSR and BIC. The preliminary income-share regressions instead use OLS t-statistics with significance stars.",
154 + "units_stated_in_notes": true,
155 + "standard_errors_description": "mixed"
156 + },
157 + "figures": {
158 + "uncertainty_shown": "unclear"
159 + },
160 + "conclusion": {
161 + "bounded_claims": true,
162 + "policy_implications": false,
163 + "open_questions": false,
164 + "restates_contribution": true
165 + },
166 + "hedging": {
167 + "level": "moderate",
168 + "typical_devices": [
169 + "'consistent with' / 'commensurate with' framing for theory links",
170 + "explicit acknowledgment of portfolios the factor fails to price",
171 + "'our results suggest' phrasing"
172 + ]
173 + },
174 + "claim_evidence": [
175 + {
176 + "claim": "Stock wealth is highly concentrated, with the top 5% holding most of the market value",
177 + "evidence_type": "table",
178 + "evidence_ref": "Table I"
179 + },
180 + {
181 + "claim": "Top stock-wealth groups' income shares co-move with the capital share",
182 + "evidence_type": "table",
183 + "evidence_ref": "Table II"
184 + },
185 + {
186 + "claim": "Capital share betas explain average returns across equity characteristic portfolios with high fit",
187 + "evidence_type": "figure",
188 + "evidence_ref": "Figure 1; Table III"
189 + },
190 + {
191 + "claim": "Capital share exposure subsumes SMB and HML in pricing equity portfolios",
192 + "evidence_type": "table",
193 + "evidence_ref": "Table IV"
194 + },
195 + {
196 + "claim": "Capital share risk competes favorably with intermediary factors for equities and other asset classes",
197 + "evidence_type": "table",
198 + "evidence_ref": "Tables V-VI"
199 + },
200 + {
201 + "claim": "A top-10% consumption proxy from household data has similar explanatory power",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table VIII"
204 + },
205 + {
206 + "claim": "The factor fails to price industry, FX, commodity and momentum portfolios",
207 + "evidence_type": "none_stated",
208 + "evidence_ref": "Introduction paragraph 5"
209 + }
210 + ],
211 + "structural_summary": "The introduction opens with the search for an empirically relevant SDF and immediately states the headline finding. It then challenges representative-agent pricing using wealth-concentration facts and sketches the redistributive-shock mechanism. Results are previewed via a figure, the paper is positioned against the limited-participation and redistributive-risk literatures, and the authors openly state the factor's failures before giving a roadmap. The body proceeds through data with preliminary distributional evidence, the econometric SDF/beta setup, and a results section of sequential horse races, beta spreads and a household-data SDF. A short conclusion restates the findings and their theoretical consistency.",
212 + "writing_profile_keywords": [
213 + "single-factor asset pricing",
214 + "capital share",
215 + "Fama-MacBeth",
216 + "bootstrapped confidence intervals",
217 + "horse race against existing factors",
218 + "heterogeneous-agent motivation",
219 + "candid limitations"
220 + ],
221 + "confidence": 0.8,
222 + "notes": "The Econometric Model section is assigned the empirical_design role rather than theory because it sets up the estimation framework. Figure uncertainty is marked unclear because the captions do not mention error bands. The 'other' move in paragraph 5 marks a discussion of the factor's limitations."
223 + },
224 + "deterministic_checks": {
225 + "roadmap_regex": true,
226 + "contribution_regex_in_intro": false,
227 + "magnitude_in_abstract": false,
228 + "abstract_n_sentences": 2,
229 + "intro_n_paragraphs": 7,
230 + "intro_n_words": 956,
231 + "causal_words_in_abstract": 0,
232 + "hedge_words_in_abstract": 0,
233 + "stars_mentioned": true,
234 + "clustered_se_mentioned": false,
235 + "internet_appendix_mentioned": true,
236 + "section_roles_rule": {
237 + "0": "introduction",
238 + "1": "data",
239 + "2": "theory",
240 + "3": "results",
241 + "4": "conclusion"
242 + }
243 + },
244 + "schema_valid": true,
245 + "code_version": "87c73cd-dirty"
246 +}
\ No newline at end of file
added data/annotations/documents/jof_12773.json +278 −0
@@ -0,0 +1,278 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12773",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:03+00:00",
9 + "source_sha256": "e4a89a970d199988ea1b48ac292797ca71f8fed821c3591af2615058cdbae2fb",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 25568,
12 + "usage": {
13 + "input": 8847,
14 + "output": 2097,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.078122,
19 + "request_id": "req_011CggqQGkkh38ZFtxi7NPsr",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "macro_finance",
23 + "subfield_secondary": "corporate_finance (optimal contracting, firm dynamics)",
24 + "primary_method": "calibration_simulation",
25 + "identification": {
26 + "estimand": "Association between industry-level idiosyncratic volatility and the average firm-level capital-income-to-sales ratio; model-implied effect of volatility and rents on aggregate versus average capital share.",
27 + "strategy": "Stylized facts from Compustat/CRSP, a calibrated industry equilibrium model with entry and exit and worker insurance, comparative statics across stationary equilibria, and within-industry regressions of capital share on idiosyncratic volatility.",
28 + "key_assumptions": [
29 + "Shareholders are diversified while workers are risk-averse, so optimal contracts insure workers.",
30 + "Comparisons are between stationary equilibria rather than transition dynamics.",
31 + "Compustat public firms are representative for the mechanism."
32 + ],
33 + "threats_addressed": [
34 + "Pay-for-performance exposure in compensation (Appendix D).",
35 + "Public versus private firm differences (Appendix B).",
36 + "Timing mismatch between stationary comparisons and observed data dynamics."
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": 2
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 3
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": 1,
79 + "move": "related_literature_positioning"
80 + },
81 + {
82 + "paragraph_index": 2,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "mechanism_or_theory_sketch"
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": "contribution_statement"
104 + },
105 + {
106 + "paragraph_index": 6,
107 + "move": "related_literature_positioning"
108 + }
109 + ],
110 + "contribution_paragraph_index": 6,
111 + "contribution_types": [
112 + "new_evidence",
113 + "new_theory_or_model"
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": "data"
126 + },
127 + {
128 + "idx": 2,
129 + "role": "theory"
130 + },
131 + {
132 + "idx": 3,
133 + "role": "theory"
134 + },
135 + {
136 + "idx": 4,
137 + "role": "theory"
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 + "idx": 11,
165 + "role": "appendix"
166 + }
167 + ],
168 + "results_presentation": {
169 + "economic_significance_discussed": true,
170 + "statistical_significance_language": "unclear",
171 + "magnitude_statements_in_intro": true
172 + },
173 + "robustness": {
174 + "location": "appendix",
175 + "types": [
176 + "alternative_specification",
177 + "subsample",
178 + "other"
179 + ]
180 + },
181 + "tables": {
182 + "convention_notes": "Few tables. A calibration table pairs data targets with model moments, a table compares data changes with model-predicted changes across periods, and an industry regression table follows. Notes describe sample filters and variable construction.",
183 + "units_stated_in_notes": true,
184 + "standard_errors_description": "unclear"
185 + },
186 + "figures": {
187 + "uncertainty_shown": "unclear"
188 + },
189 + "conclusion": {
190 + "bounded_claims": false,
191 + "policy_implications": false,
192 + "open_questions": false,
193 + "restates_contribution": true
194 + },
195 + "hedging": {
196 + "level": "moderate",
197 + "typical_devices": [
198 + "consistent with",
199 + "largely",
200 + "is associated with",
201 + "caveat that calibration compares stationary equilibria"
202 + ]
203 + },
204 + "claim_evidence": [
205 + {
206 + "claim": "Aggregate capital share of public firms rose from about 40% to 60% since 1980.",
207 + "evidence_type": "figure",
208 + "evidence_ref": "Figure 1"
209 + },
210 + {
211 + "claim": "Aggregate factor shares are driven by mega-firms, while the average firm's capital share fell.",
212 + "evidence_type": "figure",
213 + "evidence_ref": "Figures 2, 3, 9"
214 + },
215 + {
216 + "claim": "Doubling rents and raising volatility from 20% to 40% replicates the rise in aggregate and the fall in average capital share.",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table II"
219 + },
220 + {
221 + "claim": "Within-industry increases in idiosyncratic volatility are associated with lower average capital-income-to-sales ratios.",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table III"
224 + },
225 + {
226 + "claim": "Dispersion effect is strongest in health and technology industries, which saw large volatility increases.",
227 + "evidence_type": "figure",
228 + "evidence_ref": "Figure 10"
229 + },
230 + {
231 + "claim": "Higher volatility lowers the exit threshold and fattens the right tail of the stationary productivity distribution.",
232 + "evidence_type": "equation_or_proof",
233 + "evidence_ref": "Section III, Figure 7, Appendix C"
234 + }
235 + ],
236 + "structural_summary": "The paper opens with a divergence puzzle: aggregate capital share rose while firm-level shares fell. A stylized-facts section documents this divergence. A dynamic industry equilibrium model with worker insurance and endogenous exit follows, along with its equilibrium analysis. The model is then calibrated and run through comparative statics across stationary equilibria. Supporting empirical evidence comes from size-quantile, industry and regression analyses. Appendices cover data, private firms, proofs, pay-for-performance and valuation.",
237 + "writing_profile_keywords": [
238 + "stylized facts first",
239 + "model as laboratory",
240 + "selection mechanism",
241 + "calibration",
242 + "national income accounting",
243 + "supporting evidence",
244 + "differentiation from Autor et al."
245 + ],
246 + "confidence": 0.8,
247 + "notes": "Empirical evidence is framed as consistent with the model rather than causal. The type of standard errors is not visible. Uncertainty bands in figures cannot be assessed from the captions."
248 + },
249 + "deterministic_checks": {
250 + "roadmap_regex": false,
251 + "contribution_regex_in_intro": true,
252 + "magnitude_in_abstract": false,
253 + "abstract_n_sentences": 7,
254 + "intro_n_paragraphs": 7,
255 + "intro_n_words": 1292,
256 + "causal_words_in_abstract": 0,
257 + "hedge_words_in_abstract": 0,
258 + "stars_mentioned": false,
259 + "clustered_se_mentioned": false,
260 + "internet_appendix_mentioned": false,
261 + "section_roles_rule": {
262 + "0": "introduction",
263 + "1": "other",
264 + "2": "theory",
265 + "3": "theory",
266 + "4": "theory",
267 + "5": "results",
268 + "6": "conclusion",
269 + "7": "appendix",
270 + "8": "appendix",
271 + "9": "appendix",
272 + "10": "appendix",
273 + "11": "appendix"
274 + }
275 + },
276 + "schema_valid": true,
277 + "code_version": "87c73cd-dirty"
278 +}
\ No newline at end of file
added data/annotations/documents/jof_12775.json +316 −0
@@ -0,0 +1,316 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [],
4 + "doc_id": "jof_12775",
5 + "ontology_version": "1.0",
6 + "prompt_version": "1.0",
7 + "model": "claude-opus-5-5",
8 + "effort": "medium",
9 + "annotated_at": "2026-11-03T23:26:54+00:00",
10 + "source_sha256": "729d504e28650806963dc88e59312792af0d708c2d8c97a5c6bd0e7ab7ea42e9",
11 + "parser": "pymupdf-jof-heuristics-1.3",
12 + "payload_chars": 26481,
13 + "usage": {
14 + "input": 8945,
15 + "output": 3195,
16 + "cache_read": 3968,
17 + "cache_write": 0
18 + },
19 + "cost_usd": 0.100474,
20 + "request_id": "req_011Cggrq1N7mxu2wVfywSrJN",
21 + "annotation": {
22 + "genre": "mixed_theory_empirical",
23 + "subfield": "asset_pricing",
24 + "subfield_secondary": "household_finance",
25 + "primary_method": "reduced_form_regression",
26 + "identification": {
27 + "estimand": "Predictive relation between a model-implied income hedging demand (IHD) measure and future HML factor returns, at the national and state levels.",
28 + "strategy": "A two-investor equilibrium model motivates a covariance-based IHD measure estimated with 10-year rolling windows. The measure is checked against retail holdings and mutual fund flows, then used in time-series and within-state panel predictive regressions with factor and macro controls. Long-short state portfolios are formed on the predictions.",
29 + "key_assumptions": [
30 + "Hedging demand is captured by the covariance of asset returns with income growth, scaled by the return covariance matrix",
31 + "Rolling-window conditional moments track the time-varying hedging potential of assets",
32 + "State-level income risk maps to demand for locally owned stocks"
33 + ],
34 + "threats_addressed": [
35 + "Macroeconomic shocks driving both income and returns",
36 + "Estimation biases in the predictive regressions",
37 + "Standard factor predictors as confounders",
38 + "Whether the effect is specific to hedging (occupation, top-earner and local-ownership contrasts)"
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": true,
62 + "sentence_index": 4
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": "related_literature_positioning"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "mechanism_or_theory_sketch"
82 + },
83 + {
84 + "paragraph_index": 2,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 3,
89 + "move": "related_literature_positioning"
90 + },
91 + {
92 + "paragraph_index": 4,
93 + "move": "preview_of_approach"
94 + },
95 + {
96 + "paragraph_index": 5,
97 + "move": "preview_of_approach"
98 + },
99 + {
100 + "paragraph_index": 6,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 7,
105 + "move": "preview_of_results"
106 + },
107 + {
108 + "paragraph_index": 8,
109 + "move": "preview_of_results"
110 + },
111 + {
112 + "paragraph_index": 8,
113 + "move": "robustness_mention"
114 + },
115 + {
116 + "paragraph_index": 9,
117 + "move": "preview_of_results"
118 + },
119 + {
120 + "paragraph_index": 10,
121 + "move": "identification_defense"
122 + },
123 + {
124 + "paragraph_index": 11,
125 + "move": "identification_defense"
126 + },
127 + {
128 + "paragraph_index": 12,
129 + "move": "contribution_statement"
130 + },
131 + {
132 + "paragraph_index": 12,
133 + "move": "related_literature_positioning"
134 + },
135 + {
136 + "paragraph_index": 13,
137 + "move": "roadmap"
138 + }
139 + ],
140 + "contribution_paragraph_index": 12,
141 + "contribution_types": [
142 + "new_evidence",
143 + "new_data_or_measure",
144 + "new_theory_or_model"
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": "theory"
157 + },
158 + {
159 + "idx": 2,
160 + "role": "data"
161 + },
162 + {
163 + "idx": 3,
164 + "role": "results"
165 + },
166 + {
167 + "idx": 4,
168 + "role": "empirical_design"
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 + "idx": 9,
188 + "role": "appendix"
189 + },
190 + {
191 + "idx": 10,
192 + "role": "appendix"
193 + },
194 + {
195 + "idx": 11,
196 + "role": "appendix"
197 + }
198 + ],
199 + "results_presentation": {
200 + "economic_significance_discussed": true,
201 + "statistical_significance_language": "unclear",
202 + "magnitude_statements_in_intro": true
203 + },
204 + "robustness": {
205 + "location": "mixed",
206 + "types": [
207 + "alternative_specification",
208 + "subsample",
209 + "alternative_measure",
210 + "falsification",
211 + "other"
212 + ]
213 + },
214 + "tables": {
215 + "convention_notes": "Table notes are detailed. They define how each variable is built (rolling 40-quarter covariances, value-weighted state HML portfolios), state the regression design (time-series predictive or within-state panel) and refer back to the control sets listed in an earlier summary table.",
216 + "units_stated_in_notes": true,
217 + "standard_errors_description": "unclear"
218 + },
219 + "figures": {
220 + "uncertainty_shown": "no"
221 + },
222 + "conclusion": {
223 + "bounded_claims": true,
224 + "policy_implications": false,
225 + "open_questions": true,
226 + "restates_contribution": true
227 + },
228 + "hedging": {
229 + "level": "moderate",
230 + "typical_devices": [
231 + "'suggest' and 'is associated with' wording for results",
232 + "'potential' and 'could' when stating the conjecture",
233 + "'at least part of' when attributing predictability",
234 + "Qualifiers on the horizon, such as 'in the short run'"
235 + ]
236 + },
237 + "claim_evidence": [
238 + {
239 + "claim": "Model-implied IHD correlates positively with retail investors' relative value-growth portfolio weights.",
240 + "evidence_type": "table",
241 + "evidence_ref": "Table IV"
242 + },
243 + {
244 + "claim": "IHD is positively related to flows into value funds and negatively to flows into growth funds.",
245 + "evidence_type": "table",
246 + "evidence_ref": "Table IV"
247 + },
248 + {
249 + "claim": "A one-SD increase in national IHD predicts 0.59% lower HML return next month and 0.96% lower next quarter.",
250 + "evidence_type": "table",
251 + "evidence_ref": "Table V"
252 + },
253 + {
254 + "claim": "State-level IHD predicts state-level HML returns, mainly among stocks with high local ownership.",
255 + "evidence_type": "table",
256 + "evidence_ref": "Table VI"
257 + },
258 + {
259 + "claim": "A long-short portfolio across states earns an annualized risk-adjusted return above 6%.",
260 + "evidence_type": "table",
261 + "evidence_ref": "Tables VII-VIII"
262 + },
263 + {
264 + "claim": "IHD built from high-risk occupations and top earners predicts HML returns, while IHD from low-risk occupations does not.",
265 + "evidence_type": "table",
266 + "evidence_ref": "Tables IX-X"
267 + },
268 + {
269 + "claim": "IHD varies substantially across states and over time.",
270 + "evidence_type": "figure",
271 + "evidence_ref": "Figure 1"
272 + }
273 + ],
274 + "structural_summary": "The introduction motivates the paper with household-level evidence on income hedging and conjectures that time-varying hedging demand predicts returns. It justifies the focus on HML, previews the validation tests, the predictability results (with magnitudes), the state-level evidence and mechanism-specific tests, then states contributions to the household portfolio choice and style investing literatures and gives a roadmap. The body proceeds from an analytical two-investor model (proof in the appendix) to estimation of IHD, validation, aggregate and state-level predictive regressions, portfolio performance, and additional/robustness tests. The conclusion restates the findings and suggests extensions.",
275 + "writing_profile_keywords": [
276 + "model-motivated measure",
277 + "validation tests",
278 + "predictive regressions",
279 + "state-level heterogeneity",
280 + "long-short portfolio",
281 + "household income hedging",
282 + "value factor"
283 + ],
284 + "confidence": 0.8,
285 + "notes": "The roadmap in the introduction (paragraph 13) labels both the aggregate predictability section and the state-level section as Section V. From the section list, the aggregate predictability section is Section IV (idx 4), which the roadmap misnumbers. Section idx 5 is a heading fragment: its 'heading' is the tail of a regression equation from Section IV, and the entry holds the remainder of Section IV (subsections B-D, the predictability results and their robustness). It was therefore labeled 'results'. Appendix entries idx 10 and 11 are likewise equation fragments belonging to the Proposition 1 proof. The excerpts do not show how standard errors or statistical significance are reported."
286 + },
287 + "deterministic_checks": {
288 + "roadmap_regex": true,
289 + "contribution_regex_in_intro": true,
290 + "magnitude_in_abstract": true,
291 + "abstract_n_sentences": 5,
292 + "intro_n_paragraphs": 14,
293 + "intro_n_words": 1618,
294 + "causal_words_in_abstract": 0,
295 + "hedge_words_in_abstract": 0,
296 + "stars_mentioned": false,
297 + "clustered_se_mentioned": false,
298 + "internet_appendix_mentioned": false,
299 + "section_roles_rule": {
300 + "0": "introduction",
301 + "1": "theory",
302 + "2": "other",
303 + "3": "results",
304 + "4": "other",
305 + "5": "other",
306 + "6": "other",
307 + "7": "results",
308 + "8": "conclusion",
309 + "9": "appendix",
310 + "10": "appendix",
311 + "11": "appendix"
312 + }
313 + },
314 + "schema_valid": true,
315 + "code_version": "87c73cd-dirty"
316 +}
\ No newline at end of file
added data/annotations/documents/jof_12776.json +266 −0
@@ -0,0 +1,266 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12776",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:25+00:00",
9 + "source_sha256": "4d2f5c2d549474d85cf103595900ae591b98c147f1b4e65061fb16cc32f587d9",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21783,
12 + "usage": {
13 + "input": 8060,
14 + "output": 2276,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.078554,
19 + "request_id": "req_011CggqRmRTbEKkYxM768Y6K",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "financial_econometrics",
24 + "primary_method": "other",
25 + "identification": {
26 + "estimand": "Shape of a common nonlinear function of lagged VIX that forecasts excess stock and bond returns, plus asset-specific loadings on that function",
27 + "strategy": "Sieve reduced-rank (SRR) panel predictive regressions. Each asset's excess return is regressed on sieve expansions of lagged VIX, and the coefficient matrix is reduced to rank one. Inference uses Hodrick-style reverse regressions; there is no causal design.",
28 + "key_assumptions": [
29 + "A rank-one common nonlinear factor drives expected returns across assets up to affine transformations",
30 + "Reduced-rank estimation is optimal under conditionally Gaussian errors with a fixed number of regressors",
31 + "Asymptotic validity of the proposed test statistics under overlapping observations"
32 + ],
33 + "threats_addressed": [
34 + "Omitted standard predictors (dividend yield, default spread, term spread, VRP)",
35 + "Influence of the 2008-09 crisis tail events, via a 1990-2007 subsample",
36 + "Treasury-specific volatility (MOVE) as an alternative predictor",
37 + "Overlapping-horizon inference",
38 + "Out-of-sample forecasting performance"
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": 1
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": 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": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 5,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 6,
93 + "move": "contribution_statement"
94 + },
95 + {
96 + "paragraph_index": 7,
97 + "move": "contribution_statement"
98 + },
99 + {
100 + "paragraph_index": 8,
101 + "move": "robustness_mention"
102 + },
103 + {
104 + "paragraph_index": 8,
105 + "move": "related_literature_positioning"
106 + },
107 + {
108 + "paragraph_index": 10,
109 + "move": "mechanism_or_theory_sketch"
110 + },
111 + {
112 + "paragraph_index": 11,
113 + "move": "related_literature_positioning"
114 + },
115 + {
116 + "paragraph_index": 12,
117 + "move": "roadmap"
118 + }
119 + ],
120 + "contribution_paragraph_index": 6,
121 + "contribution_types": [
122 + "new_evidence",
123 + "new_method"
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": "results"
136 + },
137 + {
138 + "idx": 2,
139 + "role": "empirical_design"
140 + },
141 + {
142 + "idx": 3,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 4,
147 + "role": "theory"
148 + },
149 + {
150 + "idx": 5,
151 + "role": "conclusion"
152 + }
153 + ],
154 + "results_presentation": {
155 + "economic_significance_discussed": true,
156 + "statistical_significance_language": "mixed",
157 + "magnitude_statements_in_intro": false
158 + },
159 + "robustness": {
160 + "location": "mixed",
161 + "types": [
162 + "alternative_specification",
163 + "alternative_sample_period",
164 + "subsample",
165 + "alternative_measure",
166 + "other"
167 + ]
168 + },
169 + "tables": {
170 + "convention_notes": "Notes state the regression equation, forecasting horizons (6/12/18 months) and sample period. Tables report Hodrick (1992) t-statistics, LM test statistics with p-values in parentheses, and significance stars in the fund flow table.",
171 + "units_stated_in_notes": false,
172 + "standard_errors_description": "mixed"
173 + },
174 + "figures": {
175 + "uncertainty_shown": "unclear"
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 + "consistent with",
187 + "suggests",
188 + "points toward",
189 + "in line with",
190 + "tend to"
191 + ]
192 + },
193 + "claim_evidence": [
194 + {
195 + "claim": "Lagged VIX forecasts stock and bond returns nonlinearly, with mirror-image shapes across the two asset classes",
196 + "evidence_type": "figure",
197 + "evidence_ref": "Figure 1, Figure 4"
198 + },
199 + {
200 + "claim": "Linear VIX predictability is insignificant, but polynomial specifications are significant across horizons",
201 + "evidence_type": "table",
202 + "evidence_ref": "Table I, Figures 2-3"
203 + },
204 + {
205 + "claim": "SRR estimates show a highly significant common nonlinear function in the full sample and in the pre-crisis sample",
206 + "evidence_type": "table",
207 + "evidence_ref": "Tables II-III"
208 + },
209 + {
210 + "claim": "Tests do not reject that stocks and bonds load on the same nonlinear function, interpreted as flight-to-safety",
211 + "evidence_type": "table",
212 + "evidence_ref": "Table IV"
213 + },
214 + {
215 + "claim": "The SRR model improves out-of-sample forecasts relative to a mean-only benchmark",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table V"
218 + },
219 + {
220 + "claim": "Mutual fund flows depend on VIX with a shape similar to the return-forecasting function, supporting an asset-manager mechanism",
221 + "evidence_type": "table",
222 + "evidence_ref": "Table VI, Figure 7"
223 + },
224 + {
225 + "claim": "Forecasting slopes line up with risk factor loadings in a dynamic asset pricing model",
226 + "evidence_type": "figure",
227 + "evidence_ref": "Figure 5"
228 + }
229 + ],
230 + "structural_summary": "The paper opens with flight-to-safety theories and the difficulty of documenting nonlinear pricing empirically. It previews a mirror-image nonlinear VIX-return relation and introduces a new panel sieve reduced-rank estimator with accompanying inference tests. Motivating univariate polynomial and nonparametric evidence comes first, followed by derivation of the estimator, main results and robustness checks. An economic-interpretation section links the results to dynamic asset pricing, mutual fund flows and intermediary VaR theories. A short conclusion restates the findings.",
231 + "writing_profile_keywords": [
232 + "return predictability",
233 + "nonparametric panel estimator",
234 + "flight-to-safety",
235 + "VIX",
236 + "mirror image",
237 + "method plus evidence",
238 + "theory-consistent interpretation"
239 + ],
240 + "confidence": 0.8,
241 + "notes": "Primary method is set to 'other' because the paper's core tool is a new sieve reduced-rank estimator rather than standard reduced-form regression. Section idx 3 is a parsing artifact (its heading is an equation fragment) that continues Section II, covering SRR estimation and robustness. The Internet Appendix holds additional results. Figure 7's caption mentions panels showing predicted fund flows, but whether figures display confidence bands is not visible."
242 + },
243 + "deterministic_checks": {
244 + "roadmap_regex": true,
245 + "contribution_regex_in_intro": false,
246 + "magnitude_in_abstract": false,
247 + "abstract_n_sentences": 4,
248 + "intro_n_paragraphs": 15,
249 + "intro_n_words": 1472,
250 + "causal_words_in_abstract": 0,
251 + "hedge_words_in_abstract": 0,
252 + "stars_mentioned": true,
253 + "clustered_se_mentioned": false,
254 + "internet_appendix_mentioned": true,
255 + "section_roles_rule": {
256 + "0": "introduction",
257 + "1": "results",
258 + "2": "results",
259 + "3": "other",
260 + "4": "other",
261 + "5": "conclusion"
262 + }
263 + },
264 + "schema_valid": true,
265 + "code_version": "87c73cd-dirty"
266 +}
\ No newline at end of file
added data/annotations/documents/jof_12777.json +256 −0
@@ -0,0 +1,256 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12777",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:56+00:00",
9 + "source_sha256": "126d05da9833d5f755e9bd390990525c6cdbef1110f374612ba849dc30f9ad40",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21904,
12 + "usage": {
13 + "input": 7618,
14 + "output": 2075,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.072766,
19 + "request_id": "req_011CggqUBFjLBYv7wWYtZ7Ei",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Effect of protection sellers' CDS-portfolio capital shocks (dollar margin payments) on weekly log changes in a firm's CDS spread",
27 + "strategy": "Panel regressions of weekly spread changes on seller and buyer capital shocks that exclude positions on same-industry firms, with firm-level credit controls and macro controls. Supplemented by placebo tests, horizon dynamics, and a 2011 Japanese tsunami case study treated as an exogenous shock to seller capital.",
28 + "key_assumptions": [
29 + "Seller capital shocks built from positions on other-industry firms are not driven by the target firm's fundamentals",
30 + "Controls such as EDF, LGD and equity returns absorb changes in fundamental credit risk",
31 + "Sellers are not fully hedged, so mark-to-market losses change their risk appetite"
32 + ],
33 + "threats_addressed": [
34 + "common fundamental shocks driving both seller losses and spreads",
35 + "hedging by sellers or buyers",
36 + "whether capital shocks are large enough to matter",
37 + "construction methodology of capital measures",
38 + "institution-wide versus desk-level constraints"
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": 0
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 1
59 + },
60 + "magnitude": {
61 + "present": true,
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": 1,
77 + "move": "mechanism_or_theory_sketch"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "related_literature_positioning"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "tension_or_puzzle"
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": 4,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 5,
101 + "move": "preview_of_results"
102 + }
103 + ],
104 + "contribution_paragraph_index": -1,
105 + "contribution_types": [
106 + "new_evidence",
107 + "new_data_or_measure"
108 + ],
109 + "roadmap_present": false,
110 + "literature_positioning_location": "inside_introduction"
111 + },
112 + "section_roles": [
113 + {
114 + "idx": 0,
115 + "role": "introduction"
116 + },
117 + {
118 + "idx": 1,
119 + "role": "data"
120 + },
121 + {
122 + "idx": 2,
123 + "role": "results"
124 + },
125 + {
126 + "idx": 3,
127 + "role": "robustness"
128 + },
129 + {
130 + "idx": 4,
131 + "role": "robustness"
132 + },
133 + {
134 + "idx": 5,
135 + "role": "conclusion"
136 + },
137 + {
138 + "idx": 6,
139 + "role": "appendix"
140 + },
141 + {
142 + "idx": 7,
143 + "role": "appendix"
144 + }
145 + ],
146 + "results_presentation": {
147 + "economic_significance_discussed": true,
148 + "statistical_significance_language": "stars_and_tstats",
149 + "magnitude_statements_in_intro": true
150 + },
151 + "robustness": {
152 + "location": "mixed",
153 + "types": [
154 + "placebo",
155 + "alternative_measure",
156 + "subsample",
157 + "alternative_sample_period",
158 + "other"
159 + ]
160 + },
161 + "tables": {
162 + "convention_notes": "Notes write out the regression equation, list the control vectors, and define variables with their units (dollar changes in $ billion). Standard errors appear below the estimates, double-clustered, with stars marking p-value thresholds.",
163 + "units_stated_in_notes": true,
164 + "standard_errors_description": "clustered"
165 + },
166 + "figures": {
167 + "uncertainty_shown": "unclear"
168 + },
169 + "conclusion": {
170 + "bounded_claims": true,
171 + "policy_implications": true,
172 + "open_questions": false,
173 + "restates_contribution": true
174 + },
175 + "hedging": {
176 + "level": "moderate",
177 + "typical_devices": [
178 + "suggest",
179 + "imply",
180 + "in part",
181 + "at least partially",
182 + "likely explanation",
183 + "potential unintended consequence"
184 + ]
185 + },
186 + "claim_evidence": [
187 + {
188 + "claim": "A $1bn seller capital loss raises CDS spreads about 2.7% per week; seller shocks explain 12% of weekly spread variation.",
189 + "evidence_type": "table",
190 + "evidence_ref": "Table III"
191 + },
192 + {
193 + "claim": "Desk-level seller shocks carry information beyond institution-wide capital constraints.",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table IV"
196 + },
197 + {
198 + "claim": "The spread effect of seller losses reverts fully after about nine weeks.",
199 + "evidence_type": "figure",
200 + "evidence_ref": "Figure 2"
201 + },
202 + {
203 + "claim": "Incumbent sellers barely adjust positions after losses, and no new sellers enter.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table VI"
206 + },
207 + {
208 + "claim": "Seller exposure to Japanese firms predicts U.S. CDS spread changes around the 2011 tsunami.",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table VII"
211 + },
212 + {
213 + "claim": "Asset managers have replaced dealers as the main net sellers of protection since 2010.",
214 + "evidence_type": "figure",
215 + "evidence_ref": "Figure 5"
216 + }
217 + ],
218 + "structural_summary": "The introduction frames frictionless capital flow as a neoclassical benchmark, then sketches intermediary and slow-moving-capital theories and notes that causal identification is hard. It previews results paragraph by paragraph, giving magnitudes, desk-specific effects, horizon decay, mechanism, and dealer versus asset manager heterogeneity. The data section documents concentration in the market. The main section builds a simple pricing framework, runs a workhorse regression, and adds internal capital market, segmentation/placebo, horizon, and mechanism analyses. Robustness and additional analyses (hedging, shock size, tsunami case study, bond market) follow. The conclusion restates the segmentation interpretation and notes possible regulatory side effects.",
219 + "writing_profile_keywords": [
220 + "proprietary regulatory data",
221 + "slow-moving capital",
222 + "market segmentation",
223 + "first-person singular",
224 + "economic magnitude emphasis",
225 + "case-study corroboration",
226 + "policy implication"
227 + ],
228 + "confidence": 0.8,
229 + "notes": "The introduction excerpt is truncated, so contribution and roadmap paragraphs may exist but are not visible. Figure captions do not say whether confidence bands are shown."
230 + },
231 + "deterministic_checks": {
232 + "roadmap_regex": false,
233 + "contribution_regex_in_intro": false,
234 + "magnitude_in_abstract": true,
235 + "abstract_n_sentences": 5,
236 + "intro_n_paragraphs": 6,
237 + "intro_n_words": 870,
238 + "causal_words_in_abstract": 0,
239 + "hedge_words_in_abstract": 1,
240 + "stars_mentioned": true,
241 + "clustered_se_mentioned": true,
242 + "internet_appendix_mentioned": false,
243 + "section_roles_rule": {
244 + "0": "introduction",
245 + "1": "data",
246 + "2": "other",
247 + "3": "robustness",
248 + "4": "robustness",
249 + "5": "conclusion",
250 + "6": "appendix",
251 + "7": "appendix"
252 + }
253 + },
254 + "schema_valid": true,
255 + "code_version": "87c73cd-dirty"
256 +}
\ No newline at end of file
added data/annotations/documents/jof_12778.json +277 −0
@@ -0,0 +1,277 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.figure_note"
5 + ],
6 + "doc_id": "jof_12778",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:32+00:00",
12 + "source_sha256": "85bcbf56ab250dd56974ed96df408887269d3b924ce7145db89be56be8d97845",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 24041,
15 + "usage": {
16 + "input": 8562,
17 + "output": 2160,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.078242,
22 + "request_id": "req_011CggqRYnguqXyWf2iMEUMR",
23 + "annotation": {
24 + "genre": "mixed_theory_empirical",
25 + "subfield": "asset_pricing",
26 + "subfield_secondary": "financial_econometrics",
27 + "primary_method": "theoretical_model_analytical",
28 + "identification": {
29 + "estimand": "Conditional expected (excess and excess-of-market) returns on individual stocks implied by option-based risk-neutral variances",
30 + "strategy": "Derive a parameter-free formula from the growth-optimal SDF, then test predicted coefficients in pooled and fixed-effects panel forecasting regressions and in out-of-sample R2 comparisons against benchmarks",
31 + "key_assumptions": [
32 + "Theoretical conditions linking the growth-optimal return to option-implied variances hold approximately",
33 + "Option prices accurately measure risk-neutral variance",
34 + "Bootstrap inference is valid given overlapping observations and cross-sectional correlation"
35 + ],
36 + "threats_addressed": [
37 + "Data snooping / free parameters (formula has none)",
38 + "Overlapping returns and panel dependence in standard errors (block bootstrap)",
39 + "Omitted characteristics (beta, size, book-to-market, momentum)",
40 + "Out-of-sample validity"
41 + ],
42 + "causal_language_strength": 1,
43 + "uses_causal_claim": false
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": true,
56 + "sentence_index": 0
57 + },
58 + "results": {
59 + "present": true,
60 + "sentence_index": 2
61 + },
62 + "magnitude": {
63 + "present": false,
64 + "sentence_index": -1
65 + },
66 + "contribution": {
67 + "present": true,
68 + "sentence_index": 3
69 + }
70 + },
71 + "introduction": {
72 + "moves": [
73 + {
74 + "paragraph_index": 0,
75 + "move": "preview_of_approach"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "contribution_statement"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "contribution_statement"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "related_literature_positioning"
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": "roadmap"
108 + },
109 + {
110 + "paragraph_index": 6,
111 + "move": "related_literature_positioning"
112 + },
113 + {
114 + "paragraph_index": 7,
115 + "move": "related_literature_positioning"
116 + },
117 + {
118 + "paragraph_index": 8,
119 + "move": "related_literature_positioning"
120 + },
121 + {
122 + "paragraph_index": 9,
123 + "move": "related_literature_positioning"
124 + },
125 + {
126 + "paragraph_index": 10,
127 + "move": "related_literature_positioning"
128 + }
129 + ],
130 + "contribution_paragraph_index": 2,
131 + "contribution_types": [
132 + "new_theory_or_model",
133 + "new_evidence",
134 + "new_data_or_measure"
135 + ],
136 + "roadmap_present": true,
137 + "literature_positioning_location": "inside_introduction"
138 + },
139 + "section_roles": [
140 + {
141 + "idx": 0,
142 + "role": "introduction"
143 + },
144 + {
145 + "idx": 1,
146 + "role": "theory"
147 + },
148 + {
149 + "idx": 2,
150 + "role": "data"
151 + },
152 + {
153 + "idx": 3,
154 + "role": "results"
155 + },
156 + {
157 + "idx": 4,
158 + "role": "results"
159 + },
160 + {
161 + "idx": 5,
162 + "role": "results"
163 + },
164 + {
165 + "idx": 6,
166 + "role": "conclusion"
167 + },
168 + {
169 + "idx": 7,
170 + "role": "appendix"
171 + },
172 + {
173 + "idx": 8,
174 + "role": "appendix"
175 + }
176 + ],
177 + "results_presentation": {
178 + "economic_significance_discussed": true,
179 + "statistical_significance_language": "unclear",
180 + "magnitude_statements_in_intro": false
181 + },
182 + "robustness": {
183 + "location": "mixed",
184 + "types": [
185 + "subsample",
186 + "alternative_specification",
187 + "other"
188 + ]
189 + },
190 + "tables": {
191 + "convention_notes": "Notes describe the sample (S&P 100/500 firms, monthly data from 1996 to 2014), the regression specification and its panels. Coefficient tests are framed as hypotheses on theory-implied parameter values. Out-of-sample R2 is defined explicitly in the notes.",
192 + "units_stated_in_notes": false,
193 + "standard_errors_description": "bootstrap"
194 + },
195 + "figures": {
196 + "uncertainty_shown": "unclear"
197 + },
198 + "conclusion": {
199 + "bounded_claims": false,
200 + "policy_implications": false,
201 + "open_questions": false,
202 + "restates_contribution": true
203 + },
204 + "hedging": {
205 + "level": "low",
206 + "typical_devices": [
207 + "'suggest' in abstract",
208 + "'in principle'",
209 + "quoted 'loses' acknowledging a competitor win"
210 + ]
211 + },
212 + "claim_evidence": [
213 + {
214 + "claim": "Expected stock return can be expressed via market risk-neutral variance and stock excess risk-neutral variance with no free parameters",
215 + "evidence_type": "equation_or_proof",
216 + "evidence_ref": "Section I"
217 + },
218 + {
219 + "claim": "Formula's panel forecasting coefficients are consistent with theory-implied values",
220 + "evidence_type": "table",
221 + "evidence_ref": "Tables II-V"
222 + },
223 + {
224 + "claim": "Formula beats out-of-sample competitors at horizons from 3 to 24 months",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table IX"
227 + },
228 + {
229 + "claim": "Formula mostly beats competitors that use in-sample information on characteristics",
230 + "evidence_type": "table",
231 + "evidence_ref": "Table X"
232 + },
233 + {
234 + "claim": "Model implies more time-series and cross-sectional variation in expected returns than CAPM",
235 + "evidence_type": "figure",
236 + "evidence_ref": "Figures 1, 9"
237 + }
238 + ],
239 + "structural_summary": "The paper opens by stating a derived formula and its option-based inputs. It then lists distinctive features and contrasts them with the CAPM and factor models. The introduction previews in-sample and out-of-sample results, illustrates with firm examples, and positions the work against the idiosyncratic-volatility and option-forecasting literatures. The theory section is followed by variance measurement, tests of the theory-implied coefficients, a check against characteristics, out-of-sample R2 comparisons, and a conclusion restating the distinctive features. Appendices cover implied correlation and the block bootstrap.",
240 + "writing_profile_keywords": [
241 + "parameter-free theory",
242 + "option-implied variance",
243 + "cross-section of expected returns",
244 + "out-of-sample R2",
245 + "CAPM comparison",
246 + "confident tone"
247 + ],
248 + "confidence": 0.8,
249 + "notes": "The roadmap and the related-literature discussion are merged in the same introduction paragraph. Significance-test conventions are unclear from the excerpts. Robustness evidence is spread across subsample estimates (Figure 7), characteristic controls (Section IV), and the appendix and Internet Appendix."
250 + },
251 + "deterministic_checks": {
252 + "roadmap_regex": false,
253 + "contribution_regex_in_intro": false,
254 + "magnitude_in_abstract": false,
255 + "abstract_n_sentences": 4,
256 + "intro_n_paragraphs": 11,
257 + "intro_n_words": 1371,
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": "theory",
266 + "2": "other",
267 + "3": "theory",
268 + "4": "other",
269 + "5": "data",
270 + "6": "conclusion",
271 + "7": "appendix",
272 + "8": "appendix"
273 + }
274 + },
275 + "schema_valid": true,
276 + "code_version": "87c73cd-dirty"
277 +}
\ No newline at end of file
added data/annotations/documents/jof_12779.json +261 −0
@@ -0,0 +1,261 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12779",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:14+00:00",
9 + "source_sha256": "52ff2e9b20d30f3d5040cc9fc4dfcb9daaf7a9a12426871b6afe0891f743576c",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19090,
12 + "usage": {
13 + "input": 6454,
14 + "output": 2039,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06739,
19 + "request_id": "req_011CggqVT8jCnp6CQwYbpJjv",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "information economics / strategic voting and corporate governance",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model); model outcome is firm value, defined as the probability of a correct vote decision, with versus without a proxy advisor",
27 + "strategy": "not applicable; strategic voting model with endogenous information acquisition and a monopolistic information seller, solved by backward induction",
28 + "key_assumptions": [
29 + "shareholders have aligned (common) preferences",
30 + "monopolistic advisor sets its fee to maximize profit",
31 + "private signals have independent (or imperfectly correlated) errors",
32 + "the advisor's signal is common to all who buy it",
33 + "cost of private information exceeds a cutoff in parts of the analysis"
34 + ],
35 + "threats_addressed": [
36 + "equilibrium multiplicity and equilibrium selection",
37 + "correlated mistakes in private signals",
38 + "free access to recommendations",
39 + "communication or sale of information among shareholders",
40 + "acquisition of both signals in equilibrium"
41 + ],
42 + "causal_language_strength": 1,
43 + "uses_causal_claim": false
44 + },
45 + "abstract_moves": {
46 + "question": {
47 + "present": true,
48 + "sentence_index": 0
49 + },
50 + "gap": {
51 + "present": false,
52 + "sentence_index": -1
53 + },
54 + "method": {
55 + "present": true,
56 + "sentence_index": 0
57 + },
58 + "results": {
59 + "present": true,
60 + "sentence_index": 1
61 + },
62 + "magnitude": {
63 + "present": false,
64 + "sentence_index": -1
65 + },
66 + "contribution": {
67 + "present": true,
68 + "sentence_index": 4
69 + }
70 + },
71 + "introduction": {
72 + "moves": [
73 + {
74 + "paragraph_index": 0,
75 + "move": "motivation"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "policy_or_broader_implication"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "mechanism_or_theory_sketch"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "preview_of_results"
92 + },
93 + {
94 + "paragraph_index": 4,
95 + "move": "related_literature_positioning"
96 + },
97 + {
98 + "paragraph_index": 5,
99 + "move": "related_literature_positioning"
100 + },
101 + {
102 + "paragraph_index": 6,
103 + "move": "contribution_statement"
104 + },
105 + {
106 + "paragraph_index": 6,
107 + "move": "related_literature_positioning"
108 + }
109 + ],
110 + "contribution_paragraph_index": 6,
111 + "contribution_types": [
112 + "new_theory_or_model"
113 + ],
114 + "roadmap_present": false,
115 + "literature_positioning_location": "inside_introduction"
116 + },
117 + "section_roles": [
118 + {
119 + "idx": 0,
120 + "role": "introduction"
121 + },
122 + {
123 + "idx": 1,
124 + "role": "theory"
125 + },
126 + {
127 + "idx": 2,
128 + "role": "theory"
129 + },
130 + {
131 + "idx": 3,
132 + "role": "theory"
133 + },
134 + {
135 + "idx": 4,
136 + "role": "other"
137 + },
138 + {
139 + "idx": 5,
140 + "role": "theory"
141 + },
142 + {
143 + "idx": 6,
144 + "role": "robustness"
145 + },
146 + {
147 + "idx": 7,
148 + "role": "other"
149 + },
150 + {
151 + "idx": 8,
152 + "role": "conclusion"
153 + },
154 + {
155 + "idx": 9,
156 + "role": "appendix"
157 + }
158 + ],
159 + "results_presentation": {
160 + "economic_significance_discussed": false,
161 + "statistical_significance_language": "unclear",
162 + "magnitude_statements_in_intro": false
163 + },
164 + "robustness": {
165 + "location": "mixed",
166 + "types": [
167 + "other"
168 + ]
169 + },
170 + "tables": {
171 + "convention_notes": "No tables. Results are stated as propositions and lemmas, and proofs sit in Appendix A and the Internet Appendix. Numerical illustrations appear in figures plotting equilibrium objects against the fee and against signal precision.",
172 + "units_stated_in_notes": false,
173 + "standard_errors_description": "none"
174 + },
175 + "figures": {
176 + "uncertainty_shown": "no"
177 + },
178 + "conclusion": {
179 + "bounded_claims": true,
180 + "policy_implications": true,
181 + "open_questions": false,
182 + "restates_contribution": true
183 + },
184 + "hedging": {
185 + "level": "moderate",
186 + "typical_devices": [
187 + "conditional statements tied to parameter regions (only if precision is sufficiently high)",
188 + "'can have negative effects' / 'may' modality",
189 + "acknowledging the model is stylized",
190 + "in our model qualifiers"
191 + ]
192 + },
193 + "claim_evidence": [
194 + {
195 + "claim": "The advisor raises firm value only if its information is sufficiently precise, for any fee.",
196 + "evidence_type": "equation_or_proof",
197 + "evidence_ref": "Proposition 3; Appendix A"
198 + },
199 + {
200 + "claim": "With imprecise advisor information, shareholders over-rely on it and private research is crowded out.",
201 + "evidence_type": "equation_or_proof",
202 + "evidence_ref": "Section II-III; Figure 2"
203 + },
204 + {
205 + "claim": "With precise information, the monopolist rations recommendations, so the information is underused.",
206 + "evidence_type": "equation_or_proof",
207 + "evidence_ref": "Section III; Figure 3"
208 + },
209 + {
210 + "claim": "Reducing advisor market power or litigation pressure can lower decision quality.",
211 + "evidence_type": "equation_or_proof",
212 + "evidence_ref": "Section IV"
213 + },
214 + {
215 + "claim": "Proxy advisor recommendations strongly affect voting outcomes.",
216 + "evidence_type": "prior_literature",
217 + "evidence_ref": "Introduction footnote 1"
218 + }
219 + ],
220 + "structural_summary": "The introduction opens with institutional motivation (growth of proxy advisors, regulatory attention) and a policy debate. It then sketches the externality mechanism: correlated mistakes when shareholders follow a common signal. Next it previews the main threshold result and the inefficiency created by monopoly pricing, and positions the paper against corporate voting, strategic voting, sale-of-information, and information-externality literatures. The body builds a strategic voting model with a benchmark, solves it by backward induction for voting and acquisition given a fee, endogenizes the fee, and evaluates regulatory proposals. Later sections endogenize advisor quality, discuss robustness of assumptions, and derive empirical implications. The conclusion restates the results and the policy findings; proofs are in the appendix.",
221 + "writing_profile_keywords": [
222 + "applied theory",
223 + "strategic voting",
224 + "information sale",
225 + "policy evaluation",
226 + "backward induction",
227 + "propositions and proofs",
228 + "institutional motivation",
229 + "externalities"
230 + ],
231 + "confidence": 0.85,
232 + "notes": "Pure theory paper with no data. Robustness is handled verbally in Section VI and in the Internet Appendix. There is an empirical implications section but no tests. No roadmap paragraph is visible in the excerpted introduction."
233 + },
234 + "deterministic_checks": {
235 + "roadmap_regex": false,
236 + "contribution_regex_in_intro": true,
237 + "magnitude_in_abstract": false,
238 + "abstract_n_sentences": 5,
239 + "intro_n_paragraphs": 7,
240 + "intro_n_words": 1238,
241 + "causal_words_in_abstract": 1,
242 + "hedge_words_in_abstract": 1,
243 + "stars_mentioned": false,
244 + "clustered_se_mentioned": false,
245 + "internet_appendix_mentioned": true,
246 + "section_roles_rule": {
247 + "0": "introduction",
248 + "1": "theory",
249 + "2": "other",
250 + "3": "other",
251 + "4": "other",
252 + "5": "other",
253 + "6": "robustness",
254 + "7": "other",
255 + "8": "conclusion",
256 + "9": "appendix"
257 + }
258 + },
259 + "schema_valid": true,
260 + "code_version": "87c73cd-dirty"
261 +}
\ No newline at end of file
added data/annotations/documents/jof_12780.json +279 −0
@@ -0,0 +1,279 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12780",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:09+00:00",
9 + "source_sha256": "94cf0a1c3495d43fd153a6803395d6164711488734c800bacbf0aa6d3461144e",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 26100,
12 + "usage": {
13 + "input": 8773,
14 + "output": 2352,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.082926,
19 + "request_id": "req_011CggqUyzLMpPtf6uLRAVKr",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "international_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Cross-sectional risk premium (portfolio alpha) of firms in low versus high shipping-cost industries, and the sign of the price of foreign productivity shock risk",
27 + "strategy": "Portfolio sorts on industry shipping costs and weight-to-value, with five-factor Fama-French alphas; conditional sorts on size and profitability; betas on Chinese import growth; checks of earnings-announcement returns and analyst forecast errors; a calibrated Melitz-type GE model",
28 + "key_assumptions": [
29 + "Shipping costs capture structural, slow-moving barriers to trade that determine exposure to globalization",
30 + "Five-factor alphas reflect compensation for priced risk rather than mispricing",
31 + "Chinese import growth proxies for foreign productivity shocks"
32 + ],
33 + "threats_addressed": [
34 + "loadings on known risk factors",
35 + "mispricing or a wedge between ex post realized and ex ante expected returns (tested with earnings announcements and analyst forecasts)",
36 + "sample specificity (European stocks, subperiods)",
37 + "weighting scheme (value-weighted portfolios)"
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": 2
58 + },
59 + "magnitude": {
60 + "present": true,
61 + "sentence_index": 2
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": 1,
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": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 3,
88 + "move": "preview_of_results"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "robustness_mention"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "tension_or_puzzle"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "mechanism_or_theory_sketch"
105 + },
106 + {
107 + "paragraph_index": 6,
108 + "move": "preview_of_results"
109 + },
110 + {
111 + "paragraph_index": 7,
112 + "move": "mechanism_or_theory_sketch"
113 + },
114 + {
115 + "paragraph_index": 8,
116 + "move": "related_literature_positioning"
117 + },
118 + {
119 + "paragraph_index": 8,
120 + "move": "contribution_statement"
121 + },
122 + {
123 + "paragraph_index": 9,
124 + "move": "contribution_statement"
125 + },
126 + {
127 + "paragraph_index": 10,
128 + "move": "related_literature_positioning"
129 + },
130 + {
131 + "paragraph_index": 11,
132 + "move": "related_literature_positioning"
133 + },
134 + {
135 + "paragraph_index": 12,
136 + "move": "related_literature_positioning"
137 + },
138 + {
139 + "paragraph_index": 13,
140 + "move": "roadmap"
141 + }
142 + ],
143 + "contribution_paragraph_index": 9,
144 + "contribution_types": [
145 + "new_evidence",
146 + "new_theory_or_model"
147 + ],
148 + "roadmap_present": true,
149 + "literature_positioning_location": "inside_introduction"
150 + },
151 + "section_roles": [
152 + {
153 + "idx": 0,
154 + "role": "introduction"
155 + },
156 + {
157 + "idx": 1,
158 + "role": "results"
159 + },
160 + {
161 + "idx": 2,
162 + "role": "theory"
163 + },
164 + {
165 + "idx": 3,
166 + "role": "conclusion"
167 + }
168 + ],
169 + "results_presentation": {
170 + "economic_significance_discussed": true,
171 + "statistical_significance_language": "unclear",
172 + "magnitude_statements_in_intro": true
173 + },
174 + "robustness": {
175 + "location": "within_results",
176 + "types": [
177 + "subsample",
178 + "alternative_measure",
179 + "alternative_sample_period",
180 + "alternative_specification",
181 + "sensitivity_to_parameters"
182 + ]
183 + },
184 + "tables": {
185 + "convention_notes": "Table notes define how shipping costs and weight-to-value are constructed, give the sample (exchanges, SIC range, period), and state that monthly alphas are multiplied by 12 to annualize. Panels split the shipping-cost and weight-to-value sorts. The calibration tables give parameter sources and report simulated moments with Newey-West standard errors.",
186 + "units_stated_in_notes": true,
187 + "standard_errors_description": "mixed"
188 + },
189 + "figures": {
190 + "uncertainty_shown": "unclear"
191 + },
192 + "conclusion": {
193 + "bounded_claims": true,
194 + "policy_implications": false,
195 + "open_questions": true,
196 + "restates_contribution": true
197 + },
198 + "hedging": {
199 + "level": "moderate",
200 + "typical_devices": [
201 + "suggesting that",
202 + "can be consistent with",
203 + "we hope to stimulate",
204 + "we leave to future research",
205 + "may help rationalize"
206 + ]
207 + },
208 + "claim_evidence": [
209 + {
210 + "claim": "Shipping costs vary substantially across industries, persist over time, and correlate negatively with trade flows",
211 + "evidence_type": "table",
212 + "evidence_ref": "Tables I-III"
213 + },
214 + {
215 + "claim": "The long high-SC, short low-SC portfolio earns about -7% per year, and five-factor alphas are large",
216 + "evidence_type": "table",
217 + "evidence_ref": "Tables IV-V"
218 + },
219 + {
220 + "claim": "The premium also holds in European stocks",
221 + "evidence_type": "table",
222 + "evidence_ref": "Table VI"
223 + },
224 + {
225 + "claim": "No evidence of mispricing from earnings-announcement returns or analyst forecast errors",
226 + "evidence_type": "table",
227 + "evidence_ref": "Tables VII-VIII"
228 + },
229 + {
230 + "claim": "The premium is concentrated among small, less profitable firms, which load negatively on Chinese import growth",
231 + "evidence_type": "table",
232 + "evidence_ref": "Tables IX-X"
233 + },
234 + {
235 + "claim": "The premium is stronger in industries with high trade elasticity and strong business-stealing effects, as the model predicts",
236 + "evidence_type": "table",
237 + "evidence_ref": "Tables XI-XII"
238 + },
239 + {
240 + "claim": "With perfect risk-sharing the model fails; with limited risk-sharing consumption falls after foreign shocks",
241 + "evidence_type": "figure",
242 + "evidence_ref": "Figures 1-3, Table XIV"
243 + }
244 + ],
245 + "structural_summary": "The introduction moves from the mixed domestic effects of globalization to an asset-pricing lens. It previews the shipping-cost measure, the portfolio results and robustness, and the displacement interpretation, which it frames as a puzzle. It then sketches the model, previews the calibration outcome, positions the paper in four literatures, and closes with a roadmap. Section I builds the measure and estimates the premium, with robustness and mispricing checks folded into the results. Section II develops and calibrates a Melitz-type model and tests its cross-sectional predictions. A short conclusion restates the findings and leaves alternative frictions to future work.",
246 + "writing_profile_keywords": [
247 + "portfolio sorts",
248 + "risk premium",
249 + "puzzle framing",
250 + "trade model with asset prices",
251 + "calibration",
252 + "limited risk-sharing",
253 + "shipping costs proxy"
254 + ],
255 + "confidence": 0.8,
256 + "notes": "The empirical part relies on portfolio sorts with factor alphas; reduced_form_regression is the closest method category. The excerpts do not show how significance is reported in the main tables."
257 + },
258 + "deterministic_checks": {
259 + "roadmap_regex": true,
260 + "contribution_regex_in_intro": true,
261 + "magnitude_in_abstract": true,
262 + "abstract_n_sentences": 6,
263 + "intro_n_paragraphs": 14,
264 + "intro_n_words": 1887,
265 + "causal_words_in_abstract": 0,
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": "other",
273 + "2": "theory",
274 + "3": "conclusion"
275 + }
276 + },
277 + "schema_valid": true,
278 + "code_version": "87c73cd-dirty"
279 +}
\ No newline at end of file
added data/annotations/documents/jof_12782.json +258 −0
@@ -0,0 +1,258 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped abstract_moves.notes"
5 + ],
6 + "doc_id": "jof_12782",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:32+00:00",
12 + "source_sha256": "dc086b8118129c5441e2a0f7479305f0e0383a5fbfd92f4f04136099cafd25c9",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 25163,
15 + "usage": {
16 + "input": 8085,
17 + "output": 2145,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.076034,
22 + "request_id": "req_011CggqPEHuN4Fnve9xYkmT2",
23 + "annotation": {
24 + "genre": "survey_or_address",
25 + "subfield": "corporate_finance",
26 + "subfield_secondary": "dynamic capital structure theory; collateral and commitment",
27 + "primary_method": "theoretical_model_analytical",
28 + "identification": {
29 + "estimand": "not applicable (theoretical address)",
30 + "strategy": "Stylized dynamic trade-off model comparing equilibrium leverage policies and shareholder value under full commitment, no commitment, and collateral-based commitment",
31 + "key_assumptions": [
32 + "firm cannot commit to future leverage choices",
33 + "unsecured creditors' claims are nonexclusive",
34 + "net funding cost advantage of debt (tax shields or pricing differentials)",
35 + "frequent, costless trade in securities",
36 + "equity value maximized at all times"
37 + ],
38 + "threats_addressed": [],
39 + "causal_language_strength": 1,
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": false,
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": 3
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "tension_or_puzzle"
73 + },
74 + {
75 + "paragraph_index": 0,
76 + "move": "preview_of_results"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "contribution_statement"
81 + },
82 + {
83 + "paragraph_index": 1,
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": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "related_literature_positioning"
97 + },
98 + {
99 + "paragraph_index": 5,
100 + "move": "contribution_statement"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "policy_or_broader_implication"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "motivation"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "preview_of_approach"
113 + },
114 + {
115 + "paragraph_index": 7,
116 + "move": "preview_of_results"
117 + },
118 + {
119 + "paragraph_index": 9,
120 + "move": "preview_of_results"
121 + },
122 + {
123 + "paragraph_index": 20,
124 + "move": "roadmap"
125 + }
126 + ],
127 + "contribution_paragraph_index": 1,
128 + "contribution_types": [
129 + "new_theory_or_model",
130 + "synthesis"
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": "theory"
147 + },
148 + {
149 + "idx": 3,
150 + "role": "conclusion"
151 + }
152 + ],
153 + "results_presentation": {
154 + "economic_significance_discussed": false,
155 + "statistical_significance_language": "unclear",
156 + "magnitude_statements_in_intro": false
157 + },
158 + "robustness": {
159 + "location": "none_seen",
160 + "types": []
161 + },
162 + "tables": {
163 + "convention_notes": "No tables. Results are shown through equations and numerical model figures whose captions state parameter assumptions.",
164 + "units_stated_in_notes": false,
165 + "standard_errors_description": "none"
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 + "explicit disclaimer that the goal is reframing, not a definitive theory",
180 + "modal verbs (may, likely, could)",
181 + "acknowledging simplified/stylized setting",
182 + "'of course' concessions about alternative mechanisms"
183 + ]
184 + },
185 + "claim_evidence": [
186 + {
187 + "claim": "Without commitment and with frequent trade, leverage policy is indeterminate and does not raise shareholder value, an MM-like irrelevance result",
188 + "evidence_type": "equation_or_proof",
189 + "evidence_ref": "Section I; Figures 4-5"
190 + },
191 + {
192 + "claim": "Under full commitment the firm would rely almost entirely on debt, which counterfactually fits the data",
193 + "evidence_type": "equation_or_proof",
194 + "evidence_ref": "Section I.A; Figure 2"
195 + },
196 + {
197 + "claim": "Secured debt against collateral lets the firm capture the funding cost advantage of leverage",
198 + "evidence_type": "equation_or_proof",
199 + "evidence_ref": "Section II; Figure 8"
200 + },
201 + {
202 + "claim": "Three qualitatively different leverage policies are all equilibrium-optimal and give the same initial equity value",
203 + "evidence_type": "figure",
204 + "evidence_ref": "Figure 1"
205 + },
206 + {
207 + "claim": "Collateral reduces debt overhang and expands the range of shareholder-beneficial investment",
208 + "evidence_type": "figure",
209 + "evidence_ref": "Figure 7"
210 + },
211 + {
212 + "claim": "A regulatory debt cap can induce firms to borrow up to the cap because it acts as external commitment",
213 + "evidence_type": "figure",
214 + "evidence_ref": "Figure 9; Section III.B"
215 + },
216 + {
217 + "claim": "Collateral's role in shielding creditors from dilution builds on prior work on collateral and nonexclusivity",
218 + "evidence_type": "prior_literature",
219 + "evidence_ref": "Donaldson, Gromb, and Piacentino (2017, 2018)"
220 + }
221 + ],
222 + "structural_summary": "Presidential address that recasts dynamic capital structure as a commitment problem. The introduction opens with the unresolved empirical puzzle of leverage dynamics and sets out the thesis. It positions the paper against the collateral literature, lists empirical challenges, and previews key results through a single illustrative figure and bulleted findings, ending with a roadmap. Section I builds a stylized model of cash-flow-based borrowing, moving from deterministic to stochastic cases, and shows that leverage is irrelevant without commitment. Section II adds investment and collateral and shows that secured debt restores the benefits of leverage. Section III discusses alternative commitment mechanisms and regulatory caps, then sets out implications for theory, empirics and policy.",
223 + "writing_profile_keywords": [
224 + "presidential address",
225 + "first-person voice",
226 + "conceptual reframing",
227 + "stylized model",
228 + "numerical illustration figures",
229 + "no empirical tests",
230 + "bulleted result summaries",
231 + "policy implications",
232 + "research agenda"
233 + ],
234 + "confidence": 0.85,
235 + "notes": "Genre set to survey_or_address because the article is a presidential address. Its content is a theoretical model that relies on related work for general results. Introduction paragraph indices after 8 are fragmented bullet lines, and the roadmap sits in paragraph 20."
236 + },
237 + "deterministic_checks": {
238 + "roadmap_regex": true,
239 + "contribution_regex_in_intro": false,
240 + "magnitude_in_abstract": false,
241 + "abstract_n_sentences": 6,
242 + "intro_n_paragraphs": 23,
243 + "intro_n_words": 1452,
244 + "causal_words_in_abstract": 0,
245 + "hedge_words_in_abstract": 1,
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": "other",
253 + "3": "conclusion"
254 + }
255 + },
256 + "schema_valid": true,
257 + "code_version": "87c73cd-dirty"
258 +}
\ No newline at end of file
added data/annotations/documents/jof_12783.json +253 −0
@@ -0,0 +1,253 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12783",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:40+00:00",
9 + "source_sha256": "9b9e2d50d1d94c7055e781d54405e24c8aa0772eadd51ecbaf82201c017a6f81",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 21896,
12 + "usage": {
13 + "input": 7232,
14 + "output": 3008,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.089882,
19 + "request_id": "req_011CggqbAHP9NWTM8AiVrTbS",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "entrepreneurial_finance_private_equity",
23 + "subfield_secondary": "asset_pricing (investor skill / performance evaluation)",
24 + "primary_method": "other",
25 + "identification": {
26 + "estimand": "Cross-sectional variance of a latent LP-specific skill component in private equity fund returns, and the IRR change from a one-standard-deviation increase in that skill.",
27 + "strategy": "Model-free bootstrap tests compare the dispersion of LP persistence and returns with chance. An extended Korteweg–Sorensen Bayesian random-effects model, estimated by MCMC, then quantifies skill. Alternative explanations are probed with LP-type subsamples, public-LP controls, return-distribution quartiles and first-time or follow-on fund samples.",
28 + "key_assumptions": [
29 + "Under the null of no skill, LPs draw returns from a common distribution; the bootstrap restrictions mimic actual investment opportunities.",
30 + "GP abilities are not competed away by fundraising (no strong Berk–Green effect), so GP quality is persistent.",
31 + "Within an LP type, risk preferences are similar.",
32 + "The parametric latent-skill structure and its priors are adequate."
33 + ],
34 + "threats_addressed": [
35 + "Differences in risk tolerance across LPs",
36 + "Political pressure on public pension funds",
37 + "Differential access to high-quality GPs (first-time funds, reinvestment)",
38 + "Chance variation in returns"
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": 2
55 + },
56 + "results": {
57 + "present": true,
58 + "sentence_index": 1
59 + },
60 + "magnitude": {
61 + "present": true,
62 + "sentence_index": 2
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": 1,
81 + "move": "motivation"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 2,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 3,
93 + "move": "identification_defense"
94 + },
95 + {
96 + "paragraph_index": 4,
97 + "move": "identification_defense"
98 + },
99 + {
100 + "paragraph_index": 5,
101 + "move": "related_literature_positioning"
102 + },
103 + {
104 + "paragraph_index": 5,
105 + "move": "contribution_statement"
106 + }
107 + ],
108 + "contribution_paragraph_index": 5,
109 + "contribution_types": [
110 + "new_evidence",
111 + "new_method"
112 + ],
113 + "roadmap_present": false,
114 + "literature_positioning_location": "inside_introduction"
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": "robustness"
136 + },
137 + {
138 + "idx": 5,
139 + "role": "conclusion"
140 + },
141 + {
142 + "idx": 6,
143 + "role": "appendix"
144 + }
145 + ],
146 + "results_presentation": {
147 + "economic_significance_discussed": true,
148 + "statistical_significance_language": "unclear",
149 + "magnitude_statements_in_intro": true
150 + },
151 + "robustness": {
152 + "location": "mixed",
153 + "types": [
154 + "alternative_specification",
155 + "subsample",
156 + "alternative_sample_period",
157 + "other"
158 + ]
159 + },
160 + "tables": {
161 + "convention_notes": "Tables run from summary statistics to bootstrap comparisons of actual versus simulated distributions, then to Bayesian posterior estimates by period and LP type. Notes describe sample restrictions, panels and model variants (odd versus even columns for Model 1/2), and cross-reference variable definitions in Table V.",
162 + "units_stated_in_notes": false,
163 + "standard_errors_description": "unclear"
164 + },
165 + "figures": {
166 + "uncertainty_shown": "unclear"
167 + },
168 + "conclusion": {
169 + "bounded_claims": true,
170 + "policy_implications": false,
171 + "open_questions": false,
172 + "restates_contribution": true
173 + },
174 + "hedging": {
175 + "level": "moderate",
176 + "typical_devices": [
177 + "suggest/suggesting",
178 + "potentially reflects",
179 + "it does not appear that",
180 + "unlikely to be the main factor",
181 + "explicit limitations subsection"
182 + ]
183 + },
184 + "claim_evidence": [
185 + {
186 + "claim": "Dispersion of LP persistence in top-half funds exceeds what bootstrapped no-skill samples would produce.",
187 + "evidence_type": "table",
188 + "evidence_ref": "Table II, Table III, Figure 1"
189 + },
190 + {
191 + "claim": "A one-SD increase in LP skill raises annual IRR by roughly 1-2 pp, and by more in venture funds.",
192 + "evidence_type": "table",
193 + "evidence_ref": "Table V, Figure 2"
194 + },
195 + {
196 + "claim": "The skill effect is stronger in earlier vintages and declines over the sample period.",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table V panels B-D"
199 + },
200 + {
201 + "claim": "Risk preferences do not explain the results: variation persists within LP types, and high-skill LPs outperform across the return distribution.",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table VII, Table VIII, Table X"
204 + },
205 + {
206 + "claim": "Public pension funds do not earn systematically different returns, which argues against political pressure as the driver.",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table IX"
209 + },
210 + {
211 + "claim": "Skill estimates hold in first-time-fund and reinvestment samples, addressing the access explanation.",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table XI, Table XII"
214 + }
215 + ],
216 + "structural_summary": "The paper motivates the question with the economic importance of institutional investors and notes a gap in measuring their skill, focusing on private equity fund selection. After describing a merged LP-investment dataset, it runs model-free bootstrap tests of persistence and return dispersion. It then quantifies skill with an extended Bayesian latent-skill model estimated by MCMC, with a variance decomposition and individual skill estimates. A separate section weighs alternative interpretations (risk tolerance, political pressure, access) and limitations. The conclusion restates findings with hedged interpretation, and an appendix details the MCMC algorithm and priors.",
217 + "writing_profile_keywords": [
218 + "investor skill",
219 + "bootstrap null distribution",
220 + "Bayesian MCMC",
221 + "persistence",
222 + "alternative explanations",
223 + "private equity LPs",
224 + "variance decomposition"
225 + ],
226 + "confidence": 0.8,
227 + "notes": "Primary method is coded 'other' because the design combines bootstrap tests with a Bayesian hierarchical random-effects model (extended Korteweg–Sorensen); structural_estimation is a defensible alternative. Section II is labelled 'theory' by the rule but contains empirical model-free tests, so it is coded as results. No explicit roadmap is visible in the excerpt."
228 + },
229 + "deterministic_checks": {
230 + "roadmap_regex": false,
231 + "contribution_regex_in_intro": false,
232 + "magnitude_in_abstract": false,
233 + "abstract_n_sentences": 5,
234 + "intro_n_paragraphs": 6,
235 + "intro_n_words": 933,
236 + "causal_words_in_abstract": 1,
237 + "hedge_words_in_abstract": 0,
238 + "stars_mentioned": false,
239 + "clustered_se_mentioned": false,
240 + "internet_appendix_mentioned": false,
241 + "section_roles_rule": {
242 + "0": "introduction",
243 + "1": "data",
244 + "2": "theory",
245 + "3": "results",
246 + "4": "other",
247 + "5": "conclusion",
248 + "6": "appendix"
249 + }
250 + },
251 + "schema_valid": true,
252 + "code_version": "87c73cd-dirty"
253 +}
\ No newline at end of file
added data/annotations/documents/jof_12784.json +228 −0
@@ -0,0 +1,228 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12784",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:30+00:00",
9 + "source_sha256": "50340a5bb4f1fffa5d370d9ba2d5fb3990c373d87b8146486189d717a67ed9d9",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 12790,
12 + "usage": {
13 + "input": 4705,
14 + "output": 1858,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.056774,
19 + "request_id": "req_011CggqWrPF5w93NWWWPqFq8",
20 + "annotation": {
21 + "genre": "replication_or_correction",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "financial_econometrics",
24 + "primary_method": "calibration_simulation",
25 + "identification": {
26 + "estimand": "Bias and coverage of FDR estimates of the proportions of zero-, negative- and positive-alpha mutual funds, compared with known true proportions",
27 + "strategy": "Monte Carlo simulation of fund returns from a factor model with data-informed DGPs for alpha; the FDR estimator is applied to each sample and its estimates are compared with the true proportions",
28 + "key_assumptions": [
29 + "Simulated DGPs (alpha distributions, sample sizes, error correlation) are representative of real fund data",
30 + "Linear factor model of returns as in BSW"
31 + ],
32 + "threats_addressed": [
33 + "Discrete alpha distribution with point masses far from zero",
34 + "Sample size, both number of funds and observations per fund",
35 + "Unbalanced panel",
36 + "Cross-sectional error correlation"
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": 1
57 + },
58 + "magnitude": {
59 + "present": true,
60 + "sentence_index": 2
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": "preview_of_approach"
80 + },
81 + {
82 + "paragraph_index": 0,
83 + "move": "preview_of_results"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "mechanism_or_theory_sketch"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "tension_or_puzzle"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "mechanism_or_theory_sketch"
96 + },
97 + {
98 + "paragraph_index": 3,
99 + "move": "preview_of_approach"
100 + }
101 + ],
102 + "contribution_paragraph_index": 0,
103 + "contribution_types": [
104 + "replication_or_correction",
105 + "new_evidence"
106 + ],
107 + "roadmap_present": false,
108 + "literature_positioning_location": "inside_introduction"
109 + },
110 + "section_roles": [
111 + {
112 + "idx": 0,
113 + "role": "introduction"
114 + },
115 + {
116 + "idx": 1,
117 + "role": "theory"
118 + },
119 + {
120 + "idx": 2,
121 + "role": "results"
122 + },
123 + {
124 + "idx": 3,
125 + "role": "conclusion"
126 + }
127 + ],
128 + "results_presentation": {
129 + "economic_significance_discussed": true,
130 + "statistical_significance_language": "confidence_intervals",
131 + "magnitude_statements_in_intro": true
132 + },
133 + "robustness": {
134 + "location": "within_results",
135 + "types": [
136 + "sensitivity_to_parameters",
137 + "alternative_specification"
138 + ]
139 + },
140 + "tables": {
141 + "convention_notes": "Each table reports simulated FDR estimates against true proportions. Rows vary the true group proportions and columns vary the alpha distance or dispersion. Alphas are given as annualized percentages. Results are averaged across 1,000 simulations, and one table reports actual coverage of 95% confidence intervals.",
142 + "units_stated_in_notes": true,
143 + "standard_errors_description": "none"
144 + },
145 + "figures": {
146 + "uncertainty_shown": "unclear"
147 + },
148 + "conclusion": {
149 + "bounded_claims": true,
150 + "policy_implications": false,
151 + "open_questions": false,
152 + "restates_contribution": true
153 + },
154 + "hedging": {
155 + "level": "moderate",
156 + "typical_devices": [
157 + "'can materially change' phrasing",
158 + "'potentially unrealistic' qualifiers",
159 + "'under certain conditions' scoping",
160 + "'question applicability' framing"
161 + ]
162 + },
163 + "claim_evidence": [
164 + {
165 + "claim": "The FDR estimator overestimates the proportion of zero-alpha funds and underestimates nonzero-alpha funds under data-informed DGPs",
166 + "evidence_type": "table",
167 + "evidence_ref": "Tables I, III"
168 + },
169 + {
170 + "claim": "FDR 95% confidence intervals almost never contain the true proportions",
171 + "evidence_type": "table",
172 + "evidence_ref": "Table II"
173 + },
174 + {
175 + "claim": "Accuracy does not improve with more funds; bias stems from low signal-to-noise ratio and low test power",
176 + "evidence_type": "table",
177 + "evidence_ref": "Table IV"
178 + },
179 + {
180 + "claim": "Low-power tests spread alternative p-values across [0,1], so the null proportion is overestimated",
181 + "evidence_type": "figure",
182 + "evidence_ref": "Figure 1"
183 + },
184 + {
185 + "claim": "Bias persists in unbalanced panels and with correlated errors",
186 + "evidence_type": "table",
187 + "evidence_ref": "Tables V, VI"
188 + },
189 + {
190 + "claim": "FDR and 'no-luck' approaches compared across true alphas and proportions",
191 + "evidence_type": "figure",
192 + "evidence_ref": "Figure 2"
193 + }
194 + ],
195 + "structural_summary": "Short comment on an influential prior method. The introduction summarizes the original findings and their wide adoption, then explains intuitively how the method's assumption of high test power can fail in fund data, and previews the expanded simulations. Section I sets out the FDR mechanics. Section II runs simulations that vary the alpha distribution, sample size and error correlation, documenting bias and poor interval coverage. The conclusion attributes the original method's accuracy to unrealistic simulation assumptions and questions its use in low-power settings.",
196 + "writing_profile_keywords": [
197 + "comment",
198 + "simulation-based critique",
199 + "statistical power",
200 + "false discovery rate",
201 + "mutual fund alpha",
202 + "replication"
203 + ],
204 + "confidence": 0.85,
205 + "notes": "Labelled as a corrigendum/replication-type comment. No empirical identification; the evidence is entirely Monte Carlo. The original BSW results are cited from that paper rather than re-estimated."
206 + },
207 + "deterministic_checks": {
208 + "roadmap_regex": false,
209 + "contribution_regex_in_intro": true,
210 + "magnitude_in_abstract": true,
211 + "abstract_n_sentences": 6,
212 + "intro_n_paragraphs": 4,
213 + "intro_n_words": 599,
214 + "causal_words_in_abstract": 0,
215 + "hedge_words_in_abstract": 0,
216 + "stars_mentioned": false,
217 + "clustered_se_mentioned": false,
218 + "internet_appendix_mentioned": false,
219 + "section_roles_rule": {
220 + "0": "introduction",
221 + "1": "other",
222 + "2": "other",
223 + "3": "conclusion"
224 + }
225 + },
226 + "schema_valid": true,
227 + "code_version": "87c73cd-dirty"
228 +}
\ No newline at end of file
added data/annotations/documents/jof_12785.json +254 −0
@@ -0,0 +1,254 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12785",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:47+00:00",
9 + "source_sha256": "063d49a3724bc880ad82e96d83160d30a39aab4e8260f177002cbcd74bbb4914",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24331,
12 + "usage": {
13 + "input": 7817,
14 + "output": 2138,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.074822,
19 + "request_id": "req_011CggqTVVZVWmDnsKQRaApB",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "behavioral_finance",
23 + "subfield_secondary": "asset_pricing; household_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "How differently investors' selling decisions, reinvestment choices, and analysts' price forecasts respond to the dividend component of performance versus the price-change component; time variation in aggregate dividend demand",
27 + "strategy": "Split position performance into a price-change part and a dividend part, then compare selling propensities (disposition and rank effects) under each measure. Benchmark observed holding changes against a mechanical full-reinvestment counterfactual. Regress analyst target-price errors and dividend interim returns on dividend yield, interest rates and past market returns.",
28 + "key_assumptions": [
29 + "Trading responses reveal how investors evaluate performance",
30 + "Absence of share increases matching the dividend amount indicates non-reinvestment",
31 + "The marketwide analysis relies on stronger assumptions drawn from prior literature, which the authors acknowledge"
32 + ],
33 + "threats_addressed": [
34 + "Alternative rational explanations such as taxes and trading costs",
35 + "Alternative behavioral explanations such as self-control",
36 + "Investor heterogeneity, handled with account fixed effects and controls",
37 + "Split of investors into those showing a disposition effect and those not"
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": 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": 6,
76 + "move": "tension_or_puzzle"
77 + },
78 + {
79 + "paragraph_index": 7,
80 + "move": "preview_of_approach"
81 + },
82 + {
83 + "paragraph_index": 7,
84 + "move": "preview_of_results"
85 + },
86 + {
87 + "paragraph_index": 8,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 9,
92 + "move": "preview_of_results"
93 + },
94 + {
95 + "paragraph_index": 10,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 11,
100 + "move": "contribution_statement"
101 + },
102 + {
103 + "paragraph_index": 12,
104 + "move": "related_literature_positioning"
105 + }
106 + ],
107 + "contribution_paragraph_index": 11,
108 + "contribution_types": [
109 + "new_evidence",
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": "results"
127 + },
128 + {
129 + "idx": 3,
130 + "role": "results"
131 + },
132 + {
133 + "idx": 4,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 5,
138 + "role": "results"
139 + },
140 + {
141 + "idx": 6,
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": "mixed",
152 + "types": [
153 + "subsample",
154 + "alternative_measure"
155 + ]
156 + },
157 + "tables": {
158 + "convention_notes": "Each table pits a price-change version against a returns-including-dividends version, with panels by investor type (individuals, mutual funds, institutions). The dependent variable is typically a Sell dummy. Notes define variables and sample periods.",
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": true,
168 + "open_questions": true,
169 + "restates_contribution": true
170 + },
171 + "hedging": {
172 + "level": "moderate",
173 + "typical_devices": [
174 + "'trade as if' framing",
175 + "'consistent with'",
176 + "'appear to'",
177 + "'it seems more likely'",
178 + "explicit admission that the marketwide tests are less cleanly identified"
179 + ]
180 + },
181 + "claim_evidence": [
182 + {
183 + "claim": "Investors' disposition effect is driven by price changes rather than by total returns including dividends.",
184 + "evidence_type": "table",
185 + "evidence_ref": "Table II"
186 + },
187 + {
188 + "claim": "Rank-based selling responds to price-change ranks but little to return ranks.",
189 + "evidence_type": "table",
190 + "evidence_ref": "Table III"
191 + },
192 + {
193 + "claim": "Individuals judge reinvested positions' gain/loss status without including dividends.",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table IV"
196 + },
197 + {
198 + "claim": "Dividend payers are sold less often, and their sales are less sensitive to price changes.",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table V; Figure 1"
201 + },
202 + {
203 + "claim": "Funds and institutions rarely reinvest dividends, which causes portfolio weights of dividend payers to drift downward.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table VI; Figures 2-4"
206 + },
207 + {
208 + "claim": "Analysts' target prices are too optimistic for high-dividend-yield stocks.",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table VII"
211 + },
212 + {
213 + "claim": "Dividend demand rises when interest rates and past market returns are low.",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table VIII; Figure 5"
216 + }
217 + ],
218 + "structural_summary": "The paper opens with a press epigraph and a statement of Miller-Modigliani dividend irrelevance, then argues that the price-dividend tradeoff may not be salient to investors. It previews tests in which performance is split into price-change and dividend components. Results proceed through several sections: trading behavior (disposition, rank and rolled-disposition effects, plus price sensitivity), dividend reinvestment and portfolio drift, analyst forecast errors, and marketwide time-varying dividend demand. The conclusion restates the mental-accounting interpretation, argues rival theories cannot explain the findings together, and ends with an open question about investor education.",
219 + "writing_profile_keywords": [
220 + "mental accounting",
221 + "free dividends fallacy",
222 + "epigraph opening",
223 + "unifying explanation",
224 + "multiple investor types",
225 + "behavioral trading"
226 + ],
227 + "confidence": 0.8,
228 + "notes": "No roadmap paragraph is visible in the excerpts. Intro paragraphs 0-5 are fragments of the epigraph quote. Significance conventions and standard-error treatment are not visible in the excerpted table notes. The paper explicitly labels its marketwide analysis as less cleanly identified."
229 + },
230 + "deterministic_checks": {
231 + "roadmap_regex": false,
232 + "contribution_regex_in_intro": false,
233 + "magnitude_in_abstract": false,
234 + "abstract_n_sentences": 5,
235 + "intro_n_paragraphs": 13,
236 + "intro_n_words": 1306,
237 + "causal_words_in_abstract": 1,
238 + "hedge_words_in_abstract": 0,
239 + "stars_mentioned": false,
240 + "clustered_se_mentioned": false,
241 + "internet_appendix_mentioned": false,
242 + "section_roles_rule": {
243 + "0": "introduction",
244 + "1": "data",
245 + "2": "other",
246 + "3": "other",
247 + "4": "other",
248 + "5": "other",
249 + "6": "conclusion"
250 + }
251 + },
252 + "schema_valid": true,
253 + "code_version": "87c73cd-dirty"
254 +}
\ No newline at end of file
added data/annotations/documents/jof_12786.json +176 −0
@@ -0,0 +1,176 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12786",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:37+00:00",
9 + "source_sha256": "494a7f1f7905e626cbe3dd72f42524711acf04d42498ab422559638b94f2e312",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 3394,
12 + "usage": {
13 + "input": 1240,
14 + "output": 1346,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.032674,
19 + "request_id": "req_011CggqTADwX9bNcbmhS1Tt5",
20 + "annotation": {
21 + "genre": "replication_or_correction",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "return predictability / learning models (corrigendum)",
24 + "primary_method": "none",
25 + "identification": {
26 + "estimand": "not applicable (correction of coding errors in prior paper's predictability estimates)",
27 + "strategy": "not applicable",
28 + "key_assumptions": [],
29 + "threats_addressed": [],
30 + "causal_language_strength": 1,
31 + "uses_causal_claim": false
32 + },
33 + "abstract_moves": {
34 + "question": {
35 + "present": false,
36 + "sentence_index": -1
37 + },
38 + "gap": {
39 + "present": false,
40 + "sentence_index": -1
41 + },
42 + "method": {
43 + "present": false,
44 + "sentence_index": -1
45 + },
46 + "results": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "magnitude": {
51 + "present": false,
52 + "sentence_index": -1
53 + },
54 + "contribution": {
55 + "present": false,
56 + "sentence_index": -1
57 + }
58 + },
59 + "introduction": {
60 + "moves": [
61 + {
62 + "paragraph_index": 0,
63 + "move": "motivation"
64 + },
65 + {
66 + "paragraph_index": 0,
67 + "move": "other"
68 + },
69 + {
70 + "paragraph_index": 5,
71 + "move": "other"
72 + },
73 + {
74 + "paragraph_index": 12,
75 + "move": "preview_of_results"
76 + },
77 + {
78 + "paragraph_index": 12,
79 + "move": "other"
80 + }
81 + ],
82 + "contribution_paragraph_index": -1,
83 + "contribution_types": [
84 + "replication_or_correction"
85 + ],
86 + "roadmap_present": false,
87 + "literature_positioning_location": "none"
88 + },
89 + "section_roles": [
90 + {
91 + "idx": 0,
92 + "role": "introduction"
93 + }
94 + ],
95 + "results_presentation": {
96 + "economic_significance_discussed": false,
97 + "statistical_significance_language": "unclear",
98 + "magnitude_statements_in_intro": true
99 + },
100 + "robustness": {
101 + "location": "none_seen",
102 + "types": []
103 + },
104 + "tables": {
105 + "convention_notes": "No tables are visible in the excerpts. The text says the corrected tables are reported in the erratum, with nontrivial changes underlined.",
106 + "units_stated_in_notes": false,
107 + "standard_errors_description": "none"
108 + },
109 + "figures": {
110 + "uncertainty_shown": "no_figures"
111 + },
112 + "conclusion": {
113 + "bounded_claims": true,
114 + "policy_implications": false,
115 + "open_questions": false,
116 + "restates_contribution": false
117 + },
118 + "hedging": {
119 + "level": "low",
120 + "typical_devices": [
121 + "assertion that the corrections leave the original conclusions unchanged",
122 + "flagging which changes are nontrivial"
123 + ]
124 + },
125 + "claim_evidence": [
126 + {
127 + "claim": "Corrected coding leaves the original paper's statements, findings and conclusions unchanged",
128 + "evidence_type": "none_stated",
129 + "evidence_ref": "Introduction para 0 and 12"
130 + },
131 + {
132 + "claim": "Out-of-sample R2 of the dividend model falls after correction but stays statistically above competing models",
133 + "evidence_type": "table",
134 + "evidence_ref": "Tables III and V (updated)"
135 + },
136 + {
137 + "claim": "Return predictability R2 of the full learning model changes modestly overall, rising in expansions and falling in recessions",
138 + "evidence_type": "table",
139 + "evidence_ref": "Tables IX and XI (updated)"
140 + },
141 + {
142 + "claim": "Data and code used to generate the results are made available",
143 + "evidence_type": "none_stated",
144 + "evidence_ref": "Internet Appendix"
145 + }
146 + ],
147 + "structural_summary": "Short corrigendum with a single unnumbered section. It states that coding inconsistencies were found and claims the paper's conclusions are unaffected. It then lists four specific errors as bullets: mixed overlapping and nonoverlapping estimation, a mis-lagged regressor, an unflagged de-meaning in one figure, and a mis-timed inflation series. Next it reports the nontrivial revised R2 statistics, reaffirms that the main conclusions hold, and points to the Internet Appendix for data and code.",
148 + "writing_profile_keywords": [
149 + "erratum",
150 + "bulleted error list",
151 + "reassurance of unchanged conclusions",
152 + "revised statistics",
153 + "code transparency"
154 + ],
155 + "confidence": 0.85,
156 + "notes": "No abstract, conclusion, tables or figures appear in the excerpts, and paragraph segmentation reflects line breaks rather than true paragraphs. The original paper's methods (Kalman filter, learning model) are mentioned only in passing. Method fields are therefore set to none or not applicable."
157 + },
158 + "deterministic_checks": {
159 + "roadmap_regex": false,
160 + "contribution_regex_in_intro": false,
161 + "magnitude_in_abstract": false,
162 + "abstract_n_sentences": 0,
163 + "intro_n_paragraphs": 13,
164 + "intro_n_words": 342,
165 + "causal_words_in_abstract": 0,
166 + "hedge_words_in_abstract": 0,
167 + "stars_mentioned": false,
168 + "clustered_se_mentioned": false,
169 + "internet_appendix_mentioned": true,
170 + "section_roles_rule": {
171 + "0": "introduction"
172 + }
173 + },
174 + "schema_valid": true,
175 + "code_version": "87c73cd-dirty"
176 +}
\ No newline at end of file
added data/annotations/documents/jof_12817.json +248 −0
@@ -0,0 +1,248 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12817",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:27+00:00",
9 + "source_sha256": "0bc2f57351a517efa2480f45a1a9093ec6db45c92935301f8d802a3a576d6a54",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23743,
12 + "usage": {
13 + "input": 7734,
14 + "output": 2062,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07297,
19 + "request_id": "req_011CggqWQoNfKa6nntCsaErj",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "international_finance",
24 + "primary_method": "other",
25 + "identification": {
26 + "estimand": "Diversification and performance gains, measured by correlations, Sharpe ratios, spanning-test statistics and certainty-equivalent returns, from adding trade-weighted developed-market portfolios that target emerging economies (EE indices) to a developed-market benchmark.",
27 + "strategy": "Build EE indices by weighting developed-market industry portfolios by their exports to each emerging country, then benchmark them against EM-mimicking portfolios using DCC correlations, factor alphas, mean-variance and step-down spanning tests, out-of-sample minimum-variance portfolios, certainty-equivalent rates, and firm-level FactSet Revere reconstructions.",
28 + "key_assumptions": [
29 + "Industry exports to an emerging country proxy for exposure to that country's economic activity.",
30 + "Past return moments are informative about diversification benefits.",
31 + "EM-mimicking portfolios are the appropriate benchmark."
32 + ],
33 + "threats_addressed": [
34 + "Industry-level data may be too coarse (addressed with firm-level data).",
35 + "Comtrade data mix public and private firms.",
36 + "Performance differences driven by alphas (addressed with demeaned spanning tests).",
37 + "In-sample bias (addressed with out-of-sample minimum-variance portfolios)."
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": true,
49 + "sentence_index": 1
50 + },
51 + "method": {
52 + "present": true,
53 + "sentence_index": 2
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": "literature_gap"
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": 2,
88 + "move": "related_literature_positioning"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "contribution_statement"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "robustness_mention"
105 + },
106 + {
107 + "paragraph_index": 6,
108 + "move": "policy_or_broader_implication"
109 + },
110 + {
111 + "paragraph_index": 7,
112 + "move": "roadmap"
113 + }
114 + ],
115 + "contribution_paragraph_index": 2,
116 + "contribution_types": [
117 + "new_method",
118 + "new_evidence",
119 + "new_data_or_measure"
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": "data"
132 + },
133 + {
134 + "idx": 2,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 3,
139 + "role": "robustness"
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": "mixed",
153 + "types": [
154 + "alternative_measure",
155 + "alternative_specification",
156 + "subsample"
157 + ]
158 + },
159 + "tables": {
160 + "convention_notes": "Notes open with 'This table reports...', define the test and benchmark assets, write out the regression models explicitly, and organize content into panels. Comparisons are made across EE, EM-mimicking and MSCI DM portfolios.",
161 + "units_stated_in_notes": false,
162 + "standard_errors_description": "unclear"
163 + },
164 + "figures": {
165 + "uncertainty_shown": "no"
166 + },
167 + "conclusion": {
168 + "bounded_claims": false,
169 + "policy_implications": false,
170 + "open_questions": true,
171 + "restates_contribution": true
172 + },
173 + "hedging": {
174 + "level": "moderate",
175 + "typical_devices": [
176 + "'suggest' for interpreting results",
177 + "'may offer' benefits",
178 + "'we expect' for extensions",
179 + "'potentially' in firm-level framing"
180 + ]
181 + },
182 + "claim_evidence": [
183 + {
184 + "claim": "Listed firms represent a small share of emerging economies.",
185 + "evidence_type": "figure",
186 + "evidence_ref": "Figure 1"
187 + },
188 + {
189 + "claim": "EE portfolio returns load more strongly on emerging-economy activity proxies than EM-mimicking indices do.",
190 + "evidence_type": "table",
191 + "evidence_ref": "Table II"
192 + },
193 + {
194 + "claim": "EE index correlations with developed markets remain low and stable over time.",
195 + "evidence_type": "figure",
196 + "evidence_ref": "Figures 2-3"
197 + },
198 + {
199 + "claim": "EE portfolios have higher Sharpe ratios than DM and EM-mimicking portfolios.",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table III, Figure 4"
202 + },
203 + {
204 + "claim": "EE indices significantly shift the mean-variance frontier.",
205 + "evidence_type": "table",
206 + "evidence_ref": "Tables V-VI"
207 + },
208 + {
209 + "claim": "Firm-level EE indices show low correlations and higher Sharpe ratios.",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table X, Figures 6-7"
212 + }
213 + ],
214 + "structural_summary": "The paper opens by noting that equity indices understate diversification gains because emerging markets' listed sectors are small. It proposes trade-weighted developed-market portfolios (EE indices) as a new route to emerging-economy exposure and positions this against the EM-mimicking approach in prior work. It then validates that EE indices track economic activity. Results present correlations, return moments, alphas, spanning tests, out-of-sample portfolios and certainty equivalents. A firm-level section confirms the findings, and the conclusion restates the 'best of both worlds' message and points to applications to other nontraded assets.",
215 + "writing_profile_keywords": [
216 + "portfolio construction",
217 + "international diversification",
218 + "benchmark comparison",
219 + "spanning tests",
220 + "method generalization",
221 + "numbered findings"
222 + ],
223 + "confidence": 0.8,
224 + "notes": "The primary method is portfolio construction with performance evaluation, coded as 'other'. Statistical significance conventions are not visible in the excerpts. Robustness is spread across Section III and an Internet Appendix."
225 + },
226 + "deterministic_checks": {
227 + "roadmap_regex": true,
228 + "contribution_regex_in_intro": false,
229 + "magnitude_in_abstract": false,
230 + "abstract_n_sentences": 6,
231 + "intro_n_paragraphs": 8,
232 + "intro_n_words": 1179,
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": "other",
241 + "2": "results",
242 + "3": "other",
243 + "4": "conclusion"
244 + }
245 + },
246 + "schema_valid": true,
247 + "code_version": "87c73cd-dirty"
248 +}
\ No newline at end of file
added data/annotations/documents/jof_12818.json +265 −0
@@ -0,0 +1,265 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12818",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:50+00:00",
9 + "source_sha256": "a853565873bb6c9a7eec5f941c4b0bcbb6c5c481a4f37ae75423292b9574470f",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22450,
12 + "usage": {
13 + "input": 8132,
14 + "output": 2184,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.077002,
19 + "request_id": "req_011CggqThnVdAkgQomuh7jLL",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Average excess stock returns by week of the FOMC cycle (even versus odd weeks), and how much of the equity premium is attributable to Fed monetary policy news.",
27 + "strategy": "Regress daily excess returns on FOMC-cycle dummies. Rule out non-Fed explanations with controls for macro releases, earnings and reserve periods. Link the pattern to the Fed through the timing of intermeeting target changes, Fed funds futures changes, Board meeting interactions, Fed-put predictability regressions and changes in equity premium bounds.",
28 + "key_assumptions": [
29 + "The irregular FOMC calendar is unrelated to other news calendars.",
30 + "Board meetings and intermeeting changes proxy for the timing of Fed news.",
31 + "The Martin (2017) bounds measure the equity premium."
32 + ],
33 + "threats_addressed": [
34 + "Macroeconomic data release timing",
35 + "Corporate earnings announcements",
36 + "Reserve maintenance periods",
37 + "Data mining / small sample, via out-of-sample period 2014–2016",
38 + "Formal Fed releases and speeches as an alternative channel"
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": 1
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": 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": 0,
85 + "move": "preview_of_results"
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": "mechanism_or_theory_sketch"
98 + },
99 + {
100 + "paragraph_index": 3,
101 + "move": "preview_of_results"
102 + },
103 + {
104 + "paragraph_index": 4,
105 + "move": "policy_or_broader_implication"
106 + },
107 + {
108 + "paragraph_index": 4,
109 + "move": "preview_of_results"
110 + }
111 + ],
112 + "contribution_paragraph_index": 0,
113 + "contribution_types": [
114 + "new_evidence"
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": "results"
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": "other"
143 + },
144 + {
145 + "idx": 6,
146 + "role": "conclusion"
147 + }
148 + ],
149 + "results_presentation": {
150 + "economic_significance_discussed": true,
151 + "statistical_significance_language": "stars_and_tstats",
152 + "magnitude_statements_in_intro": false
153 + },
154 + "robustness": {
155 + "location": "within_results",
156 + "types": [
157 + "alternative_sample_period",
158 + "subsample",
159 + "other"
160 + ]
161 + },
162 + "tables": {
163 + "convention_notes": "Notes define FOMC-cycle week dummies, give units in percent (0.1 = 10 bps per day), report heteroskedasticity-robust t-statistics in parentheses with stars, state data sources, and include a narrative table of leak examples.",
164 + "units_stated_in_notes": true,
165 + "standard_errors_description": "robust"
166 + },
167 + "figures": {
168 + "uncertainty_shown": "unclear"
169 + },
170 + "conclusion": {
171 + "bounded_claims": true,
172 + "policy_implications": true,
173 + "open_questions": false,
174 + "restates_contribution": true
175 + },
176 + "hedging": {
177 + "level": "moderate",
178 + "typical_devices": [
179 + "'evidence suggests'",
180 + "'make it likely'",
181 + "'appears'",
182 + "'we argue'",
183 + "'suggests that'"
184 + ]
185 + },
186 + "claim_evidence": [
187 + {
188 + "claim": "Since 1994 the equity premium is earned entirely in even weeks of FOMC cycle time.",
189 + "evidence_type": "figure",
190 + "evidence_ref": "Figure 1; Table I"
191 + },
192 + {
193 + "claim": "The pattern is also present in international stock indices.",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table II"
196 + },
197 + {
198 + "claim": "Non-Fed news (macro releases, earnings) does not explain the even-week pattern.",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table III"
201 + },
202 + {
203 + "claim": "Intermeeting target changes and Fed funds futures declines concentrate in even weeks.",
204 + "evidence_type": "table",
205 + "evidence_ref": "Tables IV, V"
206 + },
207 + {
208 + "claim": "The even-week effect is driven disproportionately by days following Board of Governors meetings.",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table VI"
211 + },
212 + {
213 + "claim": "Even-week returns mean-revert after market declines and predict target changes (Fed put).",
214 + "evidence_type": "table",
215 + "evidence_ref": "Tables VII–IX; Figure 4"
216 + },
217 + {
218 + "claim": "Even-week reductions in the equity premium account for most of the high even-week returns.",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table X; Figure 5"
221 + },
222 + {
223 + "claim": "Official Fed releases and speeches do not account for the even-week returns; informal communication does.",
224 + "evidence_type": "table",
225 + "evidence_ref": "Tables XI, XII; Figure 6"
226 + }
227 + ],
228 + "structural_summary": "Opens with questions about the Fed's impact on stocks and notes that prior work focuses on announcement days. Section I presents the new fact (an even-week pattern), with statistical significance, out-of-sample checks, trading strategies, international evidence and cross-sectional evidence. Section II rules out non-Fed explanations, then ties the pattern to the Fed through several pieces of evidence. Section III argues the mechanism runs through reductions in the equity premium (a downside-risk channel). Section IV examines formal versus informal communication channels and Fed motives. The paper closes with a policy discussion and a conclusion.",
229 + "writing_profile_keywords": [
230 + "new stylized fact",
231 + "calendar-time anomaly",
232 + "monetary policy and stocks",
233 + "narrative evidence",
234 + "Fed put",
235 + "informal communication",
236 + "causal linking via multiple evidence pieces"
237 + ],
238 + "confidence": 0.8,
239 + "notes": "Causal claims rest on a triangulation of several pieces of evidence rather than a single quasi-experiment. Section IV's rule label of 'robustness' is overridden here as results, since it reports channel analysis rather than robustness checks."
240 + },
241 + "deterministic_checks": {
242 + "roadmap_regex": false,
243 + "contribution_regex_in_intro": false,
244 + "magnitude_in_abstract": false,
245 + "abstract_n_sentences": 4,
246 + "intro_n_paragraphs": 5,
247 + "intro_n_words": 818,
248 + "causal_words_in_abstract": 0,
249 + "hedge_words_in_abstract": 0,
250 + "stars_mentioned": true,
251 + "clustered_se_mentioned": false,
252 + "internet_appendix_mentioned": false,
253 + "section_roles_rule": {
254 + "0": "introduction",
255 + "1": "other",
256 + "2": "other",
257 + "3": "other",
258 + "4": "robustness",
259 + "5": "conclusion",
260 + "6": "conclusion"
261 + }
262 + },
263 + "schema_valid": true,
264 + "code_version": "87c73cd-dirty"
265 +}
\ No newline at end of file
added data/annotations/documents/jof_12819.json +295 −0
@@ -0,0 +1,295 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped figures.notes",
5 + "dropped figures.uncertainty_note"
6 + ],
7 + "doc_id": "jof_12819",
8 + "ontology_version": "1.0",
9 + "prompt_version": "1.0",
10 + "model": "claude-opus-5-5",
11 + "effort": "medium",
12 + "annotated_at": "2026-11-03T23:26:32+00:00",
13 + "source_sha256": "a4a309539cb09f8728fc6419a11c9e6731aaf907e1d29e51cbe1e90f8ab8c399",
14 + "parser": "pymupdf-jof-heuristics-1.3",
15 + "payload_chars": 28584,
16 + "usage": {
17 + "input": 9421,
18 + "output": 2272,
19 + "cache_read": 3968,
20 + "cache_write": 0
21 + },
22 + "cost_usd": 0.083918,
23 + "request_id": "req_011CggqVTdFnMLSuv1RpheyD",
24 + "annotation": {
25 + "genre": "empirical",
26 + "subfield": "household_finance",
27 + "subfield_secondary": "behavioral_finance",
28 + "primary_method": "reduced_form_regression",
29 + "identification": {
30 + "estimand": "Effect of locally experienced house price changes and volatility, and of personal unemployment, on individuals' expectations about national house prices and unemployment",
31 + "strategy": "Cross-sectional regressions of survey expectations on local (ZIP/MSA/state) house price experiences with time fixed effects and demographic controls; within-individual panel variation in employment status for the unemployment experience",
32 + "key_assumptions": [
33 + "Local house price variation is not correlated with unobserved determinants of national expectations beyond controls and time fixed effects",
34 + "Job transitions within the panel are not driven by contemporaneous changes in beliefs about aggregate unemployment"
35 + ],
36 + "threats_addressed": [
37 + "common aggregate shocks absorbed by time fixed effects",
38 + "measurement error in local house price experience (simulation)",
39 + "multicollinearity across geographic levels and lags (lasso)",
40 + "optimal use of limited information (informativeness tests)",
41 + "risk-adjusted probability reporting (homeownership heterogeneity)",
42 + "recall versus actual price changes"
43 + ],
44 + "causal_language_strength": 4,
45 + "uses_causal_claim": true
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": 0
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": 0
71 + }
72 + },
73 + "introduction": {
74 + "moves": [
75 + {
76 + "paragraph_index": 0,
77 + "move": "motivation"
78 + },
79 + {
80 + "paragraph_index": 0,
81 + "move": "literature_gap"
82 + },
83 + {
84 + "paragraph_index": 0,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "preview_of_approach"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "preview_of_results"
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": "contribution_statement"
106 + },
107 + {
108 + "paragraph_index": 3,
109 + "move": "related_literature_positioning"
110 + },
111 + {
112 + "paragraph_index": 3,
113 + "move": "policy_or_broader_implication"
114 + }
115 + ],
116 + "contribution_paragraph_index": 3,
117 + "contribution_types": [
118 + "new_evidence",
119 + "new_data_or_measure"
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": "data"
132 + },
133 + {
134 + "idx": 2,
135 + "role": "empirical_design"
136 + },
137 + {
138 + "idx": 3,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 4,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 5,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 6,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 7,
155 + "role": "results"
156 + },
157 + {
158 + "idx": 8,
159 + "role": "conclusion"
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 + "idx": 12,
175 + "role": "appendix"
176 + }
177 + ],
178 + "results_presentation": {
179 + "economic_significance_discussed": true,
180 + "statistical_significance_language": "stars_and_tstats",
181 + "magnitude_statements_in_intro": true
182 + },
183 + "robustness": {
184 + "location": "mixed",
185 + "types": [
186 + "alternative_measure",
187 + "alternative_specification",
188 + "subsample",
189 + "other"
190 + ]
191 + },
192 + "tables": {
193 + "convention_notes": "Table notes refer each specification to the paper's baseline regression equation. They state the dependent variable and its units (percentage points), the geographic level of the experience measure in each column, the fixed effects and controls included, and the clustering level. Significance stars are used, with standard errors shown in parentheses.",
194 + "units_stated_in_notes": true,
195 + "standard_errors_description": "clustered"
196 + },
197 + "figures": {
198 + "uncertainty_shown": "unclear"
199 + },
200 + "conclusion": {
201 + "bounded_claims": false,
202 + "policy_implications": false,
203 + "open_questions": false,
204 + "restates_contribution": true
205 + },
206 + "hedging": {
207 + "level": "moderate",
208 + "typical_devices": [
209 + "'suggests' for mechanism interpretations",
210 + "'plausible foundation' and 'one possible explanation' for broader implications",
211 + "'unlikely to explain' when ruling out alternatives"
212 + ]
213 + },
214 + "claim_evidence": [
215 + {
216 + "claim": "Prior-year local house price changes raise expected U.S. house price changes",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table II, Figure 1"
219 + },
220 + {
221 + "claim": "Personally experiencing unemployment raises the stated likelihood of higher U.S. unemployment by about 1.44 pp",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table VII"
224 + },
225 + {
226 + "claim": "Higher experienced local price volatility is associated with wider expected national price distributions",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table VI"
229 + },
230 + {
231 + "claim": "The extent of extrapolation does not depend on how informative local prices are for national prices",
232 + "evidence_type": "table",
233 + "evidence_ref": "Tables III, V; Figure 2"
234 + },
235 + {
236 + "claim": "Less sophisticated respondents extrapolate more",
237 + "evidence_type": "table",
238 + "evidence_ref": "Tables VIII, X"
239 + },
240 + {
241 + "claim": "Domain-specific experiences do not shift expectations about other aggregate outcomes",
242 + "evidence_type": "table",
243 + "evidence_ref": "Table XI"
244 + },
245 + {
246 + "claim": "Elicited expectations predict job loss and attitudes toward housing investment",
247 + "evidence_type": "table",
248 + "evidence_ref": "Tables XII, XIII"
249 + }
250 + ],
251 + "structural_summary": "The introduction motivates the paper with prior work on experience-based expectations and poses questions about what counts as a relevant personal experience. It then previews two designs: local house-price variation and within-person job transitions. Data are described next, followed by a framework separating full-information from limited-information interpretations. House-price results come first, covering prior-year effects, informativeness, geographic levels, weighted histories and volatility. Unemployment results follow, then heterogeneity used to adjudicate between mechanisms, then cross-domain expectations, and finally a check that expectations predict outcomes. The conclusion restates the findings and three contributions. Appendices provide a measurement-error simulation, lasso estimation and recall robustness.",
252 + "writing_profile_keywords": [
253 + "survey expectations",
254 + "extrapolation",
255 + "experience effects",
256 + "mechanism discrimination",
257 + "panel within-person variation",
258 + "heterogeneity by sophistication",
259 + "house prices",
260 + "unemployment"
261 + ],
262 + "confidence": 0.8,
263 + "notes": "The abstract does not report magnitudes; the intro reports the 1.44 pp unemployment effect. Figure captions are truncated, so whether figures show uncertainty is unclear. No roadmap paragraph appears in the visible intro excerpts."
264 + },
265 + "deterministic_checks": {
266 + "roadmap_regex": false,
267 + "contribution_regex_in_intro": true,
268 + "magnitude_in_abstract": false,
269 + "abstract_n_sentences": 6,
270 + "intro_n_paragraphs": 4,
271 + "intro_n_words": 755,
272 + "causal_words_in_abstract": 1,
273 + "hedge_words_in_abstract": 0,
274 + "stars_mentioned": true,
275 + "clustered_se_mentioned": true,
276 + "internet_appendix_mentioned": true,
277 + "section_roles_rule": {
278 + "0": "introduction",
279 + "1": "data",
280 + "2": "other",
281 + "3": "other",
282 + "4": "other",
283 + "5": "other",
284 + "6": "other",
285 + "7": "other",
286 + "8": "conclusion",
287 + "9": "appendix",
288 + "10": "appendix",
289 + "11": "appendix",
290 + "12": "appendix"
291 + }
292 + },
293 + "schema_valid": true,
294 + "code_version": "87c73cd-dirty"
295 +}
\ No newline at end of file
added data/annotations/documents/jof_12820.json +282 −0
@@ -0,0 +1,282 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped abstract_moves.other"
5 + ],
6 + "doc_id": "jof_12820",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:32+00:00",
12 + "source_sha256": "7066ae292ab0c96d2fbbeedd487c2529bbbbb76bfd302df6d759cbcedaffccd6",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 26456,
15 + "usage": {
16 + "input": 9214,
17 + "output": 2253,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.08271,
22 + "request_id": "req_011CggqWjkPiLmRJbaqij4mP",
23 + "annotation": {
24 + "genre": "mixed_theory_empirical",
25 + "subfield": "corporate_finance",
26 + "subfield_secondary": "financial_intermediation_banking",
27 + "primary_method": "reduced_form_regression",
28 + "identification": {
29 + "estimand": "Effect of issuer-paid rating optimism (S&P rating above investor-paid EJR rating) on secondary-market bond spreads and on the probability of choosing private over public debt",
30 + "strategy": "Validation of the overrating measure through default and incentive-conflict regressions, hazard models and a DiD on broker mergers/closures. Symmetric one-notch split-rating comparisons of bond spreads. 2SLS using analyst coverage shocks. Becker-Ivashina conditional-issuance approach to debt choice. Spread changes around private debt issuances.",
31 + "key_assumptions": [
32 + "Without inflation, S&P and EJR ratings are equally informative about default",
33 + "Brokerage mergers/closures exogenously raise S&P's leeway to inflate ratings",
34 + "Conditioning on new debt issuance holds the firm's demand for debt fixed",
35 + "Rating differences reflect incentive conflicts rather than structural methodological differences"
36 + ],
37 + "threats_addressed": [
38 + "Structural differences between agencies' rating methods",
39 + "Selection in primary-market pricing, addressed by using secondary markets",
40 + "Endogeneity of ratings inflation, addressed by the exogenous analyst-coverage shock",
41 + "Confounding of demand and supply in debt choice"
42 + ],
43 + "causal_language_strength": 3,
44 + "uses_causal_claim": true
45 + },
46 + "abstract_moves": {
47 + "question": {
48 + "present": false,
49 + "sentence_index": -1
50 + },
51 + "gap": {
52 + "present": true,
53 + "sentence_index": 0
54 + },
55 + "method": {
56 + "present": false,
57 + "sentence_index": -1
58 + },
59 + "results": {
60 + "present": true,
61 + "sentence_index": 1
62 + },
63 + "magnitude": {
64 + "present": false,
65 + "sentence_index": -1
66 + },
67 + "contribution": {
68 + "present": true,
69 + "sentence_index": 3
70 + }
71 + },
72 + "introduction": {
73 + "moves": [
74 + {
75 + "paragraph_index": 0,
76 + "move": "motivation"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "literature_gap"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_results"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 1,
92 + "move": "preview_of_approach"
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": "identification_defense"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "preview_of_approach"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "mechanism_or_theory_sketch"
113 + },
114 + {
115 + "paragraph_index": 7,
116 + "move": "preview_of_results"
117 + },
118 + {
119 + "paragraph_index": 8,
120 + "move": "preview_of_results"
121 + },
122 + {
123 + "paragraph_index": 9,
124 + "move": "related_literature_positioning"
125 + }
126 + ],
127 + "contribution_paragraph_index": 9,
128 + "contribution_types": [
129 + "new_evidence",
130 + "new_theory_or_model"
131 + ],
132 + "roadmap_present": false,
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": "empirical_design"
147 + },
148 + {
149 + "idx": 3,
150 + "role": "results"
151 + },
152 + {
153 + "idx": 4,
154 + "role": "results"
155 + },
156 + {
157 + "idx": 5,
158 + "role": "conclusion"
159 + },
160 + {
161 + "idx": 6,
162 + "role": "appendix"
163 + },
164 + {
165 + "idx": 7,
166 + "role": "appendix"
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 + ]
183 + },
184 + "tables": {
185 + "convention_notes": "Table notes state the model type (linear probability, OLS, Cox hazard, 2SLS second stage), the dependent variable definition, the sample panel and period (1999-2012), and event windows. Some coefficients are rescaled by 100 for exposition. Table I and Table AI tabulate model outcomes.",
186 + "units_stated_in_notes": true,
187 + "standard_errors_description": "unclear"
188 + },
189 + "figures": {
190 + "uncertainty_shown": "unclear"
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 + "results suggest",
202 + "consistent with",
203 + "evidence is more consistent with",
204 + "may lead"
205 + ]
206 + },
207 + "claim_evidence": [
208 + {
209 + "claim": "S&P overrating is more frequent for issuers contributing more rating revenue and with stronger rating-shopping incentives",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table IV"
212 + },
213 + {
214 + "claim": "S&P converges toward EJR faster when it was initially more optimistic, indicating lower timeliness",
215 + "evidence_type": "table",
216 + "evidence_ref": "Tables V-VI"
217 + },
218 + {
219 + "claim": "Loss of analyst coverage from broker mergers or closures raises the likelihood of S&P overrating",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table VII"
222 + },
223 + {
224 + "claim": "Bonds of firms S&P overrates carry higher spreads than symmetric split-rated counterparts",
225 + "evidence_type": "table",
226 + "evidence_ref": "Tables VIII-IX"
227 + },
228 + {
229 + "claim": "Overrated firms are more likely to issue private debt than public bonds",
230 + "evidence_type": "table",
231 + "evidence_ref": "Table X"
232 + },
233 + {
234 + "claim": "The overrating spread premium shrinks after private debt issuance",
235 + "evidence_type": "table",
236 + "evidence_ref": "Tables XI-XII"
237 + },
238 + {
239 + "claim": "Inflation in issuer-paid ratings pools risks and worsens information asymmetry even when investors price it",
240 + "evidence_type": "equation_or_proof",
241 + "evidence_ref": "Section I, Appendix A"
242 + }
243 + ],
244 + "structural_summary": "Opens with the role of rating agencies and prior evidence on issuer-pay bias. A simple model (proofs in the appendix) motivates the tests. A long validation section establishes S&P-minus-EJR overrating as a quality measure, using default regressions, conflict proxies, convergence dynamics and a broker-merger DiD. The paper then tests bond-spread pricing via symmetric split ratings and 2SLS, followed by debt-choice substitution and spread changes after private debt issuance. A short conclusion restates the findings.",
245 + "writing_profile_keywords": [
246 + "model-motivated empirics",
247 + "measure validation",
248 + "split ratings",
249 + "quasi-natural experiment",
250 + "debt choice",
251 + "lemons discount"
252 + ],
253 + "confidence": 0.8,
254 + "notes": "Standard error type and significance notation are not visible in the excerpts. Robustness checks with alternative quality measures are mentioned only briefly in the introduction."
255 + },
256 + "deterministic_checks": {
257 + "roadmap_regex": false,
258 + "contribution_regex_in_intro": true,
259 + "magnitude_in_abstract": false,
260 + "abstract_n_sentences": 4,
261 + "intro_n_paragraphs": 10,
262 + "intro_n_words": 1556,
263 + "causal_words_in_abstract": 0,
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": "theory",
271 + "2": "other",
272 + "3": "other",
273 + "4": "other",
274 + "5": "conclusion",
275 + "6": "appendix",
276 + "7": "appendix",
277 + "8": "appendix"
278 + }
279 + },
280 + "schema_valid": true,
281 + "code_version": "87c73cd-dirty"
282 +}
\ No newline at end of file
added data/annotations/documents/jof_12821.json +296 −0
@@ -0,0 +1,296 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12821",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:08:56+00:00",
9 + "source_sha256": "8ef34f94d1cb69fc5e1735f56a2dd14afec7ff5be98e68a43faa7ffd8948b87f",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 32022,
12 + "usage": {
13 + "input": 10593,
14 + "output": 2469,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.092546,
19 + "request_id": "req_011CggqU1ZKgq8Gy1WyCxH3S",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "instrumental_variables",
25 + "identification": {
26 + "estimand": "Causal effect of a foreclosure on a resetting ARM on default and foreclosure probability among neighboring properties within 0.10 miles (foreclosure spillover)",
27 + "strategy": "IV/2SLS in which the within-month interest rate shock at ARM reset, driven by index choice (LIBOR vs Treasury) and lookback period, instruments for the foreclosure of the resetting loan; first stage, reduced form and IV estimates are contrasted with OLS",
28 + "key_assumptions": [
29 + "Within-month reset rate variation from index and lookback terms is unrelated to other determinants of loan performance",
30 + "Exclusion restriction: the resetting loan's rate shock affects neighbors' defaults only through its foreclosure",
31 + "Neighboring loans have different contract terms, so they are not directly hit by the same shock"
32 + ],
33 + "threats_addressed": [
34 + "Common local shocks such as a plant closure, addressed by showing effects are concentrated at very small distances",
35 + "Pre-existing neighborhood distress, addressed with a placebo on prior foreclosures",
36 + "Borrower selection into index type",
37 + "Demand versus supply in refinancing, addressed with mortgage inquiries"
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": 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": 1,
76 + "move": "mechanism_or_theory_sketch"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "literature_gap"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "contribution_statement"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "preview_of_approach"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "identification_defense"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 3,
100 + "move": "related_literature_positioning"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "preview_of_approach"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "identification_defense"
109 + },
110 + {
111 + "paragraph_index": 5,
112 + "move": "preview_of_results"
113 + },
114 + {
115 + "paragraph_index": 6,
116 + "move": "preview_of_results"
117 + },
118 + {
119 + "paragraph_index": 7,
120 + "move": "preview_of_results"
121 + },
122 + {
123 + "paragraph_index": 8,
124 + "move": "preview_of_results"
125 + },
126 + {
127 + "paragraph_index": 9,
128 + "move": "preview_of_results"
129 + },
130 + {
131 + "paragraph_index": 10,
132 + "move": "related_literature_positioning"
133 + },
134 + {
135 + "paragraph_index": 11,
136 + "move": "roadmap"
137 + }
138 + ],
139 + "contribution_paragraph_index": 1,
140 + "contribution_types": [
141 + "new_identification",
142 + "new_evidence",
143 + "new_data_or_measure"
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": "data"
156 + },
157 + {
158 + "idx": 2,
159 + "role": "results"
160 + },
161 + {
162 + "idx": 3,
163 + "role": "results"
164 + },
165 + {
166 + "idx": 4,
167 + "role": "results"
168 + },
169 + {
170 + "idx": 5,
171 + "role": "robustness"
172 + },
173 + {
174 + "idx": 6,
175 + "role": "conclusion"
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": "mixed",
185 + "types": [
186 + "placebo",
187 + "subsample",
188 + "falsification",
189 + "alternative_measure"
190 + ]
191 + },
192 + "tables": {
193 + "convention_notes": "Table notes state the regression equations (first stage, reduced form, 2SLS) and describe the samples (Deeds vs BlackBox). Heterogeneity and placebo tables reuse the main IV specification. Summary statistics are split by index type.",
194 + "units_stated_in_notes": true,
195 + "standard_errors_description": "unclear"
196 + },
197 + "figures": {
198 + "uncertainty_shown": "unclear"
199 + },
200 + "conclusion": {
201 + "bounded_claims": true,
202 + "policy_implications": true,
203 + "open_questions": false,
204 + "restates_contribution": true
205 + },
206 + "hedging": {
207 + "level": "moderate",
208 + "typical_devices": [
209 + "consistent with",
210 + "suggest",
211 + "likely reflects",
212 + "may help to explain",
213 + "potentially point to",
214 + "highly unlikely ex ante"
215 + ]
216 + },
217 + "claim_evidence": [
218 + {
219 + "claim": "A 100bp higher reset rate raises the 12-month foreclosure probability by about 2.5 points against an 8% baseline",
220 + "evidence_type": "table",
221 + "evidence_ref": "Table II; Figure 6"
222 + },
223 + {
224 + "claim": "Each instrumented foreclosure raises neighbor default probability by about 2.1%, or 0.3-0.6 additional foreclosures",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table II, Table IV"
227 + },
228 + {
229 + "claim": "LIBOR-Treasury spread and lookback volatility produce sizable within-month payment differences",
230 + "evidence_type": "figure",
231 + "evidence_ref": "Figures 1-4"
232 + },
233 + {
234 + "claim": "Neighboring sale prices fall about 1-3% after a resetting ARM forecloses",
235 + "evidence_type": "figure",
236 + "evidence_ref": "Figure 7"
237 + },
238 + {
239 + "claim": "Neighbor refinancing falls by one-third while mortgage inquiries do not, indicating a bank supply response",
240 + "evidence_type": "table",
241 + "evidence_ref": "Table III; Figure 8"
242 + },
243 + {
244 + "claim": "Spillovers are strongest in areas with few prior foreclosures and are concentrated at short distances",
245 + "evidence_type": "table",
246 + "evidence_ref": "Table VI; Figure 10"
247 + },
248 + {
249 + "claim": "Tract-level foreclosure activity raises mortgage denials, citing collateral reasons",
250 + "evidence_type": "table",
251 + "evidence_ref": "Table VII; Figure 12"
252 + },
253 + {
254 + "claim": "Placebo: the instrument does not predict prior neighbor foreclosures",
255 + "evidence_type": "table",
256 + "evidence_ref": "Table IX"
257 + }
258 + ],
259 + "structural_summary": "The paper opens with crisis-scale motivation and lists candidate spillover channels. It then introduces two contract-detail instruments, previews first-stage, reduced-form and IV magnitudes, and walks through mechanism results (prices, refinancing supply, borrower learning) before aggregate HMDA evidence. Literature positioning sits inside the introduction, followed by a roadmap. The body covers data and strategy, main IV results, mechanisms, macro channels, and a short placebo robustness section. The conclusion restates the findings, favours an informational interpretation, and closes with bank-run analogies and policy implications.",
260 + "writing_profile_keywords": [
261 + "instrumental variables",
262 + "contract-detail instruments",
263 + "peer effects",
264 + "foreclosure spillovers",
265 + "mechanism decomposition",
266 + "single-author first person",
267 + "geocoded administrative data"
268 + ],
269 + "confidence": 0.8,
270 + "notes": "The roadmap labels Section V as the conclusion, but in the section list Section V is Robustness and VI is the Conclusion. The abstract is truncated and has no explicit gap or contribution sentence. Standard error treatment is not visible in the excerpts. Mechanism and macro sections were classified as results despite the rule-based robustness label."
271 + },
272 + "deterministic_checks": {
273 + "roadmap_regex": true,
274 + "contribution_regex_in_intro": true,
275 + "magnitude_in_abstract": true,
276 + "abstract_n_sentences": 4,
277 + "intro_n_paragraphs": 12,
278 + "intro_n_words": 1773,
279 + "causal_words_in_abstract": 1,
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": "data",
287 + "2": "other",
288 + "3": "robustness",
289 + "4": "robustness",
290 + "5": "robustness",
291 + "6": "conclusion"
292 + }
293 + },
294 + "schema_valid": true,
295 + "code_version": "87c73cd-dirty"
296 +}
\ No newline at end of file
added data/annotations/documents/jof_12822.json +276 −0
@@ -0,0 +1,276 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12822",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:06+00:00",
9 + "source_sha256": "43c2cbd5c198bb30da33fa802eb969ea8752d591dc233d8edc1f31d31f41c086",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 24390,
12 + "usage": {
13 + "input": 7671,
14 + "output": 2139,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.074258,
19 + "request_id": "req_011CggqUmKbfb79TRYA2e7De",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "behavioral_finance",
23 + "subfield_secondary": "household_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "How the inviter-invitee network is structured and how fast the idea spreads through it; how invested amounts (relative to income) relate to sponsor-investor differences in characteristics",
27 + "strategy": "Direct observation of word-of-mouth transmission through invitation-only membership, with power-law fitting and comparison to a simulated Poisson random network plus a calibrated Ponzi model; cross-sectional OLS with personal controls and inviter fixed effects",
28 + "key_assumptions": [
29 + "New investors learned of the scheme only from their inviter, since no public information existed",
30 + "Police interview data are free of the usual reporting and selection biases",
31 + "The simulated random network matches the real network on every dimension except degree distribution"
32 + ],
33 + "threats_addressed": [
34 + "survey reporting and selection bias",
35 + "extreme tails of invested amounts (5% trimming)",
36 + "sponsor-level unobservables via inviter fixed effects"
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": 0
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 1,
75 + "move": "preview_of_approach"
76 + },
77 + {
78 + "paragraph_index": 2,
79 + "move": "other"
80 + },
81 + {
82 + "paragraph_index": 3,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 3,
87 + "move": "identification_defense"
88 + },
89 + {
90 + "paragraph_index": 4,
91 + "move": "literature_gap"
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": "preview_of_results"
104 + },
105 + {
106 + "paragraph_index": 7,
107 + "move": "preview_of_results"
108 + },
109 + {
110 + "paragraph_index": 8,
111 + "move": "preview_of_results"
112 + },
113 + {
114 + "paragraph_index": 9,
115 + "move": "contribution_statement"
116 + },
117 + {
118 + "paragraph_index": 10,
119 + "move": "mechanism_or_theory_sketch"
120 + },
121 + {
122 + "paragraph_index": 11,
123 + "move": "related_literature_positioning"
124 + },
125 + {
126 + "paragraph_index": 12,
127 + "move": "related_literature_positioning"
128 + },
129 + {
130 + "paragraph_index": 13,
131 + "move": "related_literature_positioning"
132 + },
133 + {
134 + "paragraph_index": 14,
135 + "move": "related_literature_positioning"
136 + },
137 + {
138 + "paragraph_index": 15,
139 + "move": "roadmap"
140 + }
141 + ],
142 + "contribution_paragraph_index": 9,
143 + "contribution_types": [
144 + "new_evidence",
145 + "new_data_or_measure"
146 + ],
147 + "roadmap_present": true,
148 + "literature_positioning_location": "inside_introduction"
149 + },
150 + "section_roles": [
151 + {
152 + "idx": 0,
153 + "role": "introduction"
154 + },
155 + {
156 + "idx": 1,
157 + "role": "data"
158 + },
159 + {
160 + "idx": 2,
161 + "role": "results"
162 + },
163 + {
164 + "idx": 3,
165 + "role": "results"
166 + },
167 + {
168 + "idx": 4,
169 + "role": "results"
170 + },
171 + {
172 + "idx": 5,
173 + "role": "conclusion"
174 + }
175 + ],
176 + "results_presentation": {
177 + "economic_significance_discussed": true,
178 + "statistical_significance_language": "unclear",
179 + "magnitude_statements_in_intro": true
180 + },
181 + "robustness": {
182 + "location": "none_seen",
183 + "types": []
184 + },
185 + "tables": {
186 + "convention_notes": "Long descriptive notes define the panels and variables. The early tables are descriptive comparisons of sponsors versus nonsponsors and the actual versus simulated network; later tables report cross-sectional OLS of invested amount scaled by income. Currency (euros, thousands) and measurement dates are stated.",
187 + "units_stated_in_notes": true,
188 + "standard_errors_description": "unclear"
189 + },
190 + "figures": {
191 + "uncertainty_shown": "unclear"
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 + "suggests",
203 + "consistent with",
204 + "may be",
205 + "indicates"
206 + ]
207 + },
208 + "claim_evidence": [
209 + {
210 + "claim": "The number of sponsored investors per sponsor approximately follows a power law, i.e., the network is scale-free",
211 + "evidence_type": "figure",
212 + "evidence_ref": "Figure 4; KS tests"
213 + },
214 + {
215 + "claim": "The scale-free network spreads the idea far faster than a random network (15 vs 161 steps)",
216 + "evidence_type": "table",
217 + "evidence_ref": "Table III"
218 + },
219 + {
220 + "claim": "The cumulative number of investors follows an S-curve in network distance",
221 + "evidence_type": "figure",
222 + "evidence_ref": "Figure 5"
223 + },
224 + {
225 + "claim": "Investors invest more when their sponsor has relatively higher age, education, and income",
226 + "evidence_type": "table",
227 + "evidence_ref": "Table VIII"
228 + },
229 + {
230 + "claim": "The marginal effect of sponsor income peaks just above the investor's own income",
231 + "evidence_type": "figure",
232 + "evidence_ref": "Figure 7"
233 + },
234 + {
235 + "claim": "Sponsors differ from nonsponsors in personal characteristics",
236 + "evidence_type": "table",
237 + "evidence_ref": "Table V"
238 + }
239 + ],
240 + "structural_summary": "The paper opens with Shiller's social-epidemic view and introduces a unique Ponzi-scheme dataset from police investigation documents. Because entry was invitation-only, word-of-mouth transmission is directly observable, and the introduction defends the data quality. It previews network results (power law, simulation comparison, S-curve) and then investment-behavior regressions. Contributions are tied to the epidemic-diffusion, network-modeling, and peer-effects literatures before a roadmap. The body proceeds from background and data, to network analysis with simulation and calibration, to sponsor characteristics, to OLS regressions of invested amounts. The conclusion restates the finding that diffusion runs through a few hubs and notes welfare implications.",
241 + "writing_profile_keywords": [
242 + "unique data",
243 + "social network",
244 + "scale-free",
245 + "Ponzi scheme",
246 + "word-of-mouth",
247 + "simulation comparison",
248 + "descriptive-heavy"
249 + ],
250 + "confidence": 0.8,
251 + "notes": "Robustness checks are not visible in the excerpts, though an Internet Appendix exists. The regression analysis is associational. Section 1 is labeled literature by the rule but functions as background and data."
252 + },
253 + "deterministic_checks": {
254 + "roadmap_regex": true,
255 + "contribution_regex_in_intro": true,
256 + "magnitude_in_abstract": false,
257 + "abstract_n_sentences": 4,
258 + "intro_n_paragraphs": 16,
259 + "intro_n_words": 1610,
260 + "causal_words_in_abstract": 0,
261 + "hedge_words_in_abstract": 1,
262 + "stars_mentioned": false,
263 + "clustered_se_mentioned": false,
264 + "internet_appendix_mentioned": false,
265 + "section_roles_rule": {
266 + "0": "introduction",
267 + "1": "literature",
268 + "2": "other",
269 + "3": "other",
270 + "4": "other",
271 + "5": "conclusion"
272 + }
273 + },
274 + "schema_valid": true,
275 + "code_version": "87c73cd-dirty"
276 +}
\ No newline at end of file
added data/annotations/documents/jof_12823.json +271 −0
@@ -0,0 +1,271 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped hedging.typical_devices_note"
5 + ],
6 + "doc_id": "jof_12823",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:32+00:00",
12 + "source_sha256": "586ebb40875fdd079c00f469d7ed9f7bd433d038aaec782f2af0734ec8328f02",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 18112,
15 + "usage": {
16 + "input": 6147,
17 + "output": 2203,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.069442,
22 + "request_id": "req_011CggqWCDpcyqTTXrKNNrps",
23 + "annotation": {
24 + "genre": "mixed_theory_empirical",
25 + "subfield": "asset_pricing",
26 + "subfield_secondary": "delegated portfolio management / mutual funds / managerial compensation",
27 + "primary_method": "reduced_form_regression",
28 + "identification": {
29 + "estimand": "How a fund's mid-year distance from its benchmark affects second-half risk-shifting (relative tracking-error volatility), and whether this differs for managers with versus without performance-based contracts",
30 + "strategy": "Median quantile regressions of the risk adjustment ratio on Distance and excess return. Contract types are hand-collected from SAI filings. Treated funds (performance contracts) are matched to control funds (no performance contracts) on observables. A placebo test randomly reassigns benchmarks.",
31 + "key_assumptions": [
32 + "After matching on observables, contract type is uncorrelated with unobservables",
33 + "The self-designated prospectus benchmark is the relevant compensation benchmark",
34 + "The ratio of tracking-error volatility captures managerial risk-shifting"
35 + ],
36 + "threats_addressed": [
37 + "selection into contract type on observables",
38 + "spurious benchmark effects (placebo randomization)",
39 + "competing flow-performance tournament incentives",
40 + "managerial ownership, active share and tenure as confounds",
41 + "alternative risk-shifting measures"
42 + ],
43 + "causal_language_strength": 4,
44 + "uses_causal_claim": true
45 + },
46 + "abstract_moves": {
47 + "question": {
48 + "present": false,
49 + "sentence_index": -1
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": false,
65 + "sentence_index": -1
66 + },
67 + "contribution": {
68 + "present": true,
69 + "sentence_index": 3
70 + }
71 + },
72 + "introduction": {
73 + "moves": [
74 + {
75 + "paragraph_index": 0,
76 + "move": "motivation"
77 + },
78 + {
79 + "paragraph_index": 0,
80 + "move": "literature_gap"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "mechanism_or_theory_sketch"
85 + },
86 + {
87 + "paragraph_index": 1,
88 + "move": "related_literature_positioning"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "preview_of_approach"
93 + },
94 + {
95 + "paragraph_index": 2,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 2,
100 + "move": "identification_defense"
101 + },
102 + {
103 + "paragraph_index": 3,
104 + "move": "preview_of_results"
105 + },
106 + {
107 + "paragraph_index": 4,
108 + "move": "robustness_mention"
109 + },
110 + {
111 + "paragraph_index": 4,
112 + "move": "contribution_statement"
113 + },
114 + {
115 + "paragraph_index": 4,
116 + "move": "related_literature_positioning"
117 + },
118 + {
119 + "paragraph_index": 5,
120 + "move": "related_literature_positioning"
121 + },
122 + {
123 + "paragraph_index": 6,
124 + "move": "roadmap"
125 + }
126 + ],
127 + "contribution_paragraph_index": 4,
128 + "contribution_types": [
129 + "new_evidence",
130 + "new_data_or_measure",
131 + "resolution_of_puzzle"
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": "theory"
144 + },
145 + {
146 + "idx": 2,
147 + "role": "data"
148 + },
149 + {
150 + "idx": 3,
151 + "role": "results"
152 + },
153 + {
154 + "idx": 4,
155 + "role": "conclusion"
156 + },
157 + {
158 + "idx": 5,
159 + "role": "appendix"
160 + }
161 + ],
162 + "results_presentation": {
163 + "economic_significance_discussed": false,
164 + "statistical_significance_language": "unclear",
165 + "magnitude_statements_in_intro": false
166 + },
167 + "robustness": {
168 + "location": "mixed",
169 + "types": [
170 + "placebo",
171 + "alternative_measure",
172 + "subsample",
173 + "alternative_specification"
174 + ]
175 + },
176 + "tables": {
177 + "convention_notes": "Table notes define dependent variables (risk adjustment ratio as a second-half versus first-half tracking-error volatility ratio) and give the quantile-regression specification. Pooled OLS and median quantile regressions are reported side by side. A bootstrap placebo table summarizes 500 benchmark randomizations.",
178 + "units_stated_in_notes": false,
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": true,
188 + "restates_contribution": true
189 + },
190 + "hedging": {
191 + "level": "moderate",
192 + "typical_devices": [
193 + "'our results suggest'",
194 + "stating that the contract and tournament channels are not mutually exclusive",
195 + "declaring agnosticism on contract optimality",
196 + "explicitly stating the matching assumption"
197 + ]
198 + },
199 + "claim_evidence": [
200 + {
201 + "claim": "An asymmetric bonus contract is valued like an exchange option whose vega peaks when the portfolio's return is close to the benchmark's",
202 + "evidence_type": "equation_or_proof",
203 + "evidence_ref": "Section I; Appendix A (Margrabe 1978)"
204 + },
205 + {
206 + "claim": "Funds performing closer to the benchmark at mid-year increase relative risk more in the second half",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table II"
209 + },
210 + {
211 + "claim": "Risk-shifting is concentrated among funds with a clear compensation benchmark and is absent without performance pay",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table III"
214 + },
215 + {
216 + "claim": "Matched treated-control comparison supports a causal effect of contracts on risk-shifting",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table IV"
219 + },
220 + {
221 + "claim": "Manager ownership reduces risk-shifting, active share increases it, and tenure mitigates it; Distance remains important",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table VI"
224 + },
225 + {
226 + "claim": "Random benchmark assignment does not reproduce the effect",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table VII"
229 + },
230 + {
231 + "claim": "The contract-based incentive dominates flow-performance tournament incentives",
232 + "evidence_type": "table",
233 + "evidence_ref": "Section III.G"
234 + }
235 + ],
236 + "structural_summary": "The paper opens with an agency-and-contracting motivation, then sketches an option-based theory of the manager's bonus that predicts a hypothesis, priced in the appendix. After describing the data, one long empirical section moves through baseline quantile regressions, hand-collected contract heterogeneity, a matched-sample causal analysis, cross-sectional moderators, a placebo, a horse race against flow incentives, and robustness checks. A brief conclusion bounds the interpretation.",
237 + "writing_profile_keywords": [
238 + "option-pricing hypothesis",
239 + "quantile regression",
240 + "hand-collected compensation data",
241 + "matching for causality",
242 + "placebo bootstrap",
243 + "tournament versus contract"
244 + ],
245 + "confidence": 0.8,
246 + "notes": "The empirical section is a single large section with robustness checks as subsections, so robustness location is coded as mixed. The theory is light: one hypothesis, with an appendix derivation."
247 + },
248 + "deterministic_checks": {
249 + "roadmap_regex": true,
250 + "contribution_regex_in_intro": true,
251 + "magnitude_in_abstract": false,
252 + "abstract_n_sentences": 4,
253 + "intro_n_paragraphs": 7,
254 + "intro_n_words": 953,
255 + "causal_words_in_abstract": 0,
256 + "hedge_words_in_abstract": 0,
257 + "stars_mentioned": false,
258 + "clustered_se_mentioned": false,
259 + "internet_appendix_mentioned": false,
260 + "section_roles_rule": {
261 + "0": "introduction",
262 + "1": "theory",
263 + "2": "data",
264 + "3": "other",
265 + "4": "conclusion",
266 + "5": "appendix"
267 + }
268 + },
269 + "schema_valid": true,
270 + "code_version": "87c73cd-dirty"
271 +}
\ No newline at end of file
added data/annotations/documents/jof_12824.json +186 −0
@@ -0,0 +1,186 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12824",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:39+00:00",
9 + "source_sha256": "c19c5e21e17b94e82de3974aeb29b8e246b77e6c56b1d91388aa83477deb9b63",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 13255,
12 + "usage": {
13 + "input": 6343,
14 + "output": 1646,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.059086,
19 + "request_id": "req_011CggqXZ8h5Z2aYB5aR2ZEW",
20 + "annotation": {
21 + "genre": "replication_or_correction",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "portfolio choice / covariance estimation (theoretical correction)",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (analytical correction of a proof)",
27 + "strategy": "not applicable; corrected KKT-based derivation of the adjusted covariance matrix and constrained MLE equivalence",
28 + "key_assumptions": [
29 + "returns i.i.d. jointly normal (for Proposition 2)",
30 + "adjusted covariance matrix nonsingular",
31 + "sample covariance matrix positive semidefinite"
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": true,
44 + "sentence_index": 0
45 + },
46 + "method": {
47 + "present": false,
48 + "sentence_index": -1
49 + },
50 + "results": {
51 + "present": false,
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": "related_literature_positioning"
68 + },
69 + {
70 + "paragraph_index": 0,
71 + "move": "literature_gap"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "contribution_statement"
76 + },
77 + {
78 + "paragraph_index": 8,
79 + "move": "other"
80 + },
81 + {
82 + "paragraph_index": 25,
83 + "move": "other"
84 + },
85 + {
86 + "paragraph_index": 57,
87 + "move": "policy_or_broader_implication"
88 + }
89 + ],
90 + "contribution_paragraph_index": 0,
91 + "contribution_types": [
92 + "replication_or_correction"
93 + ],
94 + "roadmap_present": false,
95 + "literature_positioning_location": "inside_introduction"
96 + },
97 + "section_roles": [
98 + {
99 + "idx": 0,
100 + "role": "theory"
101 + }
102 + ],
103 + "results_presentation": {
104 + "economic_significance_discussed": false,
105 + "statistical_significance_language": "unclear",
106 + "magnitude_statements_in_intro": false
107 + },
108 + "robustness": {
109 + "location": "none_seen",
110 + "types": []
111 + },
112 + "tables": {
113 + "convention_notes": "No tables; the note consists of an optimization setup, KKT conditions, restated propositions and step-by-step proofs with numbered equations.",
114 + "units_stated_in_notes": false,
115 + "standard_errors_description": "none"
116 + },
117 + "figures": {
118 + "uncertainty_shown": "no_figures"
119 + },
120 + "conclusion": {
121 + "bounded_claims": true,
122 + "policy_implications": false,
123 + "open_questions": false,
124 + "restates_contribution": true
125 + },
126 + "hedging": {
127 + "level": "low",
128 + "typical_devices": [
129 + "explicit scope condition: original proofs valid only under nonnegativity constraints",
130 + "stated assumptions (normality, nonsingularity) in propositions"
131 + ]
132 + },
133 + "claim_evidence": [
134 + {
135 + "claim": "The original equation omitted a term, so the original proofs hold only with no-short-sale constraints and not with upper bounds on weights.",
136 + "evidence_type": "equation_or_proof",
137 + "evidence_ref": "Corrected equation (10) vs. original equation (8)"
138 + },
139 + {
140 + "claim": "With the corrected adjustment, the shrunk covariance matrix is symmetric and PSD, and the constrained minimum variance portfolio is one of its GMV portfolios.",
141 + "evidence_type": "equation_or_proof",
142 + "evidence_ref": "Proposition 1 and proof"
143 + },
144 + {
145 + "claim": "Under normality, the adjusted matrix is the constrained MLE whose implied GMV portfolio satisfies the weight constraints, and conversely.",
146 + "evidence_type": "equation_or_proof",
147 + "evidence_ref": "Proposition 2 and proof; Morrison (2005); Zwiernik, Uhler, and Richards (2017)"
148 + },
149 + {
150 + "claim": "The omitted term shifts all variances and covariances equally, so the original shrinkage interpretation is unchanged.",
151 + "evidence_type": "equation_or_proof",
152 + "evidence_ref": "Closing discussion after Proposition 2"
153 + }
154 + ],
155 + "structural_summary": "A short corrigendum in a single unnumbered section. It identifies the published propositions and pinpoints the omitted term in the original equation. It states the scope of the error (valid only without upper bounds), then restates the constrained minimum-variance problem and its KKT conditions. It re-proves Proposition 1 with the corrected adjusted covariance matrix and re-proves Proposition 2 on constrained-MLE equivalence. It closes by noting that the original shrinkage interpretation survives the correction.",
156 + "writing_profile_keywords": [
157 + "corrigendum",
158 + "proof correction",
159 + "KKT conditions",
160 + "portfolio constraints",
161 + "covariance shrinkage",
162 + "constrained MLE",
163 + "equation-dense"
164 + ],
165 + "confidence": 0.85,
166 + "notes": "Abstract is truncated and blends into the body; the body is mostly equations split into fragments that are treated as paragraphs. Paragraph indices for intro moves are approximate, given the fragmented equation text. There are no separate conclusion paragraphs; the conclusion fields are inferred from the closing remarks on the shrinkage interpretation."
167 + },
168 + "deterministic_checks": {
169 + "roadmap_regex": false,
170 + "contribution_regex_in_intro": false,
171 + "magnitude_in_abstract": false,
172 + "abstract_n_sentences": 2,
173 + "intro_n_paragraphs": 58,
174 + "intro_n_words": 2132,
175 + "causal_words_in_abstract": 0,
176 + "hedge_words_in_abstract": 0,
177 + "stars_mentioned": false,
178 + "clustered_se_mentioned": false,
179 + "internet_appendix_mentioned": false,
180 + "section_roles_rule": {
181 + "0": "introduction"
182 + }
183 + },
184 + "schema_valid": true,
185 + "code_version": "87c73cd-dirty"
186 +}
\ No newline at end of file
added data/annotations/documents/jof_12827.json +250 −0
@@ -0,0 +1,250 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12827",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:35+00:00",
9 + "source_sha256": "5254c1e1d0d04b63d8df38df3b0e2f854bba8b2ede41cd6dfc2e0d35b0787307",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20507,
12 + "usage": {
13 + "input": 7427,
14 + "output": 1849,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.067482,
19 + "request_id": "req_011CggqbdrpRSgxehFaNq2LC",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "market_microstructure",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Incremental share of variation in corporate bond yield spread changes, and in their common unexplained component, accounted for by systematic OTC friction proxies",
27 + "strategy": "Bond-by-bond time-series regressions extending the CDGM structural-variable model with marketwide proxies for inventory, search and bargaining frictions; principal component analysis of residuals to assess the common factor",
28 + "key_assumptions": [
29 + "Constructed proxies capture the intensity of systematic inventory, search and bargaining frictions",
30 + "CDGM structural variables adequately control for fundamental drivers"
31 + ],
32 + "threats_addressed": [
33 + "stale dealer quotes, handled by using transaction prices",
34 + "crisis-period dominance, handled by excluding the crisis",
35 + "other potential drivers of yield spread changes",
36 + "alternative definitions of CDGM variables"
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": true,
48 + "sentence_index": 0
49 + },
50 + "method": {
51 + "present": true,
52 + "sentence_index": 1
53 + },
54 + "results": {
55 + "present": true,
56 + "sentence_index": 1
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": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 0,
79 + "move": "preview_of_approach"
80 + },
81 + {
82 + "paragraph_index": 0,
83 + "move": "preview_of_results"
84 + },
85 + {
86 + "paragraph_index": 1,
87 + "move": "mechanism_or_theory_sketch"
88 + },
89 + {
90 + "paragraph_index": 1,
91 + "move": "preview_of_results"
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": "related_literature_positioning"
108 + },
109 + {
110 + "paragraph_index": 5,
111 + "move": "related_literature_positioning"
112 + },
113 + {
114 + "paragraph_index": 6,
115 + "move": "roadmap"
116 + }
117 + ],
118 + "contribution_paragraph_index": 0,
119 + "contribution_types": [
120 + "new_evidence",
121 + "new_data_or_measure"
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": "data"
134 + },
135 + {
136 + "idx": 2,
137 + "role": "empirical_design"
138 + },
139 + {
140 + "idx": 3,
141 + "role": "data"
142 + },
143 + {
144 + "idx": 4,
145 + "role": "results"
146 + },
147 + {
148 + "idx": 5,
149 + "role": "conclusion"
150 + }
151 + ],
152 + "results_presentation": {
153 + "economic_significance_discussed": true,
154 + "statistical_significance_language": "unclear",
155 + "magnitude_statements_in_intro": true
156 + },
157 + "robustness": {
158 + "location": "mixed",
159 + "types": [
160 + "alternative_sample_period",
161 + "alternative_specification",
162 + "alternative_measure"
163 + ]
164 + },
165 + "tables": {
166 + "convention_notes": "Notes spell out the regression equation and its variable vectors, the sample filter (bonds with at least 25 monthly observations) and the data source and period. Results come from bond-level regressions summarized across bonds.",
167 + "units_stated_in_notes": true,
168 + "standard_errors_description": "unclear"
169 + },
170 + "figures": {
171 + "uncertainty_shown": "no"
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' theory",
183 + "'suggests'/'indicates'",
184 + "'associated with'",
185 + "'support the implications of'"
186 + ]
187 + },
188 + "claim_evidence": [
189 + {
190 + "claim": "Systematic inventory, search and bargaining frictions explain 23.4% of the variation in the common component of yield spread changes",
191 + "evidence_type": "table",
192 + "evidence_ref": "Table IX"
193 + },
194 + {
195 + "claim": "The CDGM model on transaction prices leaves a large common unexplained factor",
196 + "evidence_type": "table",
197 + "evidence_ref": "Tables II, III"
198 + },
199 + {
200 + "claim": "Higher interdealer network connectivity and longer intermediation chains are associated with narrower yield spreads",
201 + "evidence_type": "table",
202 + "evidence_ref": "Table VII"
203 + },
204 + {
205 + "claim": "Search frictions capture 6.3% of the common component",
206 + "evidence_type": "table",
207 + "evidence_ref": "Table VII"
208 + },
209 + {
210 + "claim": "Results hold when the crisis period is excluded",
211 + "evidence_type": "table",
212 + "evidence_ref": "Table X"
213 + }
214 + ],
215 + "structural_summary": "The paper opens with the puzzle of a large unexplained common factor in yield spread changes and introduces OTC friction theory as a candidate explanation. It then previews the friction proxies and results, theory by theory, and positions itself against the liquidity and market-making literatures. The body describes the TRACE data, replicates the CDGM benchmark, and constructs systematic friction measures. It then tests each friction group separately and jointly, followed by robustness checks. A brief conclusion restates the key magnitudes.",
216 + "writing_profile_keywords": [
217 + "puzzle-driven",
218 + "benchmark replication",
219 + "theory-guided proxies",
220 + "explained variance",
221 + "OTC frictions",
222 + "network measures"
223 + ],
224 + "confidence": 0.8,
225 + "notes": "Sections III and IV are labelled 'other' by the role rule; Section III is mainly measure construction, so it is coded as data. Standard error conventions are not visible in the table notes."
226 + },
227 + "deterministic_checks": {
228 + "roadmap_regex": true,
229 + "contribution_regex_in_intro": false,
230 + "magnitude_in_abstract": true,
231 + "abstract_n_sentences": 7,
232 + "intro_n_paragraphs": 7,
233 + "intro_n_words": 1121,
234 + "causal_words_in_abstract": 0,
235 + "hedge_words_in_abstract": 0,
236 + "stars_mentioned": false,
237 + "clustered_se_mentioned": false,
238 + "internet_appendix_mentioned": false,
239 + "section_roles_rule": {
240 + "0": "introduction",
241 + "1": "data",
242 + "2": "theory",
243 + "3": "other",
244 + "4": "other",
245 + "5": "conclusion"
246 + }
247 + },
248 + "schema_valid": true,
249 + "code_version": "87c73cd-dirty"
250 +}
\ No newline at end of file
added data/annotations/documents/jof_12828.json +273 −0
@@ -0,0 +1,273 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped hedging.typical_devices_note"
5 + ],
6 + "doc_id": "jof_12828",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:32+00:00",
12 + "source_sha256": "ca4b28a8c13c90780b51e3debe66b974f4b0e5f5a3690fe847cb7efa23fa6701",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 22026,
15 + "usage": {
16 + "input": 7557,
17 + "output": 2199,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.075002,
22 + "request_id": "req_011CggqZshU4Ub4QVJn4becS",
23 + "annotation": {
24 + "genre": "mixed_theory_empirical",
25 + "subfield": "household_finance",
26 + "subfield_secondary": "behavioral_finance",
27 + "primary_method": "calibration_simulation",
28 + "identification": {
29 + "estimand": "How savings and dissavings behavior relates to survival optimism (subjective minus actuarial survival probability), alongside model-implied effects of subjective versus actuarial mortality beliefs on life-cycle saving.",
30 + "strategy": "Survey elicitation of beliefs; a canonical life-cycle model calibrated to the elicited beliefs and compared with an actuarial benchmark, with fit checked by SMM; ordered logit regressions with controls; Altonji-Elder-Taber/Oster coefficient bounding; a salience-based belief measure presumed orthogonal to confounders.",
31 + "key_assumptions": [
32 + "Elicited survey beliefs reflect true subjective survival probabilities",
33 + "Controls capture sophistication and other confounders",
34 + "Weight placed on salient risk factors is orthogonal to health status and general optimism",
35 + "Selection on observables is informative about selection on unobservables (Oster bounds)"
36 + ],
37 + "threats_addressed": [
38 + "omitted variable bias from sophistication or financial literacy",
39 + "confounding by income, education and demographics",
40 + "general optimism or health status",
41 + "risk preferences"
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": 2
58 + },
59 + "results": {
60 + "present": true,
61 + "sentence_index": 1
62 + },
63 + "magnitude": {
64 + "present": true,
65 + "sentence_index": 3
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": 1,
80 + "move": "motivation"
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": 1,
92 + "move": "mechanism_or_theory_sketch"
93 + },
94 + {
95 + "paragraph_index": 4,
96 + "move": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "tension_or_puzzle"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "preview_of_approach"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "preview_of_results"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "preview_of_results"
113 + },
114 + {
115 + "paragraph_index": 7,
116 + "move": "identification_defense"
117 + },
118 + {
119 + "paragraph_index": 8,
120 + "move": "related_literature_positioning"
121 + },
122 + {
123 + "paragraph_index": 9,
124 + "move": "related_literature_positioning"
125 + },
126 + {
127 + "paragraph_index": 9,
128 + "move": "contribution_statement"
129 + },
130 + {
131 + "paragraph_index": 10,
132 + "move": "roadmap"
133 + }
134 + ],
135 + "contribution_paragraph_index": 9,
136 + "contribution_types": [
137 + "new_evidence",
138 + "new_data_or_measure",
139 + "resolution_of_puzzle"
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": "data"
152 + },
153 + {
154 + "idx": 2,
155 + "role": "theory"
156 + },
157 + {
158 + "idx": 3,
159 + "role": "results"
160 + },
161 + {
162 + "idx": 4,
163 + "role": "results"
164 + },
165 + {
166 + "idx": 5,
167 + "role": "conclusion"
168 + }
169 + ],
170 + "results_presentation": {
171 + "economic_significance_discussed": true,
172 + "statistical_significance_language": "unclear",
173 + "magnitude_statements_in_intro": true
174 + },
175 + "robustness": {
176 + "location": "within_results",
177 + "types": [
178 + "alternative_specification",
179 + "other"
180 + ]
181 + },
182 + "tables": {
183 + "convention_notes": "Notes define key variables, e.g., survival optimism as subjective minus SSA actuarial survival probability. Regression tables report ordered logit log-odds. Model tables list calibration parameters and their survey sources.",
184 + "units_stated_in_notes": true,
185 + "standard_errors_description": "unclear"
186 + },
187 + "figures": {
188 + "uncertainty_shown": "unclear"
189 + },
190 + "conclusion": {
191 + "bounded_claims": true,
192 + "policy_implications": true,
193 + "open_questions": true,
194 + "restates_contribution": true
195 + },
196 + "hedging": {
197 + "level": "moderate",
198 + "typical_devices": [
199 + "explicit caution about causal interpretation of reduced-form regressions",
200 + "'suggests' / 'gets closer to understanding'",
201 + "'presumably orthogonal'",
202 + "'associated with' wording for regression results"
203 + ]
204 + },
205 + "claim_evidence": [
206 + {
207 + "claim": "Young respondents underestimate survival and older respondents overestimate it relative to actuarial data.",
208 + "evidence_type": "figure",
209 + "evidence_ref": "Figure 1"
210 + },
211 + {
212 + "claim": "Salience of cohort-specific causes of death shapes mortality belief distortions.",
213 + "evidence_type": "figure",
214 + "evidence_ref": "Table II; Figures 2-3"
215 + },
216 + {
217 + "claim": "With subjective beliefs, the young undersave by about 26% and retirees draw down assets about 27% slower than under actuarial calibration.",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table IV; Figure 5"
220 + },
221 + {
222 + "claim": "Calibrating to subjective beliefs substantially improves the model's fit to consumption and net worth data.",
223 + "evidence_type": "figure",
224 + "evidence_ref": "Figure 6"
225 + },
226 + {
227 + "claim": "Survival optimism correlates with savings rates after controlling for literacy, risk tolerance and socioeconomic factors.",
228 + "evidence_type": "table",
229 + "evidence_ref": "Table V; Figure 7"
230 + },
231 + {
232 + "claim": "Survival beliefs relate to retirement dissavings, with an association comparable to that of bequest motives.",
233 + "evidence_type": "table",
234 + "evidence_ref": "Table VI"
235 + }
236 + ],
237 + "structural_summary": "The paper opens with behavioral motivation about overweighting rare, salient events. It then documents a new survey of mortality beliefs and their salience-based formation. Next, it presents a canonical life-cycle model calibrated to the elicited beliefs and compares its results with an actuarial benchmark, including SMM fit to data and comparison with alternative mechanisms. Reduced-form ordered logit regressions on the survey then validate the model's predictions, with an explicit omitted-variable defense. The conclusion restates the findings, discusses policy relevance and points to general-equilibrium and insurance questions for future work.",
238 + "writing_profile_keywords": [
239 + "new survey",
240 + "subjective expectations",
241 + "life-cycle calibration",
242 + "salience",
243 + "household finance puzzles",
244 + "model-plus-regression validation",
245 + "cautious causality"
246 + ],
247 + "confidence": 0.8,
248 + "notes": "Section IV is labeled as validating model predictions; it is assigned the results role. Standard error type is not visible in the excerpts."
249 + },
250 + "deterministic_checks": {
251 + "roadmap_regex": true,
252 + "contribution_regex_in_intro": true,
253 + "magnitude_in_abstract": true,
254 + "abstract_n_sentences": 6,
255 + "intro_n_paragraphs": 11,
256 + "intro_n_words": 1279,
257 + "causal_words_in_abstract": 1,
258 + "hedge_words_in_abstract": 0,
259 + "stars_mentioned": false,
260 + "clustered_se_mentioned": false,
261 + "internet_appendix_mentioned": true,
262 + "section_roles_rule": {
263 + "0": "introduction",
264 + "1": "other",
265 + "2": "theory",
266 + "3": "theory",
267 + "4": "other",
268 + "5": "conclusion"
269 + }
270 + },
271 + "schema_valid": true,
272 + "code_version": "87c73cd-dirty"
273 +}
\ No newline at end of file
added data/annotations/documents/jof_12829.json +267 −0
@@ -0,0 +1,267 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12829",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:15+00:00",
9 + "source_sha256": "0cbacb29f214ffd6a2f74802af25322bdccbf5d7785620b274e105fb68a38918",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19334,
12 + "usage": {
13 + "input": 6571,
14 + "output": 2179,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.070658,
19 + "request_id": "req_011CggqZq7gDxtEMcy7VJ9W2",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "financial_intermediation_banking",
23 + "subfield_secondary": "economic history / law and finance; household finance",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of granting married women property rights (end of coverture) on household portfolio shares of moveable assets, credit supply (interest rates, deposits, loans), and sectoral labor allocation",
27 + "strategy": "Staggered adoption of married women's property laws across U.S. states (1850-1920) used in a difference-in-differences / event-time framework, plus border-county-pair 'difference-in-discontinuities' designs and randomization (placebo timing) tests",
28 + "key_assumptions": [
29 + "Timing of women's rights is not driven by omitted factors that also move financial and economic outcomes",
30 + "No reverse causality from economic outcomes to adoption of rights",
31 + "Openness of capital markets across states, so local supply shocks affect local credit outcomes",
32 + "Neighboring border counties are comparable apart from the change in rights"
33 + ],
34 + "threats_addressed": [
35 + "omitted variable bias",
36 + "reverse causality",
37 + "geographic variation in economic conditions (border pairs)",
38 + "concurrent banking regulation (usury laws, reserve requirements, double liability)",
39 + "pre-trends (event-time leads)",
40 + "rights affecting marriage composition (balancing test)",
41 + "chance findings (randomization of adoption dates)"
42 + ],
43 + "causal_language_strength": 4,
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": 2
58 + },
59 + "results": {
60 + "present": true,
61 + "sentence_index": 2
62 + },
63 + "magnitude": {
64 + "present": false,
65 + "sentence_index": -1
66 + },
67 + "contribution": {
68 + "present": true,
69 + "sentence_index": 3
70 + }
71 + },
72 + "introduction": {
73 + "moves": [
74 + {
75 + "paragraph_index": 0,
76 + "move": "other"
77 + },
78 + {
79 + "paragraph_index": 5,
80 + "move": "motivation"
81 + },
82 + {
83 + "paragraph_index": 5,
84 + "move": "preview_of_approach"
85 + },
86 + {
87 + "paragraph_index": 5,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 6,
92 + "move": "preview_of_approach"
93 + },
94 + {
95 + "paragraph_index": 7,
96 + "move": "identification_defense"
97 + },
98 + {
99 + "paragraph_index": 8,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 8,
104 + "move": "related_literature_positioning"
105 + },
106 + {
107 + "paragraph_index": 9,
108 + "move": "related_literature_positioning"
109 + },
110 + {
111 + "paragraph_index": 9,
112 + "move": "contribution_statement"
113 + },
114 + {
115 + "paragraph_index": 10,
116 + "move": "roadmap"
117 + }
118 + ],
119 + "contribution_paragraph_index": 9,
120 + "contribution_types": [
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": "data"
138 + },
139 + {
140 + "idx": 3,
141 + "role": "empirical_design"
142 + },
143 + {
144 + "idx": 4,
145 + "role": "results"
146 + },
147 + {
148 + "idx": 5,
149 + "role": "results"
150 + },
151 + {
152 + "idx": 6,
153 + "role": "conclusion"
154 + },
155 + {
156 + "idx": 7,
157 + "role": "appendix"
158 + }
159 + ],
160 + "results_presentation": {
161 + "economic_significance_discussed": true,
162 + "statistical_significance_language": "stars_and_tstats",
163 + "magnitude_statements_in_intro": true
164 + },
165 + "robustness": {
166 + "location": "mixed",
167 + "types": [
168 + "placebo",
169 + "alternative_specification",
170 + "subsample",
171 + "falsification"
172 + ]
173 + },
174 + "tables": {
175 + "convention_notes": "Notes open with the clustering level and the *, **, *** p-value thresholds, then define variables, fixed effects and weights; clustering varies by design (state-year, state, border pair); summary tables report mean, median, SD and 10th/90th percentiles; event-time leads are shown as coefficient rows.",
176 + "units_stated_in_notes": true,
177 + "standard_errors_description": "clustered"
178 + },
179 + "figures": {
180 + "uncertainty_shown": "unclear"
181 + },
182 + "conclusion": {
183 + "bounded_claims": false,
184 + "policy_implications": true,
185 + "open_questions": false,
186 + "restates_contribution": true
187 + },
188 + "hedging": {
189 + "level": "low",
190 + "typical_devices": [
191 + "'we argue that' framing of causal interpretation",
192 + "'our estimates suggest'",
193 + "reporting ranges of estimates across specifications",
194 + "'may well have' in historical discussion"
195 + ]
196 + },
197 + "claim_evidence": [
198 + {
199 + "claim": "After rights were granted, households shifted their portfolios toward moveable assets, by roughly 1 to 6 percentage points.",
200 + "evidence_type": "table",
201 + "evidence_ref": "Tables III and IV"
202 + },
203 + {
204 + "claim": "Granting rights did not alter marriage outcomes, supporting the balance of the sample composition.",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table II"
207 + },
208 + {
209 + "claim": "Rights were followed by a positive credit supply shock: lower interest rates and higher deposits and loans.",
210 + "evidence_type": "figure",
211 + "evidence_ref": "Figure 3, Table VI"
212 + },
213 + {
214 + "claim": "Credit-market effects are not reproduced when adoption dates are randomly assigned.",
215 + "evidence_type": "figure",
216 + "evidence_ref": "Figure 4"
217 + },
218 + {
219 + "claim": "Rights led to a dynamic reallocation of male labor toward nonagriculture, with no significant pre-trends.",
220 + "evidence_type": "table",
221 + "evidence_ref": "Tables VII and VIII"
222 + },
223 + {
224 + "claim": "Within nonagriculture, employment shifted toward capital-intensive industries.",
225 + "evidence_type": "table",
226 + "evidence_ref": "Table IX"
227 + }
228 + ],
229 + "structural_summary": "Opens with a historical parliamentary quotation and frames the end of coverture as a major extension of property rights and investor protection. A brief theory section derives four testable predictions. Long data and empirical-strategy sections then defend the use of staggered state adoption as exogenous, drawing on the historical record. Results proceed sequentially through portfolios, credit markets, sectoral labor and capital intensity, combining nationwide and border-pair designs with randomization tests. A short conclusion restates the findings and suggests broader applicability to groups still denied economic rights.",
230 + "writing_profile_keywords": [
231 + "economic history",
232 + "law and finance",
233 + "staggered adoption",
234 + "border-pair design",
235 + "prediction-by-prediction results",
236 + "historical narrative",
237 + "investor protection"
238 + ],
239 + "confidence": 0.8,
240 + "notes": "Section idx 5 is a parsing artifact: it continues the Results section, covering credit markets and subsections C-D. Table VI is referenced but not excerpted. The intro uses 'difference-in-discontinuities' for the border exercise. Whether figures show confidence bands is not visible in the captions."
241 + },
242 + "deterministic_checks": {
243 + "roadmap_regex": true,
244 + "contribution_regex_in_intro": false,
245 + "magnitude_in_abstract": false,
246 + "abstract_n_sentences": 5,
247 + "intro_n_paragraphs": 11,
248 + "intro_n_words": 1017,
249 + "causal_words_in_abstract": 0,
250 + "hedge_words_in_abstract": 0,
251 + "stars_mentioned": true,
252 + "clustered_se_mentioned": true,
253 + "internet_appendix_mentioned": true,
254 + "section_roles_rule": {
255 + "0": "introduction",
256 + "1": "theory",
257 + "2": "data",
258 + "3": "empirical_design",
259 + "4": "results",
260 + "5": "other",
261 + "6": "conclusion",
262 + "7": "appendix"
263 + }
264 + },
265 + "schema_valid": true,
266 + "code_version": "87c73cd-dirty"
267 +}
\ No newline at end of file
added data/annotations/documents/jof_12830.json +258 −0
@@ -0,0 +1,258 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [],
4 + "doc_id": "jof_12830",
5 + "ontology_version": "1.0",
6 + "prompt_version": "1.0",
7 + "model": "claude-opus-5-5",
8 + "effort": "medium",
9 + "annotated_at": "2026-11-03T23:26:52+00:00",
10 + "source_sha256": "3936845c62aeb2be7028d99acb0177a93f57d07b10b54d5120017229fe2b56b5",
11 + "parser": "pymupdf-jof-heuristics-1.3",
12 + "payload_chars": 20457,
13 + "usage": {
14 + "input": 6672,
15 + "output": 2187,
16 + "cache_read": 3968,
17 + "cache_write": 0
18 + },
19 + "cost_usd": 0.071222,
20 + "request_id": "req_011CggrqJqancvwEzDPnSa8D",
21 + "annotation": {
22 + "genre": "empirical",
23 + "subfield": "household_finance",
24 + "subfield_secondary": "behavioral_finance",
25 + "primary_method": "reduced_form_regression",
26 + "identification": {
27 + "estimand": "Effect of average local coworker ESPP participation/early selling on an employee's own participation/selling decision (endogenous peer effect)",
28 + "strategy": "Linear probability regressions comparing employees within the same firm and election window, with metro area (CBSA) fixed effects; peer means are measured in an excluded random half-sample; IV built from local office demographic composition (Duflo-Saez/Case-Katz style); cross-sectional tests using informed vs. uninformed peers to identify the learning mechanism",
29 + "key_assumptions": [
30 + "Office-level shocks within a firm are captured by location and demographic controls",
31 + "Exclusion restriction: the local distribution of employee characteristics affects own ESPP choices only through peer choices, conditional on own traits",
32 + "Informativeness proxies (occupation, self-reported experience, income) separate initially informed from uninformed workers"
33 + ],
34 + "threats_addressed": [
35 + "Correlated office-level shocks not captured by controls",
36 + "Endogenous sorting of workers into offices",
37 + "Reflection problem / mechanical correlation, handled with an excluded half-sample",
38 + "Uninformed mimicry versus learning",
39 + "Office size and headquarters-state effects"
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": 1
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": 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": "identification_defense"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "robustness_mention"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "mechanism_or_theory_sketch"
103 + },
104 + {
105 + "paragraph_index": 3,
106 + "move": "preview_of_results"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "contribution_statement"
111 + },
112 + {
113 + "paragraph_index": 4,
114 + "move": "related_literature_positioning"
115 + },
116 + {
117 + "paragraph_index": 5,
118 + "move": "related_literature_positioning"
119 + },
120 + {
121 + "paragraph_index": 6,
122 + "move": "related_literature_positioning"
123 + },
124 + {
125 + "paragraph_index": 6,
126 + "move": "contribution_statement"
127 + }
128 + ],
129 + "contribution_paragraph_index": 4,
130 + "contribution_types": [
131 + "new_evidence",
132 + "new_data_or_measure"
133 + ],
134 + "roadmap_present": false,
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": "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 + "subsample",
164 + "alternative_specification",
165 + "other"
166 + ]
167 + },
168 + "tables": {
169 + "convention_notes": "Table notes define the sample (a named 'Analysis Sample' of random half-observations), the dependent indicator variables, and constructed peer means from an excluded half-sample. Controls are referenced back to earlier tables. Summary and descriptive tables come before the baseline/IV and cross-sectional tables.",
170 + "units_stated_in_notes": true,
171 + "standard_errors_description": "unclear"
172 + },
173 + "figures": {
174 + "uncertainty_shown": "no_figures"
175 + },
176 + "conclusion": {
177 + "bounded_claims": true,
178 + "policy_implications": false,
179 + "open_questions": false,
180 + "restates_contribution": true
181 + },
182 + "hedging": {
183 + "level": "moderate",
184 + "typical_devices": [
185 + "'consistent with' framing of mechanisms",
186 + "'suggests the potential' in implications",
187 + "explicit acknowledgment that the exclusion restriction cannot be tested",
188 + "'associated with' for magnitudes"
189 + ]
190 + },
191 + "claim_evidence": [
192 + {
193 + "claim": "Employees underparticipate in ESPPs despite sizable forgone gains",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table I; Table II"
196 + },
197 + {
198 + "claim": "Local coworker participation raises own participation; 10pp peer rise linked to about 1.5pp own rise",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table IV"
201 + },
202 + {
203 + "claim": "Peer effect results hold using demographic-based instruments",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table IV (IV columns)"
206 + },
207 + {
208 + "claim": "Peer effects are weaker for uninformed workers when no informed colleague is present, supporting learning",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table VIII; Table IX; Table X"
211 + },
212 + {
213 + "claim": "Coworkers' early-sale behavior predicts own early-sale decisions",
214 + "evidence_type": "table",
215 + "evidence_ref": "Table XI; Table XII"
216 + },
217 + {
218 + "claim": "Peer effects vary with worker demographics, ESPP gains and population density",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table V; Table VI"
221 + }
222 + ],
223 + "structural_summary": "Opens with a motivation about retail investor mistakes and workplace plans as a learning channel. Previews novel ESPP data, low participation, the baseline peer effect magnitude, the information mechanism and selling-timing results. Defends identification with fixed effects, demographic IVs and sorting checks, then outlines informed/uninformed proxy tests of learning. Positions against finance, retirement-plan and labor peer-effect literatures. Main body: data/institutional section, then one large results section (baseline, heterogeneity, mechanism, selling). The conclusion restates findings.",
224 + "writing_profile_keywords": [
225 + "peer effects",
226 + "workplace networks",
227 + "ESPP",
228 + "information diffusion",
229 + "fixed effects",
230 + "instrumental variables",
231 + "mechanism tests",
232 + "household finance"
233 + ],
234 + "confidence": 0.8,
235 + "notes": "Section 1 is labeled literature by the rule but is actually data/institutional background. No roadmap is visible. The excerpts do not show how standard errors are computed. The final abstract sentence hedges a policy implication (investor education), but the conclusion excerpt is truncated before any policy discussion."
236 + },
237 + "deterministic_checks": {
238 + "roadmap_regex": false,
239 + "contribution_regex_in_intro": true,
240 + "magnitude_in_abstract": false,
241 + "abstract_n_sentences": 6,
242 + "intro_n_paragraphs": 7,
243 + "intro_n_words": 1269,
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": "literature",
252 + "2": "other",
253 + "3": "conclusion"
254 + }
255 + },
256 + "schema_valid": true,
257 + "code_version": "87c73cd-dirty"
258 +}
\ No newline at end of file
added data/annotations/documents/jof_12831.json +256 −0
@@ -0,0 +1,256 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12831",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:11:12+00:00",
9 + "source_sha256": "793bcb451d37485c6b9a7731115e803fce2efe4ee82367af641a3b07e4d576f4",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 17801,
12 + "usage": {
13 + "input": 6198,
14 + "output": 1997,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.065526,
19 + "request_id": "req_011CggqeGKrAH8p6avJDdciw",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; equilibrium analysis of a three-date model with incumbent pledgeability choice, debt, and auctions to experts and outsiders",
28 + "key_assumptions": [
29 + "Two periods with industry states G and B; state G for sure in period 2",
30 + "Experts are rational and never pay more than fundamental value",
31 + "Incumbent sets pledgeability, which is subject to moral hazard",
32 + "Debt contracts are fixed and pledgeability effects are hard to contract upon"
33 + ],
34 + "threats_addressed": [
35 + "Robustness to dynamic effects",
36 + "Alternative assumptions on ability loss",
37 + "Long-term debt, public equity, and state-contingent contracts"
38 + ],
39 + "causal_language_strength": 1,
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": 1
66 + }
67 + },
68 + "introduction": {
69 + "moves": [
70 + {
71 + "paragraph_index": 0,
72 + "move": "tension_or_puzzle"
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": "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": 3,
96 + "move": "contribution_statement"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "preview_of_results"
101 + },
102 + {
103 + "paragraph_index": 5,
104 + "move": "related_literature_positioning"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "policy_or_broader_implication"
109 + },
110 + {
111 + "paragraph_index": 6,
112 + "move": "roadmap"
113 + }
114 + ],
115 + "contribution_paragraph_index": 3,
116 + "contribution_types": [
117 + "new_theory_or_model"
118 + ],
119 + "roadmap_present": true,
120 + "literature_positioning_location": "both"
121 + },
122 + "section_roles": [
123 + {
124 + "idx": 0,
125 + "role": "introduction"
126 + },
127 + {
128 + "idx": 1,
129 + "role": "theory"
130 + },
131 + {
132 + "idx": 2,
133 + "role": "other"
134 + },
135 + {
136 + "idx": 3,
137 + "role": "theory"
138 + },
139 + {
140 + "idx": 4,
141 + "role": "robustness"
142 + },
143 + {
144 + "idx": 5,
145 + "role": "literature"
146 + },
147 + {
148 + "idx": 6,
149 + "role": "conclusion"
150 + },
151 + {
152 + "idx": 7,
153 + "role": "appendix"
154 + }
155 + ],
156 + "results_presentation": {
157 + "economic_significance_discussed": false,
158 + "statistical_significance_language": "unclear",
159 + "magnitude_statements_in_intro": false
160 + },
161 + "robustness": {
162 + "location": "main_text_section",
163 + "types": [
164 + "sensitivity_to_parameters",
165 + "other"
166 + ]
167 + },
168 + "tables": {
169 + "convention_notes": "No tables. Results are carried by propositions and lemmas, with proofs in the appendix. Figures depict states of nature, timing, equilibrium functions, and two descriptive time series used to discuss empirical implications.",
170 + "units_stated_in_notes": false,
171 + "standard_errors_description": "none"
172 + },
173 + "figures": {
174 + "uncertainty_shown": "no"
175 + },
176 + "conclusion": {
177 + "bounded_claims": true,
178 + "policy_implications": false,
179 + "open_questions": true,
180 + "restates_contribution": true
181 + },
182 + "hedging": {
183 + "level": "moderate",
184 + "typical_devices": [
185 + "modal verbs (can, may, could)",
186 + "seems",
187 + "explicit acknowledgment that a dynamic extension is needed",
188 + "consistent-with appeals to prior evidence"
189 + ]
190 + },
191 + "claim_evidence": [
192 + {
193 + "claim": "High anticipated liquidity raises the debt a firm can credibly repay, because wealthy experts bid full value at resale.",
194 + "evidence_type": "equation_or_proof",
195 + "evidence_ref": "Section III.A"
196 + },
197 + {
198 + "claim": "High anticipated liquidity and the resulting high debt reduce the incumbent's incentive to raise pledgeability.",
199 + "evidence_type": "equation_or_proof",
200 + "evidence_ref": "Section III.A.2"
201 + },
202 + {
203 + "claim": "In downturns after booms, firms may be sold to less productive outsiders, who raise pledgeability and later resell to experts.",
204 + "evidence_type": "prior_literature",
205 + "evidence_ref": "Eisfeldt and Rampini (2006, 2008)"
206 + },
207 + {
208 + "claim": "Higher anticipated liquidity can reduce current access to finance.",
209 + "evidence_type": "equation_or_proof",
210 + "evidence_ref": "Section III.B.2, Figure 4"
211 + },
212 + {
213 + "claim": "Proxies for weak pledgeability, such as weak internal controls and covenant-lite loans, vary over the cycle.",
214 + "evidence_type": "figure",
215 + "evidence_ref": "Figures 6 and 7"
216 + }
217 + ],
218 + "structural_summary": "The paper opens with a macro puzzle about protracted downturns after booms and critiques debt-overhang explanations. It then sketches the mechanism by which liquidity, leverage, and pledgeability interact, positions itself against key prior work, and gives a roadmap. Next come the model framework and two motivating examples. Formal solutions are organized by industry case (after prosperity, normal times, after distress). These are followed by a robustness section on alternative contracts and assumptions, and a section deriving empirical implications with suggestive proxies and related literature. The conclusion recaps the cycle and outlines future work.",
219 + "writing_profile_keywords": [
220 + "applied theory",
221 + "financing cycles",
222 + "pledgeability",
223 + "verbal mechanism exposition",
224 + "motivating examples",
225 + "proofs in appendix",
226 + "empirical implications section"
227 + ],
228 + "confidence": 0.85,
229 + "notes": "Theory paper, so empirical-only fields are set to not applicable or none. Figures 6 and 7 are descriptive time series supporting the empirical implications, not tests. Only one equation is counted, probably reflecting parsing limitations."
230 + },
231 + "deterministic_checks": {
232 + "roadmap_regex": true,
233 + "contribution_regex_in_intro": false,
234 + "magnitude_in_abstract": false,
235 + "abstract_n_sentences": 6,
236 + "intro_n_paragraphs": 7,
237 + "intro_n_words": 1071,
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": "theory",
246 + "2": "other",
247 + "3": "theory",
248 + "4": "robustness",
249 + "5": "literature",
250 + "6": "conclusion",
251 + "7": "appendix"
252 + }
253 + },
254 + "schema_valid": true,
255 + "code_version": "87c73cd-dirty"
256 +}
\ No newline at end of file
added data/annotations/documents/jof_12832.json +235 −0
@@ -0,0 +1,235 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12832",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:02+00:00",
9 + "source_sha256": "2939c0d1a175d8c75f0cb50fed8f23c79747317c486f5fc0c1ce55028d8be304",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19529,
12 + "usage": {
13 + "input": 6704,
14 + "output": 1934,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06629,
19 + "request_id": "req_011CggqZ5VbKfALxaf53ubk2",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of an exogenous rise in the relative cost of very short-term bank debt on firms' short-term borrowing, cash, trade credit, and investment, and how this effect differs with industry access to trade credit",
27 + "strategy": "Difference-in-differences around Colombia's 2011 BAD tax change. Treated firms are those in the top quintile of pre-2011 very short-term debt use; the bottom two quintiles are controls. Matching is applied, the sample is restricted to firms above a sales threshold, and a triple interaction uses industry trade-credit access measured from U.S. Compustat firms.",
28 + "key_assumptions": [
29 + "Treated and control firms would have followed parallel trends absent the tax change",
30 + "The tax shock affects very short-term bank credit without spilling over into banks' long-term credit role or into investment opportunities",
31 + "Trade-credit access measured from U.S. Compustat firms captures inherent industry access in Colombia"
32 + ],
33 + "threats_addressed": [
34 + "Pre-trend differences, examined with year-coefficient plots",
35 + "Confounds from broad credit shocks such as the financial crisis",
36 + "Tax avoidance behavior",
37 + "Covariate imbalance, addressed with a matched sample"
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": 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": 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": 0,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 1,
84 + "move": "preview_of_results"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "mechanism_or_theory_sketch"
89 + },
90 + {
91 + "paragraph_index": 2,
92 + "move": "contribution_statement"
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": "literature_gap"
105 + },
106 + {
107 + "paragraph_index": 4,
108 + "move": "identification_defense"
109 + },
110 + {
111 + "paragraph_index": 5,
112 + "move": "roadmap"
113 + }
114 + ],
115 + "contribution_paragraph_index": 3,
116 + "contribution_types": [
117 + "new_evidence",
118 + "new_identification"
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": "other"
131 + },
132 + {
133 + "idx": 2,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 3,
138 + "role": "conclusion"
139 + }
140 + ],
141 + "results_presentation": {
142 + "economic_significance_discussed": true,
143 + "statistical_significance_language": "unclear",
144 + "magnitude_statements_in_intro": false
145 + },
146 + "robustness": {
147 + "location": "none_seen",
148 + "types": []
149 + },
150 + "tables": {
151 + "convention_notes": "Each regression table note restates the difference-in-differences setup: the Treatment × Post 2011 interaction of interest, how treatment and control groups are defined, and outcomes scaled by beginning-of-year assets. Table I compares prematch and postmatch samples.",
152 + "units_stated_in_notes": true,
153 + "standard_errors_description": "unclear"
154 + },
155 + "figures": {
156 + "uncertainty_shown": "unclear"
157 + },
158 + "conclusion": {
159 + "bounded_claims": true,
160 + "policy_implications": false,
161 + "open_questions": false,
162 + "restates_contribution": true
163 + },
164 + "hedging": {
165 + "level": "moderate",
166 + "typical_devices": [
167 + "our results suggest",
168 + "likely occurs",
169 + "plausibly exogenous",
170 + "explicit acknowledgment of external validity limits"
171 + ]
172 + },
173 + "claim_evidence": [
174 + {
175 + "claim": "Exposed firms sharply reduce very short-term bank borrowing after the 2011 tax change",
176 + "evidence_type": "table",
177 + "evidence_ref": "Table II; Figure 6"
178 + },
179 + {
180 + "claim": "Exposed firms increase both cash and net accounts payable",
181 + "evidence_type": "table",
182 + "evidence_ref": "Tables III, IV"
183 + },
184 + {
185 + "claim": "Some treated firms cut capital expenditures, R&D, and inventory investment",
186 + "evidence_type": "table",
187 + "evidence_ref": "Table V"
188 + },
189 + {
190 + "claim": "In high trade-credit-access industries, firms substitute into payables with no change in cash or investment; in low-access industries, cash rises and investment falls",
191 + "evidence_type": "table",
192 + "evidence_ref": "Tables VI-VIII"
193 + },
194 + {
195 + "claim": "The tax raised the cost of short-maturity loans more than that of long-maturity loans",
196 + "evidence_type": "figure",
197 + "evidence_ref": "Figure 2"
198 + }
199 + ],
200 + "structural_summary": "The paper opens with a broad question about how firms manage liquidity and previews the findings. It then sketches a two-regime tax mechanism and positions itself against the trade-credit and bank-liquidity literatures inside the introduction. A dedicated institutional section explains why the Colombian BAD tax provides a clean shock. One combined section covers design, data, and four numbered results, ending with heterogeneity by industry trade-credit access. A short conclusion restates the contribution and acknowledges external validity limits.",
201 + "writing_profile_keywords": [
202 + "natural experiment",
203 + "tax shock",
204 + "difference-in-differences",
205 + "trade credit",
206 + "liquidity management",
207 + "emerging market",
208 + "numbered results",
209 + "institutional detail"
210 + ],
211 + "confidence": 0.8,
212 + "notes": "No robustness section is visible in the excerpts. Matching and pre-trend plots appear within the design. The standard errors clustering level is not visible."
213 + },
214 + "deterministic_checks": {
215 + "roadmap_regex": true,
216 + "contribution_regex_in_intro": true,
217 + "magnitude_in_abstract": false,
218 + "abstract_n_sentences": 6,
219 + "intro_n_paragraphs": 6,
220 + "intro_n_words": 1120,
221 + "causal_words_in_abstract": 0,
222 + "hedge_words_in_abstract": 0,
223 + "stars_mentioned": false,
224 + "clustered_se_mentioned": false,
225 + "internet_appendix_mentioned": false,
226 + "section_roles_rule": {
227 + "0": "introduction",
228 + "1": "other",
229 + "2": "data",
230 + "3": "conclusion"
231 + }
232 + },
233 + "schema_valid": true,
234 + "code_version": "87c73cd-dirty"
235 +}
\ No newline at end of file
added data/annotations/documents/jof_12833.json +262 −0
@@ -0,0 +1,262 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12833",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:55+00:00",
9 + "source_sha256": "a873fab10ec11b01baf0f01eedd915aa344f183707db057fe6a76ce244863423",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19720,
12 + "usage": {
13 + "input": 6904,
14 + "output": 2073,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.06987,
19 + "request_id": "req_011CggqYdTS7WFZxfDL1WwFb",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Degree of overreaction in analysts' long-term growth expectations (representativeness parameter theta) and its implications for return predictability across LTG-sorted portfolios",
27 + "strategy": "Portfolio sorts on analyst LTG; forecast-error regressions of the Coibion-Gorodnichenko type; estimation of a diagnostic Kalman filter model by simulated method of moments; cross-sectional tests of kernel-of-truth predictions",
28 + "key_assumptions": [
29 + "Analyst LTG forecasts proxy for market beliefs",
30 + "Earnings follow the assumed latent-fundamentals process",
31 + "Diagnostic expectations are the correct belief-formation model"
32 + ],
33 + "threats_addressed": [
34 + "Bayesian learning as an alternative explanation",
35 + "Adaptive/mechanical extrapolation as an alternative explanation"
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": 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": true,
63 + "sentence_index": 2
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": 1,
90 + "move": "roadmap"
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": "mechanism_or_theory_sketch"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "contribution_statement"
107 + },
108 + {
109 + "paragraph_index": 5,
110 + "move": "related_literature_positioning"
111 + }
112 + ],
113 + "contribution_paragraph_index": 1,
114 + "contribution_types": [
115 + "new_evidence",
116 + "new_theory_or_model",
117 + "resolution_of_puzzle"
118 + ],
119 + "roadmap_present": true,
120 + "literature_positioning_location": "inside_introduction"
121 + },
122 + "section_roles": [
123 + {
124 + "idx": 0,
125 + "role": "introduction"
126 + },
127 + {
128 + "idx": 1,
129 + "role": "data"
130 + },
131 + {
132 + "idx": 2,
133 + "role": "results"
134 + },
135 + {
136 + "idx": 3,
137 + "role": "theory"
138 + },
139 + {
140 + "idx": 4,
141 + "role": "theory"
142 + },
143 + {
144 + "idx": 5,
145 + "role": "results"
146 + },
147 + {
148 + "idx": 6,
149 + "role": "results"
150 + },
151 + {
152 + "idx": 7,
153 + "role": "conclusion"
154 + }
155 + ],
156 + "results_presentation": {
157 + "economic_significance_discussed": true,
158 + "statistical_significance_language": "unclear",
159 + "magnitude_statements_in_intro": false
160 + },
161 + "robustness": {
162 + "location": "none_seen",
163 + "types": []
164 + },
165 + "tables": {
166 + "convention_notes": "Table notes describe the decile portfolio formation on LTG each December from 1981 to 2015, equal weighting, and the variable definitions. Significance is marked with a letter superscript (e.g., 'a' for 1%). Year fixed effects are included but not shown. Model parameters are defined at the quarterly scale.",
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 + "'does a reasonable job' fit language",
183 + "'this is just a start' framing",
184 + "'in line with' phrasing for predictions"
185 + ]
186 + },
187 + "claim_evidence": [
188 + {
189 + "claim": "With 20 extra years of data, low-LTG stocks still outperform high-LTG stocks",
190 + "evidence_type": "figure",
191 + "evidence_ref": "Figure 1"
192 + },
193 + {
194 + "claim": "High-LTG firms show fast past EPS growth that slows afterward",
195 + "evidence_type": "figure",
196 + "evidence_ref": "Figure 2"
197 + },
198 + {
199 + "claim": "Analyst forecast errors are predictable, consistent with overreaction",
200 + "evidence_type": "table",
201 + "evidence_ref": "Table II"
202 + },
203 + {
204 + "claim": "The SMM-estimated diagnostic model reproduces key data patterns",
205 + "evidence_type": "table",
206 + "evidence_ref": "Table III; Figure 6"
207 + },
208 + {
209 + "claim": "The LLTG-minus-HLTG return spread widens for firms with more volatile or persistent fundamentals",
210 + "evidence_type": "table",
211 + "evidence_ref": "Table IV"
212 + },
213 + {
214 + "claim": "High-LTG group has a fat right tail of top performers, and analysts overstate its frequency",
215 + "evidence_type": "figure",
216 + "evidence_ref": "Figures 7-8"
217 + },
218 + {
219 + "claim": "Expectations revert even without news, reflecting true mean reversion in earnings growth",
220 + "evidence_type": "figure",
221 + "evidence_ref": "Figure 9"
222 + }
223 + ],
224 + "structural_summary": "Opens by revisiting a known return anomaly and frames it as a puzzle. Lists three innovations: a portable psychological model, a joint account of fundamentals, expectations and returns, and new predictions that set the model apart from adaptive expectations. Data and stylized facts come first. Next, a diagnostic Kalman filter model is developed and mapped to those facts. Overreaction is then quantified with forecast-error regressions and SMM estimation. Kernel-of-truth predictions are tested cross-sectionally. The conclusion restates portability and points to extensions.",
225 + "writing_profile_keywords": [
226 + "behavioral asset pricing",
227 + "representativeness",
228 + "diagnostic expectations",
229 + "portfolio sorts",
230 + "kernel of truth",
231 + "SMM estimation",
232 + "analyst forecasts"
233 + ],
234 + "confidence": 0.8,
235 + "notes": "Primary method is ambiguous: the paper mixes portfolio-sort evidence with SMM estimation. Bootstrapped means are mentioned in the figure captions, but whether confidence bands are plotted is not visible."
236 + },
237 + "deterministic_checks": {
238 + "roadmap_regex": false,
239 + "contribution_regex_in_intro": false,
240 + "magnitude_in_abstract": false,
241 + "abstract_n_sentences": 5,
242 + "intro_n_paragraphs": 6,
243 + "intro_n_words": 791,
244 + "causal_words_in_abstract": 0,
245 + "hedge_words_in_abstract": 1,
246 + "stars_mentioned": true,
247 + "clustered_se_mentioned": false,
248 + "internet_appendix_mentioned": false,
249 + "section_roles_rule": {
250 + "0": "introduction",
251 + "1": "data",
252 + "2": "other",
253 + "3": "theory",
254 + "4": "theory",
255 + "5": "other",
256 + "6": "other",
257 + "7": "conclusion"
258 + }
259 + },
260 + "schema_valid": true,
261 + "code_version": "87c73cd-dirty"
262 +}
\ No newline at end of file
added data/annotations/documents/jof_12834.json +221 −0
@@ -0,0 +1,221 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12834",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:11:40+00:00",
9 + "source_sha256": "d8812a99c1cfac8df8e9180a4db4c03a11667543f6689ba281b718b603a1d06b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 9321,
12 + "usage": {
13 + "input": 3306,
14 + "output": 1692,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.047858,
19 + "request_id": "req_011CggqgU1dCjbFbp3AVf6A9",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "corporate governance / organizational economics (cheap talk, control rights)",
24 + "primary_method": "theoretical_model_analytical",
25 + "identification": {
26 + "estimand": "not applicable (theoretical model)",
27 + "strategy": "not applicable; equilibrium analysis of a principal-agent cheap-talk model with costly intervention",
28 + "key_assumptions": [
29 + "incomplete contracts: actions and messages not contractible",
30 + "top-down information structure with privately informed principal",
31 + "cheap talk communication a la Crawford and Sobel",
32 + "agent biased toward overinvestment",
33 + "costly intervention by principal"
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": 1
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": 2
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": "other"
70 + },
71 + {
72 + "paragraph_index": 1,
73 + "move": "motivation"
74 + },
75 + {
76 + "paragraph_index": 1,
77 + "move": "literature_gap"
78 + },
79 + {
80 + "paragraph_index": 1,
81 + "move": "preview_of_results"
82 + },
83 + {
84 + "paragraph_index": 1,
85 + "move": "policy_or_broader_implication"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "preview_of_approach"
90 + }
91 + ],
92 + "contribution_paragraph_index": 1,
93 + "contribution_types": [
94 + "new_theory_or_model"
95 + ],
96 + "roadmap_present": false,
97 + "literature_positioning_location": "inside_introduction"
98 + },
99 + "section_roles": [
100 + {
101 + "idx": 0,
102 + "role": "introduction"
103 + },
104 + {
105 + "idx": 1,
106 + "role": "theory"
107 + },
108 + {
109 + "idx": 2,
110 + "role": "theory"
111 + },
112 + {
113 + "idx": 3,
114 + "role": "other"
115 + },
116 + {
117 + "idx": 4,
118 + "role": "conclusion"
119 + },
120 + {
121 + "idx": 5,
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": "main_text_section",
132 + "types": [
133 + "other"
134 + ]
135 + },
136 + "tables": {
137 + "convention_notes": "No tables. Results are stated as propositions, with proofs in Appendix A and additional material in an Internet Appendix. Figures plot numerical examples at stated parameter values.",
138 + "units_stated_in_notes": false,
139 + "standard_errors_description": "none"
140 + },
141 + "figures": {
142 + "uncertainty_shown": "no"
143 + },
144 + "conclusion": {
145 + "bounded_claims": false,
146 + "policy_implications": true,
147 + "open_questions": false,
148 + "restates_contribution": true
149 + },
150 + "hedging": {
151 + "level": "low",
152 + "typical_devices": [
153 + "conditional statements tied to parameter regions (e.g., low intervention cost, severe agency problem)",
154 + "modal verbs such as 'can harm' and 'can reduce'"
155 + ]
156 + },
157 + "claim_evidence": [
158 + {
159 + "claim": "A principal's power to intervene limits her ability to communicate private information effectively.",
160 + "evidence_type": "equation_or_proof",
161 + "evidence_ref": "Section I.C; Appendix A"
162 + },
163 + {
164 + "claim": "Intervention can harm the principal, especially when the intervention cost is low or the agency problem is severe.",
165 + "evidence_type": "equation_or_proof",
166 + "evidence_ref": "Section I.D.1; Figure 2"
167 + },
168 + {
169 + "claim": "Welfare effects of intervention on the agent are characterized.",
170 + "evidence_type": "figure",
171 + "evidence_ref": "Section I.D.2; Figure 3"
172 + },
173 + {
174 + "claim": "Expected intervention is lower in equilibrium with effective communication than without communication.",
175 + "evidence_type": "equation_or_proof",
176 + "evidence_ref": "Internet Appendix Section I.D"
177 + },
178 + {
179 + "claim": "Pay-for-performance contracts alter the principal's welfare from intervention.",
180 + "evidence_type": "figure",
181 + "evidence_ref": "Section II.A; Figure 4"
182 + }
183 + ],
184 + "structural_summary": "The paper opens with an epigraph, then motivates the topic with governance practice and notes that the interplay of communication and control is poorly understood. It states the main result and its implications up front, then describes the cheap-talk model. The main Analysis section covers setup, equilibrium, the effect of intervention on communication, and welfare. Extensions relax assumptions: incentive pay, an agent's intervention cost, and an informed agent. The paper then derives testable predictions and applies them to managerial leadership, boards, private equity, and activism. A brief conclusion restates the result, and formal proofs are in the appendix.",
185 + "writing_profile_keywords": [
186 + "cheap talk",
187 + "principal-agent",
188 + "control rights",
189 + "intervention",
190 + "corporate governance",
191 + "counterintuitive result",
192 + "propositions and proofs",
193 + "applications section"
194 + ],
195 + "confidence": 0.85,
196 + "notes": "Pure theory paper. Empirical predictions are offered in Section III but not tested. The introduction excerpt is truncated, so later intro moves (literature positioning, roadmap) are not visible; the roadmap is set to false based on what is visible."
197 + },
198 + "deterministic_checks": {
199 + "roadmap_regex": false,
200 + "contribution_regex_in_intro": false,
201 + "magnitude_in_abstract": false,
202 + "abstract_n_sentences": 5,
203 + "intro_n_paragraphs": 2,
204 + "intro_n_words": 231,
205 + "causal_words_in_abstract": 1,
206 + "hedge_words_in_abstract": 0,
207 + "stars_mentioned": false,
208 + "clustered_se_mentioned": false,
209 + "internet_appendix_mentioned": true,
210 + "section_roles_rule": {
211 + "0": "introduction",
212 + "1": "other",
213 + "2": "robustness",
214 + "3": "other",
215 + "4": "conclusion",
216 + "5": "appendix"
217 + }
218 + },
219 + "schema_valid": true,
220 + "code_version": "87c73cd-dirty"
221 +}
\ No newline at end of file
added data/annotations/documents/jof_12835.json +286 −0
@@ -0,0 +1,286 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [],
4 + "doc_id": "jof_12835",
5 + "ontology_version": "1.0",
6 + "prompt_version": "1.0",
7 + "model": "claude-opus-5-5",
8 + "effort": "medium",
9 + "annotated_at": "2026-11-03T23:27:07+00:00",
10 + "source_sha256": "82d626ab449347878f92e774556359ef408100d9c56ef91da3c53c684dcfc7c1",
11 + "parser": "pymupdf-jof-heuristics-1.3",
12 + "payload_chars": 21871,
13 + "usage": {
14 + "input": 7359,
15 + "output": 2666,
16 + "cache_read": 3968,
17 + "cache_write": 0
18 + },
19 + "cost_usd": 0.08355,
20 + "request_id": "req_011CggrrBzrjTiphppSbjAMY",
21 + "annotation": {
22 + "genre": "empirical",
23 + "subfield": "behavioral_finance",
24 + "subfield_secondary": "household_finance",
25 + "primary_method": "survey_or_experiment",
26 + "identification": {
27 + "estimand": "Difference in subjective return expectations when the question format (forecast returns vs. prices) and the chart format (return bar charts vs. price line charts) are varied.",
28 + "strategy": "Randomized 2x2 between-subject experimental design across three studies: lab subjects, lab subjects with incentives and added explanation tests, and finance professionals with real-world information. Treatment effects are estimated with panel regressions including demographic controls.",
29 + "key_assumptions": [
30 + "Random assignment to task and stimulus treatments",
31 + "Format changes are superficial and leave the information content unchanged",
32 + "Elicited forecasts reflect subjects' true expectations"
33 + ],
34 + "threats_addressed": [
35 + "Confounding of chart format with question format, separated by the 2x2 design",
36 + "Lack of incentives, addressed with performance-based incentives in Study 2",
37 + "Lack of expertise or real-world information, addressed with professionals in Study 3",
38 + "Alternative explanations such as dividend yield, risk premium and simple heuristics",
39 + "Price level of the charts"
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": false,
51 + "sentence_index": -1
52 + },
53 + "method": {
54 + "present": true,
55 + "sentence_index": 2
56 + },
57 + "results": {
58 + "present": true,
59 + "sentence_index": 3
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": "tension_or_puzzle"
79 + },
80 + {
81 + "paragraph_index": 0,
82 + "move": "preview_of_approach"
83 + },
84 + {
85 + "paragraph_index": 0,
86 + "move": "preview_of_results"
87 + },
88 + {
89 + "paragraph_index": 1,
90 + "move": "preview_of_results"
91 + },
92 + {
93 + "paragraph_index": 1,
94 + "move": "mechanism_or_theory_sketch"
95 + },
96 + {
97 + "paragraph_index": 2,
98 + "move": "policy_or_broader_implication"
99 + },
100 + {
101 + "paragraph_index": 3,
102 + "move": "roadmap"
103 + }
104 + ],
105 + "contribution_paragraph_index": 0,
106 + "contribution_types": [
107 + "new_evidence"
108 + ],
109 + "roadmap_present": true,
110 + "literature_positioning_location": "none"
111 + },
112 + "section_roles": [
113 + {
114 + "idx": 0,
115 + "role": "introduction"
116 + },
117 + {
118 + "idx": 1,
119 + "role": "other"
120 + },
121 + {
122 + "idx": 2,
123 + "role": "empirical_design"
124 + },
125 + {
126 + "idx": 3,
127 + "role": "results"
128 + },
129 + {
130 + "idx": 4,
131 + "role": "results"
132 + },
133 + {
134 + "idx": 5,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 6,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 7,
143 + "role": "conclusion"
144 + },
145 + {
146 + "idx": 8,
147 + "role": "appendix"
148 + },
149 + {
150 + "idx": 9,
151 + "role": "appendix"
152 + },
153 + {
154 + "idx": 10,
155 + "role": "appendix"
156 + },
157 + {
158 + "idx": 11,
159 + "role": "appendix"
160 + },
161 + {
162 + "idx": 12,
163 + "role": "appendix"
164 + },
165 + {
166 + "idx": 13,
167 + "role": "appendix"
168 + },
169 + {
170 + "idx": 14,
171 + "role": "appendix"
172 + }
173 + ],
174 + "results_presentation": {
175 + "economic_significance_discussed": true,
176 + "statistical_significance_language": "mixed",
177 + "magnitude_statements_in_intro": true
178 + },
179 + "robustness": {
180 + "location": "mixed",
181 + "types": [
182 + "alternative_specification",
183 + "subsample",
184 + "other"
185 + ]
186 + },
187 + "tables": {
188 + "convention_notes": "Table notes define the dependent variable and the treatment dummies. They specify the estimator (fixed or random effects panel regressions, probit marginal effects) and list demographic controls. Descriptive tables use nonparametric Mann-Whitney tests to compare samples.",
189 + "units_stated_in_notes": true,
190 + "standard_errors_description": "clustered"
191 + },
192 + "figures": {
193 + "uncertainty_shown": "mixed"
194 + },
195 + "conclusion": {
196 + "bounded_claims": false,
197 + "policy_implications": true,
198 + "open_questions": true,
199 + "restates_contribution": true
200 + },
201 + "hedging": {
202 + "level": "low",
203 + "typical_devices": [
204 + "'can be explained in part'",
205 + "'might help explain'",
206 + "'consistent across all studies'"
207 + ]
208 + },
209 + "claim_evidence": [
210 + {
211 + "claim": "Forecasting returns rather than prices raises expectations; return charts rather than price charts lower them",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table III, Figure 6"
214 + },
215 + {
216 + "claim": "Format effects are larger for subjects relying on intuition (low cognitive reflection)",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table IV"
219 + },
220 + {
221 + "claim": "Format use in the field varies widely and arbitrarily, including after KIID regulation",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table I, Figure 1"
224 + },
225 + {
226 + "claim": "Price charts induce extrapolation of recent returns; return charts plus return questions draw on full history",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table VIII"
229 + },
230 + {
231 + "claim": "Reluctance to state negative numbers contributes to the format effects",
232 + "evidence_type": "table",
233 + "evidence_ref": "Table V, Table VI"
234 + },
235 + {
236 + "claim": "Chart and question formats affect forecast accuracy",
237 + "evidence_type": "table",
238 + "evidence_ref": "Table IX, Figure 8"
239 + }
240 + ],
241 + "structural_summary": "The paper opens with normative invariance as the benchmark and poses whether price and return formats are equivalent. It previews an experimental design that separates chart format from question format and reports effect magnitudes in the introduction, then notes implications for robo advisors and regulation. Field evidence on how formats are used comes first, followed by the design of the three studies. Results proceed from main treatment effects to mechanisms (intuition and number processing), then attention to past returns, then forecast accuracy. The conclusion closes with policy recommendations.",
242 + "writing_profile_keywords": [
243 + "experimental finance",
244 + "between-subject design",
245 + "format effects",
246 + "expectations",
247 + "cognitive reflection",
248 + "policy relevance",
249 + "robo advisors"
250 + ],
251 + "confidence": 0.8,
252 + "notes": "No explicit literature-positioning paragraph is visible in the excerpted introduction, so literature_positioning_location is set to 'none'. The introduction text is truncated, so it may in fact contain literature positioning. Appendix headings indicate further robustness checks, and an Internet Appendix is referenced."
253 + },
254 + "deterministic_checks": {
255 + "roadmap_regex": true,
256 + "contribution_regex_in_intro": true,
257 + "magnitude_in_abstract": false,
258 + "abstract_n_sentences": 6,
259 + "intro_n_paragraphs": 4,
260 + "intro_n_words": 653,
261 + "causal_words_in_abstract": 1,
262 + "hedge_words_in_abstract": 0,
263 + "stars_mentioned": false,
264 + "clustered_se_mentioned": true,
265 + "internet_appendix_mentioned": true,
266 + "section_roles_rule": {
267 + "0": "introduction",
268 + "1": "other",
269 + "2": "other",
270 + "3": "other",
271 + "4": "other",
272 + "5": "other",
273 + "6": "other",
274 + "7": "conclusion",
275 + "8": "appendix",
276 + "9": "appendix",
277 + "10": "appendix",
278 + "11": "appendix",
279 + "12": "appendix",
280 + "13": "appendix",
281 + "14": "appendix"
282 + }
283 + },
284 + "schema_valid": true,
285 + "code_version": "87c73cd-dirty"
286 +}
\ No newline at end of file
added data/annotations/documents/jof_12836.json +305 −0
@@ -0,0 +1,305 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12836",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:33+00:00",
9 + "source_sha256": "2cd048a43f0ac728090e3b88fca9249baed456c9fc07d991aae82e24fc66f344",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 28574,
12 + "usage": {
13 + "input": 9727,
14 + "output": 2317,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.086042,
19 + "request_id": "req_011CggqbFPiUXWGbNJTiJhvW",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Return predictability and risk-exposure associations of the market-to-value and value-to-book components of market-to-book",
27 + "strategy": "RRV valuation-model decomposition of market-to-book, followed by portfolio sorts, Fama-MacBeth regressions, beta-spread comparisons and second-pass pricing tests on component-sorted portfolios",
28 + "key_assumptions": [
29 + "Industry multiples-based valuation model captures fundamental value",
30 + "Covariances and characteristics are estimated with precision (joint hypothesis)"
31 + ],
32 + "threats_addressed": [
33 + "joint hypothesis problem acknowledged",
34 + "microstructure noise in equal-weighted returns (RW/VW weighting)",
35 + "small-stock dominance (size-conditional sorts)",
36 + "industry classification choice (alternatives for robustness)"
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": "tension_or_puzzle"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "contribution_statement"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_approach"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 3,
91 + "move": "mechanism_or_theory_sketch"
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": "preview_of_results"
104 + },
105 + {
106 + "paragraph_index": 7,
107 + "move": "preview_of_results"
108 + },
109 + {
110 + "paragraph_index": 8,
111 + "move": "preview_of_results"
112 + },
113 + {
114 + "paragraph_index": 9,
115 + "move": "preview_of_results"
116 + },
117 + {
118 + "paragraph_index": 10,
119 + "move": "preview_of_results"
120 + },
121 + {
122 + "paragraph_index": 11,
123 + "move": "preview_of_results"
124 + },
125 + {
126 + "paragraph_index": 12,
127 + "move": "preview_of_results"
128 + },
129 + {
130 + "paragraph_index": 13,
131 + "move": "contribution_statement"
132 + },
133 + {
134 + "paragraph_index": 13,
135 + "move": "other"
136 + },
137 + {
138 + "paragraph_index": 14,
139 + "move": "roadmap"
140 + }
141 + ],
142 + "contribution_paragraph_index": 1,
143 + "contribution_types": [
144 + "new_evidence"
145 + ],
146 + "roadmap_present": true,
147 + "literature_positioning_location": "both"
148 + },
149 + "section_roles": [
150 + {
151 + "idx": 0,
152 + "role": "introduction"
153 + },
154 + {
155 + "idx": 1,
156 + "role": "literature"
157 + },
158 + {
159 + "idx": 2,
160 + "role": "data"
161 + },
162 + {
163 + "idx": 3,
164 + "role": "results"
165 + },
166 + {
167 + "idx": 4,
168 + "role": "results"
169 + },
170 + {
171 + "idx": 5,
172 + "role": "results"
173 + },
174 + {
175 + "idx": 6,
176 + "role": "conclusion"
177 + },
178 + {
179 + "idx": 7,
180 + "role": "appendix"
181 + }
182 + ],
183 + "results_presentation": {
184 + "economic_significance_discussed": true,
185 + "statistical_significance_language": "mixed",
186 + "magnitude_statements_in_intro": true
187 + },
188 + "robustness": {
189 + "location": "mixed",
190 + "types": [
191 + "alternative_specification",
192 + "alternative_measure",
193 + "subsample"
194 + ]
195 + },
196 + "tables": {
197 + "convention_notes": "Notes state the sample period, the estimation method (Fama-MacBeth, ordered logit, second-pass OLS), the weighting scheme (RW/VW), units (monthly returns in percent) and the panel structure. P-values appear in parentheses for some tables.",
198 + "units_stated_in_notes": true,
199 + "standard_errors_description": "unclear"
200 + },
201 + "figures": {
202 + "uncertainty_shown": "unclear"
203 + },
204 + "conclusion": {
205 + "bounded_claims": true,
206 + "policy_implications": false,
207 + "open_questions": true,
208 + "restates_contribution": true
209 + },
210 + "hedging": {
211 + "level": "moderate",
212 + "typical_devices": [
213 + "joint hypothesis caveat",
214 + "'unlikely explanation' phrasing",
215 + "'consistent with' language",
216 + "explicit disclaimer of not taking a stance on whether characteristics are priced",
217 + "'mixed evidence' labeling"
218 + ]
219 + },
220 + "claim_evidence": [
221 + {
222 + "claim": "Market-to-value drives the entire value strategy return; value-to-book has insignificant return spread",
223 + "evidence_type": "table",
224 + "evidence_ref": "Table II, Table III, Figure 1"
225 + },
226 + {
227 + "claim": "IST shock exposure of value strategy stems mainly from value-to-book",
228 + "evidence_type": "table",
229 + "evidence_ref": "Table VI"
230 + },
231 + {
232 + "claim": "Assets-in-place/operating leverage differences arise from value-to-book only",
233 + "evidence_type": "table",
234 + "evidence_ref": "Table VII"
235 + },
236 + {
237 + "claim": "Cash flow duration differences derive from unpriced value-to-book",
238 + "evidence_type": "table",
239 + "evidence_ref": "Table VIII"
240 + },
241 + {
242 + "claim": "Analyst risk rating association with market-to-book driven by value-to-book",
243 + "evidence_type": "table",
244 + "evidence_ref": "Table IX"
245 + },
246 + {
247 + "claim": "Cash flow risk evidence is mixed depending on approach",
248 + "evidence_type": "table",
249 + "evidence_ref": "Table IV"
250 + },
251 + {
252 + "claim": "Ultimate consumption risk in returns is linked to market-to-value; Parker-Julliard model prices test assets",
253 + "evidence_type": "table",
254 + "evidence_ref": "Table V, Table X, Figure 2"
255 + },
256 + {
257 + "claim": "Earnings announcement surprises reverse around portfolio formation for market-to-value, consistent with overextrapolation",
258 + "evidence_type": "table",
259 + "evidence_ref": "Table XI"
260 + },
261 + {
262 + "claim": "Market-to-value premium concentrates where limits to arbitrage bind and when arbitrage capital is low",
263 + "evidence_type": "table",
264 + "evidence_ref": "Tables XII-XIV"
265 + }
266 + ],
267 + "structural_summary": "The paper motivates with the value premium and the unresolved debate over its source. It applies the RRV decomposition to split market-to-book into a priced component and an unpriced component. It then systematically tests whether characteristics and risk exposures proposed by prominent theories load on the priced component. Results proceed in three blocks: return predictability, aggregate risk and production-based explanations, and mispricing/limits to arbitrage. The intro previews each theory's test in turn, and the paper closes with a joint-hypothesis caveat.",
268 + "writing_profile_keywords": [
269 + "decomposition",
270 + "theory horse-race",
271 + "value premium",
272 + "portfolio sorts",
273 + "replication of prior tests",
274 + "limits to arbitrage",
275 + "joint hypothesis caveat"
276 + ],
277 + "confidence": 0.85,
278 + "notes": "Standard errors are not described in the visible table notes. Section IV is labeled 'theory' by the role rule, but it consists of empirical tests of theories, so it is annotated as results."
279 + },
280 + "deterministic_checks": {
281 + "roadmap_regex": true,
282 + "contribution_regex_in_intro": false,
283 + "magnitude_in_abstract": false,
284 + "abstract_n_sentences": 5,
285 + "intro_n_paragraphs": 15,
286 + "intro_n_words": 1777,
287 + "causal_words_in_abstract": 0,
288 + "hedge_words_in_abstract": 0,
289 + "stars_mentioned": false,
290 + "clustered_se_mentioned": false,
291 + "internet_appendix_mentioned": true,
292 + "section_roles_rule": {
293 + "0": "introduction",
294 + "1": "literature",
295 + "2": "data",
296 + "3": "other",
297 + "4": "theory",
298 + "5": "other",
299 + "6": "conclusion",
300 + "7": "appendix"
301 + }
302 + },
303 + "schema_valid": true,
304 + "code_version": "87c73cd-dirty"
305 +}
\ No newline at end of file
added data/annotations/documents/jof_12837.json +260 −0
@@ -0,0 +1,260 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12837",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:20+00:00",
9 + "source_sha256": "a999f0a530169795fdf2654df34e26b08f0c4fe343810fcac008a57ce37312c5",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 18909,
12 + "usage": {
13 + "input": 6634,
14 + "output": 2158,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07049,
19 + "request_id": "req_011CggqaJqYiE3XnKPqZDv5j",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Sensitivity of net equity repurchases and net debt issuance to lagged debt- and equity-market expected excess returns (valuation measures)",
27 + "strategy": "Firm-level panel and aggregate time-series regressions of financing flows on lagged valuation proxies with fundamental controls; logit models for simultaneous issuance/repurchase; return-forecasting regressions of bond returns on firm actions controlling for hedge-ratio-adjusted stock returns; additional tests using market-specific sentiment and supply shocks (e.g., forecast errors)",
28 + "key_assumptions": [
29 + "Valuation proxies capture expected excess returns rather than fundamentals",
30 + "Control variables adequately absorb fundamental drivers of financing such as cash flows, investment demand and the business cycle",
31 + "Hedge-ratio benchmarks correctly capture the comovement of bonds and stocks implied by fundamentals"
32 + ],
33 + "threats_addressed": [
34 + "Fundamentals-driven financing, addressed through controls",
35 + "Alternative definitions of simultaneous issuance (above-average thresholds)",
36 + "Firm heterogeneity, addressed through sorts by size, profitability and KZ index"
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": false,
48 + "sentence_index": -1
49 + },
50 + "method": {
51 + "present": false,
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": 0
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 1,
75 + "move": "literature_gap"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "contribution_statement"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_results"
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": "robustness_mention"
100 + },
101 + {
102 + "paragraph_index": 5,
103 + "move": "preview_of_results"
104 + },
105 + {
106 + "paragraph_index": 6,
107 + "move": "related_literature_positioning"
108 + },
109 + {
110 + "paragraph_index": 7,
111 + "move": "roadmap"
112 + }
113 + ],
114 + "contribution_paragraph_index": 1,
115 + "contribution_types": [
116 + "new_evidence"
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": "results"
137 + },
138 + {
139 + "idx": 4,
140 + "role": "results"
141 + },
142 + {
143 + "idx": 5,
144 + "role": "results"
145 + },
146 + {
147 + "idx": 6,
148 + "role": "conclusion"
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 + "alternative_measure",
160 + "subsample"
161 + ]
162 + },
163 + "tables": {
164 + "convention_notes": "Table notes write out the estimating equation (panel, logit or time-series) and define the dependent variables, which are normalized by lagged assets. Panels split results by outcome or by firm group. The motivating table uses a firm example with units stated in millions of dollars.",
165 + "units_stated_in_notes": true,
166 + "standard_errors_description": "unclear"
167 + },
168 + "figures": {
169 + "uncertainty_shown": "unclear"
170 + },
171 + "conclusion": {
172 + "bounded_claims": false,
173 + "policy_implications": false,
174 + "open_questions": true,
175 + "restates_contribution": true
176 + },
177 + "hedging": {
178 + "level": "moderate",
179 + "typical_devices": [
180 + "'to the extent that' qualifiers",
181 + "'helps account for' rather than 'explains'",
182 + "'tend to'",
183 + "framing returns as possibly understating firm gains"
184 + ]
185 + },
186 + "claim_evidence": [
187 + {
188 + "claim": "Net equity repurchases and net debt issuance both rise when expected excess debt returns are low or expected equity returns are high",
189 + "evidence_type": "table",
190 + "evidence_ref": "Table III"
191 + },
192 + {
193 + "claim": "Simultaneous issuance in one market and repurchase in the other is more likely under the corresponding relative valuations",
194 + "evidence_type": "table",
195 + "evidence_ref": "Table IV"
196 + },
197 + {
198 + "claim": "Cross-market spillovers are strongest among large, unconstrained firms",
199 + "evidence_type": "table",
200 + "evidence_ref": "Table V"
201 + },
202 + {
203 + "claim": "After firms issue debt and buy back equity, debt subsequently underperforms hedge-ratio-adjusted equity by about 0.5-0.6 pp per year",
204 + "evidence_type": "table",
205 + "evidence_ref": "Table VI"
206 + },
207 + {
208 + "claim": "Aggregate financing flows show the same cross-market valuation patterns",
209 + "evidence_type": "table",
210 + "evidence_ref": "Table VII"
211 + },
212 + {
213 + "claim": "A large share of financing comes from firms simultaneously issuing in one market and repurchasing in another",
214 + "evidence_type": "figure",
215 + "evidence_ref": "Figure 1, Table I"
216 + },
217 + {
218 + "claim": "Market-specific valuation shocks, such as bond-yield forecast errors, drive financing decisions",
219 + "evidence_type": "table",
220 + "evidence_ref": "Table VIII"
221 + }
222 + ],
223 + "structural_summary": "The introduction opens with established single-market timing evidence and pivots to a multi-market gap. It then previews firm-level and aggregate results, including return magnitudes, positions the paper briefly against prior work and closes with a roadmap. The body moves from motivating stylized facts to a conceptual framework that generates predictions. Firm-level tests come next (baseline regressions, simultaneous-issuance logits, heterogeneity, return forecasting), followed by the aggregate analogue and additional evidence on sentiment and supply shocks. A short conclusion restates the findings and points to future research.",
224 + "writing_profile_keywords": [
225 + "market timing",
226 + "cross-market arbitrage",
227 + "capital structure",
228 + "debt-equity substitution",
229 + "firm-level and aggregate",
230 + "prediction-guided empirics",
231 + "return predictability"
232 + ],
233 + "confidence": 0.8,
234 + "notes": "Standard-error and significance conventions are not visible in the excerpts. The conceptual framework is verbal or organizing rather than formal; equations appear only in the empirical sections."
235 + },
236 + "deterministic_checks": {
237 + "roadmap_regex": true,
238 + "contribution_regex_in_intro": true,
239 + "magnitude_in_abstract": false,
240 + "abstract_n_sentences": 5,
241 + "intro_n_paragraphs": 8,
242 + "intro_n_words": 950,
243 + "causal_words_in_abstract": 0,
244 + "hedge_words_in_abstract": 0,
245 + "stars_mentioned": false,
246 + "clustered_se_mentioned": false,
247 + "internet_appendix_mentioned": false,
248 + "section_roles_rule": {
249 + "0": "introduction",
250 + "1": "other",
251 + "2": "theory",
252 + "3": "results",
253 + "4": "results",
254 + "5": "results",
255 + "6": "conclusion"
256 + }
257 + },
258 + "schema_valid": true,
259 + "code_version": "87c73cd-dirty"
260 +}
\ No newline at end of file
added data/annotations/documents/jof_12838.json +254 −0
@@ -0,0 +1,254 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12838",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:57+00:00",
9 + "source_sha256": "cedb6938a7499baf406f885c250a2fc267bbc11fc64238eb7958f2663237635b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 19830,
12 + "usage": {
13 + "input": 6824,
14 + "output": 3044,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.08897,
19 + "request_id": "req_011CggqXyTSCGQhDCwQ9nDa7",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "international_finance",
24 + "primary_method": "other",
25 + "identification": {
26 + "estimand": "How strongly dealers' quote changes co-move after efficient-price shocks (network links), and how a dealer's network centrality relates to how informative its quotes are, how fast it absorbs shocks, and how wide its spreads are.",
27 + "strategy": "Hasbrouck-style VECM with a common efficient-price trend. Partial correlations of multivariate impulse responses form edges, and price inefficiencies give node positions. The authors run Granger causality tests between centrality and inefficiency, condition on public news releases, and use the surprise 2015 removal of the EUR/CHF peg as a natural experiment for how the network forms.",
28 + "key_assumptions": [
29 + "Dealer and EBS quotes are cointegrated with a single common stochastic trend that can be read as the efficient price.",
30 + "Quote synchronicity after shocks reflects information linkages rather than physical trading connections.",
31 + "The removal of the peg was an exogenous move from a no-learning regime to a learning regime."
32 + ],
33 + "threats_addressed": [
34 + "Confusing co-activity networks with networks based on physical trade links.",
35 + "Choice of impulse-response horizon for the network snapshot."
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": 2
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": 1,
74 + "move": "motivation"
75 + },
76 + {
77 + "paragraph_index": 1,
78 + "move": "literature_gap"
79 + },
80 + {
81 + "paragraph_index": 1,
82 + "move": "contribution_statement"
83 + },
84 + {
85 + "paragraph_index": 1,
86 + "move": "preview_of_approach"
87 + },
88 + {
89 + "paragraph_index": 2,
90 + "move": "related_literature_positioning"
91 + },
92 + {
93 + "paragraph_index": 2,
94 + "move": "preview_of_approach"
95 + },
96 + {
97 + "paragraph_index": 3,
98 + "move": "preview_of_results"
99 + },
100 + {
101 + "paragraph_index": 4,
102 + "move": "identification_defense"
103 + },
104 + {
105 + "paragraph_index": 4,
106 + "move": "preview_of_results"
107 + },
108 + {
109 + "paragraph_index": 4,
110 + "move": "roadmap"
111 + }
112 + ],
113 + "contribution_paragraph_index": 1,
114 + "contribution_types": [
115 + "new_evidence",
116 + "new_method",
117 + "new_data_or_measure"
118 + ],
119 + "roadmap_present": true,
120 + "literature_positioning_location": "inside_introduction"
121 + },
122 + "section_roles": [
123 + {
124 + "idx": 0,
125 + "role": "introduction"
126 + },
127 + {
128 + "idx": 1,
129 + "role": "theory"
130 + },
131 + {
132 + "idx": 2,
133 + "role": "empirical_design"
134 + },
135 + {
136 + "idx": 3,
137 + "role": "empirical_design"
138 + },
139 + {
140 + "idx": 4,
141 + "role": "data"
142 + },
143 + {
144 + "idx": 5,
145 + "role": "results"
146 + },
147 + {
148 + "idx": 6,
149 + "role": "conclusion"
150 + },
151 + {
152 + "idx": 7,
153 + "role": "appendix"
154 + }
155 + ],
156 + "results_presentation": {
157 + "economic_significance_discussed": false,
158 + "statistical_significance_language": "unclear",
159 + "magnitude_statements_in_intro": false
160 + },
161 + "robustness": {
162 + "location": "none_seen",
163 + "types": []
164 + },
165 + "tables": {
166 + "convention_notes": "Table notes define each network or information measure (adjacency matrix as partial correlations, InfoShare lower bound, InfoSpeed, centrality thresholds 0.2/0.1) and the sample construction (five-minute subsamples, dealer markets DLR1-8 plus EBS). Titles are often phrased as hypothesis questions.",
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 + "'largely support theory'",
183 + "'in a sense'",
184 + "'to the best of our knowledge'",
185 + "'we believe'",
186 + "'hopefully inspire future researchers'"
187 + ]
188 + },
189 + "claim_evidence": [
190 + {
191 + "claim": "The information-revelation network is nontrivial, stable over time, and has a core-periphery structure.",
192 + "evidence_type": "table",
193 + "evidence_ref": "Table III; Table I"
194 + },
195 + {
196 + "claim": "Central dealers' quotes are more informative at the shock and reach the full impact faster.",
197 + "evidence_type": "figure",
198 + "evidence_ref": "Figure 4; Table II; Table IV"
199 + },
200 + {
201 + "claim": "Central dealers quote wider bid-ask spreads, consistent with recouping adverse-selection costs.",
202 + "evidence_type": "figure",
203 + "evidence_ref": "Figure 5"
204 + },
205 + {
206 + "claim": "When part of the information is released publicly via news, links weaken and revelation slows, consistent with reduced search.",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table V"
209 + },
210 + {
211 + "claim": "After the peg removal the market centralized on EBS, then gradually decentralized toward the steady-state network.",
212 + "evidence_type": "figure",
213 + "evidence_ref": "Figure 6; Table AI; Figure A1"
214 + }
215 + ],
216 + "structural_summary": "Opens with the fragmentation of trading as motivation and a gap: work on information aggregation in dealer networks is only theoretical. The paper then derives five hypotheses from that theory. Next it builds an econometric network map from a VECM, with impulse-response partial correlations as edges and price inefficiency as node positions, and details estimation and inference. It describes the hybrid FX market and the EUR/CHF quote data, and tests each hypothesis in its own results subsection. The 2015 peg-removal crash serves as a natural experiment on how the network forms. It closes by restating the findings, noting the method applies to other markets, and posing open questions about welfare and regulation.",
217 + "writing_profile_keywords": [
218 + "hypothesis-testing structure",
219 + "theory-motivated empirics",
220 + "network methodology",
221 + "cross-disciplinary analogy (neuroscience)",
222 + "natural experiment",
223 + "FX microstructure",
224 + "descriptive network maps"
225 + ],
226 + "confidence": 0.75,
227 + "notes": "Primary method is coded 'other' because the core tool is a VECM/impulse-response network construction, not standard regression. Subsection D (Hypothesis 4) is missing from the section list but is implied. Statistical inference details are in Section III.B but are not visible in the excerpts. No explicit robustness section is visible."
228 + },
229 + "deterministic_checks": {
230 + "roadmap_regex": true,
231 + "contribution_regex_in_intro": true,
232 + "magnitude_in_abstract": false,
233 + "abstract_n_sentences": 7,
234 + "intro_n_paragraphs": 5,
235 + "intro_n_words": 860,
236 + "causal_words_in_abstract": 0,
237 + "hedge_words_in_abstract": 0,
238 + "stars_mentioned": false,
239 + "clustered_se_mentioned": false,
240 + "internet_appendix_mentioned": false,
241 + "section_roles_rule": {
242 + "0": "introduction",
243 + "1": "theory",
244 + "2": "other",
245 + "3": "other",
246 + "4": "other",
247 + "5": "results",
248 + "6": "conclusion",
249 + "7": "appendix"
250 + }
251 + },
252 + "schema_valid": true,
253 + "code_version": "87c73cd-dirty"
254 +}
\ No newline at end of file
added data/annotations/documents/jof_12839.json +241 −0
@@ -0,0 +1,241 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12839",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:37+00:00",
9 + "source_sha256": "d52dbcee70f3ac04fd84ef962d76403efb08bf7cc5c9a122486bd736be233315",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 17529,
12 + "usage": {
13 + "input": 5729,
14 + "output": 2576,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07523,
19 + "request_id": "req_011CggqbGrGUfGpf6dcpqk1x",
20 + "annotation": {
21 + "genre": "theoretical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "behavioral_finance",
24 + "primary_method": "calibration_simulation",
25 + "identification": {
26 + "estimand": "not applicable (theory paper); model-implied conditional and unconditional predictability of characteristic-sorted portfolio returns",
27 + "strategy": "Dynamic behavioral model with an economy-wide disruption factor and overconfident investors, calibrated to US data from 1964 to 2014 and analyzed through simulated 50-year histories",
28 + "key_assumptions": [
29 + "risk-neutral investors discounting at a constant rate",
30 + "representative agent with common beliefs and information",
31 + "frictionless financing with Modigliani-Miller holding",
32 + "investors overconfident about their ability to assess the disruption climate",
33 + "zero unconditional return predictability over long samples"
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": 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": 2
55 + },
56 + "magnitude": {
57 + "present": false,
58 + "sentence_index": -1
59 + },
60 + "contribution": {
61 + "present": true,
62 + "sentence_index": 3
63 + }
64 + },
65 + "introduction": {
66 + "moves": [
67 + {
68 + "paragraph_index": 0,
69 + "move": "motivation"
70 + },
71 + {
72 + "paragraph_index": 0,
73 + "move": "literature_gap"
74 + },
75 + {
76 + "paragraph_index": 0,
77 + "move": "mechanism_or_theory_sketch"
78 + },
79 + {
80 + "paragraph_index": 9,
81 + "move": "preview_of_approach"
82 + },
83 + {
84 + "paragraph_index": 9,
85 + "move": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 10,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 10,
93 + "move": "other"
94 + },
95 + {
96 + "paragraph_index": 11,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 11,
101 + "move": "related_literature_positioning"
102 + },
103 + {
104 + "paragraph_index": 12,
105 + "move": "related_literature_positioning"
106 + },
107 + {
108 + "paragraph_index": 13,
109 + "move": "roadmap"
110 + }
111 + ],
112 + "contribution_paragraph_index": 0,
113 + "contribution_types": [
114 + "new_theory_or_model"
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": "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": "none_seen",
152 + "types": []
153 + },
154 + "tables": {
155 + "convention_notes": "Tables report calibrated parameters and statistics from simulated data: betas on the disruption factor, conditional Sharpe ratios, autocorrelations and Sharpe-ratio percentiles. Notes describe how each statistic is computed and point to the corresponding text sections.",
156 + "units_stated_in_notes": true,
157 + "standard_errors_description": "none"
158 + },
159 + "figures": {
160 + "uncertainty_shown": "unclear"
161 + },
162 + "conclusion": {
163 + "bounded_claims": true,
164 + "policy_implications": false,
165 + "open_questions": false,
166 + "restates_contribution": true
167 + },
168 + "hedging": {
169 + "level": "moderate",
170 + "typical_devices": [
171 + "explicit acknowledgment of stylized assumptions",
172 + "statements that the model does not capture all features of the data",
173 + "'up to' qualifiers on simulated frequencies",
174 + "framing as testable implications rather than established facts"
175 + ]
176 + },
177 + "claim_evidence": [
178 + {
179 + "claim": "Portfolios favoring newcomers load positively on disruption and those favoring incumbents load negatively.",
180 + "evidence_type": "table",
181 + "evidence_ref": "Table III"
182 + },
183 + {
184 + "claim": "Overestimating the disruption climate predicts low subsequent returns for newcomer-favoring strategies, and underestimation predicts the reverse.",
185 + "evidence_type": "table",
186 + "evidence_ref": "Table IV"
187 + },
188 + {
189 + "claim": "Characteristic-sorted portfolio returns are positively autocorrelated, so momentum across portfolios is profitable.",
190 + "evidence_type": "table",
191 + "evidence_ref": "Table V"
192 + },
193 + {
194 + "claim": "Return persistence makes high 50-year Sharpe ratios (e.g., 0.40) occur in up to 20% of simulated histories.",
195 + "evidence_type": "table",
196 + "evidence_ref": "Table VI; Figure 1"
197 + },
198 + {
199 + "claim": "The calibration broadly matches salient features of US firm data.",
200 + "evidence_type": "table",
201 + "evidence_ref": "Tables I and II"
202 + }
203 + ],
204 + "structural_summary": "The paper opens by noting the documented characteristic anomalies and the scarcity of work on their conditional behavior. It then builds a continuous-time behavioral model in which a disruption factor shifts profits from incumbents to newcomers and overconfident investors misread that factor. The model is calibrated to US data, and simulated panels are used to show portfolio exposures to disruption, conditional predictability and return autocorrelation. A final analysis asks how often historical-length samples would show high Sharpe ratios. The paper closes by acknowledging its simplifications and stating testable implications.",
205 + "writing_profile_keywords": [
206 + "behavioral asset pricing",
207 + "dynamic model",
208 + "calibration",
209 + "simulation",
210 + "overconfidence",
211 + "value/profitability/asset growth anomalies",
212 + "testable implications",
213 + "stylized model caveats"
214 + ],
215 + "confidence": 0.8,
216 + "notes": "The introduction paragraph list contains broken fragments (indices 1-8 are single words), so moves are assigned to indices 0 and 9-13. The paper is a calibrated theoretical model with no empirical identification. Statistics come from simulated samples, and standard errors are not applicable."
217 + },
218 + "deterministic_checks": {
219 + "roadmap_regex": true,
220 + "contribution_regex_in_intro": false,
221 + "magnitude_in_abstract": false,
222 + "abstract_n_sentences": 4,
223 + "intro_n_paragraphs": 14,
224 + "intro_n_words": 919,
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": "theory",
233 + "2": "theory",
234 + "3": "other",
235 + "4": "other",
236 + "5": "conclusion"
237 + }
238 + },
239 + "schema_valid": true,
240 + "code_version": "87c73cd-dirty"
241 +}
\ No newline at end of file
added data/annotations/documents/jof_12840.json +256 −0
@@ -0,0 +1,256 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12840",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:09:45+00:00",
9 + "source_sha256": "c4e38c400360d0a53d1ec91e5ccab03304b3c3ade9aa64e08bd5924743b8d570",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20034,
12 + "usage": {
13 + "input": 6561,
14 + "output": 2209,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.071218,
19 + "request_id": "req_011CggqXjPqPp19BLMSZMvX8",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "financial_intermediation_banking",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Differential likelihood of predatory selling (versus liquidity provision) by clients of brokers who intermediate fire-sale trades, especially the brokers' best clients; the resulting profits; and the added price impact borne by the liquidating fund",
27 + "strategy": "Ticket- and stock-day-manager-broker-level regressions that compare clients of aware and unaware brokers, and best and other clients, during versus before fire sales, supplemented by a natural experiment from the 2003 late-trading scandal outflows",
28 + "key_assumptions": [
29 + "Fire-sale events identified from volume Z-scores reflect liquidity-motivated rather than information-motivated selling",
30 + "Followers are not trading on a common information signal shared with the originator",
31 + "Outflows from scandal-implicated fund families are exogenous to stock fundamentals"
32 + ],
33 + "threats_addressed": [
34 + "common information signals such as recessions, earnings announcements, analyst recommendation changes and negative press news",
35 + "shared negative-signal strategies such as momentum or high short interest",
36 + "endogeneity of the liquidation source, addressed through the late-trading scandal",
37 + "alternative explanation that followers are providing liquidity rather than predating"
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": 2
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": "mechanism_or_theory_sketch"
77 + },
78 + {
79 + "paragraph_index": 1,
80 + "move": "preview_of_results"
81 + },
82 + {
83 + "paragraph_index": 2,
84 + "move": "robustness_mention"
85 + },
86 + {
87 + "paragraph_index": 2,
88 + "move": "identification_defense"
89 + },
90 + {
91 + "paragraph_index": 3,
92 + "move": "identification_defense"
93 + },
94 + {
95 + "paragraph_index": 3,
96 + "move": "contribution_statement"
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": "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 + ],
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": "data"
134 + },
135 + {
136 + "idx": 2,
137 + "role": "results"
138 + },
139 + {
140 + "idx": 3,
141 + "role": "results"
142 + },
143 + {
144 + "idx": 4,
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": "within_results",
155 + "types": [
156 + "subsample",
157 + "placebo",
158 + "other"
159 + ]
160 + },
161 + "tables": {
162 + "convention_notes": "Table notes state the level at which each regression is run, define the dependent variables and event windows, and explain how the sample is constructed (e.g., excluding the originating managers' own trades). Standard-error treatment is not visible in the excerpts.",
163 + "units_stated_in_notes": true,
164 + "standard_errors_description": "unclear"
165 + },
166 + "figures": {
167 + "uncertainty_shown": "mixed"
168 + },
169 + "conclusion": {
170 + "bounded_claims": true,
171 + "policy_implications": true,
172 + "open_questions": false,
173 + "restates_contribution": true
174 + },
175 + "hedging": {
176 + "level": "moderate",
177 + "typical_devices": [
178 + "'evidence suggests' framing",
179 + "'plausibly exogenous' qualifier",
180 + "modal verbs such as 'might' and 'can'",
181 + "'increases our confidence' phrasing"
182 + ]
183 + },
184 + "claim_evidence": [
185 + {
186 + "claim": "Clients of brokers intermediating fire sales are more likely to predate on the liquidating fund than to provide liquidity",
187 + "evidence_type": "table",
188 + "evidence_ref": "Table II"
189 + },
190 + {
191 + "claim": "Best clients of aware brokers have a net predation probability more than twice that of small clients",
192 + "evidence_type": "table",
193 + "evidence_ref": "Table III"
194 + },
195 + {
196 + "claim": "A substantial fraction (30-42%) of positions sold by other managers during fire sales is bought back within 10 days",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table V"
199 + },
200 + {
201 + "claim": "After the 2003 scandal, clients of the implicated funds' brokers sold the same stocks on the same days",
202 + "evidence_type": "table",
203 + "evidence_ref": "Table VI"
204 + },
205 + {
206 + "claim": "Predation earns roughly 25 bps over about 10 days for best clients of aware brokers",
207 + "evidence_type": "table",
208 + "evidence_ref": "Table VII; Figure 4"
209 + },
210 + {
211 + "claim": "Information leakage raises the distressed fund's liquidation costs by about 40%",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table VIII; Figure 5"
214 + },
215 + {
216 + "claim": "Predators pay higher commissions to aware brokers after events, consistent with a quid pro quo",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table IX"
219 + }
220 + ],
221 + "structural_summary": "The paper motivates the problem through predatory trading around slowly executed large orders and fire sales, and argues that brokers are privileged observers of client order flow. It identifies fire sales in institutional trade-level data and then shows that clients of intermediating brokers, especially the best clients, sell alongside the liquidator and later reverse their positions. Robustness checks exclude news- and signal-driven episodes, and a natural experiment based on the scandal outflows supports the main result. A separate section quantifies predator profits, the price impact on the originator, the persistence of broker-manager relationships, and commission payments. The conclusion discusses implications for execution practice and regulation.",
222 + "writing_profile_keywords": [
223 + "trade-level institutional data",
224 + "fire sales",
225 + "predatory trading",
226 + "broker information leakage",
227 + "natural experiment",
228 + "economic magnitudes",
229 + "policy implications"
230 + ],
231 + "confidence": 0.8,
232 + "notes": "The introduction excerpt is truncated in places. Standard-error conventions are not visible in the excerpts. Figure 2 shows confidence bands; whether the other figures show uncertainty is not stated. The intro mentions 'clients of the relevant brokers' selling after the scandal; the claim wording paraphrases this."
233 + },
234 + "deterministic_checks": {
235 + "roadmap_regex": true,
236 + "contribution_regex_in_intro": true,
237 + "magnitude_in_abstract": true,
238 + "abstract_n_sentences": 6,
239 + "intro_n_paragraphs": 6,
240 + "intro_n_words": 1087,
241 + "causal_words_in_abstract": 1,
242 + "hedge_words_in_abstract": 1,
243 + "stars_mentioned": false,
244 + "clustered_se_mentioned": false,
245 + "internet_appendix_mentioned": false,
246 + "section_roles_rule": {
247 + "0": "introduction",
248 + "1": "data",
249 + "2": "results",
250 + "3": "other",
251 + "4": "conclusion"
252 + }
253 + },
254 + "schema_valid": true,
255 + "code_version": "87c73cd-dirty"
256 +}
\ No newline at end of file
added data/annotations/documents/jof_12841.json +283 −0
@@ -0,0 +1,283 @@
1 +{
2 + "validation_errors": [],
3 + "repairs": [
4 + "dropped hedging.typical_devices_note"
5 + ],
6 + "doc_id": "jof_12841",
7 + "ontology_version": "1.0",
8 + "prompt_version": "1.0",
9 + "model": "claude-opus-5-5",
10 + "effort": "medium",
11 + "annotated_at": "2026-11-03T23:26:47+00:00",
12 + "source_sha256": "2601e7596a3fc1f68bfb412c36dd5cfadfd0290c61c02ababff9d7ae5bdafb43",
13 + "parser": "pymupdf-jof-heuristics-1.3",
14 + "payload_chars": 22766,
15 + "usage": {
16 + "input": 7356,
17 + "output": 2462,
18 + "cache_read": 3968,
19 + "cache_write": 0
20 + },
21 + "cost_usd": 0.079458,
22 + "request_id": "req_011CggqYL4ASoYxTq9jTgLev",
23 + "annotation": {
24 + "genre": "empirical",
25 + "subfield": "behavioral_finance",
26 + "subfield_secondary": "asset_pricing (mutual fund flows, sustainable investing)",
27 + "primary_method": "natural_experiment_event_study",
28 + "identification": {
29 + "estimand": "Causal effect of being assigned a given Morningstar sustainability globe rating on subsequent mutual fund net flows, and on liquidation.",
30 + "strategy": "Quasi-exogenous shock from the March 2016 publication of Morningstar globe ratings. Post-publication flows are compared across globe categories with time fixed effects. This is supplemented by regression discontinuity at globe breakpoints, nearest-neighbor matching on pre-period characteristics, and a placebo test, plus a lab/MTurk experiment on mechanisms.",
31 + "key_assumptions": [
32 + "Ratings repackage public information and do not change fund fundamentals.",
33 + "Funds in different globe categories were not systematically different in pre-publication flows.",
34 + "Within-category construction makes ratings largely uncorrelated with investment style.",
35 + "Continuity of fund characteristics around the percentile breakpoints for the RD."
36 + ],
37 + "threats_addressed": [
38 + "Pre-existing differences across globe categories (pre-period plot, matching, controlling for preperiod)",
39 + "Investors responding to continuous sustainability information rather than the discrete rating (percentile and score controls, RD)",
40 + "Placebo timing",
41 + "Institutional constraints as an alternative explanation",
42 + "Rational return expectations (performance tests)"
43 + ],
44 + "causal_language_strength": 5,
45 + "uses_causal_claim": true
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": 0
59 + },
60 + "results": {
61 + "present": true,
62 + "sentence_index": 0
63 + },
64 + "magnitude": {
65 + "present": true,
66 + "sentence_index": 0
67 + },
68 + "contribution": {
69 + "present": false,
70 + "sentence_index": -1
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": "preview_of_results"
86 + },
87 + {
88 + "paragraph_index": 1,
89 + "move": "identification_defense"
90 + },
91 + {
92 + "paragraph_index": 1,
93 + "move": "preview_of_approach"
94 + },
95 + {
96 + "paragraph_index": 2,
97 + "move": "identification_defense"
98 + },
99 + {
100 + "paragraph_index": 2,
101 + "move": "robustness_mention"
102 + },
103 + {
104 + "paragraph_index": 3,
105 + "move": "preview_of_results"
106 + },
107 + {
108 + "paragraph_index": 3,
109 + "move": "related_literature_positioning"
110 + },
111 + {
112 + "paragraph_index": 4,
113 + "move": "preview_of_approach"
114 + },
115 + {
116 + "paragraph_index": 4,
117 + "move": "preview_of_results"
118 + },
119 + {
120 + "paragraph_index": 4,
121 + "move": "mechanism_or_theory_sketch"
122 + },
123 + {
124 + "paragraph_index": 5,
125 + "move": "contribution_statement"
126 + },
127 + {
128 + "paragraph_index": 5,
129 + "move": "related_literature_positioning"
130 + },
131 + {
132 + "paragraph_index": 6,
133 + "move": "related_literature_positioning"
134 + },
135 + {
136 + "paragraph_index": 6,
137 + "move": "mechanism_or_theory_sketch"
138 + },
139 + {
140 + "paragraph_index": 7,
141 + "move": "roadmap"
142 + }
143 + ],
144 + "contribution_paragraph_index": 5,
145 + "contribution_types": [
146 + "new_evidence",
147 + "new_identification"
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": "other"
160 + },
161 + {
162 + "idx": 2,
163 + "role": "data"
164 + },
165 + {
166 + "idx": 3,
167 + "role": "results"
168 + },
169 + {
170 + "idx": 4,
171 + "role": "results"
172 + },
173 + {
174 + "idx": 5,
175 + "role": "conclusion"
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 + "placebo",
187 + "alternative_specification",
188 + "alternative_measure"
189 + ]
190 + },
191 + "tables": {
192 + "convention_notes": "Notes open by stating what the table shows. They then specify the dependent variable and regressors (globe dummies, percentile, score), panel structure (value- vs equal-weighted; MBA vs MTurk), and estimators (conventional vs bias-corrected RD; nearest-neighbor matching). Standard-error details are truncated in the excerpts.",
193 + "units_stated_in_notes": false,
194 + "standard_errors_description": "unclear"
195 + },
196 + "figures": {
197 + "uncertainty_shown": "mixed"
198 + },
199 + "conclusion": {
200 + "bounded_claims": false,
201 + "policy_implications": false,
202 + "open_questions": false,
203 + "restates_contribution": true
204 + },
205 + "hedging": {
206 + "level": "moderate",
207 + "typical_devices": [
208 + "'consistent with' interpretations",
209 + "'suggests' for mechanism evidence",
210 + "explicit statement that mechanism hypotheses cannot be definitively distinguished",
211 + "'one interpretation is'"
212 + ]
213 + },
214 + "claim_evidence": [
215 + {
216 + "claim": "Investors marketwide value sustainability: low-globe funds lost over $12B in flows, high-globe funds gained over $24B.",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table III; Section III.E"
219 + },
220 + {
221 + "claim": "Funds did not differ systematically in flows before publication of the ratings.",
222 + "evidence_type": "figure",
223 + "evidence_ref": "Figure 1; Table V"
224 + },
225 + {
226 + "claim": "Investors respond to the discrete globe categories, mainly the extremes, rather than to percentile ranks or raw scores.",
227 + "evidence_type": "table",
228 + "evidence_ref": "Table III, Table IV, Figure 4"
229 + },
230 + {
231 + "claim": "High-sustainability funds do not outperform low-sustainability funds.",
232 + "evidence_type": "table",
233 + "evidence_ref": "Table VIII"
234 + },
235 + {
236 + "claim": "Experimental subjects expect higher returns and lower risk from higher globe ratings, consistent with affect.",
237 + "evidence_type": "figure",
238 + "evidence_ref": "Figure 6; Table IX"
239 + },
240 + {
241 + "claim": "Subjects considering environmental or social factors allocate more to five-globe funds than their expectations explain.",
242 + "evidence_type": "table",
243 + "evidence_ref": "Table IX"
244 + }
245 + ],
246 + "structural_summary": "The introduction frames whether the average investor values sustainability and immediately states the causal answer. It then defends the Morningstar rating launch as a quasi-exogenous, information-free shock and previews three sets of results: flow results, discrete/extreme-category salience results, and experimental mechanism results. Literature positioning sits inside the introduction, followed by a roadmap. The body describes the rating system and the proprietary data. A large section establishes the flow effects through attention evidence, base regressions, within-globe RD, pre-period matching and economic magnitudes. A second section tests institutional, rational-return and nonpecuniary/affect explanations, using share-class splits, performance regressions and an MBA/MTurk experiment. A one-paragraph conclusion restates the findings.",
247 + "writing_profile_keywords": [
248 + "natural experiment",
249 + "salience",
250 + "fund flows",
251 + "ESG ratings",
252 + "causal framing",
253 + "mechanism decomposition",
254 + "experiment complement",
255 + "dollar magnitudes"
256 + ],
257 + "confidence": 0.85,
258 + "notes": "The format of statistical-significance reporting is not visible in the truncated table notes. Robustness checks (matching, placebo, controlling for the preperiod) are embedded as subsections of the results section. Hedging is uneven: the main causal result is stated strongly, while mechanism claims are explicitly hedged."
259 + },
260 + "deterministic_checks": {
261 + "roadmap_regex": true,
262 + "contribution_regex_in_intro": true,
263 + "magnitude_in_abstract": false,
264 + "abstract_n_sentences": 4,
265 + "intro_n_paragraphs": 8,
266 + "intro_n_words": 1311,
267 + "causal_words_in_abstract": 1,
268 + "hedge_words_in_abstract": 1,
269 + "stars_mentioned": false,
270 + "clustered_se_mentioned": false,
271 + "internet_appendix_mentioned": false,
272 + "section_roles_rule": {
273 + "0": "introduction",
274 + "1": "other",
275 + "2": "data",
276 + "3": "other",
277 + "4": "other",
278 + "5": "conclusion"
279 + }
280 + },
281 + "schema_valid": true,
282 + "code_version": "87c73cd-dirty"
283 +}
\ No newline at end of file
added data/annotations/documents/jof_12842.json +271 −0
@@ -0,0 +1,271 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12842",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:11:08+00:00",
9 + "source_sha256": "b4cb4cfee141907a732fa2afe5d3b41f56028be737adf293b975ba339573571f",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22844,
12 + "usage": {
13 + "input": 8125,
14 + "output": 2148,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.076254,
19 + "request_id": "req_011Cggqdu7ee5y7fCCxfDcH1",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "macro_finance",
24 + "primary_method": "structural_estimation",
25 + "identification": {
26 + "estimand": "Structural preference and refinancing transaction-cost parameters of a household liquidity-management model, and the model's ability to reproduce untargeted cross-sectional and time-series patterns.",
27 + "strategy": "Simulated method of moments (indirect inference). Exogenous state dynamics are estimated from data, institutional parameters are calibrated, and preference and cost parameters are estimated by matching household balance-sheet and refinancing moments. The model is then evaluated on untargeted moments, with historical paths of aggregate shocks fed in.",
28 + "key_assumptions": [
29 + "House prices, aggregate income and interest rates are exogenous (partial equilibrium)",
30 + "LTV and LTI limits follow conforming lending standards",
31 + "Households have no ex ante heterogeneity",
32 + "Idiosyncratic labor income risk is countercyclical"
33 + ],
34 + "threats_addressed": [
35 + "Interest-rate-driven refinancing as an alternative explanation for countercyclical cash-outs",
36 + "Competing explanations: self-control problems, optimistic expectations, lender moral hazard",
37 + "Lending standards looser than the baseline (model treated as a lower bound; constraints relaxed)"
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": 1
54 + },
55 + "results": {
56 + "present": true,
57 + "sentence_index": 0
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": 0,
80 + "move": "preview_of_results"
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": "preview_of_results"
97 + },
98 + {
99 + "paragraph_index": 4,
100 + "move": "related_literature_positioning"
101 + },
102 + {
103 + "paragraph_index": 4,
104 + "move": "preview_of_approach"
105 + },
106 + {
107 + "paragraph_index": 5,
108 + "move": "robustness_mention"
109 + },
110 + {
111 + "paragraph_index": 5,
112 + "move": "related_literature_positioning"
113 + },
114 + {
115 + "paragraph_index": 6,
116 + "move": "roadmap"
117 + }
118 + ],
119 + "contribution_paragraph_index": 0,
120 + "contribution_types": [
121 + "new_evidence",
122 + "new_theory_or_model"
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": "results"
135 + },
136 + {
137 + "idx": 2,
138 + "role": "theory"
139 + },
140 + {
141 + "idx": 3,
142 + "role": "empirical_design"
143 + },
144 + {
145 + "idx": 4,
146 + "role": "results"
147 + },
148 + {
149 + "idx": 5,
150 + "role": "conclusion"
151 + },
152 + {
153 + "idx": 6,
154 + "role": "appendix"
155 + },
156 + {
157 + "idx": 7,
158 + "role": "appendix"
159 + },
160 + {
161 + "idx": 8,
162 + "role": "appendix"
163 + }
164 + ],
165 + "results_presentation": {
166 + "economic_significance_discussed": true,
167 + "statistical_significance_language": "unclear",
168 + "magnitude_statements_in_intro": false
169 + },
170 + "robustness": {
171 + "location": "within_results",
172 + "types": [
173 + "sensitivity_to_parameters",
174 + "alternative_specification"
175 + ]
176 + },
177 + "tables": {
178 + "convention_notes": "Time-series regressions report standard errors in parentheses. Parameter tables are split into exogenously set and estimated panels. The moment-matching table shows data, model and SE columns side by side. Counterfactual tables vary one model feature per column.",
179 + "units_stated_in_notes": true,
180 + "standard_errors_description": "newey_west"
181 + },
182 + "figures": {
183 + "uncertainty_shown": "no"
184 + },
185 + "conclusion": {
186 + "bounded_claims": true,
187 + "policy_implications": true,
188 + "open_questions": false,
189 + "restates_contribution": true
190 + },
191 + "hedging": {
192 + "level": "moderate",
193 + "typical_devices": [
194 + "'consistent with' / 'suggests'",
195 + "model framed as a lower bound",
196 + "partial-equilibrium caveat",
197 + "explicit list of model shortcomings in the conclusion"
198 + ]
199 + },
200 + "claim_evidence": [
201 + {
202 + "claim": "Cash-out refinancing has a strongly countercyclical component not explained by interest rate movements.",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table I; Figure 1"
205 + },
206 + {
207 + "claim": "The estimated model matches targeted household balance-sheet and consumption moments.",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table IV"
210 + },
211 + {
212 + "claim": "Income risk and collateral vs debt-service constraints materially change model consumption volatility and homeownership.",
213 + "evidence_type": "table",
214 + "evidence_ref": "Table V"
215 + },
216 + {
217 + "claim": "High-leverage households experience larger consumption declines after the crash, in line with PSID estimates.",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table VI; Figure 8"
220 + },
221 + {
222 + "claim": "The model reproduces the cross-sectional refinancing and balance-sheet patterns seen in SCF and PSID data.",
223 + "evidence_type": "figure",
224 + "evidence_ref": "Figures 3-5, 7"
225 + },
226 + {
227 + "claim": "The model replicates the Mian and Sufi (2011) debt growth patterns across regions.",
228 + "evidence_type": "figure",
229 + "evidence_ref": "Figure 9"
230 + }
231 + ],
232 + "structural_summary": "The paper opens with a stylized-facts section on aggregate refinancing and equity extraction. It then builds a household model with income risk, long- and short-term mortgages and LTV/LTI constraints. Parameters are estimated by simulated moments, followed by counterfactual sensitivity exercises. The model is then evaluated on untargeted cross-sectional and boom-bust patterns. The conclusion draws policy lessons and lists where the model misfits the data.",
233 + "writing_profile_keywords": [
234 + "structural estimation",
235 + "household liquidity",
236 + "mortgage refinancing",
237 + "model-vs-data comparison",
238 + "untargeted moments",
239 + "housing boom-bust",
240 + "partial equilibrium"
241 + ],
242 + "confidence": 0.8,
243 + "notes": "Section IV is flagged 'other' by the rule but functions as model validation and results. Statistical-significance language is unclear beyond the Newey-West SEs reported in Table I."
244 + },
245 + "deterministic_checks": {
246 + "roadmap_regex": true,
247 + "contribution_regex_in_intro": false,
248 + "magnitude_in_abstract": false,
249 + "abstract_n_sentences": 4,
250 + "intro_n_paragraphs": 7,
251 + "intro_n_words": 1035,
252 + "causal_words_in_abstract": 0,
253 + "hedge_words_in_abstract": 1,
254 + "stars_mentioned": false,
255 + "clustered_se_mentioned": false,
256 + "internet_appendix_mentioned": false,
257 + "section_roles_rule": {
258 + "0": "introduction",
259 + "1": "results",
260 + "2": "theory",
261 + "3": "empirical_design",
262 + "4": "other",
263 + "5": "conclusion",
264 + "6": "appendix",
265 + "7": "appendix",
266 + "8": "appendix"
267 + }
268 + },
269 + "schema_valid": true,
270 + "code_version": "87c73cd-dirty"
271 +}
\ No newline at end of file
added data/annotations/documents/jof_12843.json +258 −0
@@ -0,0 +1,258 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12843",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:12:11+00:00",
9 + "source_sha256": "79facd154f888ddb20c4484ff17006715c75a146cb4326ac9a5735291714bf74",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 20845,
12 + "usage": {
13 + "input": 7129,
14 + "output": 2161,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.07253,
19 + "request_id": "req_011CggqiToH3haSNfn47w1Sx",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "mutual funds / taxation",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Association between a fund's prior three-year tax burden and its next-year before-tax, tax burden, and after-tax returns and alphas.",
27 + "strategy": "Annual predictive panel regressions of future fund performance on lagged average tax burden, with lagged controls and time and style fixed effects. These are complemented by a matched comparison of self-designated tax-managed funds against same-family, similar-style funds.",
28 + "key_assumptions": [
29 + "Lagged tax burden is not mechanically tied to future returns once controls and fixed effects are included.",
30 + "Matched funds share investment style and management teams, so they isolate deliberate tax management."
31 + ],
32 + "threats_addressed": [
33 + "Tax efficiency arising as a byproduct of investment style rather than deliberate management (addressed with self-designated fund matching and style controls).",
34 + "Risk or style adjustment (addressed with CAPM and multifactor alphas and style scores).",
35 + "Horizon and subperiod dependence.",
36 + "DC pension clienteles."
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": 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": "preview_of_approach"
80 + },
81 + {
82 + "paragraph_index": 1,
83 + "move": "preview_of_results"
84 + },
85 + {
86 + "paragraph_index": 2,
87 + "move": "preview_of_results"
88 + },
89 + {
90 + "paragraph_index": 2,
91 + "move": "tension_or_puzzle"
92 + },
93 + {
94 + "paragraph_index": 3,
95 + "move": "mechanism_or_theory_sketch"
96 + },
97 + {
98 + "paragraph_index": 3,
99 + "move": "preview_of_results"
100 + },
101 + {
102 + "paragraph_index": 4,
103 + "move": "identification_defense"
104 + },
105 + {
106 + "paragraph_index": 4,
107 + "move": "robustness_mention"
108 + },
109 + {
110 + "paragraph_index": 5,
111 + "move": "related_literature_positioning"
112 + },
113 + {
114 + "paragraph_index": 5,
115 + "move": "contribution_statement"
116 + }
117 + ],
118 + "contribution_paragraph_index": 5,
119 + "contribution_types": [
120 + "new_evidence",
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": "results"
146 + },
147 + {
148 + "idx": 5,
149 + "role": "robustness"
150 + },
151 + {
152 + "idx": 6,
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": "within_results",
163 + "types": [
164 + "alternative_specification",
165 + "alternative_measure",
166 + "alternative_sample_period",
167 + "subsample",
168 + "other"
169 + ]
170 + },
171 + "tables": {
172 + "convention_notes": "Regression tables are annual panel regressions on lagged tax burden with time and style fixed effects. Standard errors are fund-clustered and shown in parentheses, with stars marking 10/5/1% significance. Some notes state units (% per year), and returns in the matched-fund table are measured in excess of the market.",
173 + "units_stated_in_notes": true,
174 + "standard_errors_description": "clustered"
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 + "'is associated with'",
189 + "'indicate that'",
190 + "'seem to'",
191 + "'do not appear to'",
192 + "'consistent with'"
193 + ]
194 + },
195 + "claim_evidence": [
196 + {
197 + "claim": "Equilibrium fund performance depends on the relative size of taxable and tax-exempt clienteles under decreasing returns to scale.",
198 + "evidence_type": "equation_or_proof",
199 + "evidence_ref": "Section I; Figures 1-2"
200 + },
201 + {
202 + "claim": "A lower prior three-year tax burden predicts higher before-tax and after-tax returns over the following year.",
203 + "evidence_type": "table",
204 + "evidence_ref": "Table III, Table IV"
205 + },
206 + {
207 + "claim": "The before-tax outperformance of tax-efficient funds reflects lower trading costs, favorable style exposures and better selectivity.",
208 + "evidence_type": "table",
209 + "evidence_ref": "Table V, Table VI"
210 + },
211 + {
212 + "claim": "Self-designated tax-managed funds beat matched same-family funds after taxes without significant before-tax underperformance.",
213 + "evidence_type": "table",
214 + "evidence_ref": "Table XI"
215 + },
216 + {
217 + "claim": "Tax burden is determined by investment style, investor trading behavior and market conditions.",
218 + "evidence_type": "table",
219 + "evidence_ref": "Table II"
220 + }
221 + ],
222 + "structural_summary": "The paper opens by explaining how fund distributions are taxed and framing the trade-off in tax management: constraints that may lower returns versus manager skill. A stylized Berk–Green-type model with heterogeneous tax clienteles follows. The data section defines the tax burden, and a short section examines its determinants. The core section runs predictive regressions of future before- and after-tax performance on prior tax burden, then extends them with risk adjustment, performance decomposition, style controls, alternative horizons, subperiods, distribution types and DC ratios. A matched analysis of self-designated tax-managed funds addresses the byproduct concern. The conclusion compares the importance of taxes to fund expenses and interprets the persistence of outperformance as investor inattention.",
223 + "writing_profile_keywords": [
224 + "theory-then-evidence",
225 + "predictive panel regressions",
226 + "performance decomposition",
227 + "puzzle framing",
228 + "matched-sample check",
229 + "mutual fund taxation"
230 + ],
231 + "confidence": 0.82,
232 + "notes": "The abstract truncates into the opening of the introduction. The Determinants and Self-Designated Fund sections are labeled 'other' by the role rule; I assigned them results and robustness/identification roles. The introduction excerpt is truncated, so a roadmap may exist but is not visible."
233 + },
234 + "deterministic_checks": {
235 + "roadmap_regex": false,
236 + "contribution_regex_in_intro": true,
237 + "magnitude_in_abstract": false,
238 + "abstract_n_sentences": 6,
239 + "intro_n_paragraphs": 6,
240 + "intro_n_words": 881,
241 + "causal_words_in_abstract": 0,
242 + "hedge_words_in_abstract": 0,
243 + "stars_mentioned": true,
244 + "clustered_se_mentioned": true,
245 + "internet_appendix_mentioned": false,
246 + "section_roles_rule": {
247 + "0": "introduction",
248 + "1": "theory",
249 + "2": "data",
250 + "3": "other",
251 + "4": "other",
252 + "5": "other",
253 + "6": "conclusion"
254 + }
255 + },
256 + "schema_valid": true,
257 + "code_version": "87c73cd-dirty"
258 +}
\ No newline at end of file
added data/annotations/documents/jof_12844.json +239 −0
@@ -0,0 +1,239 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12844",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:11:25+00:00",
9 + "source_sha256": "649cb23d513bbe54ccb026c7fe6f3c75a74ccf7caa51c797553c72b14cee1a4b",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 16381,
12 + "usage": {
13 + "input": 5311,
14 + "output": 1985,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.061738,
19 + "request_id": "req_011CggqfDDcDsDd4KhZx1ghi",
20 + "annotation": {
21 + "genre": "mixed_theory_empirical",
22 + "subfield": "market_microstructure",
23 + "subfield_secondary": "asset_pricing",
24 + "primary_method": "natural_experiment_event_study",
25 + "identification": {
26 + "estimand": "Effect of the expected delay until public disclosure on informed trading intensity, time to first trade, order type choice, and the speed of price discovery during the private window.",
27 + "strategy": "Quasi-natural experiment: SEC filings were accidentally leaked to PDS subscribers before public EDGAR release. The delay is treated as exogenous, and an expected delay is predicted mainly from hour of day. Trading at the security-second and filing level is related to expected versus unexpected delay through panel regressions, tobit and multinomial logit models.",
28 + "key_assumptions": [
29 + "The length of the delay until public release is exogenous to filing content and firm characteristics.",
30 + "Informed traders can form expectations about the delay from observable factors such as time of day.",
31 + "Trades during the private window reflect trading by those with access to the leaked filings."
32 + ],
33 + "threats_addressed": [
34 + "Endogenous timing of disclosure, addressed by the accidental leak and the exogenous delay.",
35 + "Censoring of time to first trade, addressed with a two-sided tobit model.",
36 + "Expected and unexpected delay components separated."
37 + ],
38 + "causal_language_strength": 3,
39 + "uses_causal_claim": true
40 + },
41 + "abstract_moves": {
42 + "question": {
43 + "present": true,
44 + "sentence_index": 1
45 + },
46 + "gap": {
47 + "present": true,
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": 3
65 + }
66 + },
67 + "introduction": {
68 + "moves": [
69 + {
70 + "paragraph_index": 0,
71 + "move": "motivation"
72 + },
73 + {
74 + "paragraph_index": 0,
75 + "move": "identification_defense"
76 + },
77 + {
78 + "paragraph_index": 1,
79 + "move": "mechanism_or_theory_sketch"
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": 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": "contribution_statement"
100 + },
101 + {
102 + "paragraph_index": 4,
103 + "move": "related_literature_positioning"
104 + },
105 + {
106 + "paragraph_index": 5,
107 + "move": "roadmap"
108 + }
109 + ],
110 + "contribution_paragraph_index": 4,
111 + "contribution_types": [
112 + "new_evidence",
113 + "new_theory_or_model",
114 + "new_identification"
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": "data"
127 + },
128 + {
129 + "idx": 2,
130 + "role": "theory"
131 + },
132 + {
133 + "idx": 3,
134 + "role": "results"
135 + },
136 + {
137 + "idx": 4,
138 + "role": "conclusion"
139 + }
140 + ],
141 + "results_presentation": {
142 + "economic_significance_discussed": false,
143 + "statistical_significance_language": "unclear",
144 + "magnitude_statements_in_intro": false
145 + },
146 + "robustness": {
147 + "location": "none_seen",
148 + "types": []
149 + },
150 + "tables": {
151 + "convention_notes": "Table notes define the dependent variable, its unit (seconds, $1,000, basis points), the sample (Trimmed Sample defined in Figure 2), the unit of observation (security-second or filing) and the estimator (panel, tobit, multinomial logit). Standard errors are not visible in the excerpts.",
152 + "units_stated_in_notes": true,
153 + "standard_errors_description": "unclear"
154 + },
155 + "figures": {
156 + "uncertainty_shown": "unclear"
157 + },
158 + "conclusion": {
159 + "bounded_claims": true,
160 + "policy_implications": false,
161 + "open_questions": false,
162 + "restates_contribution": true
163 + },
164 + "hedging": {
165 + "level": "moderate",
166 + "typical_devices": [
167 + "'consistent with' the model's predictions or hypotheses",
168 + "'appears to' when describing the cause of the leak",
169 + "'quasi-natural experiment' framing"
170 + ]
171 + },
172 + "claim_evidence": [
173 + {
174 + "claim": "The delay until public release can be predicted mainly by the hour of the trading day.",
175 + "evidence_type": "table",
176 + "evidence_ref": "Table II; Figure 2"
177 + },
178 + {
179 + "claim": "A longer expected delay reduces informed trading intensity in both dollar value and number of trades.",
180 + "evidence_type": "table",
181 + "evidence_ref": "Table III"
182 + },
183 + {
184 + "claim": "A longer expected delay increases the time until the first trade after the leak.",
185 + "evidence_type": "table",
186 + "evidence_ref": "Table IV"
187 + },
188 + {
189 + "claim": "Limit orders are preferred over market orders when the expected delay is longer and market conditions are favorable.",
190 + "evidence_type": "table",
191 + "evidence_ref": "Table V"
192 + },
193 + {
194 + "claim": "Information disseminates faster in large-cap, high-turnover stocks and when market activity is high.",
195 + "evidence_type": "table",
196 + "evidence_ref": "Table VI"
197 + },
198 + {
199 + "claim": "Optimal insider strategy and price dynamics under a random, exponentially distributed stopping time.",
200 + "evidence_type": "equation_or_proof",
201 + "evidence_ref": "Section II"
202 + }
203 + ],
204 + "structural_summary": "The introduction opens with an institutional anecdote (an accidental leak of SEC filings) and frames it as a lab-like natural experiment. It then sketches a model adapted from Caldentey and Stacchetti, previews the empirical measures and Kyle-consistent findings, and positions the paper against the information-transmission and leakage literatures before giving a roadmap. The body proceeds from institutional background and data, to the model, to empirical tests. A short conclusion restates the findings as consistent with the hypotheses.",
205 + "writing_profile_keywords": [
206 + "natural experiment",
207 + "informed trading",
208 + "random stopping time",
209 + "Kyle model",
210 + "trade-level data",
211 + "model-guided empirics",
212 + "institutional anecdote"
213 + ],
214 + "confidence": 0.8,
215 + "notes": "Section I is labeled 'literature' by the rule, but its content is institutional background and data. Robustness checks and standard-error conventions are not visible; an Internet Appendix exists."
216 + },
217 + "deterministic_checks": {
218 + "roadmap_regex": true,
219 + "contribution_regex_in_intro": true,
220 + "magnitude_in_abstract": false,
221 + "abstract_n_sentences": 4,
222 + "intro_n_paragraphs": 6,
223 + "intro_n_words": 1015,
224 + "causal_words_in_abstract": 1,
225 + "hedge_words_in_abstract": 0,
226 + "stars_mentioned": false,
227 + "clustered_se_mentioned": false,
228 + "internet_appendix_mentioned": false,
229 + "section_roles_rule": {
230 + "0": "introduction",
231 + "1": "literature",
232 + "2": "theory",
233 + "3": "other",
234 + "4": "conclusion"
235 + }
236 + },
237 + "schema_valid": true,
238 + "code_version": "87c73cd-dirty"
239 +}
\ No newline at end of file
added data/annotations/documents/jof_12845.json +264 −0
@@ -0,0 +1,264 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12845",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:11:09+00:00",
9 + "source_sha256": "7fb58a47efc09306eb617c2d6512b5643c32cce07d2f0054e29b4dceb00697ca",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 22948,
12 + "usage": {
13 + "input": 7582,
14 + "output": 2043,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.071982,
19 + "request_id": "req_011Cggqe1Fk5NcN8jbtr1djP",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "asset_pricing",
23 + "subfield_secondary": "market_microstructure",
24 + "primary_method": "reduced_form_regression",
25 + "identification": {
26 + "estimand": "Predictive relation between trader-type net position changes or hedging pressure and subsequent commodity futures excess returns",
27 + "strategy": "Fama-MacBeth cross-sectional predictive regressions and portfolio sorts on weekly CFTC position changes, following the Kaniel–Saar–Titman approach; the direction of predictability is used to infer who provides liquidity. Hedging pressure is split into smoothed (long-term) and short-term components.",
28 + "key_assumptions": [
29 + "The sign of short-horizon return predictability reveals liquidity provision versus consumption",
30 + "Smoothed hedging pressure captures hedging demand that evolves slowly with physical production decisions",
31 + "Controls adequately capture expected returns"
32 + ],
33 + "threats_addressed": [
34 + "Private information of commercials as an alternative explanation",
35 + "Momentum-driven trading by noncommercials",
36 + "Measurement of trader categories (DCOT data)",
37 + "Cross-sectional versus time-series predictability",
38 + "Major versus minor commodity heterogeneity"
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": false,
54 + "sentence_index": -1
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": 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": "mechanism_or_theory_sketch"
90 + },
91 + {
92 + "paragraph_index": 2,
93 + "move": "preview_of_results"
94 + },
95 + {
96 + "paragraph_index": 3,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 3,
101 + "move": "identification_defense"
102 + },
103 + {
104 + "paragraph_index": 4,
105 + "move": "related_literature_positioning"
106 + },
107 + {
108 + "paragraph_index": 4,
109 + "move": "contribution_statement"
110 + },
111 + {
112 + "paragraph_index": 5,
113 + "move": "roadmap"
114 + }
115 + ],
116 + "contribution_paragraph_index": 4,
117 + "contribution_types": [
118 + "new_evidence",
119 + "resolution_of_puzzle"
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": "data"
132 + },
133 + {
134 + "idx": 2,
135 + "role": "results"
136 + },
137 + {
138 + "idx": 3,
139 + "role": "results"
140 + },
141 + {
142 + "idx": 4,
143 + "role": "results"
144 + },
145 + {
146 + "idx": 5,
147 + "role": "results"
148 + },
149 + {
150 + "idx": 6,
151 + "role": "robustness"
152 + },
153 + {
154 + "idx": 7,
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": "main_text_section",
165 + "types": [
166 + "alternative_specification",
167 + "subsample",
168 + "alternative_measure"
169 + ]
170 + },
171 + "tables": {
172 + "convention_notes": "Table notes describe Fama-MacBeth regression setups: the dependent variable, the predictors (position changes scaled by open interest, hedging pressure) and the controls. Portfolio sort tables describe quintile and double-sort construction. The visible excerpts do not describe standard errors or significance markers.",
173 + "units_stated_in_notes": true,
174 + "standard_errors_description": "unclear"
175 + },
176 + "figures": {
177 + "uncertainty_shown": "unclear"
178 + },
179 + "conclusion": {
180 + "bounded_claims": false,
181 + "policy_implications": false,
182 + "open_questions": false,
183 + "restates_contribution": true
184 + },
185 + "hedging": {
186 + "level": "moderate",
187 + "typical_devices": [
188 + "we conjecture",
189 + "suggests",
190 + "more likely reflects",
191 + "attempt to"
192 + ]
193 + },
194 + "claim_evidence": [
195 + {
196 + "claim": "Noncommercial position changes are positively related to contemporaneous and past returns, while commercials trade as contrarians",
197 + "evidence_type": "table",
198 + "evidence_ref": "Table II"
199 + },
200 + {
201 + "claim": "Commodities bought by noncommercials subsequently underperform, consistent with commercials providing liquidity",
202 + "evidence_type": "table",
203 + "evidence_ref": "Tables III-V"
204 + },
205 + {
206 + "claim": "After controlling for short-term position changes, smoothed hedging pressure positively predicts futures returns",
207 + "evidence_type": "table",
208 + "evidence_ref": "Tables VI-VIII"
209 + },
210 + {
211 + "claim": "The cost of liquidity rises with commodity implied volatility, capital constraints and position imbalance, but not with the VIX",
212 + "evidence_type": "table",
213 + "evidence_ref": "Table IX"
214 + },
215 + {
216 + "claim": "Commercials lose on momentum-driven trades but gain on the larger nonmomentum component",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table X, Figure 3"
219 + },
220 + {
221 + "claim": "Liquidity gains offset much of the insurance premium commercials pay",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table XI"
224 + }
225 + ],
226 + "structural_summary": "The introduction frames the Keynesian normal-backwardation view as incomplete. It then previews a liquidity-provision test based on short-horizon predictability and conjectures that hedging pressure has a short-term and a long-term component. Results sections proceed in order: liquidity provision evidence, hedging pressure revisited with a smoothed measure, risk-environment and private-information checks, and a profit decomposition explaining commercials' trading against momentum. A robustness section and a conclusion that restates the two-premium finding close the paper.",
227 + "writing_profile_keywords": [
228 + "two-premium decomposition",
229 + "Fama-MacBeth",
230 + "CFTC COT data",
231 + "liquidity provision",
232 + "normal backwardation",
233 + "portfolio sorts",
234 + "puzzle resolution"
235 + ],
236 + "confidence": 0.82,
237 + "notes": "The data section contains several equations. Figure 2 is missing from the excerpts. The internet appendix is mentioned only in the conclusion."
238 + },
239 + "deterministic_checks": {
240 + "roadmap_regex": true,
241 + "contribution_regex_in_intro": true,
242 + "magnitude_in_abstract": false,
243 + "abstract_n_sentences": 4,
244 + "intro_n_paragraphs": 6,
245 + "intro_n_words": 943,
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": false,
251 + "section_roles_rule": {
252 + "0": "introduction",
253 + "1": "data",
254 + "2": "other",
255 + "3": "other",
256 + "4": "other",
257 + "5": "other",
258 + "6": "robustness",
259 + "7": "conclusion"
260 + }
261 + },
262 + "schema_valid": true,
263 + "code_version": "87c73cd-dirty"
264 +}
\ No newline at end of file
added data/annotations/documents/jof_12846.json +291 −0
@@ -0,0 +1,291 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12846",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:11:25+00:00",
9 + "source_sha256": "b5dfbca07d3c9237c18da695ac6d632cf4c1c19b892632c4427053c330357407",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 29549,
12 + "usage": {
13 + "input": 10292,
14 + "output": 3314,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.108242,
19 + "request_id": "req_011CggqePv1tCxVYwG3KSGDU",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "corporate_finance",
23 + "subfield_secondary": "law and finance / credit access and financial development",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of France's 2006 collateral reform, which allowed nonpossessory pledges of physical assets, on the borrowing, real outcomes, start-up formation and survival, and credit-access inequality of private French firms.",
27 + "strategy": "Matched difference-in-differences comparing high- versus low-fixed-assets-intensity firms before and after 2006. Firms are propensity-score matched on predetermined traits, with firm and industry-year fixed effects. Triple-difference splits by size, age and financial constraints; hazard models for bankruptcy; spatial and Gini analyses; Wurgler-style allocation-efficiency regressions.",
28 + "key_assumptions": [
29 + "Absent the reform, high- and low-fixed-assets firms would have followed parallel trends.",
30 + "Pre-reform fixed-asset intensity is a valid proxy for exposure to newly pledgeable assets.",
31 + "No contemporaneous shocks differentially affect high-fixed-assets firms; other law changes are discussed in the institutional section."
32 + ],
33 + "threats_addressed": [
34 + "Pre-reform differences between treated and control firms (matching; Table I balance)",
35 + "Pre-existing differential trends (parallel trends tests)",
36 + "Generic shocks to asset-intensive firms (placebo regressions on other Napoleonic-Code countries)",
37 + "Non-collateral channels (short-term unsecured debt used as a falsification outcome)",
38 + "Rural effect driven by industry or size composition (spatial distribution checks)"
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": 1
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": false,
66 + "sentence_index": -1
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "other"
74 + },
75 + {
76 + "paragraph_index": 4,
77 + "move": "motivation"
78 + },
79 + {
80 + "paragraph_index": 4,
81 + "move": "literature_gap"
82 + },
83 + {
84 + "paragraph_index": 4,
85 + "move": "preview_of_approach"
86 + },
87 + {
88 + "paragraph_index": 5,
89 + "move": "preview_of_results"
90 + },
91 + {
92 + "paragraph_index": 5,
93 + "move": "identification_defense"
94 + },
95 + {
96 + "paragraph_index": 6,
97 + "move": "preview_of_results"
98 + },
99 + {
100 + "paragraph_index": 7,
101 + "move": "preview_of_approach"
102 + },
103 + {
104 + "paragraph_index": 7,
105 + "move": "preview_of_results"
106 + },
107 + {
108 + "paragraph_index": 8,
109 + "move": "related_literature_positioning"
110 + },
111 + {
112 + "paragraph_index": 8,
113 + "move": "preview_of_results"
114 + },
115 + {
116 + "paragraph_index": 8,
117 + "move": "policy_or_broader_implication"
118 + },
119 + {
120 + "paragraph_index": 9,
121 + "move": "preview_of_results"
122 + },
123 + {
124 + "paragraph_index": 9,
125 + "move": "contribution_statement"
126 + },
127 + {
128 + "paragraph_index": 9,
129 + "move": "related_literature_positioning"
130 + }
131 + ],
132 + "contribution_paragraph_index": 9,
133 + "contribution_types": [
134 + "new_evidence"
135 + ],
136 + "roadmap_present": false,
137 + "literature_positioning_location": "inside_introduction"
138 + },
139 + "section_roles": [
140 + {
141 + "idx": 0,
142 + "role": "introduction"
143 + },
144 + {
145 + "idx": 1,
146 + "role": "other"
147 + },
148 + {
149 + "idx": 2,
150 + "role": "data"
151 + },
152 + {
153 + "idx": 3,
154 + "role": "results"
155 + },
156 + {
157 + "idx": 4,
158 + "role": "results"
159 + },
160 + {
161 + "idx": 5,
162 + "role": "results"
163 + },
164 + {
165 + "idx": 6,
166 + "role": "results"
167 + },
168 + {
169 + "idx": 7,
170 + "role": "robustness"
171 + },
172 + {
173 + "idx": 8,
174 + "role": "conclusion"
175 + }
176 + ],
177 + "results_presentation": {
178 + "economic_significance_discussed": true,
179 + "statistical_significance_language": "unclear",
180 + "magnitude_statements_in_intro": true
181 + },
182 + "robustness": {
183 + "location": "mixed",
184 + "types": [
185 + "placebo",
186 + "falsification",
187 + "subsample",
188 + "alternative_specification",
189 + "other"
190 + ]
191 + },
192 + "tables": {
193 + "convention_notes": "Table notes write out the estimating equation (firm and industry-year fixed effects, Post×Treated), define outcome columns and panels, and describe the matched sample. Hazard-model specifications are given in the notes where used. Descriptive tables report means before and after the reform plus differences. Standard-error treatment is not visible in the excerpts.",
194 + "units_stated_in_notes": true,
195 + "standard_errors_description": "unclear"
196 + },
197 + "figures": {
198 + "uncertainty_shown": "no"
199 + },
200 + "conclusion": {
201 + "bounded_claims": false,
202 + "policy_implications": true,
203 + "open_questions": false,
204 + "restates_contribution": true
205 + },
206 + "hedging": {
207 + "level": "moderate",
208 + "typical_devices": [
209 + "'suggests' / 'consistent with' for interpretive claims",
210 + "'may lead' for extrapolations",
211 + "confident 'we show/find' for core estimates"
212 + ]
213 + },
214 + "claim_evidence": [
215 + {
216 + "claim": "The reform raised long-term leverage and cut zero-debt incidence mainly for high-fixed-assets private firms; short-term debt was unaffected.",
217 + "evidence_type": "table",
218 + "evidence_ref": "Table II; Figures 2-4"
219 + },
220 + {
221 + "claim": "Treated firms increased investment, employment, sales and profitability and had lower profit volatility and bankruptcy hazard.",
222 + "evidence_type": "table",
223 + "evidence_ref": "Table III"
224 + },
225 + {
226 + "claim": "Small, young and financially constrained firms gained the most credit access.",
227 + "evidence_type": "table",
228 + "evidence_ref": "Tables IV-V"
229 + },
230 + {
231 + "claim": "Rural firms gained most, and departmental credit Gini indices fell sharply, especially in rural departments.",
232 + "evidence_type": "figure",
233 + "evidence_ref": "Figures 5-6"
234 + },
235 + {
236 + "claim": "Start-ups had higher at-inception leverage, located farther from large cities, and had lower exit and bankruptcy rates after the reform.",
237 + "evidence_type": "table",
238 + "evidence_ref": "Tables VI-VII; Figure 7"
239 + },
240 + {
241 + "claim": "Cross-industry capital allocation efficiency improved, more so in fixed-asset-heavy and externally dependent sectors.",
242 + "evidence_type": "table",
243 + "evidence_ref": "Tables VIII-IX"
244 + },
245 + {
246 + "claim": "Treated and control firms show parallel pre-reform trends.",
247 + "evidence_type": "table",
248 + "evidence_ref": "Table X"
249 + }
250 + ],
251 + "structural_summary": "Opens with a political epigraph, then motivates the topic with the Napoleonic possessory-ownership regime and an understanding gap. The introduction previews results in sequence: credit, real outcomes, democratization by firm type and geography, then macro allocation efficiency, closing with law-and-finance positioning. The body describes the institutional setting and identification logic, then data and the matched DID design. Results sections cover the average firm, distributional and spatial effects, start-ups, and capital allocation efficiency. A dedicated robustness section follows, and the conclusion restates findings and draws policy implications.",
252 + "writing_profile_keywords": [
253 + "natural policy experiment",
254 + "matched difference-in-differences",
255 + "law and finance",
256 + "collateral",
257 + "credit democratization",
258 + "spatial analysis",
259 + "intro magnitude previews",
260 + "policy-oriented"
261 + ],
262 + "confidence": 0.8,
263 + "notes": "Abstract text includes author footnotes, which are ignored for sentence indexing. Introduction paragraphs 0-3 are epigraph fragments. Section I is labeled theory by rule but is an institutional and identification discussion, so it is coded 'other'. Standard-error and significance conventions are not visible in the excerpts."
264 + },
265 + "deterministic_checks": {
266 + "roadmap_regex": false,
267 + "contribution_regex_in_intro": true,
268 + "magnitude_in_abstract": false,
269 + "abstract_n_sentences": 16,
270 + "intro_n_paragraphs": 11,
271 + "intro_n_words": 1282,
272 + "causal_words_in_abstract": 0,
273 + "hedge_words_in_abstract": 0,
274 + "stars_mentioned": false,
275 + "clustered_se_mentioned": false,
276 + "internet_appendix_mentioned": true,
277 + "section_roles_rule": {
278 + "0": "introduction",
279 + "1": "theory",
280 + "2": "data",
281 + "3": "other",
282 + "4": "other",
283 + "5": "other",
284 + "6": "other",
285 + "7": "robustness",
286 + "8": "conclusion"
287 + }
288 + },
289 + "schema_valid": true,
290 + "code_version": "87c73cd-dirty"
291 +}
\ No newline at end of file
added data/annotations/documents/jof_12847.json +73 −0
@@ -0,0 +1,273 @@
1 +{
2 + "validation_errors": [],
3 + "doc_id": "jof_12847",
4 + "ontology_version": "1.0",
5 + "prompt_version": "1.0",
6 + "model": "claude-opus-5-5",
7 + "effort": "medium",
8 + "annotated_at": "2026-11-03T23:10:57+00:00",
9 + "source_sha256": "90ae11182046a12a1d65940e29cd01390d022e8a82a125577b5a9400ebea9a12",
10 + "parser": "pymupdf-jof-heuristics-1.3",
11 + "payload_chars": 23852,
12 + "usage": {
13 + "input": 8168,
14 + "output": 2094,
15 + "cache_read": 3968,
16 + "cache_write": 0
17 + },
18 + "cost_usd": 0.075346,
19 + "request_id": "req_011Cggqd75DEFJ99j2rFhYPr",
20 + "annotation": {
21 + "genre": "empirical",
22 + "subfield": "household_finance",
23 + "subfield_secondary": "macro_finance (housing markets, fiscal policy evaluation)",
24 + "primary_method": "difference_in_differences",
25 + "identification": {
26 + "estimand": "Effect of the First-Time Homebuyer Credit on home sales, house prices, and individual transitions into homeownership; aggregate induced sales and stimulus effects",
27 + "strategy": "Cross-ZIP difference-in-differences using ex ante exposure (year-2000 share of first-time buyers) with CBSA fixed effects, combined with a sharp regression kink design at the income phase-out, using administrative tax data",
28 + "key_assumptions": [
29 + "No time-varying ZIP-specific shocks correlated with program exposure (parallel trends)",
30 + "Low-exposure ZIPs serve as a valid counterfactual",
31 + "Smoothness of household outcomes in income around the phase-out kink apart from the policy-induced slope change"
32 + ],
33 + "threats_addressed": [
34 + "Time-varying ZIP-level shocks correlated with exposure",
35 + "Exposure to the subprime bust",
36 + "Contemporaneous policies",
37 + "Specification choices: weighting schemes and sample definitions",
38 + "ZIP-by-time trends, addressed through age-distribution shifts and a within-time starter-home placebo"
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": 1
63 + },
64 + "contribution": {
65 + "present": true,
66 + "sentence_index": 4
67 + }
68 + },
69 + "introduction": {
70 + "moves": [
71 + {
72 + "paragraph_index": 0,
73 + "move": "motivation"

Diff truncated — file too large.

Menu

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