{
 "validation_errors": [],
 "doc_id": "jof_12833",
 "ontology_version": "1.0",
 "prompt_version": "1.0",
 "model": "claude-opus-5-5",
 "effort": "medium",
 "annotated_at": "2026-11-03T23:09:55+00:00",
 "source_sha256": "a873fab10ec11b01baf0f01eedd915aa344f183707db057fe6a76ce244863423",
 "parser": "pymupdf-jof-heuristics-1.3",
 "payload_chars": 19720,
 "usage": {
  "input": 6904,
  "output": 2073,
  "cache_read": 3968,
  "cache_write": 0
 },
 "cost_usd": 0.06987,
 "request_id": "req_011CggqYdTS7WFZxfDL1WwFb",
 "annotation": {
  "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."
 },
 "deterministic_checks": {
  "roadmap_regex": false,
  "contribution_regex_in_intro": false,
  "magnitude_in_abstract": false,
  "abstract_n_sentences": 5,
  "intro_n_paragraphs": 6,
  "intro_n_words": 791,
  "causal_words_in_abstract": 0,
  "hedge_words_in_abstract": 1,
  "stars_mentioned": true,
  "clustered_se_mentioned": false,
  "internet_appendix_mentioned": false,
  "section_roles_rule": {
   "0": "introduction",
   "1": "data",
   "2": "other",
   "3": "theory",
   "4": "theory",
   "5": "other",
   "6": "other",
   "7": "conclusion"
  }
 },
 "schema_valid": true,
 "code_version": "87c73cd-dirty"
}