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": 017 },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": false41 },42 "abstract_moves": {43 "question": {44 "present": true,45 "sentence_index": 046 },47 "gap": {48 "present": false,49 "sentence_index": -150 },51 "method": {52 "present": true,53 "sentence_index": 354 },55 "results": {56 "present": true,57 "sentence_index": 358 },59 "magnitude": {60 "present": false,61 "sentence_index": -162 },63 "contribution": {64 "present": true,65 "sentence_index": 066 }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": false142 },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": true164 },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}