1{2 "validation_errors": [],3 "doc_id": "jof_12466",4 "ontology_version": "1.0",5 "prompt_version": "1.0",6 "model": "claude-opus-5-5",7 "effort": "medium",8 "annotated_at": "2026-11-03T22:56:40+00:00",9 "source_sha256": "9d49fe27275082c2c93138bc92803f500bb95706ef4d018a546ff234d3522e92",10 "parser": "pymupdf-jof-heuristics-1.3",11 "payload_chars": 21840,12 "usage": {13 "input": 7330,14 "output": 2286,15 "cache_read": 3968,16 "cache_write": 017 },18 "cost_usd": 0.075834,19 "request_id": "req_011CggpXpfyceYJzHT1Gh13g",20 "annotation": {21 "genre": "empirical",22 "subfield": "asset_pricing",23 "subfield_secondary": "analyst research / information intermediaries (capital markets)",24 "primary_method": "reduced_form_regression",25 "identification": {26 "estimand": "Difference in forecast accuracy, all-star likelihood, and market reactions between analysts with industry-related preanalyst experience and other analysts; DiD change in information asymmetry and prices after losing an expert versus a non-expert analyst",27 "strategy": "OLS and logit regressions with controls for analyst characteristics. Difference-in-differences around plausibly exogenous coverage terminations caused by brokerage house closures. Reg FD split used to probe the private-information channel.",28 "key_assumptions": [29 "Brokerage closures are exogenous to covered firms' information environment",30 "Controls capture other analyst skill attributes (experience, portfolio complexity, broker prestige)",31 "Self-selection of analysts into related industries does not drive results"32 ],33 "threats_addressed": [34 "private information via social connections (Reg FD test)",35 "sample selection/self-selection",36 "industry classification choice",37 "alternative performance measures and fixed effects",38 "analyst effort"39 ],40 "causal_language_strength": 3,41 "uses_causal_claim": true42 },43 "abstract_moves": {44 "question": {45 "present": false,46 "sentence_index": -147 },48 "gap": {49 "present": false,50 "sentence_index": -151 },52 "method": {53 "present": true,54 "sentence_index": 055 },56 "results": {57 "present": true,58 "sentence_index": 059 },60 "magnitude": {61 "present": false,62 "sentence_index": -163 },64 "contribution": {65 "present": true,66 "sentence_index": 267 }68 },69 "introduction": {70 "moves": [71 {72 "paragraph_index": 0,73 "move": "motivation"74 },75 {76 "paragraph_index": 0,77 "move": "related_literature_positioning"78 },79 {80 "paragraph_index": 0,81 "move": "literature_gap"82 },83 {84 "paragraph_index": 1,85 "move": "mechanism_or_theory_sketch"86 },87 {88 "paragraph_index": 1,89 "move": "identification_defense"90 },91 {92 "paragraph_index": 1,93 "move": "preview_of_results"94 },95 {96 "paragraph_index": 2,97 "move": "preview_of_results"98 },99 {100 "paragraph_index": 3,101 "move": "preview_of_results"102 },103 {104 "paragraph_index": 4,105 "move": "contribution_statement"106 },107 {108 "paragraph_index": 5,109 "move": "roadmap"110 }111 ],112 "contribution_paragraph_index": 4,113 "contribution_types": [114 "new_evidence",115 "new_data_or_measure"116 ],117 "roadmap_present": true,118 "literature_positioning_location": "both"119 },120 "section_roles": [121 {122 "idx": 0,123 "role": "introduction"124 },125 {126 "idx": 1,127 "role": "literature"128 },129 {130 "idx": 2,131 "role": "data"132 },133 {134 "idx": 3,135 "role": "results"136 },137 {138 "idx": 4,139 "role": "results"140 },141 {142 "idx": 5,143 "role": "robustness"144 },145 {146 "idx": 6,147 "role": "conclusion"148 },149 {150 "idx": 7,151 "role": "appendix"152 },153 {154 "idx": 8,155 "role": "appendix"156 },157 {158 "idx": 9,159 "role": "appendix"160 }161 ],162 "results_presentation": {163 "economic_significance_discussed": true,164 "statistical_significance_language": "unclear",165 "magnitude_statements_in_intro": true166 },167 "robustness": {168 "location": "mixed",169 "types": [170 "alternative_measure",171 "alternative_specification",172 "subsample",173 "other"174 ]175 },176 "tables": {177 "convention_notes": "Table notes name the estimator (OLS, logistic, DiD means) and give dependent variable definitions in detail, pointing to Appendix C for other variables. Data sources and the sample period are stated. Notes are truncated, so the reporting of standard errors cannot be seen.",178 "units_stated_in_notes": true,179 "standard_errors_description": "unclear"180 },181 "figures": {182 "uncertainty_shown": "no_figures"183 },184 "conclusion": {185 "bounded_claims": true,186 "policy_implications": false,187 "open_questions": false,188 "restates_contribution": true189 },190 "hedging": {191 "level": "moderate",192 "typical_devices": [193 "plausibly exogenous",194 "consistent with",195 "cannot completely rule out",196 "unlikely to be a primary explanation",197 "suggesting"198 ]199 },200 "claim_evidence": [201 {202 "claim": "Analysts with preanalyst experience in related industries forecast more accurately; the effect holds only for related-industry firms.",203 "evidence_type": "table",204 "evidence_ref": "Table II"205 },206 {207 "claim": "Odds of all-star status are 75% higher for related-experienced analysts than for unrelated-experienced analysts.",208 "evidence_type": "table",209 "evidence_ref": "Table III"210 },211 {212 "claim": "Forecast revisions by related-experienced analysts trigger about 0.43–0.45% stronger abnormal reactions than those of inexperienced analysts.",213 "evidence_type": "table",214 "evidence_ref": "Table IV"215 },216 {217 "claim": "Losing an industry-expert analyst through broker closure raises information asymmetry more than losing other analysts.",218 "evidence_type": "table",219 "evidence_ref": "Table VI"220 },221 {222 "claim": "The 3-day CAR difference-in-differences is 0.77% lower when an expert analyst is lost, and it does not reverse.",223 "evidence_type": "table",224 "evidence_ref": "Table VII"225 },226 {227 "claim": "The industry-experience effect persists after Reg FD, so a private-information channel is unlikely to be the main driver.",228 "evidence_type": "none_stated",229 "evidence_ref": "Introduction paragraph 1"230 },231 {232 "claim": "Results are robust to GICS classification and to alternative accuracy measures.",233 "evidence_type": "table",234 "evidence_ref": "Tables VIII, IX"235 },236 {237 "claim": "Recommendations of expert analysts have higher investment value.",238 "evidence_type": "table",239 "evidence_ref": "Table X"240 }241 ],242 "structural_summary": "The paper opens by citing practitioner surveys that rank industry knowledge as the most important analyst trait, then notes the lack of systematic evidence on it. It introduces hand-collected biographical data and previews results in sequence: forecast accuracy, a Reg FD channel test, all-star career outcomes, market reactions to revisions, and DiD tests on broker-closure terminations. A contribution paragraph and a roadmap close the introduction. The body then runs through a literature section, data, main results, analyst allocation and real effects, and robustness (largely in the Internet Appendix). The conclusion restates the findings.",243 "writing_profile_keywords": [244 "sell-side analysts",245 "hand-collected data",246 "practitioner motivation",247 "industry expertise",248 "difference-in-differences",249 "broker closures",250 "Internet Appendix robustness",251 "sequential results preview"252 ],253 "confidence": 0.8,254 "notes": "Paragraphs of the introduction appear to be partially missing between paragraphs 0 and 1, where the data and accuracy results are presumably introduced. The form of significance reporting is not visible in the excerpts. The Reg FD finding is stated in the introduction without a table reference, so its evidence type is recorded as none_stated."255 },256 "deterministic_checks": {257 "roadmap_regex": true,258 "contribution_regex_in_intro": true,259 "magnitude_in_abstract": false,260 "abstract_n_sentences": 3,261 "intro_n_paragraphs": 6,262 "intro_n_words": 872,263 "causal_words_in_abstract": 2,264 "hedge_words_in_abstract": 2,265 "stars_mentioned": false,266 "clustered_se_mentioned": false,267 "internet_appendix_mentioned": true,268 "section_roles_rule": {269 "0": "introduction",270 "1": "other",271 "2": "data",272 "3": "other",273 "4": "other",274 "5": "robustness",275 "6": "conclusion",276 "7": "appendix",277 "8": "appendix",278 "9": "appendix"279 }280 },281 "schema_valid": true,282 "code_version": "feb6c9e-dirty"283}