1{2 "validation_errors": [],3 "doc_id": "jof_12893",4 "ontology_version": "1.0",5 "prompt_version": "1.0",6 "model": "claude-opus-5-5",7 "effort": "medium",8 "annotated_at": "2026-11-03T23:14:27+00:00",9 "source_sha256": "f541747af7a703244781b8a987973fccaa8834f8e5ea9d8b59d626fc4e2a0bbf",10 "parser": "pymupdf-jof-heuristics-1.3",11 "payload_chars": 29282,12 "usage": {13 "input": 9895,14 "output": 2389,15 "cache_read": 3968,16 "cache_write": 017 },18 "cost_usd": 0.088154,19 "request_id": "req_011CggqtHtkgXUVBLjQotgL8",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": "Determinants of CMR inclusion in loan contracts, and the association of CMR presence with CEO turnover and future firm performance",27 "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.",28 "key_assumptions": [29 "Proxies for founder status, internal recruiting, heir apparent, noncompete enforcement and retirement age capture human capital risk and retention frictions",30 "The Lee (1978) structure correctly adjusts for the joint determination of covenants and yields",31 "Controls absorb confounding firm characteristics in the turnover and performance regressions"32 ],33 "threats_addressed": [34 "simultaneity of covenant structure and loan pricing",35 "alternative explanations: debt overhang, large blockholders, takeover threats",36 "other loan features such as change-in-control covenants",37 "timing of CMR addition and removal"38 ],39 "causal_language_strength": 3,40 "uses_causal_claim": true41 },42 "abstract_moves": {43 "question": {44 "present": false,45 "sentence_index": -146 },47 "gap": {48 "present": false,49 "sentence_index": -150 },51 "method": {52 "present": false,53 "sentence_index": -154 },55 "results": {56 "present": true,57 "sentence_index": 158 },59 "magnitude": {60 "present": true,61 "sentence_index": 262 },63 "contribution": {64 "present": true,65 "sentence_index": 166 }67 },68 "introduction": {69 "moves": [70 {71 "paragraph_index": 0,72 "move": "motivation"73 },74 {75 "paragraph_index": 6,76 "move": "literature_gap"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": "motivation"89 },90 {91 "paragraph_index": 9,92 "move": "mechanism_or_theory_sketch"93 },94 {95 "paragraph_index": 10,96 "move": "preview_of_results"97 },98 {99 "paragraph_index": 11,100 "move": "preview_of_results"101 },102 {103 "paragraph_index": 12,104 "move": "preview_of_results"105 },106 {107 "paragraph_index": 13,108 "move": "preview_of_results"109 },110 {111 "paragraph_index": 14,112 "move": "preview_of_results"113 },114 {115 "paragraph_index": 15,116 "move": "preview_of_results"117 },118 {119 "paragraph_index": 16,120 "move": "preview_of_results"121 },122 {123 "paragraph_index": 17,124 "move": "policy_or_broader_implication"125 },126 {127 "paragraph_index": 18,128 "move": "robustness_mention"129 },130 {131 "paragraph_index": 19,132 "move": "contribution_statement"133 },134 {135 "paragraph_index": 19,136 "move": "related_literature_positioning"137 },138 {139 "paragraph_index": 20,140 "move": "related_literature_positioning"141 },142 {143 "paragraph_index": 21,144 "move": "roadmap"145 }146 ],147 "contribution_paragraph_index": 19,148 "contribution_types": [149 "new_evidence",150 "new_data_or_measure"151 ],152 "roadmap_present": true,153 "literature_positioning_location": "inside_introduction"154 },155 "section_roles": [156 {157 "idx": 0,158 "role": "introduction"159 },160 {161 "idx": 1,162 "role": "data"163 },164 {165 "idx": 2,166 "role": "results"167 },168 {169 "idx": 3,170 "role": "results"171 },172 {173 "idx": 4,174 "role": "robustness"175 },176 {177 "idx": 5,178 "role": "conclusion"179 },180 {181 "idx": 6,182 "role": "appendix"183 }184 ],185 "results_presentation": {186 "economic_significance_discussed": true,187 "statistical_significance_language": "stars_and_tstats",188 "magnitude_statements_in_intro": true189 },190 "robustness": {191 "location": "mixed",192 "types": [193 "alternative_specification",194 "other"195 ]196 },197 "tables": {198 "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.",199 "units_stated_in_notes": false,200 "standard_errors_description": "clustered"201 },202 "figures": {203 "uncertainty_shown": "unclear"204 },205 "conclusion": {206 "bounded_claims": true,207 "policy_implications": false,208 "open_questions": true,209 "restates_contribution": true210 },211 "hedging": {212 "level": "moderate",213 "typical_devices": [214 "suggesting that",215 "tend to",216 "consistent with",217 "may",218 "is related to / association language",219 "some evidence"220 ]221 },222 "claim_evidence": [223 {224 "claim": "About 8.5% of sample firms, and 17.2% of small firms, have a CMR in at least one loan.",225 "evidence_type": "descriptive_statistic",226 "evidence_ref": "Section I; Table I"227 },228 {229 "claim": "CMRs are used more for small, risky, low-tangibility borrowers and for those with founder CEOs.",230 "evidence_type": "table",231 "evidence_ref": "Tables I-III"232 },233 {234 "claim": "CMR use rises when CEO retention frictions are higher, e.g., weak noncompete enforcement or a CEO near retirement age.",235 "evidence_type": "table",236 "evidence_ref": "Table IV"237 },238 {239 "claim": "CMR inclusion relates to a lower predicted loan yield, which is inconsistent with the collusion hypothesis.",240 "evidence_type": "table",241 "evidence_ref": "Table V"242 },243 {244 "claim": "CEO turnover probability falls by about 55% relative to the unconditional rate during a CMR loan term.",245 "evidence_type": "table",246 "evidence_ref": "Table VI; Figure 1"247 },248 {249 "claim": "When firms face retention frictions, CMRs are positively related to future Tobin's Q and operating performance.",250 "evidence_type": "table",251 "evidence_ref": "Table VII"252 }253 ],254 "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.",255 "writing_profile_keywords": [256 "hand-collected contract data",257 "competing hypotheses",258 "question-driven sections",259 "probit regressions",260 "creditor governance",261 "human capital",262 "epigraph opening"263 ],264 "confidence": 0.85,265 "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."266 },267 "deterministic_checks": {268 "roadmap_regex": true,269 "contribution_regex_in_intro": true,270 "magnitude_in_abstract": false,271 "abstract_n_sentences": 3,272 "intro_n_paragraphs": 22,273 "intro_n_words": 2477,274 "causal_words_in_abstract": 0,275 "hedge_words_in_abstract": 0,276 "stars_mentioned": true,277 "clustered_se_mentioned": true,278 "internet_appendix_mentioned": true,279 "section_roles_rule": {280 "0": "introduction",281 "1": "data",282 "2": "other",283 "3": "other",284 "4": "robustness",285 "5": "conclusion",286 "6": "appendix"287 }288 },289 "schema_valid": true,290 "code_version": "87c73cd-dirty"291}