1{2 "validation_errors": [],3 "doc_id": "jof_12585",4 "ontology_version": "1.0",5 "prompt_version": "1.0",6 "model": "claude-opus-5-5",7 "effort": "medium",8 "annotated_at": "2026-11-03T23:00:11+00:00",9 "source_sha256": "7347d19ef06d3680c172b9fc5aa92d4f48efe74f95d493e222d4af4d6d1fbae0",10 "parser": "pymupdf-jof-heuristics-1.3",11 "payload_chars": 22129,12 "usage": {13 "input": 7666,14 "output": 2450,15 "cache_read": 3968,16 "cache_write": 017 },18 "cost_usd": 0.080458,19 "request_id": "req_011CggpoHp3AX4XaoHuxAaUo",20 "annotation": {21 "genre": "empirical",22 "subfield": "financial_intermediation_banking",23 "subfield_secondary": "corporate_finance; political economy of government credit in China",24 "primary_method": "instrumental_variables",25 "identification": {26 "estimand": "Effect of CDB government credit (industrial SOE loans and infrastructure loans) on private firms' and SOEs' assets, employment, sales and efficiency, in the same industry and in downstream industries",27 "strategy": "2SLS using the predicted timing of municipal party-secretary turnover cycles, interacted with each city's predetermined focal SOE industry, as instruments for province-industry CDB loan flows; OLS baseline; input-output matrix used to identify downstream industries",28 "key_assumptions": [29 "Predicted political turnover timing is exogenous to local economic conditions",30 "Exclusion: turnover cycles affect firms only through CDB credit, not through other channels such as other-bank borrowing, land sales, fiscal transfers or tax enforcement",31 "Focal industry is predetermined and stable over time",32 "Effects of SOE loans are separable from those of infrastructure loans"33 ],34 "threats_addressed": [35 "Turnover timing correlated with local conditions (hazard regressions)",36 "Alternative political channels (other banks, land sales, transfers, tax enforcement)",37 "Confounding by infrastructure loans",38 "Coincidence with national political cycles (off-cycle cities)",39 "Unobserved investment-opportunity shocks over the turnover cycle"40 ],41 "causal_language_strength": 4,42 "uses_causal_claim": true43 },44 "abstract_moves": {45 "question": {46 "present": true,47 "sentence_index": 048 },49 "gap": {50 "present": false,51 "sentence_index": -152 },53 "method": {54 "present": true,55 "sentence_index": 456 },57 "results": {58 "present": true,59 "sentence_index": 160 },61 "magnitude": {62 "present": true,63 "sentence_index": 264 },65 "contribution": {66 "present": false,67 "sentence_index": -168 }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": "literature_gap"83 },84 {85 "paragraph_index": 0,86 "move": "preview_of_approach"87 },88 {89 "paragraph_index": 0,90 "move": "preview_of_results"91 },92 {93 "paragraph_index": 1,94 "move": "preview_of_approach"95 },96 {97 "paragraph_index": 1,98 "move": "preview_of_results"99 },100 {101 "paragraph_index": 2,102 "move": "preview_of_results"103 },104 {105 "paragraph_index": 2,106 "move": "contribution_statement"107 },108 {109 "paragraph_index": 3,110 "move": "identification_defense"111 },112 {113 "paragraph_index": 4,114 "move": "related_literature_positioning"115 },116 {117 "paragraph_index": 4,118 "move": "contribution_statement"119 },120 {121 "paragraph_index": 5,122 "move": "roadmap"123 }124 ],125 "contribution_paragraph_index": 2,126 "contribution_types": [127 "new_evidence",128 "new_data_or_measure",129 "new_identification",130 "resolution_of_puzzle"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": "other"143 },144 {145 "idx": 2,146 "role": "data"147 },148 {149 "idx": 3,150 "role": "results"151 },152 {153 "idx": 4,154 "role": "conclusion"155 },156 {157 "idx": 5,158 "role": "appendix"159 }160 ],161 "results_presentation": {162 "economic_significance_discussed": true,163 "statistical_significance_language": "unclear",164 "magnitude_statements_in_intro": true165 },166 "robustness": {167 "location": "mixed",168 "types": [169 "subsample",170 "falsification",171 "alternative_specification"172 ]173 },174 "tables": {175 "convention_notes": "Notes give unit of observation, sample coverage (cities, provinces, industries, years), estimator (OLS, 2SLS, Cox hazard, Probit) and instrument definitions. An appendix table defines variables and gives loan units (100 million RMB). Appendix tables report first-stage results and an off-national-cycle subsample.",176 "units_stated_in_notes": true,177 "standard_errors_description": "unclear"178 },179 "figures": {180 "uncertainty_shown": "unclear"181 },182 "conclusion": {183 "bounded_claims": true,184 "policy_implications": true,185 "open_questions": false,186 "restates_contribution": true187 },188 "hedging": {189 "level": "moderate",190 "typical_devices": [191 "'suggestive evidence' for the political career-concern channel",192 "'alleviate the concern' when defending the exclusion restriction",193 "'could explain' when linking results to mixed prior findings",194 "'not likely to be confounded'"195 ]196 },197 "claim_evidence": [198 {199 "claim": "CDB SOE industrial loans crowd out private firms in the same industry, reducing their assets, employment and sales",200 "evidence_type": "table",201 "evidence_ref": "Table IV, Table V"202 },203 {204 "claim": "CDB SOE loans to upstream focal industries raise assets and sales of downstream private firms",205 "evidence_type": "table",206 "evidence_ref": "Table VI"207 },208 {209 "claim": "CDB loans to cities rise early in a city secretary's term, giving a zig-zag borrowing pattern",210 "evidence_type": "table",211 "evidence_ref": "Table III, Figure 3, Table AIII"212 },213 {214 "claim": "Other channels of politician influence are uncorrelated with the turnover cycles",215 "evidence_type": "table",216 "evidence_ref": "Table VII"217 },218 {219 "claim": "Infrastructure loans crowd in private firms, while industrial SOE loans have the opposite effect",220 "evidence_type": "table",221 "evidence_ref": "Table IX, Figure A2"222 },223 {224 "claim": "Turnover timing is not predicted by local economic conditions",225 "evidence_type": "table",226 "evidence_ref": "Table II"227 },228 {229 "claim": "Turnover-cycle effects are stronger in cities with better access to CDB credit",230 "evidence_type": "table",231 "evidence_ref": "Table VIII"232 }233 ],234 "structural_summary": "The introduction frames a crowd-out versus crowd-in debate and attributes mixed prior evidence to aggregate data. It previews a supply-chain decomposition built on proprietary bank data, with an instrument from political turnover timing, and reports magnitudes for first-stage and 2SLS effects. It then defends the exclusion restriction, positions the paper in the political business cycle literature and gives a roadmap. A background section on the bank and a data section are followed by one long results section. That section moves from politicians' career concerns, to the first stage, to same-industry and downstream effects, to alternative channels, infrastructure versus SOE loans, reduced form and overall effects. A one-paragraph conclusion restates the findings and draws broader implications. Further results are in an internet appendix.",235 "writing_profile_keywords": [236 "proprietary data",237 "political turnover instrument",238 "2SLS",239 "supply-chain spillovers",240 "crowding out versus crowding in",241 "China",242 "state-owned enterprises",243 "exclusion restriction defense"244 ],245 "confidence": 0.8,246 "notes": "Section I is institutional background, so it is coded 'other' rather than literature. The excerpts do not show whether results are reported with stars or standard errors, or how standard errors are computed. Robustness checks (off-cycle cities, other channels, heterogeneity by CDB access) appear within the results section and the appendix tables. The conclusion is truncated, so policy discussion is inferred only from its broader framing of government credit's costs and benefits."247 },248 "deterministic_checks": {249 "roadmap_regex": true,250 "contribution_regex_in_intro": false,251 "magnitude_in_abstract": false,252 "abstract_n_sentences": 5,253 "intro_n_paragraphs": 6,254 "intro_n_words": 974,255 "causal_words_in_abstract": 2,256 "hedge_words_in_abstract": 0,257 "stars_mentioned": false,258 "clustered_se_mentioned": false,259 "internet_appendix_mentioned": true,260 "section_roles_rule": {261 "0": "introduction",262 "1": "literature",263 "2": "data",264 "3": "results",265 "4": "conclusion",266 "5": "appendix"267 }268 },269 "schema_valid": true,270 "code_version": "feb6c9e-dirty"271}