1{"text": "{\"genre\":\"empirical\",\"subfield\":\"household_finance\",\"subfield_secondary\":\"behavioral_finance\",\"primary_method\":\"natural_experiment_event_study\",\"identification\":{\"estimand\":\"Effect of depression-era local (region-occupation) labor market deterioration on long-run risky asset investment more than a decade later, plus the part not working through income, employment or wealth\",\"strategy\":\"Cross-sectional variation in region-occupation unemployment changes during the 1991-93 Finnish depression, conditional on region and occupation fixed effects and predepression controls; secondhand experiences of neighbors and family isolate non-wealth channels; matching as an alternative estimator\",\"key_assumptions\":[\"Conditional independence: exposure to local labor market distress is unrelated to unobserved worker characteristics given fixed effects and predepression controls\",\"Secondhand experiences do not affect the individual's wealth, income or employment\",\"Depression shocks (Soviet collapse, twin crisis) were unexpected\"],\"threats_addressed\":[\"Omitted variables / selection into occupations and regions\",\"Wealth, income and employment channels as bad controls\",\"Measurement error from recall bias in survey data\",\"Sensitivity to the labor market definition\"],\"causal_language_strength\":4,\"uses_causal_claim\":true},\"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\":2},\"magnitude\":{\"present\":false,\"sentence_index\":-1},\"contribution\":{\"present\":true,\"sentence_index\":1}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"tension_or_puzzle\"},{\"paragraph_index\":1,\"move\":\"literature_gap\"},{\"paragraph_index\":1,\"move\":\"contribution_statement\"},{\"paragraph_index\":2,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"mechanism_or_theory_sketch\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"identification_defense\"},{\"paragraph_index\":5,\"move\":\"preview_of_approach\"},{\"paragraph_index\":6,\"move\":\"identification_defense\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_approach\"},{\"paragraph_index\":9,\"move\":\"preview_of_results\"},{\"paragraph_index\":10,\"move\":\"policy_or_broader_implication\"},{\"paragraph_index\":10,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":11,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":12,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":13,\"move\":\"related_literature_positioning\"},{\"paragraph_index\":14,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":1,\"contribution_types\":[\"new_evidence\",\"new_identification\",\"new_method\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"empirical_design\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"stars_and_tstats\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"within_results\",\"types\":[\"alternative_specification\",\"alternative_measure\",\"falsification\",\"subsample\",\"other\"]},\"tables\":{\"convention_notes\":\"Notes report coefficients with t-values and spell out the dependent variable, control sets and sample. Income and wealth are in 2005 CPI-adjusted terms. Later tables cross-reference the control sets of earlier columns. Table I pairs a falsification test with descriptive statistics.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"clustered\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":false,\"open_questions\":true,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"plausibly exogenous\",\"suggest\",\"likely\",\"unlikely to explain\",\"interpret as lower bound\",\"candidate explanations\"]},\"claim_evidence\":[{\"claim\":\"A one-SD worse labor market experience lowers stock market participation by about 2.8-3.6 pp, against a 22% base rate\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table II\"},{\"claim\":\"Depression-era labor conditions do not predict predepression stock market participation (falsification)\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 3; Table I\"},{\"claim\":\"Labor market conditions affect later income, unemployment and net worth\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"Results hold with family fixed effects, parental controls and alternative labor market definitions\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table IV\"},{\"claim\":\"Matching adversely affected workers to less affected ones yields similar results\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"},{\"claim\":\"Adverse experiences also reduce holdings of fixed income funds, balanced funds and derivatives\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VI\"},{\"claim\":\"Neighbors' and family members' adverse experiences reduce risky investment by 0.4-0.6 pp, so at least 11% of the effect is not via wealth or income\",\"evidence_type\":\"table\",\"evidence_ref\":\"Tables VII, VIII\"}],\"structural_summary\":\"The introduction frames portfolio heterogeneity as a puzzle that twin studies leave open. It then organizes the design around three identification challenges, each answered by a feature of the Finnish depression setting, and previews effect sizes with clustered t-values. Four literature strands follow, then a roadmap. Section I covers the timeline, register data and a conditional independence defense. Section II gives baseline firsthand-experience regressions followed by robustness checks: controls, labor market definitions, regional analyses, matching and alternative assets. Section III uses secondhand experiences of neighbors and family to separate preference and belief channels from wealth channels. The conclusion restates the contribution, bounds the mechanism to preferences, beliefs or trust, and suggests extensions.\",\"writing_profile_keywords\":[\"natural experiment\",\"register data\",\"identification challenges enumerated\",\"falsification test\",\"secondhand experiences\",\"lower-bound interpretation\",\"clustered t-values\",\"mechanism isolation\"],\"confidence\":0.85,\"notes\":\"Section II carries the main results; the 'other' role assigned by the rule was reinterpreted as results. Section III was labeled robustness by the rule but serves as channel analysis, so it is classified as results. Whether figures show uncertainty cannot be determined from the captions.\"}", "json": {"genre": "empirical", "subfield": "household_finance", "subfield_secondary": "behavioral_finance", "primary_method": "natural_experiment_event_study", "identification": {"estimand": "Effect of depression-era local (region-occupation) labor market deterioration on long-run risky asset investment more than a decade later, plus the part not working through income, employment or wealth", "strategy": "Cross-sectional variation in region-occupation unemployment changes during the 1991-93 Finnish depression, conditional on region and occupation fixed effects and predepression controls; secondhand experiences of neighbors and family isolate non-wealth channels; matching as an alternative estimator", "key_assumptions": ["Conditional independence: exposure to local labor market distress is unrelated to unobserved worker characteristics given fixed effects and predepression controls", "Secondhand experiences do not affect the individual's wealth, income or employment", "Depression shocks (Soviet collapse, twin crisis) were unexpected"], "threats_addressed": ["Omitted variables / selection into occupations and regions", "Wealth, income and employment channels as bad controls", "Measurement error from recall bias in survey data", "Sensitivity to the labor market definition"], "causal_language_strength": 4, "uses_causal_claim": true}, "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": 2}, "magnitude": {"present": false, "sentence_index": -1}, "contribution": {"present": true, "sentence_index": 1}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "tension_or_puzzle"}, {"paragraph_index": 1, "move": "literature_gap"}, {"paragraph_index": 1, "move": "contribution_statement"}, {"paragraph_index": 2, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "mechanism_or_theory_sketch"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "identification_defense"}, {"paragraph_index": 5, "move": "preview_of_approach"}, {"paragraph_index": 6, "move": "identification_defense"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_approach"}, {"paragraph_index": 9, "move": "preview_of_results"}, {"paragraph_index": 10, "move": "policy_or_broader_implication"}, {"paragraph_index": 10, "move": "related_literature_positioning"}, {"paragraph_index": 11, "move": "related_literature_positioning"}, {"paragraph_index": 12, "move": "related_literature_positioning"}, {"paragraph_index": 13, "move": "related_literature_positioning"}, {"paragraph_index": 14, "move": "roadmap"}], "contribution_paragraph_index": 1, "contribution_types": ["new_evidence", "new_identification", "new_method"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "empirical_design"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "stars_and_tstats", "magnitude_statements_in_intro": true}, "robustness": {"location": "within_results", "types": ["alternative_specification", "alternative_measure", "falsification", "subsample", "other"]}, "tables": {"convention_notes": "Notes report coefficients with t-values and spell out the dependent variable, control sets and sample. Income and wealth are in 2005 CPI-adjusted terms. Later tables cross-reference the control sets of earlier columns. Table I pairs a falsification test with descriptive statistics.", "units_stated_in_notes": true, "standard_errors_description": "clustered"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": false, "open_questions": true, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["plausibly exogenous", "suggest", "likely", "unlikely to explain", "interpret as lower bound", "candidate explanations"]}, "claim_evidence": [{"claim": "A one-SD worse labor market experience lowers stock market participation by about 2.8-3.6 pp, against a 22% base rate", "evidence_type": "table", "evidence_ref": "Table II"}, {"claim": "Depression-era labor conditions do not predict predepression stock market participation (falsification)", "evidence_type": "figure", "evidence_ref": "Figure 3; Table I"}, {"claim": "Labor market conditions affect later income, unemployment and net worth", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "Results hold with family fixed effects, parental controls and alternative labor market definitions", "evidence_type": "table", "evidence_ref": "Table IV"}, {"claim": "Matching adversely affected workers to less affected ones yields similar results", "evidence_type": "table", "evidence_ref": "Table V"}, {"claim": "Adverse experiences also reduce holdings of fixed income funds, balanced funds and derivatives", "evidence_type": "table", "evidence_ref": "Table VI"}, {"claim": "Neighbors' and family members' adverse experiences reduce risky investment by 0.4-0.6 pp, so at least 11% of the effect is not via wealth or income", "evidence_type": "table", "evidence_ref": "Tables VII, VIII"}], "structural_summary": "The introduction frames portfolio heterogeneity as a puzzle that twin studies leave open. It then organizes the design around three identification challenges, each answered by a feature of the Finnish depression setting, and previews effect sizes with clustered t-values. Four literature strands follow, then a roadmap. Section I covers the timeline, register data and a conditional independence defense. Section II gives baseline firsthand-experience regressions followed by robustness checks: controls, labor market definitions, regional analyses, matching and alternative assets. Section III uses secondhand experiences of neighbors and family to separate preference and belief channels from wealth channels. The conclusion restates the contribution, bounds the mechanism to preferences, beliefs or trust, and suggests extensions.", "writing_profile_keywords": ["natural experiment", "register data", "identification challenges enumerated", "falsification test", "secondhand experiences", "lower-bound interpretation", "clustered t-values", "mechanism isolation"], "confidence": 0.85, "notes": "Section II carries the main results; the 'other' role assigned by the rule was reinterpreted as results. Section III was labeled robustness by the rule but serves as channel analysis, so it is classified as results. Whether figures show uncertainty cannot be determined from the captions."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.78, "usage": {"input": 7744, "output": 2385, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.07947, "request_id": "req_011CggpSEW2QrqobxsXayFPw", "cached": false, "ts": "2026-11-03T22:55:25+00:00", "effort": "medium"}