1{"text": "{\"genre\":\"empirical\",\"subfield\":\"behavioral_finance\",\"subfield_secondary\":\"household_finance\",\"primary_method\":\"reduced_form_regression\",\"identification\":{\"estimand\":\"How the inviter-invitee network is structured and how fast the idea spreads through it; how invested amounts (relative to income) relate to sponsor-investor differences in characteristics\",\"strategy\":\"Direct observation of word-of-mouth transmission through invitation-only membership, with power-law fitting and comparison to a simulated Poisson random network plus a calibrated Ponzi model; cross-sectional OLS with personal controls and inviter fixed effects\",\"key_assumptions\":[\"New investors learned of the scheme only from their inviter, since no public information existed\",\"Police interview data are free of the usual reporting and selection biases\",\"The simulated random network matches the real network on every dimension except degree distribution\"],\"threats_addressed\":[\"survey reporting and selection bias\",\"extreme tails of invested amounts (5% trimming)\",\"sponsor-level unobservables via inviter fixed effects\"],\"causal_language_strength\":2,\"uses_causal_claim\":false},\"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\":0}},\"introduction\":{\"moves\":[{\"paragraph_index\":0,\"move\":\"motivation\"},{\"paragraph_index\":1,\"move\":\"preview_of_approach\"},{\"paragraph_index\":2,\"move\":\"other\"},{\"paragraph_index\":3,\"move\":\"preview_of_approach\"},{\"paragraph_index\":3,\"move\":\"identification_defense\"},{\"paragraph_index\":4,\"move\":\"literature_gap\"},{\"paragraph_index\":4,\"move\":\"preview_of_results\"},{\"paragraph_index\":5,\"move\":\"preview_of_results\"},{\"paragraph_index\":6,\"move\":\"preview_of_results\"},{\"paragraph_index\":7,\"move\":\"preview_of_results\"},{\"paragraph_index\":8,\"move\":\"preview_of_results\"},{\"paragraph_index\":9,\"move\":\"contribution_statement\"},{\"paragraph_index\":10,\"move\":\"mechanism_or_theory_sketch\"},{\"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\":\"related_literature_positioning\"},{\"paragraph_index\":15,\"move\":\"roadmap\"}],\"contribution_paragraph_index\":9,\"contribution_types\":[\"new_evidence\",\"new_data_or_measure\"],\"roadmap_present\":true,\"literature_positioning_location\":\"inside_introduction\"},\"section_roles\":[{\"idx\":0,\"role\":\"introduction\"},{\"idx\":1,\"role\":\"data\"},{\"idx\":2,\"role\":\"results\"},{\"idx\":3,\"role\":\"results\"},{\"idx\":4,\"role\":\"results\"},{\"idx\":5,\"role\":\"conclusion\"}],\"results_presentation\":{\"economic_significance_discussed\":true,\"statistical_significance_language\":\"unclear\",\"magnitude_statements_in_intro\":true},\"robustness\":{\"location\":\"none_seen\",\"types\":[]},\"tables\":{\"convention_notes\":\"Long descriptive notes define the panels and variables. The early tables are descriptive comparisons of sponsors versus nonsponsors and the actual versus simulated network; later tables report cross-sectional OLS of invested amount scaled by income. Currency (euros, thousands) and measurement dates are stated.\",\"units_stated_in_notes\":true,\"standard_errors_description\":\"unclear\"},\"figures\":{\"uncertainty_shown\":\"unclear\"},\"conclusion\":{\"bounded_claims\":true,\"policy_implications\":true,\"open_questions\":false,\"restates_contribution\":true},\"hedging\":{\"level\":\"moderate\",\"typical_devices\":[\"suggests\",\"consistent with\",\"may be\",\"indicates\"]},\"claim_evidence\":[{\"claim\":\"The number of sponsored investors per sponsor approximately follows a power law, i.e., the network is scale-free\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 4; KS tests\"},{\"claim\":\"The scale-free network spreads the idea far faster than a random network (15 vs 161 steps)\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table III\"},{\"claim\":\"The cumulative number of investors follows an S-curve in network distance\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 5\"},{\"claim\":\"Investors invest more when their sponsor has relatively higher age, education, and income\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table VIII\"},{\"claim\":\"The marginal effect of sponsor income peaks just above the investor's own income\",\"evidence_type\":\"figure\",\"evidence_ref\":\"Figure 7\"},{\"claim\":\"Sponsors differ from nonsponsors in personal characteristics\",\"evidence_type\":\"table\",\"evidence_ref\":\"Table V\"}],\"structural_summary\":\"The paper opens with Shiller's social-epidemic view and introduces a unique Ponzi-scheme dataset from police investigation documents. Because entry was invitation-only, word-of-mouth transmission is directly observable, and the introduction defends the data quality. It previews network results (power law, simulation comparison, S-curve) and then investment-behavior regressions. Contributions are tied to the epidemic-diffusion, network-modeling, and peer-effects literatures before a roadmap. The body proceeds from background and data, to network analysis with simulation and calibration, to sponsor characteristics, to OLS regressions of invested amounts. The conclusion restates the finding that diffusion runs through a few hubs and notes welfare implications.\",\"writing_profile_keywords\":[\"unique data\",\"social network\",\"scale-free\",\"Ponzi scheme\",\"word-of-mouth\",\"simulation comparison\",\"descriptive-heavy\"],\"confidence\":0.8,\"notes\":\"Robustness checks are not visible in the excerpts, though an Internet Appendix exists. The regression analysis is associational. Section 1 is labeled literature by the rule but functions as background and data.\"}", "json": {"genre": "empirical", "subfield": "behavioral_finance", "subfield_secondary": "household_finance", "primary_method": "reduced_form_regression", "identification": {"estimand": "How the inviter-invitee network is structured and how fast the idea spreads through it; how invested amounts (relative to income) relate to sponsor-investor differences in characteristics", "strategy": "Direct observation of word-of-mouth transmission through invitation-only membership, with power-law fitting and comparison to a simulated Poisson random network plus a calibrated Ponzi model; cross-sectional OLS with personal controls and inviter fixed effects", "key_assumptions": ["New investors learned of the scheme only from their inviter, since no public information existed", "Police interview data are free of the usual reporting and selection biases", "The simulated random network matches the real network on every dimension except degree distribution"], "threats_addressed": ["survey reporting and selection bias", "extreme tails of invested amounts (5% trimming)", "sponsor-level unobservables via inviter fixed effects"], "causal_language_strength": 2, "uses_causal_claim": false}, "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": 0}}, "introduction": {"moves": [{"paragraph_index": 0, "move": "motivation"}, {"paragraph_index": 1, "move": "preview_of_approach"}, {"paragraph_index": 2, "move": "other"}, {"paragraph_index": 3, "move": "preview_of_approach"}, {"paragraph_index": 3, "move": "identification_defense"}, {"paragraph_index": 4, "move": "literature_gap"}, {"paragraph_index": 4, "move": "preview_of_results"}, {"paragraph_index": 5, "move": "preview_of_results"}, {"paragraph_index": 6, "move": "preview_of_results"}, {"paragraph_index": 7, "move": "preview_of_results"}, {"paragraph_index": 8, "move": "preview_of_results"}, {"paragraph_index": 9, "move": "contribution_statement"}, {"paragraph_index": 10, "move": "mechanism_or_theory_sketch"}, {"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": "related_literature_positioning"}, {"paragraph_index": 15, "move": "roadmap"}], "contribution_paragraph_index": 9, "contribution_types": ["new_evidence", "new_data_or_measure"], "roadmap_present": true, "literature_positioning_location": "inside_introduction"}, "section_roles": [{"idx": 0, "role": "introduction"}, {"idx": 1, "role": "data"}, {"idx": 2, "role": "results"}, {"idx": 3, "role": "results"}, {"idx": 4, "role": "results"}, {"idx": 5, "role": "conclusion"}], "results_presentation": {"economic_significance_discussed": true, "statistical_significance_language": "unclear", "magnitude_statements_in_intro": true}, "robustness": {"location": "none_seen", "types": []}, "tables": {"convention_notes": "Long descriptive notes define the panels and variables. The early tables are descriptive comparisons of sponsors versus nonsponsors and the actual versus simulated network; later tables report cross-sectional OLS of invested amount scaled by income. Currency (euros, thousands) and measurement dates are stated.", "units_stated_in_notes": true, "standard_errors_description": "unclear"}, "figures": {"uncertainty_shown": "unclear"}, "conclusion": {"bounded_claims": true, "policy_implications": true, "open_questions": false, "restates_contribution": true}, "hedging": {"level": "moderate", "typical_devices": ["suggests", "consistent with", "may be", "indicates"]}, "claim_evidence": [{"claim": "The number of sponsored investors per sponsor approximately follows a power law, i.e., the network is scale-free", "evidence_type": "figure", "evidence_ref": "Figure 4; KS tests"}, {"claim": "The scale-free network spreads the idea far faster than a random network (15 vs 161 steps)", "evidence_type": "table", "evidence_ref": "Table III"}, {"claim": "The cumulative number of investors follows an S-curve in network distance", "evidence_type": "figure", "evidence_ref": "Figure 5"}, {"claim": "Investors invest more when their sponsor has relatively higher age, education, and income", "evidence_type": "table", "evidence_ref": "Table VIII"}, {"claim": "The marginal effect of sponsor income peaks just above the investor's own income", "evidence_type": "figure", "evidence_ref": "Figure 7"}, {"claim": "Sponsors differ from nonsponsors in personal characteristics", "evidence_type": "table", "evidence_ref": "Table V"}], "structural_summary": "The paper opens with Shiller's social-epidemic view and introduces a unique Ponzi-scheme dataset from police investigation documents. Because entry was invitation-only, word-of-mouth transmission is directly observable, and the introduction defends the data quality. It previews network results (power law, simulation comparison, S-curve) and then investment-behavior regressions. Contributions are tied to the epidemic-diffusion, network-modeling, and peer-effects literatures before a roadmap. The body proceeds from background and data, to network analysis with simulation and calibration, to sponsor characteristics, to OLS regressions of invested amounts. The conclusion restates the finding that diffusion runs through a few hubs and notes welfare implications.", "writing_profile_keywords": ["unique data", "social network", "scale-free", "Ponzi scheme", "word-of-mouth", "simulation comparison", "descriptive-heavy"], "confidence": 0.8, "notes": "Robustness checks are not visible in the excerpts, though an Internet Appendix exists. The regression analysis is associational. Section 1 is labeled literature by the rule but functions as background and data."}, "model": "claude-opus-5-5", "stop_reason": "end_turn", "latency_s": 19.31, "usage": {"input": 7671, "output": 2139, "cache_read": 3968, "cache_write": 0}, "cost_usd": 0.074258, "request_id": "req_011CggqUmKbfb79TRYA2e7De", "cached": false, "ts": "2026-11-03T23:09:06+00:00", "effort": "medium"}