{"id":"dd6577c1-8e31-46d3-9141-f82a20918d06","arxiv_id":"1908.11492","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Across 83 countries, investors from cultures high in long-term orientation and indulgence show weaker disposition effects.","lead":"Using four years of trading records from 387,993 investors in 83 countries, this paper shows that the disposition effect, the tendency to sell winners and hold losers, varies across national cultures. Investors from countries high in long-term orientation and indulgence show weaker disposition effects, suggesting culture shapes trading mistakes.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Odean measure is applied to CFDs without adaptation; same-day round trips yield mechanically zero DE, so cultural coefficients may reflect trading-horizon habits rather than disposition.","rationale":"The reader's conditional verdict is built on the same measurement-validity concern: Odean's Equation (1) is imported into a CFD context that is materially different from the equity setting it was designed for. My stress-test identifies a specific mechanical failure mode: any trader with no open positions at the daily snapshot receives a DE of exactly zero. Since CFDs are frequently held intraday, this could produce a large share of artificially zero observations, pull the national averages toward zero, and create spurious associations if the propensity to day-trade correlates with cultural dimensions such as long-term orientation. The paper provides no validation that the measure behaves well on this platform, and the unavailability of data prevents the reader from checking the distribution of holding periods or the share of mechanical zeros. This is a load-bearing concern because the dependent variable is invalid for a substantial part of the sample. The test I propose would settle it directly: recompute DE on an overnight-position subsample and see whether the cultural coefficients survive. If they do, the conditional verdict should stand; if they do not, the central claim falls. The authors' own limitation paragraph only flags self-selection, not this measurement issue, and while self-selection is also acknowledged, the CFD-specific mismeasurement is more damaging because it threatens the internal validity of the outcome variable itself. I therefore keep the reader's CONDITIONAL verdict unchanged, but emphasize that satisfying the condition requires evidence on CFDs, not just broader robustness checks.","tokens_in":4942,"tokens_out":4122,"duration_ms":46489,"concrete_test":"Use the raw transaction data to recompute each trader's DE after excluding trades closed within the same calendar day (or, alternatively, after requiring at least one position open at the daily snapshot); re-estimate Model 5 on the subsample of traders with at least, say, 10 overnight positions. If the long-term orientation and indulgence coefficients become insignificant or change in sign/magnitude beyond sampling error, the headline result is an artifact of the CFD day-trading mechanism; if they survive, the measurement concern is mitigated. Also report the fraction of accounts whose DE equals exactly zero under the original measure.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The dependent variable is Odean's (1998) disposition measure, computed per Equation (1) as realized_gains/(realized_gains+paper_gains) minus realized_losses/(realized_losses+paper_losses), counting realized and paper positions 'for each trading day.' The data, however, are CFDs: leveraged derivative contracts that are frequently opened and closed intraday. For any trader who closes all positions before the daily snapshot, paper_gains and paper_losses are both zero, so both ratios mechanically equal one and DE=0. The paper reports mean DE of 0.024 across 387,993 traders, consistent with a large mass of mechanical zeros. It also never states how short CFD positions, leverage, or rollovers enter the 'paper' calculation. Odean's measure was designed around long equity positions with a cost basis; applying it to intraday CFDs changes the object being measured. If day-trading intensity or holding-period habits vary with national culture (e.g., long-term orientation), the significant negative coefficients in Model 5 could reflect cultural differences in trading style rather than in the reluctance to realize losses. The regression's dependent variable is then mismeasured, and the headline cultural correlations are not interpretable.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses trading histories from a UK-based CFD broker for 387,993 investors in 83 countries to estimate an individual-level disposition effect following Odean (1998). It then regresses this measure on Hofstede's six cultural dimensions, investor demographics, country economic conditions, and region dummies, with standard errors clustered by nationality. The central claim is that long-term orientation and indulgence are negatively associated with the disposition effect, alongside age and gender effects, providing cross-country evidence that national culture helps explain the disposition effect.","tokens_in":5289,"tokens_out":3427,"duration_ms":37536,"significance":"If the central claim holds, the paper would be one of the first large-sample cross-country demonstrations that cultural background is systematically related to the disposition effect. The dataset is unusually large and the empirical design is straightforward, with the cultural scores taken from published sources and the disposition effect computed from transaction-level data. The paper also confirms known demographic patterns (age and gender) internationally. However, the contribution depends critically on the validity of the Odean measure in a CFDs setting, which the paper does not establish. The very low R-squared (0.4%) and the country-level nature of the cultural regressors further mean that the economic importance of the effects needs careful quantification. The paper is honest about some limitations (self-selection) but does not address the measurement concern that is most load-bearing.","major_comments":[{"comment":"No explicit comment here — this is a placeholder to keep the array length. Remove this entry if not needed. Actually, we have three major comments already, which is fine. Let's keep the array with exactly these three entries.","section":"§3, Eq. (1); §4, Table 1"}],"minor_comments":[{"comment":"The text says 'GDP per capital growth' in two places; this should be 'GDP per capita growth'.","section":"§2, Table 1"},{"comment":"The formula and surrounding text contain typographical errors (e.g., '0 represents an investors') and the equation itself is not rendered cleanly. The notation for realized and paper gains/losses should be defined explicitly.","section":"§3, Eq. (1)"},{"comment":"Figure 1 includes only countries with at least 100 unique investors, but the main regressions use all 83 countries. The text should clarify that the figure is a subset and that the regression results do not change with the subset.","section":"§4, Figure 1"},{"comment":"The paper defines nationality by citizenship, but Hofstede's scores are country-level cultural values. For countries with large immigrant or multicultural populations, citizenship may not correspond to the cultural orientation measured by Hofstede; a brief discussion or robustness check using country of residence (if available) would strengthen the analysis.","section":"§2"}],"recommendation":"major_revision","confidential_remarks":"The paper is well written and the research question is interesting, but the measurement validity of the disposition effect in CFDs is the central issue. The authors need to address this head-on with additional analysis and transparent reporting of the dependent variable's distribution. I would not reject at this stage, because the problem is potentially fixable: if the authors can show that a large fraction of traders hold positions across valuation points and that results are robust to excluding intraday-only traders, the paper could be acceptable. The few-cluster inference issue also needs attention. I recommend major revision rather than reject."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper does something nobody has done directly: it compares disposition-effect magnitudes across 83 countries on a single broker's platform, using real trading records, and it connects them to Hofstede's cultural dimensions. That is a real step beyond the single-country and lab studies. The regression work is transparent, standard errors are clustered, and the robustness checks on sample size by country are sensible. If the measure is right, the negative associations with long-term orientation and indulgence are worth taking seriously.\n\nBut I have a real worry about the measure, and I think it is load-bearing. The dependent variable is Odean's (1998) ratio of realized to realized-plus-paper gains minus the same for losses. That ratio was designed around equity positions that stay open across days. The data here are CFDs, which are often opened and closed intraday. If a trader closes all positions before the daily count, paper gains and losses are both zero, both ratios collapse to one, and the disposition effect is mechanically zero. The paper reports a mean of 0.024 across 387,993 traders, which looks consistent with a large mass of mechanical zeros. The authors never explain how they handle days with zero denominators, nor how short positions, leverage, or rollovers enter the paper calculation. If country-level culture correlates with trading horizon—and long-term orientation plausibly does—then the significant negative coefficients in Model 5 may be picking up day-trading frequency rather than reluctance to realize losses. That is not a minor quibble; it goes to whether the paper is about the disposition effect at all.\n\nThere are softer spots too: only 83 clusters makes the cluster-robust inference fragile; the R-squared is 0.4%, so culture explains very little of the individual variation; and the analysis assigns Hofstede's country scores to every investor of that nationality, so within-country heterogeneity is ignored. The authors acknowledge self-selection, which is fair, but self-selection plus the CFD measurement problem makes the headline coefficient hard to interpret without additional validation.\n\nAll that said, the paper is not incoherent. It is a serious empirical attempt at a question that would matter if the measure held up. The right next step is for the authors to show that the Odean measure works on CFDs—maybe by validating on a subsample of longer-held positions, or by controlling for holding period and trading frequency. Until then, I would read the result as suggestive, not established.\n\nIf I were a desk editor, I would send this to referees. A good referee would pin the measurement issue quickly, but the paper deserves that scrutiny rather than a desk reject. I would not cite it for the cultural claim until the CFD validation is done.\n\nRecommendation: engage with it as a working paper, but treat the main result as conditional.","headline":"A genuinely new cross-country test of culture and the disposition effect, but the CFD-based dependent variable may be measuring trading horizon rather than the bias, so the headline result is conditional.","tokens_in":5653,"tokens_out":2986,"would_cite":false,"duration_ms":27136,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"National culture is associated with the disposition effect: long-term orientation and indulgence predict lower sell-winners-hold-losers behavior across 83 countries.","keywords":["disposition effect","national culture","long-term orientation","indulgence","investor behavior","loss aversion","contracts for difference","cross-country brokerage data"],"falsifier":"Run the same regressions on a subsample of investors whose CFD positions are held at least overnight, or on stock-like instruments only; if the negative coefficients on long-term orientation and indulgence disappear or flip sign, the claim that culture drives the disposition effect is not supported.","tokens_in":4760,"feed_emoji":"📉","tokens_out":7085,"duration_ms":68980,"temperature":0.7,"pith_summary":"This paper asks whether the tendency to sell winning investments too early and hold losing ones too long—the disposition effect—varies systematically with national culture. Using brokerage records of 387,993 traders from 83 countries, it finds the average disposition effect differs widely across countries, from -0.04 to 0.22, and that two cultural traits predict it: countries scoring high on long-term orientation and high on indulgence show significantly lower disposition effects. The paper argues this means cultural background is a real determinant of investor behavior, alongside age and gender. A sympathetic reader would care because it moves a well-known bias from individual psychology into the realm of cross-cultural differences, with implications for how financial advice and regulation might be tailored across societies.","feed_headline":"Culture weakens the sell-winners, hold-losers bias","feed_subtitle":"Across 83 countries, future-focused and indulgent cultures show less of this costly investor bias.","key_machinery":"The machinery is the daily realized-versus-paper gain/loss ratio, computed from Equation (1) as the fraction of realized gains minus the fraction of realized losses among all paper and realized outcomes. Equation (2) then runs an OLS regression of this investor-level measure on six national-culture dimensions (power distance, individualism, masculinity, uncertainty avoidance, long-term orientation, and indulgence), demographic dummies, economic conditions, and region, with standard errors clustered by nationality. The cultural variables carry the argument: the paper isolates their partial correlations from demographics, wealth, and region, and shows the two significant cultural coefficients survive the full specification.","core_discovery":"The central claim is that national culture helps explain international variation in the disposition effect. For each trader the paper counts realized and paper gains and losses per day and computes the standard measure: realized gains over all gains minus realized losses over all losses, so positive values mean selling winners and holding losers. Regressing this on six cultural dimensions plus demographics, economic conditions, and region, the full model shows significant negative coefficients for long-term orientation and indulgence; at the mean, a one-standard-deviation increase in each is associated with a 15.7 percent and a 25.2 percent lower disposition effect, respectively. The paper also reports that men show a smaller disposition effect and that the effect grows with age, while GDP per capita and growth are insignificant once culture is controlled.","pith_inferences":["If the negative indulgence coefficient is true, the usual story that the disposition effect stems from immediate gratification needs refinement: high-indulgence cultures might be less attached to past decisions, not more impulsive. A direct test would measure loss aversion or regret within the same sample.","Because all trades come from one broker's contracts-for-difference clients, the 83-country comparison holds the platform constant but not the instrument class; the cultural pattern could partly reflect how different cultures trade leveraged products. Replicating on a multi-country equity broker dataset would separate the two.","A natural follow-up is to trace second-generation immigrants: if culture rather than national institutions drives the effect, immigrants from high-long-term-orientation countries should retain a lower disposition effect even after living under different institutions."],"forward_implications":["If culture shapes the disposition effect, national cultural profiles can be used to predict where the sell-winners-hold-losers bias will be strongest even before trade-level data are available.","Interventions aimed at reducing the disposition effect may need cultural adaptation: future-focused and high-indulgence societies already show a milder bias, so feedback or default-based interventions would likely have the largest scope in low-scoring countries.","The demographic findings—men less affected, older investors more affected—extend single-country results to a global sample, strengthening the generalizability of gender and age effects.","Because GDP per capita and growth lose significance once culture is included, the results point toward cultural mechanisms such as loss aversion and mental accounting rather than economic development as the active channel."],"supporting_citations":[{"why":"Provides the realized-versus-paper gain/loss measure used to construct the dependent variable.","marker":"Odean (1998)"},{"why":"Defines the disposition effect and links it to loss aversion, the behavioral mechanism the authors interpret.","marker":"Shefrin and Statman (1985)"},{"why":"Supplies the six cultural dimensions, including long-term orientation and indulgence, used as the key explanatory variables.","marker":"Hofstede (2001)"},{"why":"Prior evidence that sophistication and demographics affect the disposition effect, motivating the control set.","marker":"Dhar and Zhu (2006)"},{"why":"Prior gender finding that the paper extends to an international sample.","marker":"Rau (2014)"},{"why":"Shows culture affects loss aversion, providing a cultural route to the disposition effect.","marker":"Wang, Rieger, and Hens (2017)"},{"why":"Shows mental accounting differs across cultures, another mechanism behind the disposition effect.","marker":"Banerjee, Chatterjee, Mishra, and Mishra (2019)"},{"why":"Meta-analysis showing cultural dimensions predict emotions and behavior at the individual level.","marker":"Taras, Kirkman, and Steel (2010)"}],"fun_headline_variants":["Future-focused cultures resist the sell-winner bias","Indulgence and long-term thinking slash disposition effect","Global study: culture shapes the hold-losers bias","Culture explains why some traders hold losers longer","Long-term orientation and indulgence curb costly bias"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire result rests on treating the realized-versus-paper gain/loss count from leveraged, often intraday contracts-for-difference trading as a valid measure of the disposition effect; if that measure is not comparable across instruments and holding periods, the cultural correlations cannot be interpreted.","fun_headline_variants_meta":{"raw":{"variants":["Future-focused cultures resist the sell-winner bias","Indulgence and long-term thinking slash disposition effect","Global study: culture shapes the hold-losers bias","Culture explains why some traders hold losers longer","Long-term orientation and indulgence curb costly bias"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000133,"raw_usage":{"total_tokens":1041,"prompt_tokens":755,"completion_tokens":286,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":371,"completion_tokens_details":{"reasoning_tokens":215}},"tokens_in":371,"tokens_out":286,"duration_ms":3585,"temperature":1.0,"reasoning_tokens":215,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:13:26.274771+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same regressions on a subsample of investors whose CFD positions are held at least overnight, or on stock-like instruments only; if the negative coefficients on long-term orientation and indulgence disappear or flip sign, the claim that culture drives the disposition effect is not supported.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the disposition effect and links it to loss aversion, the behavioral mechanism the authors interpret."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the six cultural dimensions, including long-term orientation and indulgence, used as the key explanatory variables."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior evidence that sophistication and demographics affect the disposition effect, motivating the control set."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior gender finding that the paper extends to an international sample."},{"cited_title":"O., & Hens, T","cited_arxiv_id":null,"evidence_quote":"Shows culture affects loss aversion, providing a cultural route to the disposition effect."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows mental accounting differs across cultures, another mechanism behind the disposition effect."},{"cited_title":"L., & Steel, P","cited_arxiv_id":null,"evidence_quote":"Meta-analysis showing cultural dimensions predict emotions and behavior at the individual level."}],"review_version":1}