🤖 AI Summary
Harvey et al. (2016) raised concerns about pervasive false discoveries in financial cross-sectional return predictability studies due to publication bias and empirical correlations among predictors.
Method: This paper develops a novel nonparametric and semiparametric framework for false discovery rate (FDR) estimation, jointly correcting for publication bias and predictor dependence. It introduces the first analytically tractable, closed-form FDR estimator with a theoretically guaranteed strict upper bound, integrating clustered bootstrap resampling, kernel density estimation, and a semiparametric mixture model.
Contribution/Results: Empirical analysis on the Chen–Zimmermann 205-factor dataset demonstrates that most published significant predictive relationships are genuine, and the estimated FDR is substantially lower than prior pessimistic assessments. This work provides the first FDR inference tool for asset pricing that combines rigorous theoretical foundations with practical implementability, significantly advancing the credibility assessment of empirical asset pricing research.
📝 Abstract
Harvey, Liu, and Zhu (2016) “argue that most claimed research findings in financial economics are likely false.” Surprisingly, their false discovery rate (FDR) estimates suggest most are true. I revisit their results by developing non- and semi-parametric FDR estimators that account for publication bias and empirical correlations. These estimators provide simple closed-form expressions and reliably produce an upper bound on the FDR in simulations that cluster-bootstrap from empirical predictor returns. Applying these estimators to the Chen-Zimmermann dataset of 205 predictors, I find that most claimed statistical findings in the cross-sectional predictability literature are likely true.
