A1078
Title: Empirical decisions and replicating anomalies: The benefit of the aggregate average
Authors: Jiaqi Guo - University of Birmingham (United Kingdom)
George Korniotis - University of Miami (United States)
Alok Kumar - University of Miami (United States)
Peng Li - University of Bath (United Kingdom) [presenting]
Abstract: The reproducibility of asset pricing anomalies is examined by assessing the effects of empirical choices in constructing samples and portfolios. To reduce the influence of methodological variation, a two-stage bootstrapping procedure is proposed that accounts for sample period and methodological variability, and averages anomaly return spreads across methods. The aggregate average approach mitigates p-hacking and false discoveries. Analyzing 173 anomalies using 96 methods, it is found that while p-hacking is possible, particularly with a t-statistic threshold of 2, around 70\% of the 173 anomalies can be successfully replicated.
