Replication crisis in finance? A unified modeling framework for reconciling pessimism and optimism

We introduce a unified modeling framework that explains why conflicting findings on the replicability of financial anomalies, defined as characteristic-sorted, zero-cost long-short portfolios that earn positive abnormal returns after controlling for a chosen benchmark, can arise from different prior beliefs. For each anomaly, the framework encodes prior skepticism about its existence as an explicit, tunable prior null probability and then updates it systematically with observed data to yield a posterior null probability, declaring a positive finding only when this probability falls below a chosen threshold. Our framework is not a new anomaly-testing model but a flexible, model‑agnostic decision layer that incorporates theory- or domain-informed priors into a context-aware study of anomalies. An empirical illustration shows that “pessimistic” versus “optimistic” conclusions reflect different points along a spectrum of prior assumptions, rather than contradictory evidence. In this way, the framework provides a transparent, posterior-based standard that enables the evidence supporting or refuting positive findings to be reproduced consistently across studies.