Constructing and Using Double-Adjusted Alphas to Analyze Mutual Fund Performance

Abstract

We propose a new approach for estimating mutual fund performance controlling for
both factor exposure and characteristics. Motivated factor models’ failure to fully adjust
returns for anomalies, our hierarchical Bayesian model separates the true alpha from
the effect of stock characteristics. It is straightforward and improves the traditional
two-pass estimation. Our double-adjusted alphas produce a different ranking of mutual
funds than the traditional alphas. Consequently, evidence of persistence in performance
increases, the link between selectivity and alpha disappears, but fund flows are related to
the true alpha and not to characteristics. Concludingly, measuring true outperformance
is crucial for understanding skill.

Keywords: Mutual fund performance, Double-adjusted performance, Firm characteristics, Hierarchical Bayes

JEL Classification: C11, G11, G23