Mutual Fund Flows, Performance Persistence, and Manager Skill

摘要

This paper adapts the model in Berk and Green (2004) to explain with reasonable success the data on mutual fund returns and flows. Using a Bayesian measure of fund-manager skill, I find that posterior estimates of skill vary substantially in the cross section and that perceived differences in ability persist through time. Consistent with the model, investor fund flows respond in a convex manner to posterior updates of manager skill scaled by functions of the expense ratio, and this result is robust to controlling for a convex function of past performance. While cross-sectional vari- ation in posterior fund manager skill estimates (which control for fund flows) has predictive power for out-of-sample subsequent fund performance, such predictability is present only in the short run. Beyond one year, high-skilled managers do not consistently out-perform low-skilled managers as skill-chasing fund flows equalize the realized abnormal net fund returns across managers. Overall, my empirical evidence is consistent with some managers possessing high ability, investors rationally chasing returns generated by those managers, and lack of long-run persistence in fund returns due to fund flows and diseconomies of scale in assets under management. Outside of the model, I show that the cross-sectional distribution of managerial ability is related to fund style and fund-manager compensation in a way that is consistent with matching marginal managerial productivity to the nature of the assets in the underlying portfolio.