Endogeneity in the mutual fund flow–performance relationship: An instrumental variables solution

Abstract

We use an instrumental variables (IV) approach to examine the effects of dynamic endogeneity when estimating the relationship between mutual fund flows and performance. Unlike the one-stage estimation approach commonly used in prior research, the IV approach allows us to address reverse causality between flow and performance. Through rigorous exclusion tests, we conclude that fund media coverage, risk ranking, and management structure win in a horse race as exogenous instruments for fund flow, while the fund turnover ratio and institutional share perform best as instruments for fund performance. We then demonstrate that endogeneity bias leads to inaccurate inferences in one-stage estimates, as evidenced by the reversals of the signs of flow and performance coefficient estimates when we switch to the IV approach. We find that careful attention to model specification allows us to resolve several widespread inconsistencies in the literature that were likely driven by model misspecification. • We examine dynamic endogeneity in mutual fund flows and performance. • The instrumental variable (IV) approach addresses reverse causality. • Flow and performance coefficients reverse signs with the IV approach. • We include rigorous exclusion, exogeneity, and identification tests. • Instruments for mutual fund flow are more robust than instruments for performance.