A Role for the Theory of Tournaments in Studies of Mutual Fund Behavior

Abstract

We develop a two-period model of a Brown, Harlow & Starks (1996) mutual fund tournament in which two fund managers with unequal midyear performances compete for new cash inflows. When one of the managers is an exogenous benchmark, a winning manager indexes and a losing manager gambles. However, when both managers are active, a winning manager is more likely to gamble - especially when the midyear performance gap is high or when stocks offer high returns and low volatility. Empirical evidence that winning managers gamble has been documented in Chevalier & Ellison (1997), who find a positive correlation between past performance and increases in unsystematic risk, and Busse (1998), who finds that likely winners increase total risk more than likely losers. With a dataset of weekly returns from 1984-1996 for 660 mutual funds, we test for the other managerial gaming effects predicted by the model. Our results suggest a role for the theory of tournaments in studies of mutual fund behavior.