This paper examines the agency conflict between mutual fund investors and mutual fund companies. Investors want the fund company to use its judgment to maximize risk-adjusted fund returns, while the fund company is motivated by its own profits and has an incentive to increase investment inflows. Using a semiparametric model, the authors estimate the shape of the flow-performance relationship for a sample of growth and growth and income funds over the 1982-1992 period. They find that mutual funds alter their portfolio riskiness between September and December in a manner consistent with these risk incentives. The analysis shows that mutual funds adjust their riskiness based on their year-to-date return, with funds that are behind the market having an incentive to gamble to catch up, and funds that are ahead having an incentive to lock in gains. The study also finds significant nonlinearities in the relationship between flow and performance, with the overall sensitivity and shape of the relationship depending on the fund's age. The results suggest that mutual funds respond to incentives in a way that is consistent with the flow-performance relationship, providing new insights into fund behavior. The paper concludes that the agency problem in the mutual fund industry is complex and that the flow-performance relationship is empirically estimable, offering a useful framework for understanding fund behavior.This paper examines the agency conflict between mutual fund investors and mutual fund companies. Investors want the fund company to use its judgment to maximize risk-adjusted fund returns, while the fund company is motivated by its own profits and has an incentive to increase investment inflows. Using a semiparametric model, the authors estimate the shape of the flow-performance relationship for a sample of growth and growth and income funds over the 1982-1992 period. They find that mutual funds alter their portfolio riskiness between September and December in a manner consistent with these risk incentives. The analysis shows that mutual funds adjust their riskiness based on their year-to-date return, with funds that are behind the market having an incentive to gamble to catch up, and funds that are ahead having an incentive to lock in gains. The study also finds significant nonlinearities in the relationship between flow and performance, with the overall sensitivity and shape of the relationship depending on the fund's age. The results suggest that mutual funds respond to incentives in a way that is consistent with the flow-performance relationship, providing new insights into fund behavior. The paper concludes that the agency problem in the mutual fund industry is complex and that the flow-performance relationship is empirically estimable, offering a useful framework for understanding fund behavior.