UC Irvine

I examine how mutual fund investor overconfidence and over-extrapolation bias affect mutual fund flow and future performance. By generalizing the model of Berk and Green (2004), I show that when investors over-extrapolate fund past return and/or are overconfident, fund flows no longer eliminate fund outperformance efficiently as predicted in their theory. Instead, fund future performance can be positively predicted by unit value-added, a variable that represents fund past performance, and negatively predicted by unit fee-added, a variable that represents fund flow activities. Using the U.S. actively managed mutual funds data, I find that the skill-competition variable constructed based on my model predicts fund future risk-adjusted performance. In addition, I find that investor overconfidence dominates over-extrapolation bias when investors make an investment decision on actively managed mutual funds.