Mutual Fund Performance at Long Horizons

We study long horizon mutual fund performance, showing that measures constructed from long horizon returns contain notably different information than measures constructed from short horizon (monthly) returns. One prominent distinction is that the distribution of long horizon fund returns is strongly positively skewed, while such skewness is not evident in short horizon returns. As a consequence, the percentage of U.S. equity mutual funds that outperform the SPY ETF decreases from 46.9% in monthly returns to 39.9% in annual returns and 29.5% over the full 1991-2018 sample. More fundamentally, we show theoretically and empirically that alphas for short horizon returns need not have the same sign as alphas for long horizon returns. Compared to a benchmark of 40.9% obtained from monthly returns, the percentage of equity mutual funds with positive alpha estimates decreases to 12.4% (increases to 56.1%) at long horizons for funds with monthly betas greater (less) than one.