| The first chapter investigates the empirical implications of style investing. Using CRSP stock and mutual fund data, we find significant excess and risk-adjusted returns for stocks in styles (e.g., large value, small growth, etc.) with the worst past returns and net inflows. Cross-sectional regression results adjusted for size, book-to-market equity, dividend yield, and past stock returns corroborate these findings. This is consistent with overreaction at the style level and with Barberis and Shleifer's (2002) style investing story. We find ancillary evidence that supports the latter. Overall, this paper suggests that the effects of style investing are evident in the data, and that style-level contrarian strategies may be profitable.; The second chapter takes a fresh look at mutual fund return persistence. Equity mutual funds are grouped into styles (e.g., large value, small growth, mid-cap growth, etc.) and are often confined to trading stocks within their style. Hence, one should measure fund performance relative to style when investigating managerial ability. Using CRSP mutual fund data and a methodology similar to Carhart (1997), we find that differences in style-adjusted fund returns persist for up to six years. Neither one-year momentum nor expense ratios explain the results. The results are also robust to controlling for size, book-to-market equity, load, and total net assets. Since manager tenure is about four years, the results suggest that the persistence is due to differences in managerial ability.; The third chapter studies the relationship between fund characteristics and gross cash inflows and outflows in open-end U.S. equity mutual funds over the period 1993–2000. The empirical results suggest that mutual fund investors purchase shares in mutual funds that have performed well in the very recent past, while the extent to which they flee bad performance is not as great. Although high expense ratios are associated with low net inflows, they are associated with high gross inflows, perhaps because they decrease search costs associated with finding mutual funds. Consistent with empirical predictions in Chordia (1996), redemption rates, as well as their variance, are lowest in funds with back-end loads. |
