A Comprehensive Investigation of Fund Performance: a New Technique

Abstract

We explore the ability of U.S. fund managers to forecast their respective style benchmarks by regressing the associated excess style index return of each fund on lagged values of the realized beta time series with lags of the predetermined publicly available information variables included to control for publicly available information. A significantly positive value of the coefficient on the lagged realized beta of the funds indicates an appropriate rebalancing of the fund portfolio. In the stock selection context the ‘realized” beta series is fed into a market model to obtain a residual alpha time series. The time varying “realized” alpha series is then regressed on the demeaned lagged publicly available predetermined information variables and a significantly positive regression intercept indicates superior stock selection ability. We find no timing ability at monthly frequencies but at quarterly frequencies we find a significant proportion of large capitalization funds time their respective style benchmarks. Using monthly data we find a significantly negative abnormal return in the first month of the quarter and a significantly positive abnormal return in the third month of the quarter for a significant portion of small capitalization funds. At quarterly frequencies, a large portion of small capitalization funds exhibit stock picking abilities.