The Right Answer to the Wrong Question: Identifying Superior Active Portfolio Management
By W. V. Harlow (Fidelity Research Institute) and Keith Brown (University of Texas, Austin), Journal of Investment Management
This study develops a selection process which helps increase the odds that an investor selects a mutual fund that will generate superior risk adjusted performance in the future. Superior performance is measured by a fund’s alpha coefficient which is defined as the actual fund return minus a fund’s expected return given the fund’s risk exposure. A positive alpha coefficient indicates that a fund manager has special skills or talents that can lead to superior investment results.
By examining over 5000 equity funds covering the period January 1979 to December 2003, the authors identify specific relationships between observable fund characteristics and a fund’s future alpha coefficient. For each fund in the study, the authors calculate the fund’s expected return for each month t based on the fund’s risk parameters prior to month t. The risk parameters are estimated using the Fama French model (i.e. the Carhart model minus the momentum factor). Then the expected return is subtracted from the actual return generated by the fund during the month to create the fund’s alpha coefficient. Using the time series of alphas, the authors examine the correlation between a fund’s future alphas coefficient and the following attributes of the fund:
-
historical alpha coefficient
-
expense ratio
-
assets under management
-
portfolio turnover
-
level of portfolio diversification
-
level of fund return volatility
The most important result of this study is that by sorting funds on their historical alpha coefficients and their expense ratios investors can substantially increase the probability of choosing the best performing funds in the future. By selecting funds with superior past performance, investors can increase the probability of choosing a fund that will generate a positive future alpha by 9%. Using low expense ratios as the selection criteria increases the probability by 3%. When combined together, a fund chosen from the high historical alpha, low expense cohort has a 60% probability of generating a positive future alpha coefficient. On average, the differential in future alpha between the high historical alpha, low expense cohort and the remaining cohorts is 300 basis points a year. To ensure the validity of the results, the authors control for several fund characteristics and various fund investing styles (i.e. growth, value, small cap, large cap, etc).
Overall, the results indicate that investors can add value to their portfolio by allocating assets to a subset of active fund managers that possess tangible security selection and market timing skills as measured by past alpha.
