The comparison of passively managed portfolio returns with those of a mutual fund with similar risk characteristics may incorrectly assess the mutual fund manager’s skills in stock selection and market timing. The authors suggest that, instead, managers’ performance should be evaluated against a self-designated benchmark. They find that, on average, mutual funds underperform their self-designated benchmarks and managers’ market-timing activities negatively affect fund performance.
Mutual fund managers are evaluated against the benchmark mentioned in the fund prospectus—a benchmark that they apparently set for themselves. The authors argue that traditional studies do not correctly estimate manager performance. To estimate the manager’s skills, traditional studies have generally examined a fund’s excess performance estimated by (1) the fund’s excess return, or the fund’s alpha, relative to the return on a passive portfolio with the same risk profile or (2) the excess return computed by adjusting for risk, size, book-to-market ratio, and momentum of the stocks in the fund portfolio. The alpha is considered to be the manager’s contribution to the fund’s return and is indicative of his or her skills.
The authors suggest that using implied benchmarks, such as the S&P 500 Index for large stocks, S&P 500 Value Index for funds with a value emphasis, or S&P 500 Growth Index for funds with growth emphasis, to assess manager performance is not correct because the manager’s investment decisions are not influenced by these benchmarks but by the benchmark stated in the fund prospectus. These implied benchmarks may have their own alphas and exposures (betas) to systematic risk factors. Using these benchmarks without accounting for their own alphas and betas will lead to an incorrect assessment of managers’ stock selection and market-timing skills.
The authors suggest incorporating a fund’s self-designated benchmark for examining managers’ stock selection skills. Using this methodology, they find that, in aggregate, the funds’ alphas are negative. But these alphas are less negative than those estimated by using the traditional methods. They find the difference in the alphas varies with investment styles. It is more significant for small-cap growth funds: –3.66% a year with the traditional approach versus –1.48% a year with authors’ approach.
The authors also find significant timing decisions by fund managers. They contend that by changing the proportion of the fund investment in their target portfolios, managers take bets against their stated benchmarks. But timing is found to have only a small effect on mutual fund performance. In aggregate, mutual funds underperform their stated benchmarks by 1.90% a year. Stock selection is responsible for much of the negative performance, and timing’s contribution is –0.65% a year.
How Did the Authors Conduct This Research?
Actively managed equity mutual fund daily return data are from the CRSP Mutual Fund database from September 1998 to June 2012. The authors also gather investment objectives of the funds. Risk factor and style portfolio returns are obtained from Kenneth French’s website. Daily benchmark returns are from Datastream. The authors assign Lipper’s objective code based on the objective of the fund as stated in its prospectus. The funds are then categorized into nine categories, from large-cap core funds to small-cap-value funds. The sample has 2,729 large-cap funds, 1,320 mid-cap funds, and 1,689 small-cap funds.
As for investment styles, there are 2,228 growth, 2,272 core, and 1,238 value funds. Using traditional methods, the authors document that mutual funds’ underperformance is consistent and statistically significant across all size groups. The alphas calculated by using the traditional method are also negative across all size groups and value/growth investment styles. Although the authors find that alphas calculated by using the self-designated benchmarks are negative in all cases, they are less negative than alphas found by using the traditional method. This result indicates that managers destroy value with their stock selection activities, but this loss is not as large as the traditional methods would suggest.
The authors examine the timing skills of the managers by computing the average deviation of funds’ exposure to the market portfolio, the capitalization factor, the value/growth factor, and the momentum factor relative to the benchmark exposures. The average deviations from the benchmark market, size, and value/growth betas are negative and statistically significant, which indicates that, in aggregate, managers take less risk than their respective benchmarks. Nonetheless, the result suggests that managers adjust their portfolios’ exposure based on their expectations about future factor performance. There is no significant difference in the average deviation from benchmark momentum exposure.
The authors decompose their data into two subperiods—pre- and post-1 January 2008—and find similar results. The alphas for both subperiods found by using the traditional method are more negative than those found by using the self-designated benchmarks. The authors’ conclusions hold when they extend their analysis to European funds.
The authors bring the academic studies of mutual fund managers’ performance closer to how performance is evaluated in practice. Rather than comparing managers’ performance against three widely used benchmarks, the authors classify mutual funds into nine categories based on their stated objective and use the managers’ stated benchmarks. Although the authors label these benchmarks as “self-designated,” I believe that the phrase “self-implied” benchmarks may be more appropriate.