Few topics have been studied as closely as selecting actively managed funds that will outperform the market. Advisors who use such funds need to be confident in their choices – and justify their methodology to clients. Here’s what the latest academic research says on this highly contentious issue.
My February article addressed the effectiveness of using actively managed funds. That investigation showed some evidence of selection skills, but not enough, on average, to overcome the low-cost advantage of indexing. In this article, I explore whether prospects of beating the averages can be improved with particular fund-selection techniques.
Practitioners attempting to identify superior fund-selection techniques face two challenges: returns need to be calculated on a risk-adjusted basis, and the results need to be statistically significant. For example, an advisor might argue that he demonstrates superior skill because the four active funds he recommends to clients all beat the S&P 500 in the past 10 years have. Superior fund-selection is a possibility, but his success could also be explained by luck or by those funds having higher risk than the S&P 500.
The most popular methods used to adjust returns for risk are factor models like Fama-French. These models split returns between systematic beta components (such as large-capitalization or small-capitalization index performance) and alpha, which represents the positive or negative value added by investment management. Funds can be compared on a risk-adjusted basis by focusing on the alphas they produce.
Factor models use regression, and the significance of the alphas can be evaluated using standard statistical tests. Producing statistically significant results requires a lot of data – for example, the performance of thousands of mutual funds over multiple decades.
The Jones and Wermers study
In 2011, Robert C. Jones and Russ Wermers surveyed the literature on the value of active management and reported their findings in the paper Active Management in Mostly Efficient Markets. Let’s look at how this overview study assessed alpha generation in four broad categories:
- Past performance : Some studies show that past performance can be a predictor of future returns over relatively short periods, but the overall evidence is mixed. The studies the authors covered utilize sophisticated statistical techniques and use multi-factor risk-adjustment models to separate alpha and beta components of returns. Below I'll discuss in more detail one of the studies cited by the authors.