|
A recurring question in the topic of active versus passive management is the degree to which active mutual fund managers who outperform their benchmark can be considered to have done so through skill versus luck. An academic study, described in an article by Mark Hulbert in the New York Times several weeks ago, answers this question through a new statistical technique.
The study’s authors are Russ Wermers, a professor of finance at the University of Maryland, Laurent Barras of the Swiss Finance Institute, and Oliver Scaillet of the University of Geneva. We spoke with Professor Wermers on July 25, 2008.
The False Discovery Rate
If all active fund managers were to choose stocks by throwing darts, inevitably some small percentage would deliver alpha, even over some large period of time. However, they would do so through random luck. Fund managers are not dart throwers, yet some percentage of them will nonetheless deliver alpha through luck. Wermers’ tool, known as the False Discovery Rate (FDR), identifies the size of the group delivering alpha through skill, and well as the size of the group failing to deliver alpha through lack of skill.
The FDR technique begins by segregating fund returns into three groups: negative alpha, zero alpha, and positive alpha. The zero alpha group consists of those funds that earn returns just sufficient to match their benchmark, net of expenses. They deliver zero alpha to their investors. Based on the number of funds that exhibit an alpha close to zero, which are almost all funds without skills, the FDR technique estimates the number of funds without skills that end up with positive (or negative) alphas simply by luck (good of bad). Then, it is simply a matter of subtracting the actual size of the positive alpha group from the expected size (based on luck alone) to determine the size of the group of funds that delivered alpha through skill. A similar procedure is used to determine the size of the group that deliver negative alpha (net of expenses) through lack of skills.
The study used the Center for Research in Securities Prices (CRSP) data, and matched it with Thomson’s CDA data for fund investment-objective information. The data is free of survivorship bias and only funds with at least 60 months of returns were included. Share classes were consolidated (dollar weighted) into a single fund. Sales loads were not modeled (if they were, it is likely an even smaller percentage of funds would have delivered alpha).
Display article as PDF for printing.
Would you like to send this article to a friend?
Remember, if you have a question or comment, send it to
. |
