Russ Wermers, one of the authors of the study cited in the article, responds:

Ron,

I agree with what you are saying--there are effective ways to use portfolio holdings to decompose and attribute short-term performance to understand much more about the skill vs. luck question--including your PODs.

Our paper did not address a complete breakdown using portfolio holdings, nor were we looking for very short-term skills, which I am sure exist.

We were concerned with whether we could simply use returns in conjunction with a well-understood and simple 4-factor model to determine whether long-term skills exist, where "long-term" means at least five years.  We found very little of
this during recent years.  This is not to say that very short-term skills do not exist--I believe that they do.  But, for an unsophisticated investor, I would still recommend that they hold either index funds or low-expense active funds since these investors really do not have the tools to do better.

Contrary to your statements, our statistical approach does separate skillful results from lucky results (or unskillful results from unlucky results).   The "False Discovery Rate" does a very good job of estimating the presence of luck vs. skill--as long as there are a lot of funds with a long history of returns to examine.

You are right that it is dangerous to examine a single successful fund manager and conclude that he or she is skilled (or to conclude that he/she is lucky).  The power of such a test is low, as well as the bias in looking at a high-flyer (I have another paper in the Journal of Finance during 2006 that addresses this).  But, if we examine the entire group of funds in the right tail of alphas, then we can control quite effectively for the presence of luck. This is the statistical “magic” of the False Discovery Rate (FDR) procedure.

One caveat, however:

As noted in the article, since we do this in a group setting, we cannot (with the FDR) identify which specific funds with high alphas are skillful vs. lucky.  We can only say that a certain percentage of the high alpha funds were skillful (and not simply lucky).  To restate, the procedure works very well in estimating the prevalence of skill in a group, but not in identifying exactly which managers are the skilled ones.  You may be thinking of this problem, which our approach admittedly does not address. As you state, other methods may be helpful for analyzing individual funds or small groups of funds.

Russ Wermers
Associate Professor of Finance
Robert H. Smith School of Business
University of Maryland

Display article as PDF for printing.

Would you like to send this article to a friend?