Wednesday, June 5, 2013

Learning to take the good with the bad

In the course of some research I came across a working paper by A. Basso and S. Funari (dated September 2001) titled "A generalized performance attribution technique for mutual funds."

Please don't let the title fool you: this isn't about performance attribution. Rather, it's about risk-adjusted performance. But putting that aside ...

In the article they generally describe measures such as the Sharpe and Treynor ratios as "numerical indexes ... that take into account an expected return indicator and a risk measure and synthesize them in a unique numerical value." I happen to think that this description is excellent. The notion that the returns and risk measures are synthesized into a unique numerical value. Isn't that an excellent description? Okay, so that's the good.

The "bad," in my view, is how the authors describe some of the results they obtained during their analysis. They used data from Italian mutual funds. During the period observed, "two bond funds ... exhibit a negative mean excess return; this entails that the values of the Sharpe, reward to half-variance and Treynor indexes for these funds are negative and, above all, meaningless. In fact, when the excess return is negative these indexes can be misleading, since in this case the index with the higher value is sometimes related to the worse return-to-risk ratio."

We've taken this matter up before; the realization that negative excess returns yield confusing Sharpe and Information ratios (as well as, apparently, other risk-adjusted measures). To refer to these results as "meaningless" and "misleading" is unfortunate, I believe. I'll confess my own struggles with this, but believe that my earlier explanation (see for example the January 2012 edition of The Spaulding Group's monthly newsletter) provides some insights into the value and interpretation of the results.

When things don't make sense, sometimes additional time is needed to reflect upon them. Again, I'll confess my own typical impatience with such things. But time spent on these events can prove beneficial.

2 comments:

  1. FROM STEVE CAMPISI:
    "When things don't make sense?" I think that's why they invented "interpretation" and "explanation." I disagree heartily and vigorously (a la Johnny Depp's "Hatter" in "Alice in Wonderland) with the idea that it's a good thing for risk and return to be synthesized into a single number. It's NOT a good thing, especially when it leads to people spouting numbers without much thought. No, I think that this is one of they biggest problems with what masquerades as performance "analysis" these days - analysts calculate something according to a stock formula and then make declarations on little more than whether one number is bigger than the other. At best, these ratios and such are merely starting points. There is no substitute for a well-trained performance analyst who understands the investment process and can explain the various numbers in a credible and representative context. We have precious little of that these days.

    Here's a painfully simple example: how many people look at the various points along the efficient frontier and puzzle over why the Sharpe ratios get "worse" as the level of return increases? Or who puzzle over whether the absolute number of an Information Ratio has any explanatory value? (Hint: it does!) We need less academics and calculations and more understanding of basic investment principles. Time much better spent.

    ReplyDelete
  2. Steve, thanks for your input.

    Interesting that you cite the "efficient frontier." Who came up with that concept? Wasn't it an academic?

    While I was merely praising the wording, I must also confess some degree of admiration for risk-adjusted return measures (my favorite being, of course, M-squared).

    If one looks at risks and returns separately, it's difficult to judge a manager vs. his/her benchmark. But, by "synthesizing" them, we have a metric that leads itself to easy comparison.

    From your note it appears you don't like Sharpe ratios, Information Ratios, etc. That is fine, and as a practitioner, you would know better than I as to their true value. But, the industry (not just academia) has embraced these measures. And again, my praise was for the elegant way of describing what occurs.

    ReplyDelete

Note: Only a member of this blog may post a comment.