More damning data

By Buttonwood

FURTHER to the latest column on hedge fund returns, Michael Edesess of Fair Advisors draws my attention to a fairly damning paper on the subject by Adam Aiken, Christopher Clifford and Jesse Ellis, three US academics. The paper tackles the problem of self-selection bias that can mark hedge fund indices (only the best funds choose to report) by looking at funds that have registered with the SEC (which requires them to report performance data). They can then compare the returns of those funds that report their numbers to the index providers.

Broadly speaking, there are two effects; first among funds that remain in commercial databases and second, in those that drop out. The authors look at the so-called alpha of funds (jargon for skill) that voluntarily report their numbers. This is a complex calculation since one has to know the portfolio allocation of managers; much of their return may be down to exposure to the S&P 500 or to corporate bonds and so would be classed as beta (simple market exposure that can be obtained at much lower costs elsewhere). Whereas, the managers that report to an index seem to show alpha of 3.5 percentage points, that is because they are the best performers. According to the authors

95% of a typical fund manager's measured skill can be explained by whether they report to a database

When things go wrong for a hedge fund, they tend to stop reporting their numbers to the commercial providers but they still have to report to the SEC. Almost half of all such funds still operate for two years after they stop reporting to the indices. Such funds produce returns that are more than 7 percentage points a year below those that continue to report to commercial providers. The authors conclude that

Our results indicate that when we exclude self-selected database funds, the average excess returns of hedge funds does not differ markedly from zero.