How Smart are the Smart Guys? A Unique View from Hedge Fund Stock Holdings.
By: John Griffin, University of Texas & Jin Xu, Zebra Capital Management
Published: August 21, 2006
Thanks go out to Mebane Faber at WorldBeta for giving us the heads up about this article. Mebane has an interesting blog that amounts to a live test of alpha-centric (or more accurately, beta-centric) investing.
Tons of research has been conducted on the return histories of hedge funds to forecast their future returns. From Bill Sharpe to David Hsieh, factor analysis has been used to determine the amount of alpha in a mutual fund or hedge fund. But according to Griffin and Xu, no one has actually lifted the hood on hedge funds to study their actual holdings in an effort to assess their investing skill. The two set out to answer the question: Do hedge fund long holdings out perform mutual fund (long) holdings?
Finding an answer to this question was no small feat. Hedge funds don’t have to report holdings semi-annually like mutual funds. So the authors used quarterly 13F filings that containing long equity positions over US$200,000 or over 10,000 shares.
No worries, say the authors, you can still get a pretty clear picture of the stock-picking skill by examining the long books of equity long/short funds. But later in the report they do point out that the correlation between the returns on these long books and the reported returns of the corresponding hedge funds averages only 0.55.
Still, the researchers make a lot of interesting observations. Many of them confirm generally-held beliefs about the industry (e.g. that hedge fund turnover is higher than mutual fund turnover, that hedge fund weightings differ from the market portfolio, that hedge funds invest in stocks with low analyst coverage, etc.). But several observations are sure to create controversy.
Specifically, the report concludes that despite their reputation as momentum players, hedge funds have higher weights in previously declining stocks. So perhaps hedge funds do serve a public good after all, (as suggested by some regulators) by providing liquidity to those who want to bail out of those losers.
The report also concludes that stocks with a high proportion of hedge fund ownership tend to perform slightly better than stocks without a large proportion of hedge funds on their shareholder lists:
“Interestingly, hedge fund breadth contains additional forecasting power for future returns, suggesting that there is information embedded in whether a hedge fund holds a security.”
But before you run out and invest alongside the nearest hedge fund, the authors point out that this relationship isn’t strong enough to actually make any money. In fact, the authors find that large bets by individual hedge funds are actually related to negative returns.
Many hedge funds – particularly short-biased funds – might use long positions simply as a hedge. So what if the long positions in a hedge fund are meant to out perform, but rather to act as a hedge against an actively managed short book? The authors acknowledge that this might affect their results:
“…it is possible that our overall finding of no performance differences between hedge and mutual funds is driven by extremely negative returns in those funds that primarily short securities and use long positions as a hedge.”
Okay, you might think, so how could reported hedge fund returns out perform mutual fund returns if hedge fund’s aren’t any better at picking long positions? The authors provide no answer to this important question. But we hypothesize that one answer might lie in this statement on page 26:
“…there is no reliable source on hedge fund leverage ratios available that we are aware of…”
So maybe hedge funds short more in down drafts. Looking only at long positions would not reveal increases or decreases in the short books (which could a. generate alpha outright, or b. amount to market-timing via net exposure management). Leverage statistics might have provided some hint about the size of the short books since shorting would have had the effect of created more leverage (assuming no changes to the long books).
Another reason why hedge funds out perform what their long positions would suggest is that there might have been window-dressing in the 13F filings (by either – or both – of the hedge funds and mutual funds).
The lack of a conclusive answer to this disparity leaves major questions. Still, this study gets full points for taking a unique approach to understanding hedge fund returns. There is also a fascinating section in this paper that describes the types of companies in which hedge funds seem to go long (page 58).