Since the dawn of hedge fund indexes, detractors have charged that it’s simply impossible to capture the aggregate performance of an industry using voluntarily-provided data. The voluntary nature of hedge fund indexes has given birth to a litany of alleged “biases” and inaccuracies from survivorship and backfilling to selective reporting and outright BS.
The challenge for investors, academics and regulators has been that there is no objective way to measure this bias. There is no “compulsory” database against which to compare the bevy of voluntary ones in existence today.
But wait, say Vikas Agarwal of Georgia State University, and Vyacheslav Fos & Wei Jiang of Columbia University. You could try to use 13-F filings to measure the unique characteristics of hedge funds before, during and after they report to voluntary databases. Then you just may be able to draw some conclusions about how funds do before and after they report to databases, and how voluntarily-reporting funds might actually differ from non-reporting databases. That’s just what they did in this paper.
Since 13-F’s only contain long positions, the trio’s paper needs to be taken with a grain of salt. But they are quick to point out that the importance of long positions to overall returns and their relative sizes in long/short portfolios means that any change in performance of longs around the time of voluntary database listing or delisting may be quite informative.
Who are you?
One thing industry watchers have always wondered about voluntary reporters is simply: “Who are you? Who, who? Who ,who? I really want to know…”
By comparing the 13-F data from voluntary reporters and non-reporters, Agarwal, Fos and Jiang determine that voluntary reporters have a higher turnover than non-reporters – suggesting that non-reporters would prefer to stick with the 13-Fs only since they are concerned about “trade secrets” getting out (think Phil Goldstein).
Curiously, voluntarily-reporting funds had a higher exposure to Fama French factors such as SMB and HML – suggesting that the more transparent group was more likely to ride market betas than their more introverted brethren. The fact that non-reporting funds have more concentrated portfolios seems to corroborate this.
Other than that, the differences between voluntarily-reporting funds and 13-F only funds weren’t that significant. Both categories had similar volatilities, returns, launch dates etc.
This is very significant since, as the authors point out, it means that hedge fund databases may not be as biased as some think (our emphasis):
“…the difference in the return performance, though slightly in favor of the non-reporting funds, is small. “
Easy come. Easy go.
But how does the decision to self-report to a hedge fund database affect performance? Or, perhaps more accurately, how is performance affected by factors that are coincident with the decision to self-report? Further, what happens to performance after funds cease to voluntarily report?
These questions have perplexed researchers in the same way scientists (and the rest of us) wonder what happens to our soul before we are born and after we die.
Too bad our souls don’t have to file 13-F’s. ‘Cause here’s what we might find. First, it seems that absolute monthly returns (of longs) take a dive after reporting ceases.
You might expect this. After all, bad performance is a leading cause of failure to report. But check out what happens to the long books before reporting commences…
Yikes. It looks like the very decision to go public with returns is a predictor of poorer performance to come – at least when it comes to attribution from longs.
ABC – Always Be Closing…
It’s commonly assumed that the primary motivation behind voluntary hedge fund database reporting is to market the fund and raise capital. So does this actually work?
According to Agarwal, Fos and Jiang, sort of. Apparently, there is a slight bump in AUM around the commencement of reporting. But this bump only lasts a few months and fund raising drops off again. This happens in part, as the authors suggest, because performance drops after reporting commences. (As you would expect AUM also takes a serious dive after reporting stops.)
Post-Script: Regular readers may recall a way-cool Venn diagram by David Hsieh and Bill Fung showing the intersections and union of 5 major databases. Well this paper contains a new version of that diagram (covering a sightly different list of databases). Click to enlarge.