Exploring Uncharted Territories of the Hedge Fund Industry

Posted by R. Christopher Small, Co-editor, HLS Forum on Corporate Governance and Financial Regulation, on Friday July 5, 2013 at 9:22 am
  • Print
  • email
  • Twitter
Editor’s Note: The following post comes to us from Daniel Edelman of Alternative Investment Solutions; William Fung, Visiting Research Professor at the London Business School and Chairman of Maple Financial Group; and David Hsieh, Professor of Finance at Duke University.

It is virtually impossible to obtain accurate historical data on the entire universe of hedge funds. In our paper, Exploring Uncharted Territories of the Hedge Fund Industry: Empirical Characteristics of Mega Hedge Fund Firms, forthcoming in the Journal of Financial Economics, we identify previously unexplored data sources whereby collecting data on fewer than four hundred large hedge fund management firms that do not participate in major commercial databases adds to the observable industry in assets under management (AUM) terms by as much as 34% in 2001 rising to 65% by the end of 2010. Towards the end of our sample period, these nonreporting firms collectively manage US $862 billion of AUM that is missing from the reported US $1,322 billion of AUM managed by firms in the three major commercial databases combined. We manually collect the names and AUMs of large hedge fund firms that do not participate in commercial databases from surveys published by Institutional Investor and Absolute Return+Alpha magazines, which are good sources of information with almost a decade of continuous history. These previously untapped sources of data provide valuable insight into the capital formation process of the industry over the past decade. While commercial databases have successfully depicted data on the growing trend of hedge fund industry’s AUM, from US $278 billion in 2001 to US $1,322 billion in 2010, there is a more important trend in the capital formation process of the industry that has not been considered in the research literature. We show that over this past decade, the AUM of nonreporting mega hedge fund firms has grown from US $118 billion (2001) to US $863 billion (2010). Results point to a rapid growth of mega hedge fund companies opting for privacy dropping out of the voluntary system of reporting to commercial databases. The empirical evidence confirms that a small group of mega hedge fund firms manages the bulk of the assets in the industry. Taken together, this implies that the assets of the hedge fund industry are concentrated in the hands of a small number of mega management firms with rising opacity as their AUM increases.

Collecting performance data of nonreporting firms and integrating them with commercially available data is a challenging task. This is especially so when performance data from reporting firms also have missing observations. We are able to collect 13,561 additional monthly returns of funds from reporting firms in our commercial dataset. This new source of performance data provide valuable insight on a long-standing empirical issue on measuring hedge fund performance when there are missing data on reporting firms in commercial databases. We refer to this as the delisting bias. By adding back the missing returns of reporting funds with greater than US $50 million AUM, we show that delisting bias is numerically small and statistically insignificant. Specifically, while the Ackerman, McEnally, and Ravenscraft (1999) conjecture that missing returns from successful firms that voluntarily stopped reporting data can exhibit marginally higher mean returns, we find that the difference is just 0.124% per annum on average and statistically not different from zero. A similar observation applies to the conjectures of Posthuma and Van der Sluis (2003) and Malkiel and Saha (2005), which imply that delisted firms heading for liquidation are likely to have inferior returns on average. While we find results consistent with this implication, at -0.022% per annum on average, it is not statistically different from zero. More important, when these two opposite reasons for firms to stop reporting are combined, the effect of missing returns is neutralized for all practical purposes.

In addition to collecting missing returns of the funds in reporting firms, we are able to collect hard-to-observe performance data on the majority of the funds from nonreporting firms. With this additional source of data we are able to construct indices of nonreporting firms using both an equally-weighted and an AUM-weighted methodology. Conventional tests fail to reject equality of the first two moments of the unconditional return distributions between nonreporting and reporting firms after controlling for size. The Kendall and Stuart (1967) tests also failed to detect significant differences in the respective histograms. To contrast the time-series properties of nonreporting and reporting firms’ returns we investigate alternative models of return serial correlation in Getmansky, Lo, and Makarov (2004). We compare return serial correlations of different strategy categories indices from hedge fund research (HFR) and Down Jones Credit Suisse (DJCS) with comparable mutual funds. We find that, for portfolios of hedge funds, return serial correlation is primarily factor-driven rather than manager-driven. Thus, we are able to estimate the risk factor exposures of the return spread series between nonreporting and reporting firms (and between nonreporting firms and indexes of hedge funds) using an extended version of the Fung, Hsieh, Naik, and Ramadorai (2008) model, that are not biased by the existence of return serial correlation. We find only a few instances in which indices of nonreporting firms’ returns cannot be inferred from reporting firms after controlling for size as well as using commonly used hedge fund indices. Overall, a high degree of similarity exists between the return characteristics of nonreporting mega firms and their observable reporting counterparts, with the largest divergence between them occurring during the 2008 financial crisis, feeding through differences in their respective exposure to the credit related factor.

The full paper is available for download here.


Add your comment below:


(required but not published)

RSS feed for comments on this post. TrackBack URI

  •  » A "Web Winner" by The Philadelphia Inquirer
  •  » A "Top Blog" by LexisNexis
  •  » A "10 out of 10" by the American Association of Law Librarians Blog
  •  » A source for "insight into the latest developments" by Directorship Magazine