The original premise of a “hedged fund,” as a financial journalist originally described the concept in 1949, was simple: A portfolio balanced between long and short positions could profit in nearly any market.
That idea may have taken a while to seep into the mainstream, but as it has over the past decade, the hedge fund industry has exploded, rocketing from $310 billion in assets under management in 2002 to more than $2 trillion today. Institutional investors and high net-worth individuals flocked to these largely unregulated, nonpublic funds in no small part because they offered access to assets and trading strategies that are all but impossible to replicate.
But new research from Nicolas Bollen, the E. Bronson Ingram Professor of Finance, says those hedge funds that are hardest to imitate—something investors look for and for which they often pay a premium—are the ones most prone to failure.
In addition, Bollen finds that these types of funds contain a significant amount of volatility, indicating that they are vulnerable to the type of risks they are supposed to guard against. “This result suggests the presence of an omitted but potentially catastrophic risk factor in funds for which standard regression analysis fails,” Bollen writes in the study, forthcoming in the Journal of Financial and Quantitative Analysis.
Those previously undetected risks raise the annual probability of failure for hard-to-replicate funds from 10 percent to 12 percent. The findings have implications for investors who rely on statistical models to screen funds for heightened risk factors as part of their due diligence process.
Determining Hedge-fund Performance
The difficulty in assessing hedge fund performance lies in the industry’s opacity. Fund managers report returns publicly at their discretion, leaving wide gaps of data about holdings, accuracy, and even whether a fund is still operating. (In October 2011, hedge funds with more than $1.5 billion in assets under management were required to start disclosing fund details to U.S. regulators, but that information will not be made public.)
As hedge funds have grown, academic researchers have developed statistical models designed to correlate hedge fund returns with known investment strategies. Using these models, along with data from a broad cross section of funds from 1994-2008, Bollen found that more than one-third of all funds cannot be correlated to known style factors. The phenomenon becomes even more pronounced in funds with short histories.
Bollen suggests those results indicate that using hedge fund regression models to learn about a fund manager’s trading style and selection of assets may be even weaker than previously thought. Further, he says it may be a Sisyphean task to try to develop a complete set of risk factors, especially those representing catastrophic losses during rare events.
Where does that leave investors? For the time being, relying more heavily on qualitative judgments about things like a fund manager’s background and strategy mix than the quantitative analyses of econometricians.
A version of this article originally appeared in VB Intelligence.