As hedge funds come under tighter scrutiny, Vanderbilt Owen Graduate School of Management professor Nicolas Bollen has identified several that could pose a fraud risk similar to the kind undertaken by Bernie Madoff, who bilked investors out of $65 billion.
In a study examining the effectiveness of five performance flags used to detect hedge fund fraud, Bollen compared 195 “problem” funds against a pool of 8,575 “non-problem” hedge funds. He found that some of the funds with no violations reported against them actually displayed characteristics signaling risk of fraud.
“Our performance flags may be indicating that some of the non-problem funds are in fact at higher risk of fraud, but have not yet been charged with any violations,” says Bollen.
The Securities and Exchange Commission has adopted screening tools similar to those tested by Bollen in the wake of the Madoff scandal, though their use could grow under the new financial reforms package. The performance flags are also similar to analytical tools used by whistleblower Mark Markopolos to alert authorities to irregularities with Madoff’s investments as early as 2000.
“The approach we’re validating for hedge fund monitoring is, in some ways, similar to the one used by the IRS to determine which tax returns to audit,” says Bollen. “By statistically parsing through funds and identifying ‘red flags,’ we demonstrate financial regulation can work without being prohibitively expensive.”
Here are five indicators for monitoring hedge fund fraud:
(1) A kink in the fund’s distribution of returns at zero: A statistical test will show a lack of smoothness—a kink—in a hedge fund’s distribution of reported returns at zero. This may indicate an effort to avoid reporting a loss by inflating returns in one month, then later reversing the overstatements. When calculating returns on a bimonthly basis, the study found that this kink disappears.
(2) A low correlation with other assets: Hedge fund returns should be correlated with a set of investment “style factors,” which researchers developed to mimic well-known hedge fund trading strategies. A low degree of correlation could be the result of a hedge fund actually making good on its promise to deliver unique returns. However, Bollen argues that if no reliable correlation exists, it is likely that the hedge fund returns are distorted, perhaps in an effort to mask risk or, as in the case of Madoff, because they were being fabricated.
(3) Artificially smooth returns: Returns that show a low level of volatility and a positive serial correlation could be the result of hedge fund managers purposely smoothing returns by reporting moving averages. Research has shown that moving averages feature lower volatility than raw observations, and will possess serial correlation even when raw observations have none.
(4) Losses reported differently than gains: When the serial correlation is conditional on a particular variable, it may point to a hedge fund manager’s desire to smooth losses by delaying reports of poor performance, while fully reporting gains when they occur in hopes of winning investor capital.
(5) Poor data quality: Portfolios can be analyzed for “man-made” data patterns that include characteristics such as too many returns exactly equal to zero, too few unique returns, too long a string of identical returns, and an extremely low percentage of negative returns. Regarding the last point, the authors say that, naturally, very few losses would be reported if returns were being fabricated, such as in a Ponzi scheme.
Bollen also notes that this so-called performance flag approach could be applied to deter fraud in a wide range of investments.
“The information we are providing also can benefit investment advisers by making them aware that pre-screening can be a very effective way to protect client portfolios.”