First, the Securities and Exchange Commission began raising questions about the way hedge funds price their portfolios.
By Hal Lux
November 2002
Institutional Investor Magazine
Now a forthcoming academic study concludes that the month-to-month performance of some hedge funds may, in fact, be too good to be true.
Written by Andrew Lo, a highly regarded quantitative investing expert and Massachusetts Institute of Technology finance professor who happens to run a hedge fund himself, the investigation has just been submitted for publication. In the study, “An Econometric Model of Performance Smoothing and Illiquidity in Hedge Fund Returns,” Lo wanted to examine the curious phenomenon of high “serial correlation” in the month-to-month returns of hedge funds. Serial correlation refers to the tendency in investment returns for one month of up performance to be followed by another up month. The ability to achieve such smooth returns -- to generate consistently up performance regardless of market conditions -- is exactly what many hedge funds are selling. But the degree to which some managers are able to pull off this difficult feat appeared fishy to Lo, so he and fellow researchers Mila Getmansky and Igor Makarov, both MIT graduate students, set out to examine the phenomenon.
“Most hedge fund managers are good, honorable people,” says Lo. “But there are probably some engaged in unsavory practices.”
Lo and his team looked at the performance reported by 900 hedge funds in the Tass hedge fund database, going back to 1977. The researchers found that statistical models that incorporated such factors as hedge fund compensation systems or different leverage strategies failed to account for serial correlation. What did work, they found, was a model that presumed that hedge funds were able to manage their reported performance by “storing” returns in good months and doling them out to make up for shortfalls in subsequent, weaker months.
In this model Lo and his co-authors assumed that the monthly returns in the database were a weighted average of the current and past two consecutive months’ true economic returns. The model then calculates a coefficient, from 0 to 1, which measures the level of smoothing, or the percentage of the true return of each fund that was reported each month and the percentage that was stored for use in future months. A coefficient of 1 means there’s no smoothing. A coefficient of 0.5 indicates that 50 percent of a month’s actual returns are being stored for the future.
Lo and his team discovered that hedge funds trading liquid securities, such as futures contracts, where there can be no debate about their price, come pretty close to 1. But hedge funds active in illiquid strategies, such as relative-value or fixed-income arbitrage, produce an average coefficient of about 0.7. Event-driven and relative-value hedge funds, as well as funds-of-hedge-funds, made up the bulk of the funds that produced the highest level of smoothing. One fund -- they are all unnamed -- produced a coefficient of 0.464, meaning that it was not, according to the model, reporting the majority of its returns each month.
Why do many funds look as if they’re managing returns? Lo offers two possible answers. The first is that brokers, who themselves don’t know the prices of extremely illiquid instruments, are using simplistic formulas to price hedge fund portfolios each month; that would mean that hedge funds are essentially priced arbitrarily. The other possibility, which is even less heartwarming, is that hedge fund managers are intentionally manipulating their returns to improve the consistency of their performance, which is illegal.
Lo emphasizes that no statistical study can prove that people are tinkering with numbers on purpose, but he adds that it is nonetheless unlikely that all of the smoothing is accidental. “The effect is so extreme,” he says. “There are some funds with returns that are almost a straight line.”
Or as the study puts it: “While our econometric model is capable of capturing a broad range of smoothing behavior, no statistical procedure can identify malice or intent on the part of hedge fund managers. Performance smoothing can be deliberate or inadvertent, and our framework is not designed to distinguish between the two.”
Interested readers will likely include regulators at the SEC. Last month Paul Roye, head of the commission’s division of investment management, told an American Bar Association conference that the agency’s ongoing investigation of the hedge fund industry had led to some concerns about the valuation of hedge fund portfolios.
