If the board of the $650 million Art Institute of Chicago endowment had taken a closer look at Integral Investment Management LP's performance data, they might not have bothered investing with the Dallas-based hedge fund.
In retrospect, that would have been wise, since that investment may have cost the endowment as much as $43 million, as well as putting much-publicized egg on board members' faces.
What they missed was that the firm's Sharpe ratio - which measures the efficiency a portfolio on a risk-adjusted basis - was sky high. To many investors, that's a huge red flag.
That's because it's easy to manipulate the measure, using a combination of out-of-the-money short puts and calls, explained William Goetzmann, Edwin J. Beinecke Professor of Finance and Management Studies and director of the Yale School of Management's International Center for Finance, New Haven, Conn.
By issuing out-of-the-money puts and calls, a hedge fund can generate a steady stream of revenue, which then could be reinvested in a money market return. That, Mr. Goetzmann said, would produce a nice return without demonstrating a whit of skill - that is, until a sudden financial crisis could shift the calls and puts into the money, forcing the manager to absorb huge losses.
"Had the Chicago Art Institute known ex ante the basis for (the) fund's high historical Sharpe ratio, they might not have lost nearly $43 million," wrote Mr. Goetzmann, Jonathan Ingersoll, Matthew Spiegel and Ivo Welch, all finance professors at the Yale School of Management, in an unpublished draft paper.
The problem is that the traditional tools pension executives use to evaluate money managers don't always work that well when it comes to hedge funds.
The messages to pension executives are: "buyer beware, and basically we need better tools to evaluate hedge funds," said Richard Herrman, director of operations at Fuller & Thaler Asset Management Inc., San Mateo, Calif., which runs a large-cap market-neutral strategy and a micro-cap long-short strategy.
Problems encountered
"You don't throw out the rulebook, but you have to use it differently," said Leola B. Ross, a research analyst at Frank Russell Co., Tacoma, Wash.
Here are a few of the problems pension officials and their consultants encounter when trying to evaluate hedge fund managers:
* Picking hedge-fund managers based on style is a perilous exercise because managers vary so much even within the same category. Manager selection is a far more bottom-up exercise than a typical top-down-oriented asset allocation process, Ms. Ross and George Oberhofer, a senior practice consultant at Frank Russell, wrote in "What the `Indexes' Don't Tell You About Hedge Funds."
* Hedge funds may rely far more on market exposure than skill than is generally believed. Added value and diversification benefits "vanish" for the broad universe and many styles of hedge funds when data is adjusted to eliminate the effects of "stale pricing," according to a paper by three hedge fund managers at AQR Capital Management LLC, New York.
* Sharpe ratios, widely used for measuring the efficiency of portfolios, also are inflated by the presence of illiquid securities. Andrew W. Lo, Harris & Harris Group Professor at the Massachusetts Institute of Technology's Sloan School of Management, Cambridge, Mass., and chief scientific officer of AlphaSimplex LLC, Cambridge, finds the annual Sharpe ratio for a hedge fund can be overstated by as much as 65%.
U.S. institutional investors account for one-sixth of the $563 billion invested in hedge funds, and are growing at a 30% annual clip, said Charles J. Gradante, managing principal of the Hennessee Group LLC, New York. With that amount of money pouring into the funds, institutions need to scrutinize hedge funds closely.
One of the biggest problems is that pension executives are accustomed to looking at peer universes to see how well a manager has performed. But this practice can be misleading when it comes to hedge funds because hedge-fund universes "are biased in favor of small, new and open funds," the Russell consultants wrote.
Except for the CSFB/Tremont Hedge Fund universe, leading hedge-fund universes equally weight results, meaning a tiny hedge fund has as much weight as a giant one. Also, many larger hedge funds don't report their data because they no longer are gathering assets.
Trickier to understand is that the traditional way of reporting returns net-of-fees can distort returns. While net returns are the standard for traditional managers, they can produce distorted results for hedge funds, which include a 20% or higher standard performance fee.
If pension funds can't rely on universe comparisons, what should they do? Some experts advocate using fund-of-funds returns for performance data. But the Russell paper argues that funds of funds have their own style biases, and data is not available on just what those biases are.
Stale pricing effects
In a study in the fall 2001 Journal of Portfolio Management, Clifford Asness, managing principal at AQR Capital Management, New York, together with AQR principals Robert Krail and John Liew, found several hedge fund styles, including convertible arbitrage and event-driven and emerging markets portfolios, have grossly understated betas, a measure of market risk.
What's more, the alpha generated by these styles is not only overstated but can be negative, the study found.
The culprit is the use of unvarnished monthly performance data from hedge fund managers. Many hedge funds hold illiquid securities or hard-to-price over-the-counter securities, the authors wrote. The net effect: There often is a lag in valuing the portfolios, which tends to reduce their volatility and correlation with traditional indexes.
And not all the delayed pricing is due to illiquidity. The authors strongly suggest some managers manipulate eturns by delaying pricing.
Instead of using simple monthly regressions, the authors look at the effects of delayed pricing over one-, two- and three-month periods. Adding those betas together with contemporaneous beta, they discover that the "summed beta" actually is 0.84 - more than double the 0.37 beta from a simple monthly regression - from 1994 through 2000.
What those findings imply is that hedge fund managers carry far more market risk than is generally believed. The Sharpe ratio of the overall index plunges to a disappointing -0.4 when the summed beta is used, from 0.8, the paper says.
In fact, the only hedge fund styles adding value were equity market-neutral, managed futures and dedicated short-biased. Even when the effect of the 1998 Russian debt crisis is removed from the data, the summed beta is a healthy 0.69.
In short, the paper recommended using hedged Sharpe ratios and lagged beta techniques when evaluating hedge fund managers.
When Frank Russell's Mr. Oberhofer and Ms. Ross applied the technique to individual managers, they were surprised to find far fewer managers than expected with high betas. That's because hedge fund managers vary so much within their categories.
"Ultimately, managers in the hedge-fund world are quite heterogeneous, and style-level analysis is unlikely to provide useful information on what can be expected from an individual manager within that style classification," they wrote.
"Hedge funds are not the ugly ogres recent reports have made them out to be," but investors must scrutinize them closely, they concluded.
Hedge funds also have their ups and downs. Some researchers have noted that hedge funds are more highly correlated with stock-market returns than generally believed, especially in bear markets.
But rather than a being function of bear markets, Mr. Oberhofer and Ms. Ross find that skewed investment performance stems from heightened volatility: By replacing data from the three months surrounding the 1998 Russian debt crisis, the distribution of returns becomes much more symmetrical, they wrote.
Problems with Sharpe ratios
MIT's Mr. Lo also finds fault with superficially applying Sharpe ratios to hedge funds. That's because monthly returns are serially correlated - there is a relationship between one month's return and the next, although the model assumes that no such correlation exists.
In "The Statistics of Sharpe Ratios," Mr. Lo wrote: "The common practice of annualizing Sharpe ratios by multiplying monthly estimates by (the square root of 12) is correct only under very special circumstances." The correct Sharpe ratio could be either much higher or lower, depending on whether the serial correlation is positive or negative. In some cases, the Sharpe ratio can be overstated by as much as 65%, he wrote.