It sounds like pretty much everybody in the investment business has come to the conclusion that hedge funds are part of an optimal investment strategy. The databases offered by various vendors seem to show that hedge funds as an industry outperform the stock market.
But what do we really know about hedge fund performance? As it turns out, the topic has garnered significant interest in academic research in the last few years. On the whole, these efforts have generated a surprising amount of evidence calling the image of hedge funds as "money machines" into question. Some of these findings are by now well known. Others, though important, have not yet been widely reported.
Today, most practitioners are aware of a few caveats, most importantly reporting biases in the data. A widely accepted version of the story goes roughly as follows: Biases may reduce absolute performance, but because of their favorably low correlations with the market, hedge funds outperform on a risk-adjusted basis. There are, in fact, empirical results compatible with this view. Academic studies find reporting biases (most importantly the survivorship bias) to overstate annualized return by about four to five percentage points. After adjusting for biases, absolute performance is roughly comparable to the stock market. Average alphas (the traditional measure of risk-adjusted returns) are estimated at six to eight percentage points on an annualized basis.
So far so good. However, recent academic literature has drilled quite a bit deeper. Here are some of the key insights:
•True risk-adjusted performance. Traditional measures are prone to generating misleading results when applied to portfolios with non-linear payoff structures, which are common in hedge funds. More accurate analyses compare hedge funds returns to those of risk-replicating dynamic strategies or employ non-linear risk factors such as options. Under these methodologies, risk-adjusted performance drops to either neutral or significantly negative.
•Concavity, tail risk and phase locking. There are indications that the discrepancy between positive alphas and neutral to negative true risk-adjusted performance stems from concavity. Concavity suggests the manager sacrifices upside return potential or amplifies downside risk in order to achieve measured alpha. Concavity can cause "fat tails" or "tail risk" (meaning higher than normal probabilities of very large losses), a common observation in hedge fund returns. And correlations with the market appear to be higher than average in bear markets and highest during periods of severe market turmoil such as the summer of 1998. This particular phenomenon is also referred to as "phase locking."
With more money chasing a limited amount of market inefficiency, pressure on managers is mounting. A recent National Bureau of Economic Research study found that a) correlations with the stock market have been creeping up over the last few years; b) illiquidity exposure is slowly increasing as well; and c) there appears to be a marked increase in liquidation probabilities — i.e., hedge funds going out of business — in the past few years.
As Alan Greenspan has observed, "Continuing efforts to seek above-average returns could create risks for which compensation is inadequate. Significant numbers of trading strategies are already destined to prove disappointing, a point that recent data on the distribution of hedge fund returns seem to be confirming."
Overall, academic research shows at least some of hedge funds' nominal performance appears to be generated by concavity-inducing strategies rather than skill. In terms of a risk-return profile, this comes down to short positions in options — not a strategy one should engage in on a large scale. The related phenomena of tail risk and phase-locking are worrisome and, if real, undermine the supposed benefits of hedge funds exactly when they are most needed. Increasing cash inflows and performance pressures are likely to exacerbate the problem.
On the other hand, the history of hedge funds as a broad phenomenon is too short for these findings to be definitive. Additionally, they describe the industry as a whole, not each individual fund or style category. It is plausible, and even likely, that there are exploitable inefficiencies as well as truly skilled and disciplined managers. Even if this is the case, it is not obvious how to determine a fund's true risk profile and distinguish between skill and "pseudo" alpha, given the lack of transparency.
From a practical portfolio management perspective, these findings suggest some preliminary warnings and guidelines:
•Don't trust your portfolio optimization software if it tells you to invest 60% of your assets in hedge funds.
•Hedge fund returns are lower than the raw data suggest, and they may carry significantly more unsystematic (that is, non-market) and systematic (that is, market) risk than is apparent.
•There is evidence that many funds are closely correlated with the stock market in severe downturns, exactly when you don't want it.
•Apply a thorough manager selection process but assume that the process isn't foolproof.
•Most importantly, try to determine the actual exposure and watch out for tail risk.
•For the time being, build an "uncertainty factor" into your models and do not invest more in any given hedge fund or even style category than you can afford to lose.