HEDGE FUND RISK, WHITE PAPERS

On Hedge Fund, Human nature and Black swans

by Marc Gross, 2008

As hedge funds seek to institutionalize their business, many investors wonder what they have to offer and why they are becoming increasingly popular solutions.  Clearly based on any measure of risk adjusted returns a diversified portfolio of hedge funds offers superior performance, but are hedge funds subject to a different type of risk?  If so, what can investors do to mitigate these extreme, unforeseen risks (Black Swans)?

One answer to their allure lies in Daniel Khaneman’s research in behavioural economics, which indicates that economic decisions are subject to framing effects that can violate the basic requirements of rationality.  These effects lead to asymmetrical sensitivities to gains and losses.  Thus human nature seems to favour steady positive returns over infrequent large gains and at the same time, infrequent large losses over steady small losses.  Thus it is clear that the human mind itself may inherently favour short gamma strategies.

This characteristic of human nature may be one of the fundamental reasons that hedge funds; which seem perfectly designed to deliver such payoff patterns, have become increasingly popular.  In short, hedge funds have developed a niche which disproportionately rewards the hedge fund manager precisely because many provide the short gamma exposure which humans tend to favour; and because standard measures are poorly designed to adequately reflect the risk of infrequent, large losses.

Ask the traditional incumbents and you’re likely to be blinded by Pseudo-science.  A vast army of naïve analysts and investment consultants will dutifully run endless Gaussian, linear regression models and solemnly proclaim that the investment has no discernable beta against “the market”.  The problem here is that this is a canonical example of what Taleb describes as a Platonic process.  They are confusing the absence of proof with proof of absence.  Just because the source of risk has not been active during the period of study should not be confused with evidence that the risk is not there.

In fact standard analytic methods, which tend to ignore outliers, contribute to this fallacy.  If a fund is uncorrelated to the market 95% of the time, but suffers a large loss at precisely the same time as the market has its worst month, what are we to conclude?  The traditional school of thought, focusing on the bulk of observations, would show a very low, insignificant level of correlation and would confirm the naïve view that there was no meaningful relationship.

In fact this single outlier observation contains a great deal of information, precisely because it appears to disprove the hypothesis that the fund and the market behave independently of one another.  Because each loss event is improbable (5%), the possibility of a single joint occurrence indicates they are unlikely to be random (.25%).  Remember that a thousand observations can never conclusively prove a hypothesis (only reinforce it), but a single contradictory observation can disprove it!  There is much value to be gained by actively seeking out “disproof”.

Because hedge funds are opaque, they provide the manager with an opportunity to gain exposure to esoteric sources of market risk (also called alternative betas) in a highly asymmetrical way.  In other words, a manager can chose from a myriad of spread bets and collect a regular risk premium (steady positive return) safe in the knowledge (or comfortable self delusion) that when the infrequent event to which they are exposed comes to pass, they will have already pocketed enough of their management and performance fees to retire and live out their days in smug opulence.

One key reason which underlies this apparent unfairness (or persistent arbitrage opportunity) is the miss-match between the short term compensation window of the manager and the very long term structure of the risk exposure taken.  If a manager is exposed to a 99th percentile risk (a source of extreme market risk which rears its head about once every eight years), then there is likely to be a good streak of performance to  attract large capital inflows, providing hefty management and performance fees before the other shoe drops.  The high water mark is of no consequence to the manager because she holds a put option on the portfolio.  If the loss is severe enough she knows she’ll not regain the high water mark for years to come and so she picks up her marbles and goes home; dissolving the fund and starting a new one (after a suitable term of sabbatical (please see my previous comments about smug opulence)).

Unknown unknowns, out of the money puts, insurance companies.

As a fiduciary or investor in these funds it is as if you are selling out of the money puts without knowing which markets you’re selling them on.  It is as if you have an ownership interest in an insurance company, collecting premiums each month and hoping that the coming disaster won’t wipe you out.  It becomes your responsibility to actively search out such unknown, asymmetrical risk exposures; to try and infer which risks you are exposed to.  This is an essential part of your job; separating the many charlatans from the few providers of genuine value.  It doesn’t behove you to believe the hype.  Without a critical, clinical process of searching out “unknown unknowns” you are a sucker in the making.

Managing risk through strategy classification weights does not provide adequate diversification.  Strategy classifications are narrative labels which can bear no relation to the underlying risk exposures of the fund.  Many (if not most) funds are poorly calibrated to their strategic benchmark and yet, seemingly unrelated strategies can under perform in lock step during a market crisis.  This is largely due to the conditional correlation shifts that systematically occur during market crisis.

So how can we incorporate these black swans into our analysis of hedge funds?

  • Steer clear of simplifying assumptions, the real world is messy (linearity, Gaussian, time locked).
  • Searching for unknown unknowns. Allow yourself to be surprised.
  • Keep an open mind
  • Regularly scan a suitably large factor set
  • Look for independent validation or contradiction of narrative explaination.
  • Don’t confuse absence of proof with proof of absence.