Put yourself in the CIO’s seat of a fund of hedge funds for a moment. Imagine describing to investors your risk-management process for existing and prospective investments. “Our risk-management process relies on the risk-adjusted returns listed on the major investors’ databases to which we subscribe. Key amongst these are the Sharpe ratios that we use to assess performance. In addition, all the funds in which we invest have an in-house risk manager and provide risk information via third-party independent providers. As we are invested in all the major strategies, we are comfortable with the diversification in our returns.” Sound familiar? Relatively standard pieces such as the one mentioned above have become familiar AI jargon.

If diversification is the name of the game, the CIO above would need to offer plenty of explanations after the months of April and May. According to the HFR database, all major strategies were down in April. Based on preliminary estimates, May will broadly continue the previous month’s pattern. Such tight correlation in performance is striking because there were no single-day shocks, defaults or other catastrophic events to blame as, for instance, during the Tequila, Asian or Russian crises. The answer has to be found elsewhere.

This article addresses a variety of common myths on diversification and liquidity, hot topics for anybody who is looking for ways to attract the interest of institutional investors. A sequel article will delve into the topic of risk management in hedge funds, another important element in a fund’s marketing process.

The oft-used magic word of diversification has several important meanings to investors. Let’s review some of the most common:

Use of past performance figures is the acid test on which all of the standard definitions above can rely. These definitions of diversification are attractive for quantitative asset allocation and limits, since they are fairly objective and can be formalised in numbers and codes to feed computer models.

Liquidity is another element that is often high on many investors’ question list. Unfortunately, as we will soon see, diversification and liquidity are generally located in two watertight compartments. Liquidity is an odd creature, in that only rough quantitative measures exist to track it. Most common examples include average daily or weekly trading volume, average bid-offer spread or concentrations of largest block traders as a proportion of total trading volume. As such, liquidity does not easily lend itself to quantitative asset allocation, and most investors address it with crude limits and guidelines, such as:

• “no more than x% of total daily trading volume”, or
• “no more than x% of total allocation into a given fund”, or
• “no more than x% of total assets of any fund”.

There are several definitions of liquidity and the semantics can be a challenge. Indeed, many if not most investors associate it with their ability to pull their investments out of a fund. Thus, the following table attempts to clarify some of the meanings that are most frequently associated with liquidity:

Amongst these, the first three are fairly straightforward. The fourth is not, since it tracks an unwelcome derivative effect of liquidity that lurks in the dark. This fairly elusive side, unfortunately, is often the flipside of the return coin for most investors. One could define it as follows: (%RP/%IP), where the numerator refers to the percentage return – over a given time horizon – achieved by security P, and the denominator indicates the return achieved by a preset reference index I for the most liquid comparable security. The classic instances that come to mind are the “on-the-run” versus “off-the-run” Treasury Notes in the U.S. government bond market. The on-the-run securities provide the benchmark against which all other comparable securities are priced. Then, the above equation assumes the value 1 for on-the-run securities, since numerator and denominator move by an equal amount. Depending on the current market trend, it will change by more or less than 1 for all other securities. This means risk for investors, who, accordingly, will require compensation in terms of an excess return. This is the liquidity premium.

Off-the-run securities will also feature wider bid-offer spreads – the definition of trading liquidity risk – than their on-the-run counterparts. Thus, the latter will be the preferred choice for investors who wish to execute quickly and inexpensively. This, in turn, will make their prices move faster than the others in the presence of large market fluctuations. Consequently, on-the-run securities will have a bias towards outperforming the others during the first move of bullish trends and will fare marginally worse at the beginning of bear markets. Correlation pairs between similar equities or credit instruments also feature a security with a risk premium and will display similar behaviours.

From a strict semantics point of view, relative-value “arbitrages” and the instability of the liquidity premium would seem contradictory. As a matter of fact, that liquidity differential is the very reason for the premium. In practice, one can view the premium as an option that the investor has sold: most of the times, he or she will earn an excess return by being long the less liquid security and short the more liquid one to hedge against directional market risk. By definition, the liquidity premium should never become negative, since the liquidity features of the benchmark security will always have some extra appeal relative to others. Sometimes, however, the more liquid security of which the investor is short will rally more than his long leg of the arbitrage. This means that the excess return – the liquidity premium – will widen and large losses will follow.

Several prestigious hedge funds have fallen prey to the vagaries of the liquidity premium. Following the default of Russia, the summer of 1998 witnessed a major rally in the U.S. Treasury Bond market. Many funds which had set up arbitrages by holding off-the-run, higher yielding Treasuries and shorting on-the-run issues were caught on the wrong side of the trade. Of course, the massive amounts of leverage added on top of these trades resulted in a major explosion of the losses.

Over the last 18 months, the importance of liquidity for returns has been the target of empirical work. In particular, one of the most advanced pieces of research to date (Getmansky, Lo & Makarov, 2003) has studied over 900 funds in the Tremont-TASS database to investigate serial correlation and illiquidity in hedge-fund returns. While preliminary, the results confirm the intuition that illiquidity is the real source of returns for the AI industry and their investors. Lo, Petrov & Wierzbicki (forthcoming) also show that portfolios with very different liquidity features can be constructed to fit standard mean-variance portfolio optimisation. Increases in negative skewness and kurtosis in portfolios with higher allocations to hedge funds and other statistical properties are also fairly well-known features that have attracted substantial research work (see, amongst others, Kat, 2001; 2002).

The aspect that is puzzling about investors’ relatively nonchalant attitude toward liquidity is that, from a macroeconomic point of view, hedge funds are machines created to probe the frontiers of theoretical finance in the real world (Goetzmann, 2004). Accordingly, they are compensated for their efforts to test the quantitative and qualitative capacity of markets to absorb new financial inventions and strategies. Eventually, liquidity – or lack thereof – is the name of the game. While being paid to test the frontier, in the form of a liquidity premium, AI managers can never really quantify the denominator of this ratio, namely, how much illiquidity they are taking.

This is the point in which the false promises of diversification come in. By diversifying as indicated in the table above, investors are indeed actively managing their concentration by manager, strategy and asset class. As demonstrated by the Getmansky, Lo & Makarov’s study, however, liquidity is the theme that cuts across virtually all strategies.

One could view it as the systematic, undiversifiable risk of the AI industry. Spreading allocations across a variety of strategies gives the investor a presumed – but illusory – diversification benefit. This way, investors are displaying a psychological weakness that is well known in behavioural finance, the 1/n bias (Benartzi & Thaler, 1998).

According to this tendency, investors are prone to increase the number of securities in which they invest regardless of their common underlying exposure. Consistent with the popular adage of not putting too many eggs into one basket, a broader range of different strategies provides psychological comfort. In the same fashion, many investors fall in the trap of believing that having several different strategies and a variety of managers will protect them from liquidity crunches. In addition, investors who rely on diversification alone are unwittingly taking a bet on the correlation of different asset classes remaining fairly tame.

The fifth definition of liquidity – strategic liquidity risk – is even more elusive for the industry but no less disquieting. One chilling statistic from the HFR database summarises it beyond doubt: over the last 14 years, the AI industry has experienced net outflows only once. That was in 1994, the worst bear market in bonds in six decades but a relatively short one. Since then, a lot of the capacity has been built in and around the AI industry over the last few years. It is legitimate to wonder how prepared the industry – and the largest funds – are for the possibility of an extended bear market in bonds.

How can one manage liquidity risk more consciously? Let me suggest some simple steps:

1) First of all, an investor must be very clear on the portfolio objectives in terms of buying or selling liquidity insurance and at what cost. Most funds will be structural liquidity insurance sellers and get paid accordingly. If you are uncomfortable with the amount of liquidity insurance that your funds are selling, you’d better get real on the returns on your return targets, too.

2) You can then proceed to classify how your different hedge fund strategies may behave under different liquidity conditions. For instance, a credit crunch in the Treasury market will likely – but not always – coincide with a credit crunch in equity markets. Under these circumstances, your current diversification policy will indeed help you.

3) Consistent with basic Risk Management, your AI managers should provide sufficient evidence of ongoing portfolio stress-testing to assess portfolio performance under a variety of liquidity conditions. Correlations can and do become “unreasonable” and “impossible” every once in a while.

4) Last but not least, there is a small segment of hedge funds that are classified as Long-Volatility – or even Long-Gamma – funds that are taking the other side of the liquidity trade. Under normal or benign liquidity conditions, these niche players will expect to incur a higher percentage of small losses than other strategies. This is natural, since they are essentially investing in liquidity insurance, an event with a relatively infrequent positive payoff. In return, however, they will tend to outperform in those 10%–20% of the instances of turmoil in which liquidity conditions and correlations tend to become most stretched. A proportional allocation to these strategies can help smooth your return profile in these particular situations. As their bets experience windfall profits only occasionally, tight position-sizing is key to their performance.

5) With respect to strategic liquidity risk – investors should assess carefully a fund’s breakeven point and revenue sensitivity. It does not take a genius to figure out the link between increased operational risk and the sudden cut-backs and layoffs that would result from an industry downturn.

Other psychological biases loom large over investors’ AI allocation process. My forthcoming book The Dark Side of Risk Management (Prentice-Hall, July 2004) explains some of their practical consequences. Let me leave you with some questions that will be revisited in detail in the sequel to this article in the next issue:

1) What do people mean by Risk Management in Alternative Investments?
2) Frequency beware: how often are you relying on data frequency as a proxy for your probability assessments?
3) Some standard risk measures, uses and misuses.

That will extend the review to some other common pitfalls in performance measurement.

References

Benartzi, S. & Thaler, R. H. (1998) Illusory diversification and retirement savings. Working paper, University of Chicago & UCLA.

Getmansky, M., Lo, A. & Makarov, I. (2003) An Econometric Model of Serial Correlation and Illiquidity In Hedge Fund Returns. MIT LFE Working paper No. 2001-1023.

Goetzmann, W. (2004) Hedge Funds and the Frontiers of Finance. Presentation at the 4th Hedge Fund Conference, Milan, 20 May 2004.

Kat, H. M. (2001) The Statistical Properties of Hedge Fund Index Returns and Their Implications for Investors. Working paper, Cass University Business School.

Kat, H. M. (2002) Taking the Sting Out of Hedge Funds. Alternative Investment Research Centre, Cass University Business School.

Lo, A. W., Petrov, C. & Wierzbicki, M. (forthcoming) It’s 11 pm – Do you know where your liquidity is? The Mean-Variance Liquidity Frontier. Journal of Investment Management.

This article was first published in the Swiss Derivatives Review. For a free subscription .please contact timo.kueng@weber-thedy.com