Portfolio Construction: Hedge fund ALM
The 2008 crisis showed how liquidity mismatches can undermine apparently robust hedge fund portfolios. Peter Meier and Jann Stoz argue that measuring returns autocorrelation can enable investors to assess mismatches using only fund of fund-level information
The perception of liquidity and liquidity risk will never be how it was before the financial crisis of 2008. Banks will have to comply with liquidity coverage ratios, and hedge fund investors are still overly sensitive to liquidity risks and avoid longer redemption frequencies and notice periods. Investors have not only to check the liquidity terms of the relevant fund of hedge fund (FoHF), but also of the underlying fund positions and the liquidity match between those two layers. A FoHF might offer monthly redemptions with five days' notice, but its constituencies can have lock-ups of three years and quarterly redemptions, leading to substantial redemption risk for the FoHF investors. This article offers a method to measure liquidity gaps between FoHFs and their underlying hedge fund positions using only return and redemption information available at the FoHF level.
FoHF liquidity risk is tricky to evaluate and monitor. Liquidity terms per se, like lock-ups or redemption frequency, tell little about the capacity of a fund to redeem cash when systemic liquidity dries up and hedge funds have to ‘gate' (putting limits on the amount an investor can withdraw at any one time) or suspend redemptions.
A comprehensive analysis of FoHF liquidity risk is far reaching. Besides the liquidity terms on the FoHF level, one has to make sure that the FoHF can control the terms of its single managers and the match between the two. Furthermore, leverage and credit facilities have to be analysed and monitored, and funds under consideration should not be ‘cannibalised' by parallel vehicles at the same managers with side letters for privileged investors, or managed accounts. Under investors' pressure to secure liquidity, many managers have opened differentiated forms of portfolios based on the same strategies, and this introduces the potential for unequal treatment, particularly when it comes to redemptions.
We were part of a team that developed a ‘total risk rating' which, among other things, encompassed a comprehensive module for analysing and scoring FoHF liquidity and credit risks. A quantitative proxy for liquidity risk is autocorrelation - the tendency of time series data to show similar patterns at different points in time. Market returns are independent from previous returns if the market is efficient and prices are marked-to-market. Positive autocorrelation or serial dependence is an indication of illiquidity, but can also have its roots in a smoothing of returns by managers in order to reduce measurable volatility and artificially increase Sharpe ratios.
While the latter source for illiquidity is unacceptable and corruptive, the former is an inherent characteristic of many hedge fund strategies that use ‘sticky' instruments like private equity or apply arbitrage strategies that take time to display convergence. Recently, academic literature has shown that autocorrelation is a useful proxy for illiquidity but far from perfect. A weakness is its failure to discriminate between inherent illiquidity and smoothing of returns (although both sources of autocorrelation are adverse characteristics).
A way to evaluate the significance of autocorrelaton is to compare it with the contractual liquidity terms of hedge funds or FoHFs. In figure 1, autocorrelation is plotted against the total redemption delay for FoHFs from the Hedgegate universe (www.hedgegate.com), which comprises all Swiss-registered FoHFs and qualified investor FoHFs.
Autocorrelation for the first time lag is measured for a 48-month period ending in February 2011, and the total redemption delay is the sum of the redemption frequency and notice period. The relation between autocorrelation and redemption is statistically established by recent research and is not rejected by the Hedgegate data for FoHFs.
A separation of the FoHFs according to hedgegate classification reveals more detailed findings. For diversified FoHFs total redemption delays span from daily liquidity to 180 days. Daily liquidity is provided by investment companies with a permanent secondary market, but obviously at the cost of discounts between the tradable prices and the NAV. The focused directional FoHFs have generally shorter redemptions since they incorporate more liquid strategies like long-short equity, global macro or managed futures. Focused non-directional FoHFs using arbitrage and event driven strategies have the longest total redemption frequencies.
If autocorrelation is supposed to be a good illiquidity measure, one might expect a tighter relationship with the redemption delays compared to the cloudy picture of figure 1. Effectively, we find FoHFs with short redemptions and high autocorrelation in the north-west corner and others with long redemptions and low correlation in the south-east corner.
The ‘north-west' FoHFs have risky liquidity structures: they have short redemption liabilities while their constituent single hedge funds follow illiquid strategies, as measured by autocorrelation. Examples are investment companies which suffered maximal discounts of more than 50% during the liquidity crunch in autumn 2008. The term structures of the ‘south-east' FoHFs are very conservative with respect to liquidity - they offer investors low liquidity, while their target funds seem to be liquid. This cross-sectional analysis allows us to classify FoHFs with respect to their relative liquidity match or mismatch without having data for the single hedge fund holdings.
Figure 2 covers a list of FoHFs with long total redemption delays and low autocorrelations - offering particularly favourable liquidity conditions with respect to the match between the FoHF and its single hedge funds. The latter are liquid, while the FoHFs have protection against short-term redemptions by their investors.
Figure 2 reveals some interesting features of the most cautious FoHFs and their managers. First, most successfully navigated through the 2008 crisis because they were not hit by the liquidity crunch. Secondly, those with the most secure liquidity matches have above-average performance ratings: 15 of the 23 FoHFs in table 1 have ratings A or B (under a Hedgegate rating system based on alpha, beta premiums and conditional value at risk), putting them in the top 30% of FoHFs.
This second finding is perhaps surprising because, in general, illiquidity is not only looked at as a downside risk, but also as a source for a premium and a contribution to returns. No investor is ready to give up liquidly without getting a pay-off for it and this is also true for FoHFs - although the illiquidity premium is not easy to show, because liquidity and performance are difficult to measure appropriately.
Our research reveals significant relationships between illiquidity (measured by redemption delays or by autocorrelation) and performance (measured by the Hedgeate performance score which consists of alpha estimates, beta premiums and expected shortfall). The liquidity premiums are more pronounced for directional FoHFs than for non-directional - and the relationship vanishes when the crisis period of 2008 is included in the analysis. Illiquid strategies like long-short equity market neutral or credit arbitrage which should incorporate high illiquidity premiums suffered most during the crisis, and therefore potential premiums disappeared. To conclude: illiquidity is a return contributor as a general rule, but it is sensitive to investment cycles and styles.
Peter Meier is head and Jann Stoz is a research assistant at the Centre of Alternative Investments at ZHAW, Zurich University of Applied Sciences