Institutional investors in general and pension funds in particular have been dramatically affected by negative stock market returns at the beginning of the millennium. In the context of a cumulative asset/liability deficit that was estimated at more than £55 billion in 2003 for the companies in the FTSE 100, institutional investors are seeking new asset classes or forms of investment management that would allow them to broaden their traditional choice of asset allocation.

Following considerable success in private wealth management, hedge funds are targeting institutional investment

An alternative investment offering has been introduced in the past several years, allowing investors to optimise the risk/return combination of their portfolio. While the original rise of hedge funds was explained by the success encountered with private clients, we now expect institutional investors to follow the lead of private banking with significant inflows of capital in the next few years. Moreover, these inflows have already begun, with the number of institutional investors using alternative investments in constant progression, particularly due to significant increases in Europe, Australia and Japan. In 2005, hedge fund use by European institutions has doubled to 48% from 24% in 2003. By 2007, European institutions are expected to dedicate 7.2% of portfolios to hedge funds, an increase from the 5.3% estimated for 2005.¹

However, the use of hedge funds in asset-liability management requires precautions

While it is certainly legitimate to seek to improve the asset/liability ratio with the help of hedge funds, a naïve approach to the question is likely to give extremely unstable results. For example, if we use the historical data on hedge fund returns without any precautions in a simple relative risk optimisation exercise (optimisation of the information ratio in relation to a stylised benchmark that is perfectly matched with the investor’s liabilities), the optimal allocation to hedge funds obtained varies from 0% to 100% depending on the period used for calibrating the model! Such a lack of robustness in the analysis suggests that a naïve approach to the place of hedge funds in ALM is totally inappropriate.

Unlike traditional asset classes (stocks and bonds), for which we can find both reliable stochastic models and estimations of long-term parameters that are indispensable for carrying out a surplus optimisation study, major obstacles actually prevent institutional investors today from considering hedge funds to be an ALM class in the same way as traditional classes. Not only does the absence of a sufficiently long historical record, together with the opaqueness of the data and the biases that characterise the available databases, make any effort at projecting long-term risk and return parameters particularly illusory, but there is no satisfactory existing model for hedge fund returns, in spite of recent progress in academic research on the subject. Hedge fund behaviour, which is typically characterised by dynamic management strategies that sometimes employ a multitude of complex products, is difficult to capture in the linear models that are classically used in finance.

A new approach to integrating hedge funds: as a complement, not a strategic allocation class

In order to cope with these difficulties in modelling hedge funds, and the risk relating to estimation of the parameters, in particular the expected return parameter, which is highly sensitive to the choice of sample, we propose in this study to adopt a more robust alternative approach to the place of hedge funds in the context of ALM. Rather than considering hedge funds to be an additional class to the traditional classes in an asset-liability exercise, we consider hedge funds to be complementary management styles for stocks and bonds.

This approach, in which hedge funds are not regarded as an asset-liability allocation class, is in fact consistent with the usual practices of institutional investors with respect to the traditional management styles. It is indeed common to model a very limited number of major asset classes in a stochastic asset-liability optimisation exercise, rather than seeking to model all the sub-components of the class. In the stock universe, for example, we typically model the return of a global large-cap equity index, to which we might possibly add a small-cap stock index, without looking to independently model sector, geographical or style (value versus growth) indices. This approach to alternative investment as a complement to traditional investment management is also consistent with the idea that hedge funds are above all considered by institutional investors to be diversification tools that allow the risks of a stock or bond portfolio to be reduced, rather than a tool for improving long-term returns. This desire to see hedge funds as a diversification tool is reinforced by recent academic research results which have shown that the diversification properties (risk reduction benefits) of hedge funds were more robust from a statistical standpoint than their capacity to provide absolute performance (return enhancement benefits).

The study presented here shows that it is possible to construct diversification benchmarks that allow the risk related to holding stock or bond portfolios to be reduced in a very significant and robust way, by i) appropriately selecting the alternative strategies and ii) optimising these with proven techniques (minimising the extreme risks, as measured by the Value-at-Risk of the overall portfolio).

These diversification benchmarks thereby allow the long-term volatility parameters of the stock and bond classes to be reduced significantly.

As a result, introducing these diversification benchmarks as complements to the traditional classes (stocks and bonds) enables ALM performance to be improved significantly, even for reasonable quantities of investment in hedge funds. In particular, we find that the introduction of only slightly more than 11% of hedge funds into the overall allocation can lead to a reduction of more than 27% in the probability of the value of the assets falling below 75% of that of the liabilities (asset-liability deficit greater than 25%). Moreover, it should be noted that an allocation to hedge funds of approximately 20% allows the probability of extreme risk to be reduced by more than 50%.

Conclusion

While most institutional investors are looking into hedge funds as a possible solution to the challenges posed by asset-liability management in the presence of serious concerns over the size of the equity and bond premium and the associated risks, very little is known about how to include these alternative investment styles in an ALM context.

This study provides evidence of the contribution of hedge funds in a surplus optimisation context. To this end, we have proposed a pragmatic approach that does not treat hedge funds as an addition but rather as a complement to traditional asset classes (stocks and bonds), which alleviates the concern over the modelling of hedge fund return distributions and parameter estimation. Our conclusion is that suitably designed hedge fund portfolios can be particularly attractive when the objective of optimising expected returns is constrained to meeting liabilities. This is due to hedge funds’ benefits in terms of diversification properties, which in turn is related to their appealing behaviours in terms of the impact on tail-distribution and extreme risks of stock and bond portfolios.

This research was sponsored by Lyxor.

Footnotes:

¹ Source: Russell Investment Group’s seventh annual report on alternative investment, September 2005.