Barely a day passes without the announcement of another significant mandate from a large institutional investor looking to diversify its portfolio of investments by making an allocation to hedge funds. For example, in April it was announced that Railpen in the UK had plans to invest £600m into hedge funds1, Nestlé is investing 18% of its pension fund in hedge funds2 and the Massachusetts State Pension awarded mandates amounting to $1.6bn to a range of fund of hedge fund providers3. The reason for the increased capital flow from large traditional institutional investors to hedge funds has been the perceived higher returns available in hedge funds in a period when both equity and bond market return expectations are sluggish, as well as to increase diversification. However, the world of hedge funds is very different from the traditional investment arena where these large institutions have roamed in the past. One important difference is that hedge funds have an absolute return objective while in the long only world performance is compared with benchmarks in order to assess the level and quality of returns. This represents a fundamental change in how traditional asset managers will have to determine and analyse investment returns. For example, if a pension fund has in the past used equity and bond benchmarks to assess the relative performance of its equity and bond exposures respectively, how can the fund determine whether its investments in hedge funds have performed acceptably relative to other investment possibilities in the hedge fund asset class, or indeed to other alternative asset classes such as private equity? Furthermore, without a valid benchmark, how can the asset manager develop adequate risk controls and risk budgeting systems?
The easiest approach when selecting a benchmark, not to mention the approach that is closest to the existing mindset of traditional asset managers, is to select one of the numerous published hedge fund indices as developed and reported by organisations such as CSFB/Tremont (www.hedgeindex.com), Hedge Fund Research (www.hfr.com), Morgan Stanley (www.morganstanley.com) or Standard & Poor’s (www.standardandpoors.com). For an inexperienced hedge fund investor this would seem to represent the safe choice, as the implementation of risk and performance controls for the hedge fund investment would be seamless. However, hedge fund indices suffer from a range of data biases and inherent construction problems that make their overall usability questionable. For many of the longer running indices the background data for the indices comes from their database of historical hedge fund returns, which is the cause for most of the inherent problems such as:
q Selection bias, which occurs due to the unregulated nature of the hedge fund industry. As each individual fund may decide whether they want to report their performance to the database providers, it will generally be funds with “reasonable” performance and that remain open for investors, that will submit their returns. The incentive for funds that are closed to submit their data is non-existent, and thus the available data that forms the index will not be representative of the hedge fund universe. This selection bias is also an issue with the newer indices, such as the S&P hedge fund indices. While designed from a pool of hedge funds on which they have accurate historical data, in many cases they also only include open hedge funds in order to incorporate investability into their indices; this makes them unrepresentative.
q Survivorship bias, which occurs due to the database providers not including information from defunct funds in their indices, thus leading to an upward bias in the reported aggregate return.
q Instant history bias, which occurs because it is up to the hedge fund itself to decide when it starts reporting data, so for example if a fund has received seed capital and goes through an incubation period, it is up to the manager whether this period will be included in the fund’s track record.
While these data biases have been the focus of intense academic research, there are further difficulties in constructing a hedge fund index. It is important to consider the fact that there is no industry-wide accepted definition of hedge fund strategies. For asset managers that choose to invest directly in hedge funds (rather than the fund of funds route) this will create a problem, as it may prove difficult to match up the fund and applicable index. This is further exacerbated by the increasing tendency of hedge funds to develop and alter their strategy dependent on market conditions and in response to increasing capital inflow into the more traditional hedge fund strategies. For fund of funds investors, this problem is insignificant, though both forms of hedge fund indices have further issues in terms of sample size and accuracy of the data. In many cases the sample size of hedge funds (or fund of funds) that are chosen to represent the index will not be more than 20–30 funds. With such a small sample size, the representation of the index is questionable, and it will further increase the effect of the data biases, as outliers in terms of performance will naturally skew the entire index. The accuracy of the data is also questionable, as in many indices the returns are reported from the funds themselves rather than an independent pricing agent such as an administrator or prime broker. This may lead to data reporting errors, and in conjunction with a small sample size in an index, the end result may be a very significant tracking error between the index’s reported performance and that of the underlying managers.
As the industry matures, the “hunt” for better hedge fund indices will continue, and several service providers, hedge funds and academics are trying to overcome the problems mentioned above and others.
However, as more traditional asset managers move slowly into the dark and uncertain world of hedge funds, perhaps it is now time for them to review how they use benchmarks. When investing in a hedge fund, what one is looking for is an absolute return, an investment that should be able to (more often than not) make positive returns independent of the direction of equity or bond markets. The natural progression would then be to formulate a minimum acceptable return (MAR) that the asset manager is expecting from the investment and an acceptable level of risk. As the capital that is allocated into hedge funds is frequently taken from traditional asset classes, it is often tempting to compare the performance between the two and base expectations on that. It is further very tempting to base a return expectation on the past performance of the hedge fund/fund of funds, but hedge funds tend to have asymmetrical return profiles; that is, their extreme returns occur somewhat more frequently than is normally assumed for performing traditional return/risk analysis. Thus the true risk of a hedge fund is often severely understated if one just looks at past performance. Using either measure, it is impossible to accurately monitor risk over time, as the funds price monthly rather than daily. Moving from a risk system where daily data is available and risk imbalances can be spotted early, to a monthly system where the risk can suddenly appear without any warning can be a harrowing experience for any asset manager.
Until any superior system has been devised, the way forward must be to:
q regard the hedge fund investment as a long-term commitment – and with generally poor liquidity among hedge funds, the investor often has no choice but to invest for the longer term;
q aim to earn a multiple of cash rates; two to three times the risk free rate should be acceptable for most large pension funds;
q focus on the non-correlated nature of hedge funds, and their consequent diversifying effect on the portfolio.
RMB International is a division of Ansbacher & Co Limited, Two London Bridge, London, SE1 9RA, United Kingdom. Ansbacher & Co Limited is authorised and regulated by the Financial Services Authority.