In response to the need of investors for greater transparency on the risk and return characteristics of funds, Edhec carried out an international consultation on the sensitive issue of FoHF reporting. The objective of this ambitious project was to define the relevant information that FoHF reports should contain so that investors could properly integrate hedge fund strategies into their global asset allocation. To this end, Edhec made a series of recommendations based on state-of-the-art risk management tools suited for hedge fund strategies. The resulting white paper was then sent out to fund managers and investors for discussion.
The findings of this consultation, which brought together the opinions of 98 institutional investors and fund managers, allow a consensus to be established on the information required for the implementation of a relevant reporting method in the field of alternative investment. Investors and fund managers agree that reports should cover the whole spectrum of risk definitions, ranging from normal risk (i.e. volatility) to loss risk (e.g. max. drawdown) and extreme risks (i.e. Value-at-Risk estimations). In the same vein, they agree that FoHF reports should better inform investors on the determinants of FoHF performance so that they can gain a good understanding of its risk factor exposures. In the remainder of this article, we will elaborate on another key finding of the Edhec FoHF Reporting Survey, namely the growing acceptance of relative returns.
As mentioned earlier, hedge fund strategies have long been seen as absolute return strategies serving as return enhancers. They were consequently integrated into the “satellite” part of investors’ portfolios to implement so-called “portable alpha” strategies. With the long bear market that followed the bursting of the internet bubble, investors desperately sought new diversification tools, which led them to consider hedge fund strategies as risk reducers. They thus started to integrate hedge funds into their “core portfolio”. This shift from an alpha logic to a beta logic is reflected in investors’ growing interest in the underlying risk factors of hedge fund strategies (see Edhec (2005). This is in turn illustrated by the growing acceptance of relative returns, as opposed to absolute returns, by market participants. As a matter of fact, while 62% of fund managers currently report the performance of their fund relative to a benchmark (see Edhec (2003), 80% of them intend to do so in the near future (see Edhec (2005). Interestingly, while only 17% of fund managers consider that relative returns are not important, and should consequently not be found in tomorrow’s FoHF reports, 42% of investors think the same. Nevertheless, as shown in Illustration 1, the majority of investors, namely 58%, still consider that it is ‘important to very important’ to find this information in FoHF reports. This is a sign that a profound revolution is taking place in the alternative arena.
This slight discrepancy between the mindset of fund managers and investors is in turn illustrated by the relative importance they grant to indicators such as year-to-date return or cumulative returns. While 97% of fund managers consider that year- to-date returns are ‘important to very important’, 79% of investors think so. Moreover, 80% of the former consider that cumulative returns are important, while 58% of the latter think so. These observations do not come as a surprise. Since the bulk of fund managers have marketed their funds as “absolute return” vehicles, investors expect to obtain what they have paid for, i.e. steady positive returns. Since the year-to date and cumulative returns of a fund that consistently posts positive returns are positive, we can easily understand that it is not essential information for investors. On the other hand, it is useful for managers to show that despite some potential negative outcomes, they manage the fund well since they (hopefully) post positive year-to-date and/or cumulative returns in the end.
Nevertheless, the recent poor performance of hedge fund strategies could put an end to this somewhat awkward situation. Many fund managers have already started to base their sales patter on relative returns and diversification properties. The goals of fund managers and investors should therefore be progressively aligned and greater attention will certainly be paid to relative returns in the future.
We have seen in the previous part of this article that most fund managers are already including an estimation of relative returns in their reports. The problem is that the benchmarks used to measure relative returns are generally inappropriate, which brings the accuracy of those relative returns into question.
A good benchmark must be representative of the fund’s underlying risk profile. A good index, on the other hand, must be representative of the investment universe it is meant to cover. Both must also be investable. Since funds and FoHF – even those following the same strategy – are typically very heterogeneous in terms of their risk profile, the probability of disposing of an index that is representative of a specific fund or FoHF’s risk profile is relatively low. In other words, a good index is not a good benchmark in most cases. It is therefore necessary to design a customised benchmark.
Unfortunately, in most cases, fund managers disclose their performance relative to a benchmark that is not meaningful at all (see Edhec (2003). In certain cases they consider the risk-free rate as a reference. However, being market neutral does not imply that there is no exposure to sources of risk other than market risk (see Amenc et al. (2003). Indeed, notwithstanding the fact that most hedge fund strategies are based on arbitrage, they generate the bulk of their returns through their exposure to a wide variety of risk factors (e.g. volatility risk, liquidity risk, credit risk, etc.). Since using the risk-free rate implicitly assumes no exposure to any source of risk, it can certainly not be representative of the risk profile of hedge fund strategies. As a consequence, it cannot be regarded as a quality benchmark. Similarly, even if hedge fund strategies provide investors with an alternative to traditional asset classes, using stock and bond indices as benchmarks is not reasonable. As mentioned earlier, hedge fund strategies are exposed to a wide range of risk sources other than market risk. Stock and bond indices cannot therefore be regarded as quality benchmarks. Finally, in certain cases, fund managers use hedge fund indices to benchmark their performance. Again, if the index is perfectly representative of the fund’s risk profile this is a good solution. However, in most cases this will not be the case. On the other hand, the hedge fund indices available on the market are generally not investable. They therefore fail to fulfil the basic criteria of a quality benchmark. As a result, investors should not take the relative returns that are disclosed in FoHF reports at face value.
Relative returns measured with inappropriate benchmarks are biased, which may in turn alter the efficiency of the whole investment process. For this reason, it is essential to make sure that fund managers are using customised benchmarks to estimate relative returns. The difficulty is that strategies implemented by hedge funds are highly sophisticated, which makes it difficult to identify the key determinants of the performance of a given fund. A set of risk factors must however be identified to design the benchmark. The construction of the benchmark is therefore subject to significant specification risk. This is all the more important in that relative returns are extremely sensitive to the choice of benchmark. To mitigate this issue, one can use pseudo-risk factors embedding the specific risk profile of hedge funds. Hedge fund indices appear to be ideal candidates to serve as pseudo-risk factors.
Fund managers should consequently construct their benchmark through a return-based style analysis as introduced in Sharpe (1988, 1992). Sharpe’s style analysis involves running a constrained regression as described below. Normal returns represent the returns of the strategy benchmark, i.e. the returns corresponding to the investment policy of the fund manager. Abnormal returns correspond to the value added through active management (i.e. tactical allocation and/or fund picking) by the fund manager over and above the return of his strategy benchmark. Relative returns are equal to the latter.
Where RFoHF,t is the performance of the FoHF at time t, bk is the factor loading of style factor k (with k = 1 to n), RFk,t is the performance of style factor k in t, a is the intercept of the regression, and et an error term.
Since the quality of the output (i.e. relative returns) of a return-based style analysis strongly relies on the quality of the inputs (i.e. hedge fund indices used to construct the strategy benchmark), Sharpe suggests using a series of style factors that are “collectively exhaustive” and “mutually exclusive”. Collectively exhaustive means that a style factor covers the whole investment universe which it is meant to represent. Mutually exclusive means that investment universes covered by the different style factors must not overlap. In other words, hedge fund indices must be representative and pure. Furthermore, since quality benchmarks must offer a viable passive alternative to actively managed funds, the style indices used in the return-based style analysis must be investable or at least easily replicable.
Finding appropriate style factors is a challenging task. On the one hand, most “non-investable” hedge fund indices available on the market lack representativity (see Amenc et al. (2004)). On the other hand, as highlighted in Géhin and Vaissié (2005), most “investable” hedge fund indices available on the market are nothing but passively managed FoHF that have improved their investability at the cost of their representativeness. Nevertheless, the latest research conducted by Edhec on this topic suggests that it is possible to construct indices that are both “investable” and representative, provided that (i) the underlying funds are selected for their beta characteristics and not for their performance and (ii) maximisation of the representativeness dimension is explicitly targeted in the index construction process (see Goltz et al. (2005)). As can be seen from Illustration 2, in this case it is possible to construct Factor Replicating Portfolios (FRP) made up of fewer than 10 funds, which mimic the behaviour of the 1st component of a principal component analysis (PC1) carried out on a database of over 3,000 hedge funds. With these third generation indices, it is therefore possible to construct quality benchmarks, and to measure relative returns properly, by subtracting normal returns from the fund’s total return.
The considerable participation of fund managers in the consultation process initiated by Edhec indicates that they have understood the strategic importance that FoHF reports will have in a context of increasingly demanding investors. However, defining relevant information is a tricky task. Fortunately, on the one hand, fund managers need to draw capital from institutional investors to maintain their market share; and on the other, institutional investors need hedge fund strategies to increase protection against market downturns. Reaching a compromise between investors’ need for information and fund managers’ propensity for secrecy is consequently of the greatest importance for both parties. The results presented in the Edhec FoHF Reporting Survey are encouraging, since fund managers and investors appear to agree globally on the content of tomorrow’s FoHF reports. Another key finding of Edhec (2005) is that market participants are progressively handling hedge fund strategies as “relative return” strategies, as opposed to “absolute return” strategies. This is a major step forward for the alternative investment industry. It derives from a better understanding of hedge funds’ return generating processes and should allow hedge fund strategies to become part of mainstream investment shortly. Market participants must however pay particular attention to the way in which relative returns are estimated. Relative returns measured with flawed benchmarks will be systematically biased. To avoid this pitfall, fund managers must design customised benchmarks that properly represent the underlying risks of the fund. This can be done with a return-based style analysis provided that the underlying style indices are not only representative of their investment universe but also investable. With third generation hedge fund indices, market participants now dispose of reliable tools enabling relative returns to be measured in the alternative arena just as they are in the traditional world. This is an encouraging point for the future development of the alternative industry.
Amenc, N., Martellini, L. and Vaissié M., 2003, Benefits and Risks of Alternative Investment Strategies, Journal of Asset Management, 4, 2, 96-118.
Amenc, N., Martellini, L., and Vaissié, M., 2004, Indexing Hedge Fund Indices, in Intelligent Hedge Fund Investing, ed. Barry Schachter, Publisher: RiskBooks.
Edhec, 2003, Edhec European Alternative Multimanagement Practices Survey Edhec, 2005, Edhec Funds of Hedge Funds Reporting Survey
Géhin, W., and Vaissié, M., 2005, Hedge Fund Strategy Benchmarks: Tricks of the light or lighthouses?, Journal of Indexes, forthcoming. Goltz, F., Martellini, L., and Vaissié, M., 2004, “Hedge Fund Indices from an Academic Perspective: Reconciling Investability and Representativity”, Working Paper, Edhec Risk and Asset Management Research Centre. Sharpe, William F., 1988, “Determining a Fund’s Effective Asset Mix”, Investment Management Review, December, p.59-69.
Sharpe, William F., 1992, “Asset Allocation: Management Style and Performance Measurement”,
Journal of Portfolio Management, Vol.18, N °2, p.7-19.