A November 2006 article in the Economist1 and a report by Merrill Lynch2 in October both referred to research by myself and colleagues on the nature of hedge fund returns and the resulting flows into different types of fund. They drew in particular on a working paper by Fung, Hsieh, Naik and Ramadorai (available for download on the BNP Paribas Hedge Fund Centre’s website, www.london.edu/hedgefunds.html). The paper is entitled “Hedge Funds: Performance, Risk, and Capital Formation”. In this short article I want to describe briefly the questions we were asking, some of the results we obtained, and to speculate on how, if we are right, the hedge fund industry might evolve.
We particularly wanted to focus firstly on whether the stereotype of hedge fund returns as uncorrelated with market-related returns is correct. Consultants, for example, recommend hedge funds as part of a defined benefit pension fund portfolio because of their supposedly low correlation with existing investments in equity and bond markets. Hedge funds also provide access, according to consultants, to active management strategies that are difficult to identify within the traditional, long-only investment management community. Secondly, we wanted to know whether the market could identify high alpha funds and whether capital tended to flow to these, or whether the distinction between pure alpha returns and market-related or “beta” returns had no effect on subsequent capital flows. Thirdly we wanted to know whether those flows into successful funds impacted their future capacity to generate high returns. Lastly, the ten years covered by our dataset – from January 1995 to December 2004 – was a turbulent period, punctuated by the collapse of LTCM in September 1998, a bull market in technology stocks that burst in March 2000, and a subsequent bear market. It also saw the enormous growth in the popularity of hedge funds and a change in clientele, from almost exclusively high net worth individuals towards institutions such as endowment and pension funds. We wanted to know therefore how the mix of returns had changed over that period, and whether in absolute terms they had grown or diminished.
This is an ambitious programme. Firstly, however, a word about our data. We used fund of hedge funds (FOFHs) data, for the simple reason that these reflect far more faithfully the true investment experience of end-investors. Hedge fund indices exist, but apparent performance is often biased upwards by various reporting (and non-reporting) influences.
To isolate beta effects from alpha, we constructed a factor model measuring the sensitivities of FOHFs to the equity and bond markets, and also to the small-cap effect and to default risk, as measured by the corporate bond spread over treasuries. To these four factors we added three momentum factors that measured a trend-following exposure in the currency, commodity and bond markets. Thus we defined beta for the purposes of this study as the cumulative effect of exposure to these seven factors, each of them relatively easily replicable in a mechanical way, and therefore unlikely to command a premium fee for managers – at least not in the long run.
We discovered that over the whole period, the proportion of FOHF returns explained by our factor model averaged 74%. Clearly this average masks variation among funds, through time, and across our different factors, but it is nevertheless a high proportion of returns. It confirms results found by others.
My co-authors and I were also interested in whether there was a statistically significant break in the patterns of returns between our three sub-periods. Our results confirmed that the period before September 1998 differed significantly – in terms of factor exposures – from the subsequent technology-led bull market up to March 2000, which itself differed from the remaining period to December 2004. Thus while it may be the case that market-related factors dominate FOHF returns, the pattern of factor exposures will vary through time. Interestingly, only during the dot.com boom from 1998 to 2000 did the average fund show statistically significant alpha.
We divided our dataset further into funds with a two-year record of significant alpha, and those without. We called them the have-alpha and the beta-only groups. The composition of each group was then revised annually on a rolling basis. We were interested to know whether the have-alpha funds attracted more new money than the beta-only group. This is indeed what we found. The market for investment management services clearly functions, in that buyers compete to identify successful managers and reward them with more business. Unfortunately, however, it appears from the data that there are diminishing returns to scale. Have-alpha funds, which expanded at an above-average rate, tended to do less well in subsequent periods than those that expanded at a below-average rate, albeit better in alpha terms than the beta-only group. There is limited capacity therefore for successful funds to expand and continue doing well.
If capacity constraints operate at the individual fund level, what happens at the industry level? After all, as we noted earlier, the industry has expanded considerably in recent years, suggesting that – on average – even high alpha funds would find it more difficult to add value. Our data show that in our first sub-period, before the LTCM crash, the average have-alpha fund added 5.6% per annum. In contrast, the same group in the post 2000 period added only 2.2% per annum.
Our results may have – indeed, may already be having – an impact on the shape of the industry. If substantial parts of hedge funds returns are not unique to the manager, but represent scalable market factors, where there are no or few capacity constraints, then there is no reason why these return patterns should not be available to investors as passive funds – passive that is in the sense that they are created mechanically. We do not mean traditional equity and bond index funds, but relatively simple long/short strategies that exploit market-wide systematic risk factors. These differ from the un-hedgeable and idiosyncratic alpha returns of our have-alpha group. Passive vehicles would have the advantage of being more transparent, though they are unlikely to command the fee levels we see in the industry today. Indeed a number of major investment banks are thought to be developing investment products along these lines already. The capacity constrained, high fee sector is likely to co-exist alongside these new products. Such differentiation along investment-character and fee-level lines has been a feature of the investment management industry for many years. The hedge fund industry is no exception.
Professor Narayan Naik also teaches on the open enrolment Hedge Fund programme which is a 3-day course at London Business School, 12–14 March 2007. For further details, visit www.london.edu/fpeur/.
1Hedge Funds: Rolling in it” Economist, 18 November, 2006
2“Replicating Hedge Fund Returns” by Bowler, Ebens, Davi, and Amanti, Merrill Lynch 18 October 2006