A few weeks ago, a friend asked me what I thought about a recently published study, entitled “The value of the hedge fund industry to investors, markets and the broader economy”. Written by the Centre for Hedge Fund Research of the Imperial College in London, it is published by KPMG and the Alternative Investment Management Association (AIMA). Given the pedigree of the study and its title, one does not really need to read it to arrive at a reasonable guess of its main thrust. Indeed, the conclusion (on page 25) leaves little doubt about the authors’ conviction that hedge funds have had an extremely positive impact on financial markets and the economy as a whole.
But what struck me was one particular claim. According to the study, hedge funds “provide superior performance, even on a risk-adjusted basis (page 4).” Given its obvious academic ambitions (there is a very long list of citations, starting with Markowitz), this is somewhat surprising. After all, there is a whole literature that largely suggests the opposite. It has not found any evidence that hedge funds, as a group, have generated superior performance, i.e., positive alpha, once the returns have been properly adjusted for risk. But this “oversight” is part of a general pattern. The possibility of biases in hedge fund index data is casually dismissed, statistical results are interpreted in strange, inconsistent ways and imprecise citations of studies that appear to be supportive abound. It really does not reflect well either on its authors or its sponsors.
But what about that claim of superior risk-adjusted returns? I mentioned in a recent post that it is always possible to use a sufficiently poor risk model to arrive at any desired result. This is exactly what has been done in this study. The authors pretend that the investment universe is limited to three types of assets: developed market equity, investment grade bonds and commodities. In other words, there are just three risky assets and three risk premiums. Thus, any asset that outperforms these three asset classes (after risk adjustment) must do so because of skill, i.e., the outperformance must be alpha, since it cannot be explained otherwise. So if we were to analyse an passively managed emerging markets fund in this framework, we would have to conclude that its outperformance over developed market equities of about nine percent per annum over the past ten years is pure alpha, clearly a nonsense. Any discussion on alpha must take into account the entire available investment universe and all available systematic risk premiums. Given that the investment universe of hedge funds is much broader, much of what the authors claim to be alpha are simply various risk premiums.
But rather than rant further about this and other “studies” that find alpha in strange places, it might be more useful and productive to discuss what might legitimately be considered as alpha and what should not. Therefore, I will look at alpha in a series of upcoming posts. The first will start with some definitions and introduce the notion of a replication portfolio and then look at alpha generation within traditional asset classes. In subsequent posts, I will discuss alpha generation across traditional asset classes and in the context of absolute return strategies. I will try to keep the mathematics and statistics to an absolute minimum.
As a teaser, here is a partial list of what I do not consider alpha: systematic style or size biases (e.g., a small cap bias in a portfolio managed to a blue chip benchmark), carry trades (credit long, government bonds short), “core plus” strategies (broad investment universe managed against a low risk benchmark), short option strategies, the systematic exploitation of anomalies and any performance track record that is shorter than, say, three years.