One of our all-time favorite alpha-centric companies, Bridgewater Associates, released an interesting edition of its “Daily Observations” Newsletter earlier this week that caught our eye when a loyal reader passed it along to us. First, here’s why we’re fans. Says the latest edition of Daily Observations:
“As you know, we generally view the move into hedge funds as part of the evolution of money management. As we have described for many years now, the investment world shouldâ€”and willâ€”evolve towards a world of separating passive investment decisions (we call them beta) from active investment decisions (alpha). Most institutional investors continue to tie together their alpha and beta decisions (i.e. an institution typically decides how much money they want in equities and then goes out and hires equity managers to manage it). This is clearly inefficient, as the two decisions need not be linked. Instead, investors should decide which asset classes they want to be in and then overlay on top of these asset classes the best alpha managers they can find, no matter which asset class they get their alpha from. This is alpha overlay and it is a better way to run a portfolio. Cutting-edge institutions have begun to manage their assets this way, and the rest of the world will eventually adopt this superior strategy.”
Later in this issue, the firm reiterates its January 2007 observations about alternative beta. While we are fans, we felt that Bridgewater didn’t address a few key issues in its analysis. Knowing first hand, however, how the like to keep to itself, we opted to bite our tongues (Ed: to clarify, each opting to bite their own tongues).
But since our friends at HedgeWorld chose to run with the story, we felt that now might be a good time to chime in.
Bridgewater makes the accurate observation that composite hedge fund indices (the largest and most diversified of all hedge fund indices), have a 90% correlation to “a package of simple betas”. Ergo, why not just buy a much cheaper market ETF, swap or futures contract instead? In their words:
“The fact that we can just naively buy a package of simple betas and replicate hedge fund indices with a 90% correlation illustrates the point that there is a tremendous amount of beta in hedge funds, and some of the most popular strategies are the worst offenders (money is flowing into emerging market hedge funds that have extremely high beta content for example).”
But with so much idiosyncratic (uncorrelated) risk embedded in the myriad of funds and strategies contained in these mega-indices, isn’t it little wonder that these randomly distributed risks cancel each other out? Several academics, notable Harry Kat of the Cass Business School in London, have held this up as an explanation of high betas for both composite and sub-strategy hedge fund indices (see related posting). With an increasing number of hedge funds in these composite indices, one might assume idiosyncratic risks would eventually disappear.
No doubt, this suggests that hedge fund composite indices are an expensive way to buy market beta. But it does not, on its own, prove that many individual hedge funds are “selling beta at alpha prices”.
The same might be said about the high correlation between the Bridgewater beta replication model and various hedge fund sub-indices. The finding that these indices also have a high amount of beta is nothing new. Lars Jaeger and Christian Wagner of Partners Group found similarly high beta correlations in this 2005 study (see related posting). Below, we compare Jaeger and Wagner’s findings with Bridgewater’s January 2007 findings and their recent updates (comparing correlation coefficients).
(Notes: we’ve square-rooted Jaeger and Wagner’s adjusted r-squares to compare the results to Bridgewater’s correlations. For a detailed list of actual factors used by Jaeger and Wagner, refer to the appendix to their paper here.)
You will quickly note some similarities and some differences in the factor correlations reported by Jaeger and Wagner and those reported by Bridgewater. Most notably, Bridgewater’s model reports much lower correlations for both long/short equity and convertible arbitrage factor models, but much higher correlations for their global macro factor model. Event driven and fixed income seem to be about the same.
Jaeger and Wagner didn’t analyze “emerging markets” hedge funds (probably because they’re not technically “hedge” funds and its no surprise that they’re correlated with their own benchmark).
Since alpha is a zero-sum game, then as proportion of all active investors represented by hedge fund indices increases, one might expect the alpha to decrease. Its quite possible that early members of these indices outperformed for some fundamental or structural reason. But today’s constituents may be a better representation of the wider (“zero-sum”) asset management community.
However, none of this means that individual hedge funds are necessarily just riding various betas. As Professor Kat points out, research by Andrew Lo of MIT suggests individual hedge funds actually contain very little market beta (see related posting). Extrapolating from the universe of hedge funds to the individual manager is like extrapolating from the overall equity market to argue that every investor is just investing in an ETF. Markets are averages of all investors, not representative of any individual investor. The reality is that many hedge funds produce negative alpha and many produce positive alpha. On average, they produce only a modest amount of alpha. Barring any further data, that’s about all we can say from this analysis.
As the Hedgeworld piece on this edition of Bridgewater Daily Observations indicates, there is a lot more to this report. So it’s definitely worth a read if you can get your hands on it.
Editor’s Note: Anonymous tips like this make the blogosphere go around. So if you come across anything you think we ought to know about, go ahead and send it to us.