While the Edhec Risk and Asset Management Research Centre is bigger, smarter and better–connected than AllAboutAlpha.com, both organizations share the same genetic blueprint. Edhec’s Lionel Martellini confirmed this fact this morning in his introduction to day two of the university-affiliated organization’s annual hedge fund conference in London. Edhec sees alternative investments as a (the) central issue in institutional portfolio management and believes that we need stronger links between research and practice.
Some would say that this overstates the importance of emerging asset classes. But Martellini points to two unique aspects of alternative investments that are fundamentally unique: non-normality and data integrity. The introduction of higher moments such as skew and kurtosis le to an explosion of data since correlation was now joined by co-skew and co-kurtosis. These co-moments add exponentially to the amount of information required for portfolio construction. To compound things, monthly data that is susceptible to “smoothing” makes alternative investment research a whole new ball game for academics and practitioners.
The problem, Martellini says, is that a growing body of academic insight is not making its way into practice as quickly as it is being generated. Enter “Edhec-Risk” (and its blog corollary AllAboutAlpha.com). Edhec’s goal is to increase the dialogue and feedback between academic and practitioner communities.
Martellini then presented Edhec’s recent research on liability driven investing (LDI). He explained how pension plans need to separate their portfolios into two parts – one that aims to match future pension liabilities (the “Liability Hedging Portfolio (LHP)”) and one that aims to generate excess returns in order to keep contributions down (the “Performance Seeking Portfolio (PSP)” ). He refers to this concept as the “separation theorem”.
Martellini explains that inflation remains one of the key risks in the liability stream than requires hedging by the LHP. So most pensions use TIPS and other inflation-linked securities to create the LHP and things like equities for the PSP. But his research suggests that pension funds may be able to use alternative investments that both hedge inflation and produce excess returns as long as the pension fund is willing to think about the long term.
Over the short term, commodities and real estate provide a very imperfect hedge against inflation. However, over the long term (10 years), they actually provide a very good hedge against inflation – while also providing their own excess risk adjusted returns. This allows pension funds to essentially kill two birds with one stone, thus allowing for more efficient use of the fund’s resources.
In the spirit of dialogue and feedback from practitioners, a panel made up of the heads of 3 of Europe’s biggest pension funds then probed Martellini’s thesis. All expressed interest in the idea – mixed with the healthy scepticism that often accompanies new academic research like this. Some, such as Sally Bridgeman of BP’s pension plan, described themselves as “purists” who were hesitant to add alternative investments to the liability-hedging portfolio (LHP). APG’s Jaap Maassen needed more convincing that real estate and commodities were an adequate inflation hedge – even in the long run. Morgan Stanley’s Joseph McDonnell noted that pensions were preoccupied at the moment with urgent issues such as funding gaps and a need to rebalance after the recent market crash.
Some Mud Sticking to Barn Wall
This afternoon featured some pretty technical material – much of which went over my head. But as they say, “if you throw enough mud against a barn wall, some of it is going to stick…”
The first afternoon workshop featured “enhancements to core-satellite investing” based on a paper that was published by Edhec in 2004. The idea is based on the premise that there are times when hedge fund investors would want to simply track the market (i.e. in up markets) and times when they should want to avoid the market like the plague (recently, for example). In this model, the market is considered to be the “core” and hedge funds are considered to be the “satellite”.
Without a crystal ball, of course, no one can know exactly when to jump horses between core and satellite. But using a form of constant proportion portfolio insurance (CPPI), you can adjust allocations in a manner that results in capturing some of the upside of the market and some of the upside of hedge funds. The explanation of the dynamic trading process that creates a protection against major drawdowns in either the core or satellite sounded vaguely like the dynamic trading that underscores the distributional replication approach developed by Harry Kat for hedge fund replication.
Finally today, Edhec’s Rene Garcia taught a session on performance measurement. He argued that most traditional approaches used for performance measurement such as various risk/return ratios, alphas, and downside-only measures all neglect higher moments such as skew and kurtosis (although some might argue the Omega ratio implicitly captures higher moments). Ergo, the non-linearity and non-normality of hedge fund returns requires a new approach.
Last year at this event, Garcia spoke about a technique he developed to locate a “kink” in the otherwise linear relationship between the market and various hedge fund strategies – in other words, a regression line on a scatter plot that has a different slope on one end than it does on the other. (Example from his recently-published paper below – although the blob of data points looks like a Rorschach Diagram to us.)
This year, Garcia amped it up some more – proposing a “stochastic discount factor” technique that he describes as an improvement on traditional measures of alpha. Without getting into the details (which flew unimpeded right over my head), the basic idea is to add a utility function to a non-linear model of hedge fund returns.
If you want more, check out the abstract contained in the program for this upcoming symposium on “hedge fund econometrics”:
“Hansen and Jagannathan (HJ, 1991) provided bounds on the volatility of Stochastic Discount Factors (SDF) to diagnose and test asset pricing models. These non-parametric bounds reflect a duality between mean-variance frontiers for SDFs and for portfolios of asset returns. We propose information bounds that minimize general convex functions of SDFs taking into account higher moments of returns. These bounds reflect a duality with finding optimal portfolios of asset returns with general HARA utility functions. We use these information bounds to diagnose consumption-based asset pricing models, to analyze the pricing of size portfolios, and to assess the performance of hedge funds.”
Brain firmly fried, I welcomed the conclusion of Edhec’s Annual Alternative Investment Days 2008. If you’re interested in any of the material covered over the past 2 days, keep an eye on Edhec’s website for conference slide ware.