More on how the ivory towers grow so tall

27 December 2007

The investment returns of university endowments remained a hot topic throughout 2007. And their managers, such as Yale’s David Swensen, often make it into Halls of Fame of the investments world. However, little is known of the goose that lays the endowments’ golden eggs. While in Boston at the “unnamed absolute return event” earlier this year (see related posting), we saw a presentation by Cristian Tiu of the State University of New York (SUNY) at Buffalo. Tiu, together with his co-authors Keith Brown and Lorenzo Garlappi, had something to say about this fabled goose. He also brings a real-world perspective to his research from previous work at The University of Texas endowment (UTIMCO).

The Troves of Academe – or how university endowments make their money

Special to AllAboutAlpha.com by Cristian Tiu, Assistant Professor, State University of New York (SUNY) Buffalo

In a recent paper, my colleagues and I asked what is, and then what explains the performance of university endowments. We found that while the average endowment performance across the years has been around 10%, the risk adjusted returns (or alpha) are not statistically significantly different from zero. Exposure to momentum stocks seems to be the main driver of the returns. Once this is accounted for, endowments don’t seem to have – on average - any alpha producing capabilities.

However, some endowments obviously perform better than others. Why? To begin with, all endowments are relatively unconstrained, tax exempt and large enough to hold different asset classes; they are near academic centers; and there is a huge volume of academic literature on asset allocation. Hence asset allocation seems to be one of the first things to look at as a possible determinant of performance. But can asset allocation itself generate alpha? We found that the passive hedge fund index we used to benchmark the performance of hedge funds as an asset class has positive and significant alpha. Therefore, an endowment invested 100% in hedge funds really should have generated alpha. So if asset allocation is “smart” enough, it could generate alpha by itself.

But does it? Is their asset allocation “smart enough”? And does it help performance?

We found that the asset allocation is indeed diverse. That is, endowments don’t all adopt the same asset class mix. But despite this heterogeneity in asset allocation, endowments all seem to have very similar performance of their policy portfolios. Apparently, they all want to hold the same policy risk.

One good way to see this would be to look at the year 1992, where the standard deviation across all endowment policy portfolios was only 1.18%, more than four times smaller than the standard deviation across the total returns (4.94%). Since the return is a combination of the policy portfolio (or passive) return and the active return, and the former is similar across all endowments, it is only the latter which is left to determine total performance. Thus, the most active endowments, and not the endowments with the “best” asset allocation are the ones at the top of total performance charts.

If you are an endowment, the more of your total returns come from your policy portfolio, the worse off you are relative to the overall endowment universe - both in terms of raw and risk-adjusted returns. And, as the figure below shows, endowments on top of performance lists (the “winners”) which stayed active remained on top of that list. By contrast, the ones which became more passive dropped off (and some even went to the bottom of the performance chart to become “losers”). In other words, how active an endowment is seems to explain not only its performance but also its performance persistence.

Are there any other determinants of endowments’ performance? Yes, but they aren’t as intriguing. Our study also documents that larger endowments and endowments that were the first to move into alternative investments tend to out perform. However, these differences are not significant when we look at alpha instead of just raw returns. The level of payouts and the institution’s status (public or private) seemed not to matter very much.

So why do endowments bother to have different policy portfolios when the returns of those portfolios are the same anyway? We speculate that while monitoring the total risk of their policy portfolio, endowments invest in those asset classes in which their alpha generating capabilities are superior.

Do endowments optimally combine their passive asset allocation with their alpha generating capabilities? We also found that endowments may benefit from tilting their portfolio from asset class indices and toward their actual investments. If endowments want their Sharpe ratios (or Sortino ratios) to be optimal, they may benefit from trading about 90% (50%) or their policy portfolio in exchange for 90% (50%) more of their actual portfolio. Such a transaction will not change the asset allocation, but only the extent of active management in the portfolio.

In conclusion, if you are an endowment manager do not try to copy the Harvards or the Yales of the world. In fact, we looked at what consequences such herding behavior would have: implementing passively the asset allocations of the top 25, 10 or even 5% performers produces a portfolio whose returns rank in the bottom 30-40% of the endowment universe. Instead, focus on those asset classes where your alpha generating capabilities are best. This doesn’t necessarily mean mean that you should take more risk, or that you should even increase your exposure to alternative investments. It only means that you would likely be best served by focusing more on what you know instead of on the accepted norm. At the end of the day, this is really very basic and reasonable advice.