Product and Strategy Notes: Sovereign Wealth Funds; Alternative Beta Funds; and Don't Miss This New Paper on Commodities

Sovereign Wealth Funds

A lot of ink has been spilled recently on the subject of so-called "sovereign wealth funds." Traditionally, countries accumulating foreign exchange reserves invested them in low risk and very liquid government securities - e.g., U.S. Treasury and Agency bills, notes and bonds. However, as the size of some countries' reserves have grown to levels well in excess of any conceivable precautionary needs, they have established new vehicles (Sovereign Wealth Funds) to invest in a wider variety of asset classes to earn higher long-term returns. Norway was among the first nations to take this approach when its North Sea oil revenues rose; a number of Persian Gulf oil exporters have also gone this route (e.g., the Kuwait Investment Office). However, it seems that the announcement that China would also take this approach (via the launch of the China Investment Corporation) set of a new wave of analyses of SWFs' likely impact on the financial markets. By far the best of these was produced by Morgan Stanley. They estimate that, in future years, the shift of foreign exchange reserves out of government bonds and into other asset classes could push up average yields on the former by 30 to 40 basis points, while reducing the equity risk premium by 80 to 110 basis points - a not insignificant amount if you believe, as we do, that the best estimate of the ex-ante ERP (what investors expect to receive, as opposed to the ex-post return they actually realize) is between 3.5% and 4.0%.

Alternative Beta Funds

In Moliere's "Bourgeois Gentilhomme," Monsieur Jourdain was surprised to discover he had been speaking prose all his life. We have had a similar reaction to some recent articles on the "new concept" of "alternative beta" funds that claim to replicate (at least to some degree) the "higher returns, lower risk" results claimed by many hedge fund sponsors through the use of investments in "non-traditional" asset classes. Reduced to its essence the main message some of the marketing literature is that a portfolio that includes more asset classes than domestic debt and equity can deliver a superior risk/return profile. And here we were thinking that for the past ten years all we had been doing was advocating the advantages of allocating investments across a wide range of broadly defined asset classes. If we'd only know how avant-garde we were!

Unfortunately, but perhaps predictably, the pitches for these "new" products leave out a critical point - the distinctions between passive and active management, and correlated and uncorrelated returns. As we have frequently noted, the returns between broadly defined asset classes will on average be relatively low - say, .6 or less (although this will vary over market cycles and regimes). Diversifying a portfolio across these asset classes will therefore reduce risk (usually by quite a bit compared to the strawman domestic bonds and equities benchmark so beloved by many fund sponsors). This diversification can be accomplished very cheaply through the use of the index products that are available across a growing number of asset classes (or, in their absence, through the use of actively managed investments like timber REITs). So far, so good.

It may (and we emphasize the may) be possible to further improve the risk/return performance of this well diversified passive portfolio by adding to it selective actively managed investments whose returns have a low correlation with the returns on the broadly defined asset classes. Collectively, these investments are often referred to as "uncorrelated alpha strategies." A small number of uncorrelated alpha strategies are available to retail investors. Most of them are so-called "equity market neutral" strategies that take long and short positions in different companies based on their expected future performance while hedging away the broad market risk, leaving only uncorrelated company-specific risk and returns. There are also a few other uncorrelated alpha strategies available to retail investors, for example, those based on foreign exchange trading.

"Uncorrelated alpha strategy" is not synonymous with "hedge fund". Some hedge funds are indeed uncorrelated alpha strategies, but some are really "alternative beta" strategies that have quite high correlations with the returns on one or more broad asset classes (for which you pay a much higher price than going the mix of low cost index fund route). So why might a rational investor be willing to pay these high fees (we'll leave aside the irrational reasons one might pay them)? The starting point here is the shape of the distribution of historical returns on most broadly defined asset classes. They are nearly normal (i.e., bell curve shaped). Technically, many are Student's T distributions, with slightly fatter "tails" (i.e., a greater proportion of extreme events) than the normal distribution. In contrast, the shape of the historical (and, presumably, expected future) distribution of most hedge fund returns is decidedly non-normal, due to the trading strategies they employ and their use of derivative instruments like options and futures. For example, a hedge fund that writes (sells) options on extreme events (i.e., provides insurance against severe financial market events) should earn steady, low risk returns under a wide variety of market conditions, unless the disaster scenario occurs.

Today, the cutting edge of the debate over replicating hedge fund returns at a much lower cost to investors is not about "alternative beta." Rather, it is about creating low cost products whose expected distribution of future returns approximate those of much higher priced "traditional" hedge funds. Broadly speaking, three approaches have been proposed: factor models, trading rules, and direct distribution replication. Factor models use long and short positions in a limited number different traded instruments (e.g. going long a value stock index and short a growth stock index) to replicate the return distribution of a given type of hedge fund. Essentially, it is an exercise in finding the best regression model to explain historical hedge fund returns, and hoping it continues to work in the future. Trading rules are just that: automatic mechanical instructions to buy this and sell that if a given set of conditions occurs that attempt to copy the dynamic trading strategy of a hedge fund manager. The underlying statistics used to discover these rules can be much more challenging than regression. Perhaps the most controversial approach is the use of options and futures trading rules to create an expected distribution of returns that has a particular shape (e.g., mean, standard deviation, skewness and kurtosis) and sometimes low correlations with returns on one or more broad asset classes. Again, the underlying statistics are daunting.

As noted above, this is the cutting edge of the debate. Some have argued that the products now coming to market (which are still more expensive than "do-it-yourself alternative beta with a bit of uncorrelated alpha") fall well short of what a "real hedge fund" should produce. Others claim these new products are the best thing since sliced bread, and will expose the extent to which "real hedge fund" managers have overcharged their investors. Time will tell which of these opinions accumulates the most supporting evidence. In the meantime, if you want to read more about this subject, we recommend the following three papers: "Alternative Routes to Hedge Fund Return Replication" by Harry Kat; "Thoughts on Hedge Fund Return Replication" by Northwater Capital Management, and the soon to be published "The Myths and Limits of Passive Hedge Fund Replication" by Amenc, Gehin, Martellini, and Meyfredi.

Don't Miss This New Paper on Commodities

In our past writing about commodities, we have noted the controversy over the nature of the underlying return generating process. In their new paper, "The Fundamentals of Commodity Futures Returns", Gorton, Hayashi and Rouwenhorst significantly increase our understanding of this asset class. The starting point for their analysis is the Theory of Storage. The authors note that "this theory provides a link between the term structure of futures prices and the level of inventories of commodities. This link, also known as 'cost of carry arbitrage', predicts that in order to induce storage, futures prices and expected spot prices of commodities have to rise sufficiently over time to compensate inventory holders for the costs associated with storage." The authors summarize previous research findings in this area, which predict "a link between the level of inventories and future spot price volatility, since inventories act as buffer stocks which can be used to absorb shocks to demand and supply, thus dampening the impact on spot prices...At low inventories, the risk of a stock-out increases and expected future spot price volatility rises." Gorton, Hayashi and Rouwenhorst extend these previous analyses by allowing for a link between the level of inventories and a risk premium embedded in futures prices. "Given that futures contracts provide insurance against price volatility, the level of inventories is negatively related to the required risk premium on commodity futures." Using a new data set covering 31 commodities between 1996 and 2006, the authors test this hypothesis and find that, as predicted, "low inventory levels for a commodity are associated with a backwardated term structure of futures prices [where futures prices are lower than spot prices], while high levels of inventories are associated with a contangoed term structure [where futures prices are higher than spot prices]." The authors also find that "the shape of the futures curve is non-linear; the slope becomes steeper as inventories decline."

Up to now, the expected return on commodity index products that are based on baskets of futures contracts has been shown to be comprised of three main parts: (1) the return earned on collateral securities, like government bonds (since futures contracts are bought on margin); (2) the so-called "roll return", which comes from selling maturing futures contracts and purchasing new ones (when futures prices are backwardated - i.e., when the longest dated futures are priced below the current spot price - roll returns are positive); and (3) unexpected changes in spot prices (which, in theory, should net out to zero over time). Gorton, Hayashi and Rouwenhorst have now shown that the roll return really amounts to "compensation earned for bearing risk during times when commodity inventories are low." Finally, this paper's findings may lead to a reexamination of the structure of commodity index funds, and either some adjustment to their weighting rules (e.g., making them more dynamic and related to inventory levels) or the introduction of new quantitative funds that take this approach.

| This Month's Letters to the Editor: Shortcomings of Academic Research; "Black Swan" Event; and Taxable vs Tax-exempt Considerations to Whole Life Insurance | Economic Update, June 2007 | Asset Class Valuation Update | Global Asset Class Returns | Product and Strategy Notes: Sovereign Wealth Funds; Alternative Beta Funds; and Don't Miss This New Paper on Commodities | Asset Allocation and the Limits to Our Knowledge | This Month's Issue: Key Points |