Benchmark Blues. (Allocation of Capital in Global Emerging Markets)

Article excerpt

Agency problems in the construction of global emerging market benchmarks and the allocation of global capital.

It is well known that the ratio of "Buy" recommendations to "Sell" recommendations proposed by Wall Street security analysts is at least 95 to 5. This phenomenon arises in part because of a classic incentive mismatch problem: The incentive structure of sell-side analysts does not neatly coincide with the interests of their clients. Two and a half trillion dollars in losses since March 2000 give investors additional cause to ponder this phenomenon.

What is perhaps less well known is that behavioral factors may influence the construction of benchmarks used by global equity investors. Changes in benchmarks frequently drive global institutional equity capital flows. A poorly constructed benchmark can do a disservice to investors and to a stable, well-functioning global capital market.

What is a benchmark? A benchmark plays many roles: a default holding in the absence of information; a proxy for an asset class; and a performance target for active portfolio managers. Suppose you are a Japanese equity investor. You believe that markets are reasonably efficient, that you have no special information, and that active managers will have a difficult time outperforming the market portfolio over time. Thus, you wish to hold the U.S. market. But what is the U.S. market? The S&P 500? The Wilshire 5000? Or the Russell 3000? Each of these potential benchmarks has different structural characteristics and implications for the investor's ultimate portfolio.

The benchmark serves not only as a proxy for the asset class, but also as a performance target for active managers. An investor can generally obtain the systematic return of an asset class through indexing--passive management. The decision to invest in an asset class--say emerging markets--is distinct from the decision to hire an active manager.

The Capital Asset Pricing Model, still a robust theory after all these years, suggests that investors are rational and that the market portfolio is the most efficient, that is, provides the highest rate of expected return for a given level of anticipated risk. Many financial economists likewise agree that the world market portfolio is the most efficient over the long term. Equity returns tend to revert to a global mean. For example, over the past thirty-one years in dollars, the Morgan Stanley Capital International (MSCI) Japan and Europe Indices and the S&P 500 have returned within 110 basis points of one another (about 12. 2 percent per annum compounded.) Despite this evidence, factors like home country bias, regulation, transaction costs, and information gaps all inhibit a world market weighting by major institutional investors. For example, by regulation, Canadian pension funds can invest only 30 percent of the book value of their assets abroad.

But whether the world market portfolio is in fact the truly efficient portfolio is not free from doubt. To begin with, as Richard Roll argued, the true market portfolio is unobservable. In addition, as Will Goetzmann and Phillipe Jorion observed, many global equity markets have suffered long periods of disruption--the Polish market under the Soviet occupation, for example. They pointed out that many emerging markets of today are in fact "re-emerging markets." As well, investors must take into account transaction costs, which today average about 130 basis points one way for an emerging markets investor. These and other phenomena led Goetzmann and Jorion to argue that: "The high equity premium obtained for U.S. equities therefore appears to be the exception rather than the rule." (The equity premium is the amount by which the return of stocks exceeds the return of bonds or cash.)

The definition of benchmarks is important because it shapes institutional investors' view of the feasible investment opportunity set. Benchmark inclusion sends a positive signal to investors, and investors make decisions based on benchmark information. …