William Vickrey, an emeritus professor at Columbia (for a research paper in 1961) and James Mirless, a professor at Cambridge. Although the popular press extolled Vickreys contribution as breaking fresh intellectual ground in fields as diverse as tax policy and government bond auctions, the professor denied the hyperbole. He said, "[its] one of my digressions into abstract economics.... At best its of minor significance in terms of human welfare." When interviewed, he talked instead about other unrelated work he had done, which he considered far more important.

The failure of the complex statistical models to provide much insight into current economic problems has resulted in a cutback in hiring of economists by Wall Street and major 11 8. Lawrence Myers, who before becoming a governor of the Federal Reserve, ran one of the nations most successful economic forecasting firms, St. Louis-based Macroeconomic Advisers, said, "In our firm we always thanked Robert Lucas for giving us a virtual monopoly. Because of Lucas and others, for two decades no graduate students were trained who were capable of competing with us by building econometric models that had a hope of explaining short-term output and price dynamics. We educated a lot of macroeconomists who were trained to do only two things-teach macroeconomics to graduate students and publish in the journals. . . . [These economists] dont care what happens out there. [They] dont try to build models which are consistent with the real world."*

Complicated statistical analysis is no different in the investment arena, nor should it be, since its another branch of economics. Simple assumptions are usually necessary as a platform for abstruse statistical methods. More complex assumptions, though far more descriptive of the real world, do not allow the development of the statistical analysis the researchers desire, or the academic journals will publish.

The assumption of total rationality is the mother lode of complex statistical analysis. It eliminates the need for any other psychological assumptions, which, though likely to provide better guidelines to investor behavior in the real world, would vastly complicate the analysis, and probably send it in directions completely away from the researchers paradigm.

Given the simple assumption of rationality, researchers in the best tradition of the Samuelson Revolution can merrily take off to examine how the totally rational investor will approach markets. They can then use the most complex differential equations or other statistical methodology to discover new results. Whether these assumptions have the remotest connection to reality is irrelevant. Who cares?

Figure 17-1

NFC Championships and the Market

1967-1997

Average Dow Return in NFC Champion Years: 19.3%

Average Dow Return in AFC Champion Years; 4.6%

Correlations Unlimited

The lack of realistic assumptions leads direcdy to the next level of error for EMH. Correlations are the lifeblood of the theory. Using complex statistical analysis, researchers often find correlations, which they take as proof of die existence of an important causal relationship. In the world as it is, unfortunately, correlations are frequendy pure chance. The derivation of beta, as we saw in chapter 14, is an example of correlation fishing. The researchers simply went back and looked at volatility that seemed to correlate with stock returns in the past, and assumed it was as immutable as Newtons law of gravity-it would work the same way in the future.

One of the important premises laid down in die physical sciences centuries ago was that correlation does not prove causation. Writers on sci-

The Vanishing Support for EMH

What about efficient markets themselves? Chapter 3 looked at how the theory was constructed and at some of the evidence presented in its support. Because of its impact on contemporary investors decisionmaking, it is worthwhile to examine how "efficiency" stands up under scrutiny.

According to Fama, if the necessary conditions for market efficiency are present-i.e. information is readily available to enough investors,

entific method warn that findings that appear to support a hypothesis often are pure chance, fhe correlation may seem convincing, as does the EMH finding that professionals do not outperform the market. But this is not evidence of a cause-and-effect relationship; it does not prove the hypothesis that rational investors keep prices in line with real value.

Some Wall Streeters have long been aware of this problem. Two slightly ridiculous examples taken from the investment world will show what I mean. Several decades back, the late Ralph A. Rotnem of Smith Barney noticed a correlation between the height of womens hemlines and the level of the Dow Jones Industrial Average. In the 1920s hemlines rose and stock prices followed; both fell sharply in the early 1930s. In the first years of the 1960s, both hemlines and stocks worked higher. With hot pants the market surged ahead in the early 1970s. Hemlines obviously dictate the course of the Dow Jones Industrial Average. The hypothesis seemed repeatedly vindicated, but I doubt anyone would back this hne of reasoning.

Another classic example of such a chance correlation is the NFL-stock market relationship. When a team from the original NFL has won the Super Bowl over its 31-year history, the market has gone up sharply the next year a large percentage of the time. The average is 19.3% over the 19 years theyve won. This is double the average rate of retum of stocks over time. When the AFC has been victorious, the market has gone up only 4.6%, less than half of the average stock retum. The correlation is so strong that the T-test, a test of the probability of it being chance, is less than 1 in 25. But no serious investor believes the Super Bowl victory was the cause of the rising market. Like the hemline indicator, beta, and hundreds of similar academic correlations, it is sheer chance.

But false correlations were but one of the critical challenges to the theory.