Insight Line—May 7, 2007

One of the most closely tracked investment indicators in the equity market today is the consensus earnings forecast. This forecast is derived by polling each analyst on his or her forward earnings expectation for every stock. The analysts’ responses are averaged to produce a consensus estimate. This number is then compared with the quarterly earnings results released by the company. In that this consensus of analyst’s estimates is now what’s expected by investors, any deviation from the consensus estimate may have dramatic implications for the price activity of the stock. Arguably then, it becomes critically important for investors to be aware of the consensus earnings estimates for companies in specific and the market in general.

However, when a group of people is asked to reach a consensus conclusion that is dependent upon a series of unknown criteria, the resulting forecast is likely to be more representative of group-think than of the actual outcome. Understanding the potential biases inherent in such forecasts is as important as understanding the information itself.

To illustrate the pitfalls underlying any consensus forecast, I share with you a contest designed by Richard Thaler, a leader in the study of behavioral psychology in investing. His “pick-a-number-game” appeared in the Financial Times, May 10, 1997.¹ Readers were asked to choose a whole number between zero and 100. The winning entry would be the closest to two-thirds of the average entry number. For example, if five people entered the contest and selected the numbers 10, 20, 30, 40, and 50 the winning number would be 20. In other words, the average of the five numbers is 30 and two-thirds of that number is 20. Thus, the contestant who submitted the number 20 would be the winner. The point of the game is to win; yet, in order to do so, one needs to make assumptions as to how the other entrants are going to play. This is arguably the same type of reasoning used by analysts as they strive to produce the most accurate earnings forecast. The game functions as a simple demonstration of how insufficient information gives rise to inaccurate conclusions.

Suppose, for example, you determine, through in depth analytical research of past random surveys, that the most likely number chosen should be 45. (This is somewhat akin to examining past corporate income statements.) Therefore, your entry would be 30, which is two-thirds of 45. However, upon further consideration, you would realize that others have access to the same information and they, too, might reach the same conclusion and submit the same entry. In that case you would change your entry to 20 to reflect the newly constructed model: that is, two thirds of 30 is 20. Unfortunately, the logical outcome of this exercise would lead you to submit the number 1, as it is the only possible outcome if all entrants utilize the same unemotional logic. (One could argue that zero is the ultimate entry, but it does not lie between zero and 100.)

Interestingly enough, the winning entry in Thaler’s contest was not 1, it was 13.

This game illustrates how most people, when making choices with less-than-perfect information, are prone to making errors. If everyone had chosen the number 1, which was the logical choice, then their choices would have demonstrated that people are not prone to making mistakes. Why? Because they thoughtfully reached a reasoned conclusion. However, thousands of entrants did not answer 1, and therefore success in this game rested more in the ability to understand the magnitude of the errors made by other players than in logically selecting the winning number.

We can draw some similarities between this pick-a-number exercise and the consensus earnings forecasting process. Analysts not only try to determine the winning number, but also the number that others might select. Ironically, despite such behavioral shortcomings, investors seemingly accept the consensus earnings forecast as a valid number.

In fact, such behavior was documented earlier in market history. In 1936, John Maynard Keynes published The General Theory of Employment, Interest and Money, supplying his own analogy of how understanding the thought processes of investors and analysts are involved in the creation of a successful investment strategy.

“Professional investment may be likened to those newspaper competitions in which competitors have to pick out the six prettiest faces from a hundred photographs, the prize being awarded to the competitor whose choice most nearly corresponds to the average preferences of the competitors as a whole; so that each competitor has to pick, not those faces which he finds the prettiest, but those which he thinks likeliest to catch the fancy of the other competitors, all of whom are looking at the problem from the same point of view,” Keynes writes. “It is not a case of choosing those, which to the best of anyone’s judgment are really the prettiest.”²

It is naïve to suggest the consensus earnings forecast shouldn’t matter because in the short run it most certainly does. Instead, I suggest that understanding the thought process used to reach the number is far more important. As with the pick-a-number game and the beauty contest, the successful utilization of the consensus earnings forecast is more dependent upon understanding what people think than correctly determining the error-free answer.

¹ As cited by Hirsch Shefrin, Beyond Greed and Fear, Harvard Business School Press, Boston, MA, ©2000, pages 6 and 270.

² Excerpt cited by Laszlo Birinyi, Jr., Money Flow, Mr. Market Speaks, Birinyi and Associates, Inc., ©2000, page 4.

This material has been prepared using sources of information generally believed to be reliable. No representation can be made as to its accuracy. The forecasts and opinions in this piece are not necessarily those of Van Kampen, and may not actually come to pass. Information in this report does not pertain to any Van Kampen product and is not a solicitation for any product.