Against the Gods: The Remarkable Story of Risk Read online

  Meir Statman began to be interested in nonrational behavior when, as a student of economics, he noted that people reveal a tendency to look at problems in pieces rather than in the aggregate. Even qualified scholars in reputable journals reached faulty conclusions by failing to recognize that the whole is the product of interaction among its parts, or what Markowitz called covariances, rather than just a collection of discrete pieces. Statman soon recognized that the distortions caused by mental accounting were by no means limited to the public at large.

  Statman cites a case that he found in a journal about a homeowner's choice between a fixed-rate mortgage and a variable-rate mortgage.8 The paper dealt with the covariance between mortgage payments and the borrower's income and concluded that variable rates were appropriate for people whose income generally keeps pace with inflation and that fixed rates were appropriate for people whose incomes is relatively constant. But Statman noted that the authors ignored the covariance between the value of the house itself and the two variables mentioned; for example, an inflationary rise in the value of the house might make a variable-rate mortgage easy enough to carry regardless of what happened to the homeowner s income.

  In 1981, Hersh Shefrin, a colleague of Statman's at Santa Clara University, showed Statman a paper titled "An Economic Theory of Self-Control," which Shefrin had written with Thaler.9 The paper made the point that people who have trouble exercising self-control deliberately limit their options. People with weight problems, for example, avoid having a cake ready at hand. The paper also noted that people choose to ignore the positive covariance between their mortgage payments and the value of their house as borrowing collateral; they view the house as a "piggy bank" that is not to be touched, even though they always have the option to borrow more against it and, thanks to home equity loans, sometimes do.* After reading this paper, Statman too was off and running.

  A year later, Shefrin and Statman collaborated on an illuminating paper on behavioral finance titled "Explaining Investor Preference for Cash Dividends,"10 which appeared in the Journal of Financial Economics in 1984.

  Why corporations pay dividends has puzzled economists for a long time. Why do they pay out their assets to stockholders, especially when they themselves are borrowing money at the same time? From 1959 to 1994, nonfinancial corporations in the United States borrowed more than $2 trillion while paying out dividends of $1.8 trillion.t They could have avoided nearly 90% of the increase in their indebtedness if they had paid no dividends at all.

  From 1959 to 1994, individuals received $2.2 trillion of the dividends distributed by all corporations, financial as well as nonfinancial, and incurred an income-tax liability on every dollar of that money. If corporations had used that money to repurchase outstanding shares in the open market instead of distributing it in dividends, earnings per share would have been larger, the number of outstanding shares would have been smaller, and the price of the shares would have been higher. The remaining stockholders could have enjoyed "home-made" dividends by selling off their appreciated shares to finance their consumption and would have paid the lower tax rate on capital gains that prevailed during most of that period. On balance, stockholders would have been wealthier than they had been.

  To explain the puzzle, Shefrin and Statman draw on mental accounting, self-control, decision regret, and loss aversion. In the spirit of Adam Smith's "impartial spectactor" and Sigmund Freud's "superego," investors resort to these deviations from rational decision-making because they believe that limiting their spending on consumption to the amount of income they receive in the form of dividends is the way to go; financing consumption by selling shares is a no-no.

  Shefrin and Statman hypothesize the existence of a split in the human psyche. One side of our personality is an internal planner with a long-term perspective, an authority who insists on decisions that weight the future more heavily than the present. The other side seeks immediate gratification. These two sides are in constant conflict.

  The planner can occasionally win the day just by emphasizing the rewards of self-denial. But when the need arises, the planner can always talk about dividends. As the light fixture "hides" the liquor bottle from the alcoholic, dividends "hide" the pool of capital that is available to finance immediate gratification. By repeatedly reciting the lesson that spending dividends is acceptable but that invading principal is sinful, the planner keeps a lid on how much is spent on consumption.

  Once that lesson is driven home, however, investors become insistent that the stocks they own pay a reliable dividend and hold out a promise of regular increases. No dividend, no money to spend. No choice. Selling a few shares of stock and the receipt of a dividend are perfect substitutes for financing consumption in theory-and selling shares even costs less in taxes-but in a setting of self-control contrivances, they are far from perfect substitutes in practice.

  Shefrin and Statman ask the reader to consider two cases. First, you take $600 of dividend income and buy a television set. Second, you sell $600 of stock and use the proceeds to buy a television set. The following week, the company becomes a takeover candidate and the stock zooms. Which case causes you more regret? In theory, you should be indifferent. You could have used the $600 of dividend income to buy more shares of the stock instead of buying the TV. So that was just as costly a decision as your decision to sell the shares to finance the TV. Either way, you are out the appreciation on $600 worth of shares.

  But oh, what a horror if dividends are cut! In 1974, when the quadrupling of oil prices forced Consolidated Edison to eliminate its dividend after 89 years of uninterrupted payments, hysteria broke out at the company's annual meeting of stockholders. Typical was one question put to the company chairman, "What are we to do? You don't know when the dividend is coming back. Who is going to pay my rent? I had a husband. Now Con Ed has to be my husband." This shareholder never gave a thought to the possibility that paying dividends out of losses would only weaken the company and might ultimately force it into bankruptcy. What kind of a husband would that be? Selling her shares to pay the rent was not one of the options she allowed herself to consider; the dividend income and the capital were kept in separate pockets as far as she was concerned. As in a good marriage, divorce was inadmissible.

  In a discussion of Shefrin and Statman's work, Merton Miller, a Nobel Laureate at the University of Chicago and one of the more formidable defenders of rational theory, made the following observation about investors who do not rely on professional advisers:

  For these investors, stocks are usually more than just the abstract "bundles of returns" of our economic models. Behind each holding may be a story of family business, family quarrels, legacies received, [and] divorce settlements ... almost totally irrelevant to our theories of portfolio selection. That we abstract from all these stories in building our models is not because the stories are uninteresting but because they may be too interesting and thereby distract us from the pervasive market forces that should be our principal concern."

  In Chapter 10, I mentioned a paper titled "Does the Stock Market Overreact?" which Thaler and one of his graduate students, Werner DeBondt, presented at the annual meeting of the American Finance Association in December 1985. There this paper served as an example of regression to the mean. It can also serve as an example of the failure of the theory of rational behavior.

  I was a discussant at the session at which Thaler and DeBondt presented their findings, and I began by saying, "At long last, the academic world has caught up with what investors have known all along."12 Their answer to the question posed by the title was an unqualified "Yes."

  As an example of Prospect Theory, Thaler and DeBondt demonstrated that, when new information arrives, investors revise their beliefs, not according to the objective methods set forth by Bayes, but by overweighting the new information and underweighting prior and longer term information. That is, they weight the probabilities of outcomes on the "distribution of impressions" rather than on an objective calculation bas
ed on historical probability distributions. As a consequence, stock prices systematically overshoot so far in either direction that their reversal is predictable regardless of what happens to earnings or dividends or any other objective factor.

  The paper provoked criticism from members of the audience who were shocked by this evidence of irrational pricing. The dispute continued over a number of years, focusing primarily on the manner in which Thaler and DeBondt had gathered and tested their data. One problem related to the calendar: an excessive proportion of the profits from selling the winners and buying the losers appeared in the one month of January; the rest of the year appeared to have been about break-even. But different tests by different folks continued to produce conflicting results.

  In May 1993, a related paper entitled "Contrarian Investment, Extrapolation, and Risk" appeared under the auspices of the prestigious National Bureau of Economic Research.13 The three academic authors, Josef Lakonishok, Andre Shleifer, and Robert Vishny, provided an elaborate statistical analysis which confirmed that so-called "value" stocks-stocks that sell at low prices relative to company earnings, dividends, or assets-tend to outperform more highly valued stocks even after adjustments for volatility and other accepted measures of risk.

  The paper was memorable for more than the conclusion it reached, which was not original by any means, nor for the thoroughness and polish of the statistical presentation. Its importance lay in its confirmation of Thaler and DeBondt's behavioral explanation of these kinds of results. In part because of fear of decision regret and in part because of myopia, investors price the stocks of troubled companies too low in the short run when regression to the mean would be likely to restore most of them to good health over the long run. By the same token, companies about which recent information has indicated sharp improvement are overpriced by investors who fail to recognize that matters cannot get better and better indefinitely.

  Lakonishok, Shleifer, and Vishny have certainly convinced themselves. In 1995, they launched their own firm to manage money in accordance with their contrarian model.

  Thaler never recovered from his early fascination with that "very interesting" disparity between prices for which people were willing to buy and sell the identical items. He coined the expression "endowment effect" to describe our tendency to set a higher selling price on what we own (are endowed with) than what we would pay for the identical item if we did not own it.*

  In a paper written in 1990 with Daniel Kahneman and another colleague, Jack Knetsch, Thaler reported on a series of classroom experiments designed to test the prevalence of the endowment effect.t4 In one experiment, some of the students were given Cornell coffee mugs and were told they could take them home; they were also shown a range of prices and asked to set the lowest price at which they would consider selling their mug. Other students were asked the highest price they would be willing to pay to buy a mug. The average owner would not sell below $5.25, while the average buyer would not pay more than $2.25. A series of additional experiments provided consistent results.

  The endowment effect is a powerful influence on investment decisions. Standard theory predicts that, since rational investors would all agree on investment values, they would all hold identical portfolios of risky assets like stocks. If that portfolio proved too risky for one of the investors, he could combine it with cash, while an investor seeking greater risk could use the portfolio as collateral for borrowings to buy more of the same.

  The real world is not like that at all. True, the leading institutional investors do hold many stocks in common because the sheer volume of dollars they must invest limits them to stocks with the highest market values-stocks like General Electric and Exxon. But smaller investors have a much wider range of choice. It is rare indeed to find two of them holding identical portfolios, or even to find significant duplication in holdings. Once something is owned, its owner does not part with it lightly, regardless of what an objective valuation might reveal.

  For example, the endowment effect arising from the nationality of the issuing company is a powerful influence on valuation. Even though international diversification of investment portfolios has increased in recent years, Americans still hold mostly shares of American companies and Japanese investors hold mostly shares of Japanese companies. Yet, at this writing, the American stock market is equal to only 35% and the Japanese to only 30% of the world market.

  One explanation for this tendency is that it is more costly to obtain information on securities in a foreign market than it is to obtain information on securities in the home market. But that explanation seems insufficient to explain such a great difference in holdings. There must be more compelling reasons why investors are so reluctant to hold securities domiciled in markets that account for 65% to 70% of the investible universe.

  A masterful study of the influence of the endowment effect on international investing was carried out in 1989 by Kenneth French, then at the University of Chicago and now at Yale, and James Poterba at MIT.15 The target of their inquiry was the absence of cross-border ownership between Japanese and American investors. At that time, Japanese investors owned just over 1% of the U.S. stock market, while American investors owned less than 1% of the Tokyo market. A good deal of activity was taking place across the borders; substantial buying and selling of American stocks went on in Japan and of Japanese stocks in the United States. But net purchases on either side were tiny.

  The result was a striking distortion of valuations across the markets. French and Poterba's calculations indicated that the small holdings of Japanese stocks by U.S. investors could be justified only if the Americans expected annual real (inflation-adjusted) returns of 8.5% in the United States and 5.1% in Japan. The small holdings of American stocks by Japanese investors could be justified only if the Japanese expected real annual returns of 8.2% in Japan and 3.9% in the United States. Neither taxation nor institutional restrictions were sufficient to explain disparities that would set von Neumann spinning in his grave.* Nor could theories of rational investor decision-making explain them. The endowment effect must be the answer.*

  The evidence presented in this chapter gives only a hint of the diligence of the Theory Police in apprehending people in the act of violating the precepts of rational behavior. The literature on that activity is large, growing, and diverse.

  Now we come to the greatest anomaly of all. Even though millions of investors would readily plead guilty to acting in defiance of rationality, the market-where it really counts-act as though rationality prevailed.

  What does it mean to say "where it really counts"? And, if that is the case, what are the consequences for managing risk?

  Keynes provides a precise definition of what it means to say "where it really counts." In a famous passage in The General Theory of Employment, Interest and Money, Keynes describes the stock market as, "... so to speak, a game of Snap, of Old Maid, of Musical Chairs-a pastime in which he is victor who says Snap neither too soon nor too late, who passes the Old Maid to his neighbor before the game is over, who secures a chair for himself when the music stops."16

  Keynes's metaphor suggests a test to determine whether the market acts as though rationality prevails, where it counts: the prevalence of nonrational behavior should provide endless opportunities for rational investors to say Snap, to pass on the Old Maid, or to seize a chair ahead of investors on the run from the Theory Police. If those opportunities do not present themselves, or are too brief to provide an advantage, we might just as well assume that the market is rational even though we recognize that many irrational forces are coursing through it. "Where it counts" means that there are very few opportunities to profit by betting against irrational investors, even though there is so much evidence of their presence in the market. Where it counts, the market's behavior conforms to the rational model.

  If all investors went through the identical rational thinking process, expected returns and adjustments for risk would look the same to everyone in possession of the same info
rmation at the same moment. In the unlikely event that a few investors succumbed to nonrational behavior, they would end up buying high and selling low as betterinformed investors were driving prices back to a rational valuation. Otherwise, prices would change only when new information became available, and new information arrives in random fashion.

  That is how a fully rational market would work. No one could outperform the market as a whole. All opportunities would be exploited. At any level of risk, all investors would earn the same rate of return.

  In the real world, investors seem to have great difficulty outperforming one another in any convincing or consistent fashion. Today's hero is often tomorrow's blockhead. Over the long run, active investment managers-investors who purport to be stock-pickers and whose portfolios differ in composition from the market as a whole-seem to lag behind market indexes like the S&P 500 or even broader indexes like the Wilshire 5000 or the Russell 3000. Over the past decade, for example, 78% of all actively managed equity funds underperformed the Vanguard Index 500 mutual fund, which tracks the unmanaged S&P 500 Composite; the data for earlier periods are not as clean, but the S&P has been a consistent winner over long periods of time.

  There is nothing new about this pattern. In 1933, Alfred Cowles, a wealthy investor and a brilliant amateur scholar, published a study covering a large number of printed financial services as well as every purchase and sale made over four years by twenty leading fire insurance companies. Cowles concluded that the best of a series of random forecasts made by drawing cards from an appropriate deck was just as good as the best of a series of actual forecasts, and that the results achieved by the insurance companies "could have been achieved through a purely random selection of stocks."" Today, with large, sophisticated, and well-informed institutional investors dominating market activity, getting ahead of the market and staying there is far more difficult than it was in the past.