Departure from Fundamental Value: Short-Term or Long-Term?
One of the most striking claims of behavioral finance is that heuristic-driven bias and frame dependence can cause prices to deviate from fundamental values for long periods. Shiller (1979, 1981) argues that there is more volatility in stock markets and bond markets than would be the case if prices were determined by fundamentals alone. His analysis vividly illustrates the length of time stock price and fundamental value can part company.
Shiller computed what the fundamental value of stocks would have been over time for an investor who had perfect foresight about the future value of dividends. He then compared fundamental value and prices.6 Figure 4-3 depicts the timelines for two indexes, starting in 1925. The figure is scaled to adjust for the long-term historical growth rate of stock prices. One line depicts the index for fundamental value, while the other shows the index for actual stock prices.
The years 1929 and 1987 were crash years, and, as I discussed in chapter 2, 1973 was the beginning of a long bear market. Consider how these events are portrayed in figure 4-3. Prior to 1929 and 1973, price lies well above fundamental value. Shortly after those years price falls below fundamental value. The lesson here is that prices move away from fundamental value for long periods, but eventually revert.
On December 3, 1996, Shiller, along with John Campbell, expressed his views about the market in joint testimony before the Board of Governors of the Federal Reserve system (Campbell and Shiller 1998). Apparently, their testimony had some influence. Two days later, on December 5, Alan Greenspan, chairman of the Federal Reserve Board, shocked global markets when he used the term “irrational exuberance” to describe the state of the U.S. stock market.7
Figure 4-3 Fundamental Value versus Actual Price, 1925–1999 Stock prices tend to stray from fundamental value for long periods of time. The period after 1994 is especially striking.
What did Shiller and Campbell tell the Federal Reserve Board? They explained that historically, when the dividend yield (D/P) has been low and the price-to-earnings ratio (P/E) high, the return to holding stocks over the subsequent ten years has tended to be low. This should not be surprising. The earnings yield is E/P, the inverse of P/E. In a rationally priced market, dividend yields and earnings yields form the basis of stock returns.8 Recall the question in chapter 2 concerning reinvested dividends and the Dow Jones Industrial Average. One of the lessons from that question is that when it comes to long-run stock returns, compounded dividends swamp stock price.9 The future course of earnings and dividends would have to be dramatically better than in the past to rationalize high subsequent stock returns in a low D/P and E/P environment.10
To place the Campbell-Shiller argument into context, let me point out that the historical mean for the dividend yield is 4.73 percent. But in late 1996, it was an extremely low 1.9 percent. The historical mean for P/E is 14.2. Moreover, for most of the time since 1872, P/E has moved in the range of 8 to 20. Until recently, its peak of 26 dated back to 1929; however, in December 1996, the P/E stood at 28. In their joint testimony, Campbell and Shiller predicted that between 1997 and 2006, the stock market would lose about 40 percent of its real value.11
Shiller’s 1981 work generated considerable controversy and has been the subject of many debates.12 The central question asks: Do stock prices only change in response to fundamentals? Most of the debate has focused on technical details, which Robert Merton described in his survey of market efficiency (see chapter 1).
Merton wrote his survey in 1986 and published it in 1987. Of course, 1987 was a propitious year for debating questions about stock prices and fundamentals, given that the stock market crashed in October. Immediately thereafter, Shiller conducted a major investor survey to identify the information that led stock prices to lose 25 percent of their value in the course of a single day. Shiller (1990) documents that the market crash of 1987 occurred in the absence of any major news about changing fundamentals.13
With this in mind, I draw your attention to the extreme right of figure 4-3. Look at the period 1995 through early 1999, during which price rose well above Shiller’s estimate of fundamental value. Note that figure 4-3 ends in January 1999, at which time the dividend yield on the S&P 500 had fallen further, to 1.26 percent.14 The trailing P/E stood at a record 32.7!15
Like Campbell and Shiller, some Wall Street strategists also placed the market’s rise of 1995 and 1996 in historical context. For example, Edward Kerschner, a strategist at PaineWebber, had been bullish during the market’s climb in 1995 and 1996 but then turned bearish. In a June 1997 article that appeared in Barron’s he stated, “In ′87 the market went to 135 percent of fair value, and in ′73 we got to 155 percent of fair value.” Kerschner went on to note that according to his P/E model, the market was 15 percent overvalued, making it the third most expensive market in a quarter of a century.16
In a related vein, Charles Lee, James Myers, and Bhaskaran Swaminathan (1999) have devised an intrinsic measure of the Dow Jones Industrial Average based upon the book-to-market ratio, the long-term return on equity, expected earnings growth, and interest rates. In mid-June of 1997, when Kerschner indicated that the market was 15 percent overvalued, the Lee-Myers-Swaminathan measure indicated that the Dow was 42 percent overvalued.
Kerschner has not been as steadfast in his stated view as some academic scholars have been. Strategists are subject to different pressures than scholars. During strong bull markets, bears become unpopular on Wall Street.17 In early 1999, as the P/E of the S&P 500 hit a record high 32, the Wall Street Journal reported Kerschner’s view as “[S]tocks may have gotten a little ahead of themselves … but he thinks the fundamentals dictate future stock gains.”18
Overconfidence
Remember the case of Royal Dutch/Shell discussed in chapter 1? The market values of Royal Dutch and Shell Transport were misaligned relative to fundamental values. Yet, in attempting to exploit the mispricing, hedge fund Long-Term Capital Management (LTCM) managed to lose heavily. There is a moral to that story. Not every instance of mispricing leads to $20 bills on the sidewalk waiting to be picked up, or even to nickels, for that matter.
In fact, there are many behavioral lessons in the saga of LTCM. It does appear that their early success can be attributed to the exploitation of mispricing. At the same time, mispricing does get reduced as investors trade to exploit the associated profit opportunities. And investors do learn, albeit slowly; thus profit opportunities that had existed in 1994 through 1996 in the derivatives markets dried up in 1997. In response, LTCM began to take other kinds of positions, such as bets on the movements of foreign currency movements. Myron Scholes is reported to have been critical of these trades, asking his LTCM colleagues questions like “What informational advantage do we have over other traders?”
Scholes asked an eminently sensible question. Here is an analogy: “How good a driver are you? Relative to the drivers you encounter on the road, are you above average, average, or below average?”
Between 65 and 80 percent of the people who answer the driver question rate themselves above average. Of course, we all want to be above average, but only half of us are! So, most people are overconfident about their driving abilities. I suspect that investors are about as overconfident of their trading abilities as they are about their driving abilities.
There are two main implications of investor overconfidence. The first is that investors take bad bets because they fail to realize that they are at an informational disadvantage. The second is that they trade more frequently than is prudent, which leads to excessive trading volume. See my work with Statman (Shefrin and Statman 1994) and Terrance Odean (1998b).
A Wall Street Journal article describing the experience of Long-Term Capital Management quotes Nobel laureate William Sharpe.
“Most of academic finance is teaching that you can’t earn 40 percent a year without some risk of losing a lot of money,” says Mr. Sharpe, the former Stanford colleague of Mr. Scholes. “In some sense
, what happened is nicely consistent with what we teach.”19
Proponents of behavioral finance, especially those who manage money, recognize that beating the market is no snap, and they try to avoid being overconfident. Russell Fuller and Richard Thaler operate Fuller and Thaler Asset Management. They manage a mutual fund, based on the De Bondt–Thaler effect, called Behavioral Value.20 It may sound paradoxical, but Fuller believes that markets are, in the main, efficient. He tells me that many of the De Bondt–Thaler losers are, in fact, properly priced, that “most should be losers.”21 What’s the lesson? Don’t think the streets are paved with gold, or at least Wall Street anyway.
One other thing: Behavioral finance offers refutable hypotheses, but it does not have all the answers. De Bondt and Thaler predicted overreaction based on representativeness. But take another look at figure 4-1. It shows that a portfolio of extreme losers does outperform the market. However, a careful inspection of the figure shows that the effect is concentrated in the month of January. I know of nothing that suggests that people rely on representativeness in some months but not others.
Summary
This chapter covered the third theme of behavioral finance, inefficient markets, which is connected with the earlier two themes by cause and effect. Heuristic-driven bias and frame dependence cause prices to stray from fundamental values. Three examples are (1) representativeness as a cause of the winner-loser effect; (2) conservatism as a cause of post-earnings-announcement drift; and (3) mental accounting as a cause of a historical equity premium that has been too high, relative to the underlying fundamentals. But as I noted, prices can stray far from fundamental value, and for very long periods.
I conclude the chapter with a word of caution. The departure of price from fundamental value does not automatically lead to risk-free profit opportunities. In fact, the “smart money” may avoid some trades, although they have identified mispricing. Why? Because of nonfundamental risk, meaning risk associated with unpredictable sentiment.
Part II Prediction
Chapter 5 Trying to Predict the Market
When it comes to predicting the market, how immune are Wall Street’s strategists from heuristic-driven bias? Are they any different from the typical investor? And what illusions, if any, bias investors’ predictions about where the market is headed?
This chapter discusses the following:
• strategists’ susceptibility to gambler’s fallacy
• evidence that strategists are overconfident
• which investors bet on trends
• how anchoring-and-adjustment, and salience, influence investors’ predictions
• the key illusions that most people have about randomness, and why these illusions bias their predictions
• the impact of inflation on strategists’ predictions of the market
Gambler’s Fallacy: A Case Study
Consider the predictions that strategists made in January 1997. Let’s start with some background. The years 1995 and 1996 had resulted in spectacular back-to-back returns of 34 percent and 20 percent on the S&P 500.1 In the wake of the 1996 national elections, the S&P 500 soared 7.5 percent during November, a move that prompted Federal Reserve chair Alan Greenspan to ask whether “irrational exuberance has unduly escalated asset values.”2
For Abby Joseph Cohen, cohead of the investment policy committee at Goldman Sachs, the market’s strong performance corroborated the bullish forecasts she had consistently been making. These forecasts had led Barron’s to describe her as the “virtual maven of the nineties’ bull market.”3
On January 1, 1997, the S&P 500 stood at 740. Cohen’s target for the end of the year was 815 to 825, an 11.5 percent increase, predicated on projected earnings growth of 10 percent. Yet, the index ended the year at 970, up 31 percent!
Cohen had plenty of professional company in misgauging the market during 1997. In its issue of December 30, 1996, Barron’s published the predictions of Wall Street’s leading strategists for the mid-year and end-of-year values of the Dow Jones Industrial Average. In addition to Abby Joseph Cohen, this esteemed group included Marshall Acuff of Salomon Smith Barney, Charles Clough of Merrill Lynch, Edward Kerschner of PaineWebber, Elizabeth Mackay of Bear Sterns, David Shulman of Salomon, and Byron Wien of Morgan Stanley. Every single one underestimated the market’s performance during 1997.
The Dow closed at 6448 in 1996 and at 7908 in 1997, a 22.6 percent increase. After the Dow surged by 33.5 percent in 1995 and 26 percent in 1996, could the strategists be faulted for not having anticipated a third spectacular year in a row? Not to my mind. But this does not mean that their predictions were free from bias.
Strategists are prone to committing gambler’s fallacy, a phenomenon whereby people inappropriately predict reversal. In chapter 2, I pointed out that gambler’s fallacy is regression to the mean gone overboard, and quoted Merrill Lynch’s Robert Farrell in this connection. Remember, Farrell had predicted that future performance would be below average. What does regression to the mean suggest about predictions in the wake of above-average performance? It implies that future performance will be closer to the mean, not that it will be below the mean in order to satisfy the law of averages.4
Let’s return to the strategists’ 1997 predictions for the Dow. In the wake of above-average performance in 1995 and 1996, did the strategists predict that 1997 would feature below-average performance? Indeed they did. They predicted that the Dow would actually decline by 0.2 percent in 1997, well below the 8.6 percent annual rate that the Dow had grown between 1972 and 1996.5
General Findings
The instances of gambler’s fallacy described above are not isolated cases. Werner De Bondt (1991) has examined the market predictions collected by the late Joseph Livingston since 1952. De Bondt reports that, in accordance with gambler’s fallacy, these predictions consistently are overly pessimistic after three-year bull markets and overly optimistic after three-year bear markets. As for accuracy, De Bondt concludes that market predictions are not “particularly useful.”6
So What? Prediction and Performance
Is there a downside to strategists succumbing to gambler’s fallacy? That depends on how their predictions get used. Strategists also make recommendations about strategic asset allocation, the proportion of a portfolio devoted to equities, bonds, cash, and other assets. These recommendations are routinely available from sources such as Dow Jones News Service and Business Week. Not surprisingly, strategists’ recommended equity allocations have a high positive correlation with their market predictions. So if strategists are unduly pessimistic in bull markets, the returns to the portfolios they recommend are less than they might have been.
Richard Bernstein, head of quantitative research at Merrill Lynch, suggests that the asset allocations recommended by Wall Street strategists do indeed reflect missed profit opportunities. In fact, he has developed an indicator based on the average recommended allocation to stocks. He issues a “buy” signal when this average drops below 50.4 percent, and a “sell” recommendation when it rises above 57.5 percent (R. Bernstein 1995). He considers the range in between as neutral. Consider the period between September 1997 and April 1998, when strategists increased their average equity allocation in a balanced portfolio from 48.1 percent to a four-year high of 54.5 percent. Wall Street Journal writer Greg Ip described Bernstein’s perspective on this change as follows:
Mr. Bernstein considers that negative for stocks, but not unduly so. Indeed, gurus aren’t so much turning more bullish, as easing their entrenched bearishness. “Over the 13 years of data that we have, the average equity allocation is only 53.4%,” says Mr. Bernstein, with the remainder in bonds or cash. But he notes that most survey participants consider a 60% stock allocation to be about neutral. “So through time, Wall Street is underweighted on equities. There’s all this talk of the ‘wall of worry.’” There it is. The people who have the greatest incentive to sell equities are habitually bearish.7
Kenneth Fisher a
nd Meir Statman (1999b) confirm Richard Bernstein’s claims. They find that for every 1 percent decline in the recommended equity allocation by strategists, the S&P 500 has increased by 26 basis points.8
Overconfidence
A study sponsored by PaineWebber, and administered by the Gallup Organization, found that experience is an important factor in investors’ expectations about the market. The results were summarized as follows in the July 8, 1998, Wall Street Journal: “As stock prices hover at or near records, a new poll indicates that inexperienced investors expect considerably higher returns on their portfolios than do longtime investors—and are more confident of their ability to beat the market.”9
This finding is very interesting. Inexperienced investors are more confident that they will beat the market than are experienced investors. Given the difficulty that many investors actually have beating the market, novice investors may be not just confident, but overconfident.
How is overconfidence reflected in the predictions people make? I discussed the general issue of overconfidence in chapter 2. In order to review the basic issue, and set the stage for a discussion of overconfident predictions, let us consider the following five-question quiz.10 The first two questions pertain to general knowledge, and the remaining three to financial predictions.
You will be asked to give your best guess in answering each of the five questions. In addition to giving your best guess, consider a range—a low guess and a high guess—so that you feel 90 percent confident that the right answer will lie between your low guess and your high guess. Try not to make the range too narrow. Otherwise, you will appear overconfident. At the same time, try not to make the range between your low guess and high guess too wide. This will make you appear underconfident. If you are well calibrated, you should expect that only one out of the five correct answers does not lie between your low guess and your high guess.
Beyond Greed and Fear Page 9