11. De Bondt-Thaler prescribes buying losers, whereas Shefrin-Statman points out that investors are reluctant to sell losers. This struck Merton as contradictory, and he stated that a “rational” investor would “find himself ‘convicted’ by his actions of one or the other cognitive failures.” I remember Bob Merton telling me that he was especially struck by the fact that the two articles appeared back-to-back in the same journal issue.
12. In Shefrin-Statman, losers are portfolio specific, and reflect initial purchase price for the investor. In De BondtThaler, losers are the worst-performing stocks during the past thirty-six months. A Shefrin-Statman loser may not be a De Bondt-Thaler loser, and vice versa.
Chapter 2
1. In case it’s not clear: Availability is the principle; judging the frequency of occurrence by the number of instances that come readily to mind is the heuristic rule of thumb; being predisposed to ease of recall resulting from distortions in media coverage is the bias; and judging homicide to be a more frequent cause of death than stroke is the error.
2. I thank Barbara Stewart, Santa Clara University, for providing these data.
3. Keep in mind that the question pertains to those who graduate college.
4. Readers will see that representativeness plays a major role in many of the issues discussed in this volume. Some of the most important applications are predicting the market, picking stocks, choosing mutual funds, selecting money managers, and investing in initial public offerings (IPOs) and seasoned equity offerings (SEOs).
5. There is a technical issue about predicting autocorrelated time series. Is the autocorrelation positive or negative? Most economic and financial series feature positive autocorrelation, whereby above-average performance tends to be followed by above-average performance; likewise, below-average performance tends to be followed by below-average performance.
6. Michael Siconolfi, Anita Raghavan, and Mitchell Pacelle, “All Bets Are Off: How the Salesmanship and Brainpower Failed at Long-Term Capital,” Wall Street Journal, 16 November 1998.
7. Jonathan Clements, “Getting Going: Behavioral Specialists Put Investors on Couch,” Wall Street Journal, 28 November 1995.
Chapter 3
1. Miller made this statement at the University of Chicago conference on behavioral finance in October 1986, on the day the Nobel Prize committee announced that Franco Modigliani was to receive the prize for that year.
2. Loss aversion implies that people have a predisposition toward avoiding a certain loss. But I note that loss aversion can be counterbalanced by panic. In an interview with Forbes magazine titled “Management, Strategies, Trends: Living with the Bull, Preparing for the Bear,” fund manager Martin Zweig compared his own behavior with that of Warren Buffett, stating: “I could never do what he does. He buys things and holds forever—correctly, because he recognizes that if you sell and pay taxes, it cuts your return. But my problem is, I can’t sit through bear markets. Buffett says, if you can’t afford to see your stock go down 30% or 40%, you shouldn’t be in it. I can’t take the pain. I built my technique based on what I can handle personally. This means I have to be in and out a lot. My main thing is, I want to survive. I want to be there, and when the big bear market comes, I don’t want to get chewed up” (Brimelow 1998).
3. “Today’s Rogue Traders Just Want to Save Face,” Associated Press, 13 September 1995.
4. I thank Bob Saltmarsh, who was treasurer at Apple during this period, for sharing his insights about the Newton with me.
5. Jim Carlton, “Apple Drops Newton, an Idea Ahead of Its Time,” Wall Street Journal, 2 March 1998.
6. Because their studies used students, Thaler and Johnson employed stakes that were smaller by a factor of 100.
7. So called because gamblers who receive a gift of chips from the house tend to be more tolerant of risk, since they begin by seeing themselves ahead.
Chapter 4
1. Michael Siconolfi, Anita Raghavan, and Mitchell Pacelle, “All Bets Are Off: How Salesmanship and Brainpower Failed at Long-Term Capital,” Wall Street Journal, 16 November 1998.
2. Ibid.
3. I thank Jacob Thomas, coauthor of Bernard and Thomas (1989) for providing me with the data for this figure.
4. The questions that I used involved stakes that were near the historical growth rates of real personal consumption expenditures.
5. In replicating the Shiller-Pound survey in 1998, I obtained a similar result, almost 40 minutes.
6. I thank Bob Shiller for providing me with this data, updated to 1999.
7. Although he did not shock his audience, most of whom were nodding off at Greenspan’s after-dinner speech at the American Enterprise Institute, according to an unnamed source from the Securities and Exchange Commission who was present that evening.
8. Along with interest rates, inflation, and tolerance for risk.
9. In Chapter 2 I indicated that the 1998 closing value of the Dow was 9181, but would have been 652,230 if dividends were taken into account.
10. As can be seen from figure 4-3, the low D/P does not stem from low D, but from high P!
11. In Chapter 2, I discussed gambler’s fallacy in connection with Robert Farrell’s predictions for below-average stock returns. Although Campbell and Shiller consider mean reversion, and predict below-average stock returns over the next ten years, they do not succumb to gambler’s fallacy. Their analysis is based on the relationship between valuation measures and subsequent returns. I hasten to emphasize that from a statistical perspective, the confidence associated with the 1996 Campbell-Shiller prediction for the 1997–2006 period is very low. This is because the value of P / E at year-end 1996 was at the extreme end of its historical range, which is where prediction confidence is lowest. It is worth noting that the Wall Street Journal reported John Campbell to have been very confident in his analysis. Campbell hedged his entire stock portfolio using futures, as a means of selling his stocks without triggering capital gains. See “The Outlook: Sometimes Stocks Go Nowhere for Years,” by David Wessel, January 13, 1997. At the end of 1996, the Dow Jones Industrial Average stood at 6448. It crossed 11,000 on May 3, 1999.
12. Shiller’s work, and that of LeRoy and Porter (1981), which was published at the same time, was based on some controversial technical assumptions. Much of the debate has centered on some of these assumptions. In addition, his measure of fundamental value is based on a time invariant expected real interest rate and the stream of future dividends. Yet, managers appear to smooth dividends over time. See Alan Kleidon (1986) and Robert Merton (1987b).
13. The Wall Street Journal makes the same point in connection with a sharp market drop that took place in August 1998 (see Chapter 5).
14. What is wrong with this picture? From year-end 1993 through year-end 1998, earnings on the S&P 500 grew at 11.9 percent a year and dividends grew at 4.2 percent a year, but the return on the S&P 500 was 24 percent a year. Earnings in 1997 and 1998 earnings grew at 3 percent and -4 percent at a time when the S&P 500 returned 33.3 percent and 28.4 percent. What dramatic differences are investors expecting in future earnings, dividends, interest rates, or rates of inflation?
15. The leading P / E ratio was 26.46.
16. The high P / E environment in the U.S. market, relative to the historical norm, reminds me of the Japanese stock market in the 1980s. In February 1999, the Nikkei stood at 14,232. But it was over three times as high a decade earlier when the P / E on Japanese stocks was over 50. During 1986 and 1987, the Nikkei had risen by more than 28 percent a year. In January, 1988 concerns about the Japanese market being in a bubble started to surface: see Kenneth French and James Poterba (1991). But despite being overvalued the Japanese market rose by a further 64 percent over the next two years. But what about the long-term return? One hundred yen invested in the Nikkei in January 1988 would have been worth 71 yen in March 1999. Yet an investor who pulled out of Japanese stocks in January, 1988, would have spent the next two years kicking himself.
17. Du
ring the market’s strong performance in 1997 and 1998, Wall $treet Week with Louis Rukeyser host Louis Rukeyser continually chided panelist Gail Dudack, the most bearish of all the panelists on his program.
18. E. S. Browning, “Stock Market’s New Year’s Party Keeps Going,” Wall Street Journal, 11 January 1999.
19. Browning, “Market’s New Year’s Party.”
20. As of January 1999, Behavioral Value had been operating for three years, during which time it posted returns of 18.3 percent, net of fees. The Russell 2000 returned 14.4 percent in the same period.
21. Russ Fuller, interview by author, February 2, 1999. Fuller and Thaler apply a screening procedure to the De Bondt-Thaler losers, in order to select stocks for the Behavioral Value fund. A generic De Bondt-Thaler strategy has an excess return (alpha) of between 3 and 4 percent. Fuller tells me that his screening procedure sifts out the most underpriced 10 percent, with the resulting alpha being 11 percent.
Chapter 5
1. Earnings had grown by 11 percent and 14 percent respectively. The yield on the 30-year Treasury bond had fallen from above 8 percent to 6.6 percent.
2. This is a much-quoted phrase from a speech given by Greenspan on December 5 at the American Enterprise Institute.
3. Jonathan Laing, “High Anxiety: As Market Gurus Debate, Investors Watch Nervously After Last Week’s Slide,” Barron’s, 10 August 1998.
4. Unless we are dealing with negative autocorrelation, which is not the case with stock market returns.
5. Abby Cohen’s predictions were certainly higher than average, but below the historical average. During her appearance on the program Wall $treet Week with Louis Rukeyser, on January 8, 1999, program host Louis Rukeyser mentioned that her actual predictions appeared to be on the low side. Cohen responded by saying that Goldman Sachs announces prediction levels that they believe to be attainable.
6. De Bondt’s study also examines the predictions for future inflation and the growth of industrial production.
7. Greg Ip, “The Final Bears May Be Giving Up: Bull’s Victory Prompts Fears,” Wall Street Journal, 13 April 1998.
8. This finding is statistically significant at the 5 percent level.
9. Greg Ip, “It’s Official: Stock Market’s Pups Are Likely to Be Bulls,” Wall Street Journal, 8 July 1998.
10. This quiz is a variant of the one found in Russo and Schoemaker (1989).
11. Lauren R. Rublin, “A Very Good Year,” Barron’s, 30 December 1996.
12. Lauren R. Rublin, “Another Chance,” Barron’s, 23 June 1997.
13. Ibid.
14. Lauren R. Rublin, “Sober Salute: It Was a Great Market Year, So Why All the Frowns?” Barron’s, 5 January 1998.
15. An analogous statement holds when the market has been declining. They believe there is only a little room left for movement on the downside, but a lot of room left for movement on the upside.
16. The data on individual investors comes from the Association of Individual Investors (AAII).
17. Steven Mufson wrote the article.
18. On the television program Wall $treet Week with Louis Rukeyser, 27 December 1996.
19. Notice that Acampora had retreated from his prediction of 10,000, in fact pushing it out to 1999.
20. Jonathan Laing, “High Anxiety: As Market Gurus Debate, Investors Watch Nervously After Last Week’s Slide,” Barron’s, 10 August 1998.
21. For a discussion of events pertaining to Asia, see Chapter 21.
22. Leslie Scism and E. S. Browning, “Heard on the Street: In Battle of the Stock Market Gurus, Bulls Win a Round,” Wall Street Journal, 6 August 1988.
23. In Chapter 6, I discuss in further detail the way that technical analysts treat measures of sentiment.
24. Greg Ip and Aaron Lucchetti, “Mood Swing: Stock Market Plunges as Negative Sentiment Spreads to Blue Chips,” Wall Street Journal, 5 August 1998.
25. Michael Santoli, “A Ray of Sunshine,” Barron’s, 31 August 1998. Those interviewed were Marshall Acuff (Salomon Smith Barney), Byron Wien (Morgan Stanley Dean Witter), Douglas Cliggott (J. P. Morgan), Abby Joseph Cohen (Goldman Sachs), and Elizabeth Mackay (Bear Stearns). Acuff stated that this was not the end of the bull market, but rather a cyclical correction. Wien described it as a temporary bottom from which to rally. Cliggott restated his target of 1065 for the S&P 500 at year-end. Cohen restated her target of 1150 for the S&P 500 at year-end. Mackay predicted that returns from stocks would lie in the historically normal range.
26. A friend of mine learned this lesson, but not by tossing coins. He recently welcomed the birth of his first son after five daughters.
27. Some may feel that a situation involving tosses of a fair coin produces the least predictability, making it an inappropriate analogy for a financial market. Here is a variant that might come closer. Let there be two coins: a gold coin where the probability of tossing a head is 80 percent, and a silver coin where the probability of tossing a tail is 80 percent. Suppose that after every head we toss the gold coin, and after every tail we toss the silver coin. That makes long runs much more likely than when a fair coin is tossed. More important, it makes short-run predictions reliable. If the last toss was a head, and you predict that the next toss will be a head, you will be right four out of five times, much better than what Wall Street strategists do. But how about the long term? Suppose I begin by tossing the gold coin. What is the probability that a head will appear on the fifth toss? It is definitely more that 50 percent—in fact, the answer is 53.9 percent. On the tenth toss, the probability is down to 50.3 percent, not much different than with a fair coin. The point is that predictability evaporates very quickly.
28. Moreover, Acampora tends to change his long-term forecasts quite frequently. He justifies these changes by saying, “People say, ‘It sounds like one day you say one thing and another day you say another thing,’ but that’s what they pay me to do: call significant turns. And we’ve had some very significant turns in a very short time. See Mufson, “Joe Investor.”
29. Santoli, “Ray of Sunshine.”
30. Lauren R. Rublin, “Another Chance,” Barron’s, 23 June 1997.
Chapter 6
1. In Chapter 5, I discussed Richard Bernstein’s work showing that an index of recommended allocations by strategists is negatively correlated with subsequent returns. However, there is no such relationship to speak of when it comes to the predictions made by newsletter writers.
2. Greg Ip, “The Final Bears May Be Giving Up: Bull’s Victory Prompts Fears,” Wall Street Journal, 13 April 1998.
3. Investors Intelligence, November 26, 1984.
4. Ip, “Final Bears.”
5. Greg Ip, “Market’s Logic,” Wall Street Journal, 17 May 1997.
6. Greg Ip’s Wall Street Journal article, “Final Bears,” mentions that sentiment indicators are usually used to time short-term moves in the market. The character of figure 6-1 is no different when it comes to short term moves as when it comes to long term rates.
7. John Dorfman, “Your Money Matters,” Wall Street Journal, 26 January 1989.
8. Clements, Jonathan, “Getting Going,” Wall Street Journal, 12 May 1998. “The Truth Investors Don’t Want to Hear on Index Funds and Market Soothsayers.”
9. Specifically, when advisors are bearish, the market is no more likely to go up than down.
10. In a fascinating article, Leda Cosmides and John Tooby (1992) differentiate this heuristic from another heuristic that efficiently tests for evidence of cheating on social contracts. In some circumstances, the latter leads to the right algorithm for testing the validity of “IfP, then Q.” For example, suppose P is “drink an alcoholic beverage” and Q is “underage.” If your job is to enforce this rule in a bar, whom must you check? People you know to be underage (not-Q) and those you know to be drinking alcoholic beverages (P).
11. The following experiment by psychologist Peter Wason (1960) illustrates the point. Imagine that I show you four cards. I tell you that each card
has a letter on one side and an integer on the other. But when I show you the cards for the first time, you get to see only one side. The four cards I show you have on them A, B, 2, and 3. Consider an if-P-then-Q hypothesis about these four cards: Every card with a vowel on one side (P) has an even number on the other side (Q). The hypothesis may or may not be true. Your task is to verify whether the hypothesis is true by turning over as few cards as possible. Most people who play this game choose A and 2, or A alone. They do so because turning these cards provides confirming evidence. But this is the wrong way to go. The efficient way of examining the hypothesis is to turn over cards that might falsify the hypothesis. Consider the falsification potential of each card in turn. Suppose we turn the card featuring an A. We will find either an even number or an odd number. If we find an even number, we have evidence supporting the hypothesis. However, if we find an odd number we know that the hypothesis is false. Next, suppose we turn over the card with the letter B. This card provides us with no evidence to judge the veracity of the hypothesis, since the hypothesis says nothing about cards featuring consonants. Next, consider the card featuring a 2. If we turn it over, we might find a vowel. This would be consistent with the hypothesis. Alternatively, we might find a consonant. That would be irrelevant to the hypothesis. Hence, this card offers no opportunity for falsification. Last, suppose we turn over the card featuring a 3. If we find a vowel, we know that the hypothesis is false. A consonant provides no information to support or falsify the hypothesis. So the two cards that offer the potential for falsification feature A and 3; i.e., P and not Q. However, as mentioned, most subjects choose A and 2, or A alone. Notice that while A allows for both support and falsification, 2 allows for support only.
Beyond Greed and Fear Page 38