This paper was prepared for John B. Taylor and Michael Woodford, Editors

17

See also Case and Shiller (1988) for a similar analysis of recent real estate booms and busts.

On the other hand, Garber (1990) analyzes some famous speculative bubbles, including the

tulipomania in the 17th century, and concludes that they may have been rational.

18

Even public expectations of a stock market crash does not prevent the stock market from rising;

there is evidence from options prices that the stock market crash of 1987 was in some sense expected

before it happened; see Bates (1991, 1995).  Lee, Shleifer and Thaler (1991) argue that investor

expectations, or rather “sentiment” can be measured by closed-end mutual fund discounts, which vary

through time.

17

average lower, not higher, than on those sold.  This appears to be evidence of over-

confidence among these investors.

Within the week of the stock market crash of October 19, 1987 I sent out questionnaires

to 2,000 wealthy individual investors and 1,000 institutional investors, asking them to recall

their thoughts and reasons for action on that day; see Shiller (1987b).  There were 605

completed responses from individuals and 284 responses from institutions.  One of the

questions I asked was:  “Did you think at any point on October 19, 1987 that you had a

pretty good idea when a rebound was to occur?”  Of individual investors, 29.2% said yes,

of institutional investors, 28.0% said yes.  These numbers seem to be surprisingly high:  one

wonders why people thought they knew what was going to happen in such an unusual

situation.  Among those who bought on that day, the numbers were even higher, 47.1% and

47.9% respectively.  The next question on the questionnaire was “If yes, what made you

think you knew when a rebound was to occur?”  Here, there was a conspicuous absence of

sensible answers; often the answers referred to “intuition” or “gut feeling.”  It would appear

that the high volume of trade on the day of the stock market crash, as well as the occurrence,

duration, and reversal of the crash was in part determined by overconfidence in such

intuitive feelings.

17

If people are not independent of each other in forming overconfident judgments about

investments, and if these judgments change collectively through time, then these “noisy”

judgments will tend to cause prices of speculative assets to deviate from their true

investment value.  Then a “contrarian” investment strategy, advocated by Graham and Dodd

(1934) and Dreman (1977) among many others, a strategy of investing in assets that are

currently out of favor by most investors, ought to be advantageous.  Indeed, there is much

evidence that such contrarian investment strategy does pay off, see for example, De Bondt

and Thaler (1985), Fama and French (1988, 1992), Fama (1991), and Lakonishok, Shleifer

and Vishny (1994).  That a simple contrarian strategy may be profitable may appear to some

to be surprising:  one might think that “smart money,” by competing with each other to

benefit from the profit opportunities, would ultimately have the effect of eliminating any

such profit opportunities.  But, there are reasons to doubt that such smart money will indeed

have this effect; see Shiller (1984), De Long et al. (1990a,b), and Shleifer and Vishny

(1996).

18

---

18

The Disjunction Effect

The disjunction effect is a tendency for people to want to wait to make decisions until

information is revealed, even if the information is not really important for the decision, and

even if they would make the same decision regardless of the information.  The disjunction

effect is a contradiction to the “sure-thing principle” of rational behavior (Savage, 1954).

Experiments showing the disjunction effect were performed by Tversky and Shafir

(1992).  They asked their subjects whether they would take one of the bets that Samuelson’s

lunch colleague, discussed above, had refused a coin toss in which one has equal chances

to win $200 or lose $100.  Those who took the one bet were then asked whether they then

wanted to take another such bet.  If they were asked after the outcome of the first bet was

known, then it was found that a majority of respondents took the second bet whether or not

they had won the first.  However, a majority would not take the bet if they had to make the

decision before the outcome of the bet was known.  This is a puzzling result:  if one’s

decision is the same regardless of the outcome of the first bet, then it would seem that one

would make the same decision before knowing the outcome.  Tversky and Shafir gave their

sense of the possible thought patterns that accompany such behavior:  if the outcome of the

first bet is known and is good, then subjects think that they have nothing to lose in taking

the second, and if the outcome is bad they want to try to recoup their losses.  But if the

outcome is not known, then they have no clear reason to accept the second bet.

The disjunction effect might help explain changes in the volatility of speculative asset

prices or changes in the volume of trade of speculative asset prices at times when

information is revealed.  Thus, for example, the disjunction effect can in principle explain

why there is sometimes low volatility and low volume of trade just before an important

announcement is made, and higher volatility or volume of trade after the announcement is

made.  Shafir and Tversky (1992) give the example of presidential elections, which

sometimes induce stock market volatility when the election outcome is known even though

many skeptics may doubt that the election outcome has any clear implications for market

value.

Gambling Behavior and Speculation

A tendency to gamble, to play games that bring on unnecessary risks, has been found to

pervade widely divergent human cultures around the world and appears to be indicative of

a basic human trait, Bolen and Boyd (1968).  Kallick et al. (1975) estimated that 61% of the

adult population in the United States participated in some form of gambling or betting in

1974.  They also estimated that 1.1% of men and 0.5% of women are “probably compulsive

gamblers,” while an additional 2.7% of men and 1% of women are “potential compulsive

gamblers.”  These figures are not trivial, and it is important to keep in mind that compulsive

gambling represents only an extreme form of the behavior that is more common.

The tendency for people to gamble has provided a puzzle for the theory of human

behavior under uncertainty, since it means that we must accommodate both risk-avoiding

behavior (as evidenced by people’s willingness to purchase insurance) with an apparent risk-

---