IV. Theories of Overreaction and UnderreactionThe chapters in the previous section document some of the anomalies that be-
havioral finance researchers identified. All are troubling to the efficient mar-
ket hypothesis, but as my Chicago colleagues like to say, “It takes a theory to
beat a theory.” This section contains three attempts by leading theorists to
offer an account of the anomalous facts, specifically, the apparent finding that
stock prices display mean reversion over long horizons but positive serial cor-
relation over short to medium horizons. The first two chapters, by Barberis,
Shleifer, and Vishny (BSV), and by Daniel, David Hirshleifer, and Avanidhar
Subrahmanyam (DHS) were written in parallel, and both adopt a representa-
tive agent formulation. In BSV agents at times believe that earnings are trend-
ing and at other times think that they are mean-reverting. The model is based
on the psychological principles of representativeness and conservatism. DHS
build their model around overconfidence, specifically overconfidence about
the validity of what investors treat as private information. The third chapter
in the section, by Harrison Hong and Jeremy Stein, followed shortly after the
first two, and instead of modeling a single representative agent they envision
a world with two different types of boundedly rational agents. “Newswatch-
ers” make forecasts based on what they consider private information but do
not condition properly on past prices. “Momentum traders” condition only
on past price movements. In the version of the paper included in this book,
Hong and Stein also report some empirical evidence supporting some of the
predictions of their model.
V.Investor BehaviorBehavioral finance during most of its first decade was more finance than be-
havior. In recent years that emphasis has begun to change as researchers
have obtained data sets of actual behavior by investors. Using such data re-
searchers have been able to show that the same biases discovered by psy-
chologists in laboratory experiments at low or zero stakes also emerge in
the field at high stakes. The pioneer of this style of research was Terry
Odean, who as a graduate student obtained a large data set of trading in-
formation from a large discount brokerage firm. Writing first on his own
and then later in collaboration with Brad Barber, he has documented how
investors behave. Barber and Odean summarize this research in Chapter
- They stress two findings: First, investors are reluctant to sell stocks that
have declined in value (compared to stocks that have appreciated) even
though the effect of taxes is to push investors to do just the opposite; sec-
ond, investors display overconfidence in the sense that they trade too much.
Women will not be surprised to learn that this overconfident behavior is
more pronounced among men.
PREFACE xv