knowledge, I am reluctant to buy if you are ready to sell. In contrast to this
prediction, the volume of trading on the world’s stock exchanges is very
high. Furthermore, studies of individuals and institutions suggest that both
groups trade more than can be justified on rational grounds.
Barber and Odean (2000) examine the trading activity from 1991 to
1996 in a large sample of accounts at a national discount brokerage firm.
They find that after taking trading costs into account, the average return of
investors in their sample is well below the return of standard benchmarks.
Put simply, these investors would do a lot better if they traded less. The
underperformance in this sample is largely due to transaction costs. How-
ever, there is also some evidence of poor security selection: in a similar data
set covering the 1987 to 1993 time period, Odean (1999) finds that the av-
erage gross return of stocks that investors buy, over the year after they buy
them, is lower than the average gross return of stocks that they sell, over
the year after they sell them.
The most prominent behavioral explanation of such excessive trading is
overconfidence: people believe that they have information strong enough to
justify a trade, whereas in fact the information is too weak to warrant any
action. This hypothesis immediately predicts that people who are more
overconfident will trade more and, because of transaction costs, earn lower
returns. Consistent with this, Barber and Odean (2000) show that the in-
vestors in their sample who trade the most earn by far the lowest average
returns. Building on evidence that men are more overconfident than
women, and using the same data as in their earlier study, Barber and Odean
(2001) predict and confirm that men trade more and earn lower returns on
average.
Working with the same data again, Barber and Odean (2002a) study the
subsample of individual investors who switch from phone-based to online
trading. They argue that for a number of reasons, the switch should be
accompanied by an increase in overconfidence. First, better access to infor-
mation and a greater degree of control—both features of an online trading
environment—have been shown to increase overconfidence. Moreover, the
investors who switch have often earned high returns prior to switching,
which may only increase their overconfidence further. If this is indeed the
case, they should trade more actively after switching and perform worse.
Barber and Odean confirm these predictions.
7.4. The Selling DecisionSeveral studies find that investors are reluctant to sell assets trading at a
loss relative to the price at which they were purchased, a phenomenon la-
belled the “disposition effect” by Shefrin and Statman (1985). Working
with the same discount brokerage data used in the Odean (1999) study
from above, Odean (1998) finds that the individual investors in his sample
A SURVEY OF BEHAVIORAL FINANCE 53