that has recently gone up in value (De Long et al. 1990b, Barberis and
Shleifer 2003). If a company’s stock price goes up this period on good earn-
ings, positive feedback traders buy the stock in the following period, caus-
ing a further price rise. On the one hand, this generates momentum and
post-earnings announcement drift. On the other hand, since the price has
now risen above what is justified by fundamentals, subsequent returns will
on average be too low, generating long-term reversals and a scaled-price
ratio effect.
The simplest way of motivating positive feedback trading is extrapolative
expectations, where investors’ expectations of future returns are based on
past returns. This in turn, may be due to representativeness and to the law
of small numbers in particular. The same argument made by BSV as to why
investors might extrapolate past cash flows too far into the future can be
applied here to explain why they might extrapolate past returnstoo far into
the future. De Long et al. (1990b) note that institutional features such as
portfolio insurance or margin calls can also generate positive feedback
trading.
Positive feedback trading also plays a central role in the model of Hong
and Stein (1999), although in this case it emerges endogenously from more
primitive assumptions. In this model, two boundedly rational groups of in-
vestors interact, where bounded rationality means that investors are only
able to process a subset of available information. “Newswatchers” make
forecasts based on private information, but do not condition on past prices.
“Momentum traders” condition only on the most recent price change.
Hong and Stein also assume that private information diffuses slowly
through the population of newswatchers. Since these investors are unable
to extract each others’ private information from prices, the slow diffusion
generates momentum. Momentum traders are then added to the mix. Given
what they are allowed to condition on, their optimal strategy is to engage
in positive feedback trading: a price increase last period is a sign that good
private information is diffusing through the economy. By buying, momen-
tum traders hope to profit from the continued diffusion of information.
This behavior preserves momentum, but also generates price reversals:
since momentum traders cannot observe the extent of news diffusion, they
keep buying even after price has reached fundamental value, generating an
overreaction that is only later reversed.
These four models differ most in their explanation of momentum. In two
of the models—BSV, and Hong and Stein (1999)—momentum is due to an
initial underreaction followed by a correction. In De Long et al. (1990b)
and DHS, it is due to an initial overreaction followed by even more overre-
action. Within each pair, the stories are different again.^29
A SURVEY OF BEHAVIORAL FINANCE 43(^29) In particular, the models make different predictions about how individual investors would
trade following certain sequences of past returns. Armed with transaction-level data, Hvidk-
jaer (2001) exploits this to provide initial evidence that may distinguish the theories.