rational, and actually manage to take advantage of the other group of
traders, the newswatchers. This distinction is closely related to the fact that
in DeLong et al, there is never any underreaction. There is short-run posi-
tive correlation of returns, but this reflects an initial overreaction, followed
by even moreoverreaction.^29
At a more general level, this chapter revisits several themes that have
been prominent in previous theoretical work. The notion that one group of
optimizing traders might create a negative informational externality, and
thereby destabilize prices even while they are making profits, also shows up
in Stein (1987). Stretching a bit further, there is an interesting analogy here
with the ideas of Banerjee (1992) and Bikhchandani, Hirshleifer, and Welch
(1992) on informational cascades. In these models, agents move sequen-
tially. In equilibrium, each rationally bases his decision on the actions of the
agent before him, even though this inflicts a negative informational exter-
nality on those that follow. Very much the same thing could be said of the
generations of momentum traders in this model.
5.ConclusionsAt the outset, we argued that any new “behavioral” theory of asset pricing
should be judged according to three criteria: (1) it should rest on assump-
tions about investor behavior that are either a priori plausible or consistent
with casual observation; (2) it should explain the existing evidence in a par-
simonious and unified way; and (3) it should make a number of further pre-
dictions that can be subject to testing and that are ultimately validated.
How well have we done on these three scores? With respect to the first, we
believe that our particular rendition of bounded rationality—as the ability to
process a small subset of the available information in an unbiased way—is
both plausible and intuitively appealing. Moreover, in our framework, this
sort of bounded rationality implies a widespread reliance by arbitrageurs on
simple momentum strategies. As we have discussed, this implication appears
to be strongly consistent with what is observed in the real world.
In terms of the parsimony/unity criterion, it should be emphasized that
everything in our model is driven by just one primitive type of shock:
slowly-diffusing news about future fundamentals. There are no other ex-
ogenous sources of investor sentiment, and no liquidity disturbances. Our
main conceptual contribution is to show that if there is ever any short-run
underreaction to this kind of news on the part of one set of traders, then
530 HONG AND STEIN
(^29) Also, the model of DeLong et al. (1990) does not rally endogenously deliver reversals.
Rather, prices are just forced back to fundamentals on a terminal date. In our model, the re-
versal phase is more endogenous, corresponding to the unwinding of momentum traders’ posi-
tions. It also involves more complex dynamics, with the sort of damped oscillations seen in the
figures.