(given the simple nature of arbitrage strategies) there must eventually be
overreaction in the longer run as well.
Finally, our model does deliver several testable auxiliary implications.
Among the most noteworthy are the following:(1) both short-run continua-
tion and long-run reversals should be more pronounced in those (small,
low-analyst-coverage) stocks where information diffuses more slowly; (2)
There may be more long-run overreaction to information that is initially
private than to public news announcements; and (3) there should be a rela-
tionship between momentum traders’ horizons and the pattern of return
autocorrelations. Evidence supportive of the first prediction is already emerg-
ing; we hope to explore some of the others in future work.
A UNIFIED THEORY OF UNDERREACTION 531