The retained earnings hypothesis predicts that stock distributions accounted for
by reducing retained earnings are a more credible signal of managerial optimism than
stock distributions that do not reduce retained earnings. This study examines the
costs of false signaling that are a necessary precondition for the hypothesis and
finds them to be generally very small. The absence of the requisite costs of false
signaling calls the validity of the hypothesis into question for most firms. However,
prior studies have reported broad-based market evidence consistent with the retained
earnings hypothesis. To resolve this apparent inconsistency, this study replicates
and extends tests of the retained earnings hypothesis contained in three prior studies.
It shows that the findings in support of the retained earnings hypothesis can be
attributed to specification and measurement choices that bias the results in favor
of the hypothesis. The support for the retained earnings hypothesis is weaker when
the sources of the bias are removed. However, some support for the hypothesis remains
for a limited set of distributing firms.