Essays on earnings persistence, earnings trends, and the predictability of post-earnings-announcement stock returns

This dissertation studies the effects of two earnings attributes, earnings persistence and earnings trends, on investors' under- and overreaction to earnings information. I first provide evidence that the association between post-earnings-announcement abnormal returns and earnings changes depends on fundamentally determined earnings persistence. The association is positive when earnings persistence is high, negative when earnings persistence is low. This evidence indicates that the market underreacts to high-persistence earnings and overreacts to low-persistence earnings. I also show that historical earnings trends affect investors' assessment of earnings persistence. When current earnings changes are consistent with past earnings trends, post-earnings-announcement drift is reduced and investors' underestimation of high-persistence earnings significantly decreases. The findings of this thesis are consistent with stock prices behaving as if investors' expectations of earnings persistence are moderated toward an average level that is insufficiently high or low relative to actual earnings persistence. Moreover, investors' assessment of earnings persistence appears to be affected by the patterns of past earnings series: investors seem to believe that earnings changes consistent with prior earnings trends are more persistent than earnings changes inconsistent with prior earnings trends.