Post-Earnings Announcement Drift?

The predictability of abnormal returns based on information contained in past earnings announcements is an anomaly that is statistically and economically significant. Neither is it illusory, nor is it an artifact of the experimental design. It may be a result of market inefficiency. Our results cannot rule out this explanation. However, we find that earnings change numbers are associated with the probability that firms leave the sample through acquisition, bankruptcy or for other reasons, or are not included in the sample in the first place. Moreover, we find that the magnitude of the post-earnings announcement effect is correlated with factors that proxy for the ex ante probability of the firm surviving to be part of the earnings surprise sample. It also appears to be related to determinants of the bid-ask spread