Please use this identifier to cite or link to this item:
Specification and Measurement Issues in Unexpected Earnings Response Regression Models
Cheung, Joseph K.
|Issue Date:||2017-11-15 14:55:06 (UTC+8)|
This research addresses three issues that have not received sufficient attention in the vast literature on the relationship between security returns and unexpected earnings. First, it is hypothesized that the magnitude, as opposed to the sign, of unexpected earnings is not necessarily a better explanatory variable of residual returns, since earnings expectations are largely unknown and thus can only be proxied with error. Second, it is observed that extant studies typically focus on earnings announcements exclusively, leading to the possibility of a missing-variable problem by ignoring other contemporaneous announcements such as dividends. Third, most studies either use a return window that is too long or do not have precise event dates, thus failing to capture the most dramatic price impact of the announcements. This study aims to address these issues by exploring the usefulness of the signs of earnings and dividend changes in explaining security returns, and by using a data set that has precise event dates. Three regressions are estimated and compared. An additive model evaluates the relationship between residual returns and the magnitudes of unexpected earnings and dividends. An interactive model evaluates the relationship between residual returns and five categorical variables defined by the signs of unexpected earnings and unexpected dividends. Finally, the third model evaluates whether both the magnitudes and the signs of the two unexpected variables are jointly significant in explaining residual returns. The evidence shows that the signs of unexpected earnings, in corroboration with those of unexpected dividends, are more useful in explaining residual returns than the magnitudes of the unexpected variables. The implications of this finding for research in this area are discussed.
|Relation:||會計評論, 27, 187-204|
|Appears in Collections:||[會計評論] 期刊論文|
Files in This Item:
All items in 學術集成 are protected by copyright, with all rights reserved.