The information content of earnings announcements with varying return parameters
Abstract
Changes in beta (systematic risk of a firm) over time have received increasing attention in the accounting and finance literatures. However, few studies on the information content of earnings announcements have considered the possibility of changing betas. It is well known that betas vary due to various macroeconomic and microeconomic events. Among the various economic events, this study concentrates on earnings announcements. We also consider alpha changes in this information content study. This study is to re-examine some issues regarding the information content of quarterly earnings announcements with the possible variability of parameters in the market model. Before proceeding to the empirical tests theoretical positive linkages between earnings changes (unexpected earnings) and beta changes is provided. We then test mainly: (1) the significance of parameter (alpha and beta) shifts during earnings announcement period, (2) the directional relationship between parameter changes and earnings changes, (3) the significance of abnormal returns when abnormal returns are measured based on varying parameters, and (4) the directional relationship between varying-parameters-based abnormal returns and unexpected earnings. The tests are conducted using the BERAB (Bayesian Estimators for Random Alpha and Beta) approach which is devised for the purpose of estimating varying parameters. Empirical results in this study report that changes in betas are positively and significantly related to changes in earnings and that alphas vary during event period of earnings announcement. They also show that, when parameter changes are considered in measuring expected rate of returns within the framework of the market model, the expected returns are closer to actual realized returns at the portfolio and individual firm levels comparing to the expected returns estimated with OLS parameters. As a result, BERAB-based abnormal returns, which is the difference between realized returns and expected returns, reduce toward zero at the levels of portfolio and individual firm. Another result shows that the directional relationship between BERAB-based abnormal returns and unexpected earnings is weaker than the directional relationship between OLS-based abnormal returns and unexpected earnings. The above results are mainly due to the fact that BERAB-based expected returns contain the information from earnings changes, while OLS-based expected returns do not. This difference between the two types of expected returns cases BERAB-based abnormal returns and OLS-based abnormal returns to behave differently.
Degree
Ph.D.
Advisors
Ro, Purdue University.
Subject Area
Accounting
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.