Stock Returns and Accounting Earnings
推导并检验了当前意外回报与意外收益的关系，该关系纳入了未来收益预测的修正，解决了传统回归中因遗漏信息导致的低解释力和系数偏误问题，对会计收益价值相关性研究有重要参考。
In this paper we derive and test a relation between current-period unexpected returns and unexpected earnings that incorporates revisions in forecasts of future earnings. Our motivation is to emphasize the misspecification in returns/earnings regressions that omit information currently available about future earnings and offer a solution. Since changes in expectations of future earnings are related strongly to unexpected returns, those regressions have low explanatory power. More important, coefficient estimates are biased because the omitted and included variables are correlated, and explanatory power is even lower when that correlation varies within the sample. The relation we derive extends the simple regression of unexpected returns on unexpected earnings, often used in studies examining the value relevance of accounting earnings (beginning with Ball and Brown [1968]), to a multiple regression. The additional regressors, which reflect information contained in forecast revisions and discount rate changes occurring during the year, are identified using the abnormal earnings valuation model.' Relative to the simple regression, the multiple regression improves explanatory power significantly (R2 values increase from