Changes versus Levels in Earnings As Explanatory Variables for Returns: Some Theoretical Considerations
从理论角度探讨了用收益变量解释同期股票回报的模型设定问题，指出传统上用收益变化作为解释变量的做法缺乏严格论证，对会计与金融研究者有参考价值。
This paper considers theoretical aspects of the specification of an earnings variable to explain contemporaneous returns. The traditional approach in market-based accounting research centers on changes in earnings, deflated by beginning-of-period price, as the benchmark explanatory variable for returns. The reasons for using this variable appear to follow a three-step argument. First, unexpected earnings determine the theoretically correct concept of the earnings variable; second, the change in earnings approximates unexpected earnings because current earnings approximate expected earnings, i.e., the random walk hypothesis applies;1 third, the beginning-of-period price deflates the earnings change variable to make it dimensionally consistent with returns.2 While this argument seems overly heuristic, we are unaware of a generally acceptable analysis that provides a more compelling argument. This lack of rigor has recently become an issue because