Our paper examines inconsistent conclusions regarding the accrual anomaly and demonstrates the importance of aligning regression specifications with hypotheses. Richardson, Sloan, Soliman, and Tuna (2005) conclude that accruals are mispriced and the mispricing seems to increase as accrual reliability decreases. Barone and Magilke (2009) and Ball, Gerakos, Linnainmaa, and Nikolaev (2016) conclude that cash flows rather than accruals are mispriced. We show that the divergent conclusions come from misalignment between the null hypothesis and regression specification in Richardson et al. (2005) . In addition, analysis of the contemporaneous relations between stock returns and components of earnings supports an initial underreaction to cash flows by investors. We fail to detect links between the reliability measures in Richardson et al. (2005) and investor behavior once we align the statistical tests with the null hypothesis. Our reexamination of prior findings benefits accounting academics, standard setters, and others interested in how investors use earnings components. JEL Classifications: M41. Data Availability: All data used in this study are publicly available from the sources identified in the text.
