Does the Accrual Anomaly End When Accruals Reverse?

Using quarterly data, we develop a two-step procedure to detect accrual initiations in a quarter and trace their reversals in a future quarter. The constructed “accrual initiations with reversals” sample allows us to directly test the investor fixation explanation for the accrual anomaly, i.e., that the anomaly ends by the time accruals reverse, and will not be present beyond the reversal point. Our tests yield two key results: a) strong evidence of the accrual anomaly over the period from the accrual initiation to reversal, but not thereafter; and b) the anomaly ends only when accruals reverse and not when an extreme change in cash from operations occurs before accruals reverse. These results provide more complete and more convincing evidence consistent with the fixation explanation than that provided by prior studies. We also show that the accrual-based trading strategy can be significantly improved by ex-ante detecting accrual reversals using our procedure.