Discretion Lost? Accrual Heterogeneity and Implications for Accrual Quality Inferences

Abstract

The empirical regularity that accruals persist into future earnings at a lower rate than cash flows is frequently interpreted as evidence of low accrual quality. We predict that accrual heterogeneity significantly impacts this empirical regularity. To test our prediction, we construct an extensive dataset of more than 100,000 accruals that reconcile net income to cash flow from operations. The accruals we identify are distinct from working capital accrual adjustments and depreciation. Our results show that these accruals vary considerably in their persistence and this variation reflects differences in: (i) business activities; (ii) accounting measurement rules; and (iii) annual recurrence. We find that some accruals that prior research assumed had higher estimation risk are persistent, while other accruals assumed to be measured reliably have low persistence. Our results highlight the importance of considering the interaction between the business activity and measurement rules before attributing differences in accrual persistence to estimation error or managerial discretion. Finally, we develop a novel firm-level accrual line item measure of reporting consistency. We show that our consistency measure captures unique aspects of reporting quality and that reporting consistency can induce both higher or lower earnings persistence depending on the measurement rules attributable to the underlying accrual.

Keywords: earnings quality, accruals, consistency, persistence, aggregation, disaggregation

JEL Classification: G12, M41