Predicting Accruals Based on Cash-Flow Properties

Our goal is to understand the extent to which cash-flow properties explain accruals. Using the Dechow, Kothari, and Watts (1998) model, we derive a negative relation between accruals and cash-flow changes, and show that the strength of the relation is linked to negative serial correlation in cash-flow changes. Dechow et al. (1998) also suggest that the strength of the relation between accruals and revenue changes relates to operating cycle length. Prior accrual models have not incorporated these theoretical relations. We show that incorporating cash-flow changes, serial correlation in cash-flow changes, and operating cycle length increases explanatory power of all accrual models considered (i.e., Jones 1991; Ball and Shivakumar 2006; McNichols 2002; Jeter and Shivakumar 1999). We find that incorporating these variables in accrual models also improves specification and power, aids detection of earnings management in AAER firms, and produces a nondiscretionary accrual estimate that better predicts future cash flows and earnings. These results suggest the importance of considering the economic role of accruals when predicting accruals.