Accruals and firm life cycle: Improving regulatory earnings management detection

Abstract

Regulators have invested considerable energy into developing analytical tools to better detect earnings management. We propose that firms in similar life cycle stages (LCSs) face similar strategic concerns, managerial pressures, growth prospects, etc., and that the commonality in these factors contribute to the “normal” accruals generating process. Consistent with this prediction, we simulate various earnings management conditions and find that accruals models are misspecified in detecting manipulation within particular LCSs; in particular, introduction, shakeout, and decline firms are over-identified as manipulators, while growth and mature firms are under-identified as manipulators when LCS is not used to estimate accruals. Weighted average performance across life cycle stages reveals that LCS estimation of discretionary accruals substantially improves successful detection and reduces Type I errors relative to other grouping alternatives. The combined improvement across both Type I and Type II errors is over 70% for both the modified Jones and discretionary revenue models of accruals-based earnings management.