Do Firms Time Changes in Accounting Estimates to Manage Earnings?†

Abstract

ABSTRACT Prior earnings management research often focuses on specific accounts or on estimations of discretionary accruals but provides only limited insight into the methods firms actually use to manage earnings. In order to begin exploration of some of the operational details regarding how earnings are managed, we investigate whether firms time their decisions to make changes in accounting estimates (CAEs) in consideration of their earnings benchmarks. Using CAE data across all accounts from 2006 to 2018, we find that 28.1% of income‐increasing CAEs are implemented in quarters where pre‐CAE earnings are below a forecasted earnings benchmark but inclusion of the CAE effectively allows the firm to meet the benchmark. We find that income‐increasing CAEs are more likely implemented when a firm's pre‐CAE earnings are further below the benchmark. We also find that firms are more likely to implement income‐decreasing CAEs under two scenarios: (i) when pre‐CAE earnings are relatively high, as a way either to smooth earnings or to “bury bad news,” and (ii) when pre‐CAE earnings are already low, as a way to take a financial “big bath” and position the firm for positive future earnings. In addition, we present evidence that firms using CAEs to achieve an earnings benchmark face financial consequences in terms of poorer immediate stock price performance and subsequent return on assets. Our conclusions hold after performing several additional analyses, including consideration of other discretionary options, addressing endogeneity concerns, and conducting falsification tests. In sum, we contribute to the earnings management literature by presenting consistent evidence that firms appear to time CAEs to meet earnings benchmarks or achieve other reporting objectives.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 13 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.