A Competing Explanation for Sources of Bias in Earnings-Based Measures of Conditional Conservatism

• In Event: Asymmetric Timeliness of Recognition

Abstract

This paper reexamines the bias in earnings-based differential timeliness (DT) coefficients that Patatoukas and Thomas (2011a) attribute to scale-related variance and loss effects and that Ball, Kothari and Nikolaev (2012b) attribute to failing to control for the covariance between expected earnings and expected returns. We show that the majority of the bias identified in both studies is attributable to operating cash flow (CFO) asymmetry, which does not capture differential verification thresholds for recognizing good and bad news about expected future cash flows. We show that accrual-based differential timeliness coefficients do not suffer from the bias documented in Patatoukas and Thomas (2011a) and Ball et al (2012b). We also show that accrual-based differential timeliness coefficients are related to factors that affect the demand for conservative reporting, both in the cross-section and over time. Our results suggest that accrual-based differential timeliness measures will be less biased than earnings-based DT measures and has the ability to detect predictable differences in conditional conservatism.

Authors

• Xiaoli (shaolee) Tian, The Ohio State University

• Daniel W Collins, The University of Iowa

• Paul Hribar, The University of Iowa