Lagged Earnings Asymmetry in a Firm-Year Measure of Accounting Conservatism

Khan and Watts (2009) develop a firm-year measure of conditional conservatism, labeled C_Score, that builds on the Basu (1997) asymmetric timeliness (AT) measure. However, recent research documents an asymmetric relation between lagged earnings and current returns, indicative of bias in the Basu measure. We demonstrate that this lagged earnings asymmetry (LEA) taints C_Score by mimicking (i.e., overstating) C_Score's relation to nearly all of the firm characteristics used by KW as validation tests. Thus, LEA represents an alternative interpretation for hypotheses tests involving C_Score as a measure of conditional conservatism. We examine two very distinct explanations for LEA identified in prior research and demonstrate that controlling for both is necessary and sufficient to yield a modified C_Score measure that is uncorrelated with LEA. We conclude that while LEA identifies a threat to the usefulness of C_Score as a firm-year measure of conservatism, the underlying causes of LEA can be adequately addressed