Estimating a Firm-Year Measure of Conditional Conservatism for Non-U.S. Firms: Evidence from Japan

The firm-year-level accounting conservatism measure known as C_Score, developed by Khan and Watts (2009), is frequently used in accounting research internationally. However, as they developed the measure was developed the measure for the United States (U.S.), it may not be appropriate for other countries. Thus, we investigate its validity in the Japanese context, where the institutional features substantially differ from those in the U.S. We begin by replicating the tests used by Khan and Watts (2009) and find that some results are consistent with those of the U.S. However, our additional tests suggest that the use of the C_Score in Japan is not necessarily appropriate. First, the coefficients of two of the three coefficients of C_Score factors—market-to-book ratio and leverage—are not statistically significant for many years, and their signs differ from the predictions. Second, we find that the relationship with each of the C_Score factors is unstable and varies over time; furthermore, firm size solely determines the C_Score for several investigated years. Finally, contrary to previous research in the U.S., we find no evidence of the predicted association between C_Score and asset impairment losses, which are typical, conditionally conservative accruals. Overall, our findings suggest that researchers should confirm the validity of the C_Score before applying it to non-U.S. samples.