차별적 손익인식 적시성 모형의 호재/악재 집단 분류변수와 편의

This study examines bias in asymmetric earnings timeliness in the Basu (1997) model by applying the sign of earnings besides the sign of returns to partition the sample into bad or good news groups. The sign of earnings as a partitioning variable can be justified by Hayn (1995) that documents asymmetric earnings response coefficient between profit and loss firms. Accordingly, I partition the sample into (1A) positive returns and earnings group, (1B) positive returns and negative earnings group, (2A) negative returns and positive earnings group, and (2B) negative returns and earnings group. Empirical finding shows that the negative returns and earnings group is the most conservative in terms of cumulative non-operating accruals and market-to-book equity ratio and that the significantly positive asymmetric timeliness is consistently observed for the negative returns and positive earnings group. Adoption of the sign of earnings as an additional partitioning variable reduces the bias by 40% and enhances the explanatory power of the model by 50%, implying that adopting the sign of earnings successfully mitigates problems in the Basu model. However, this group is not more conservative than other groups partitioned on the signs of returns and earnings, suggesting that asymmetric timeliness might represent a phenomenon qualitatively different from conditional conservatism. This study contributes to the literature by providing an implication that information users pay attention to the partitioning variables in understanding and identifying the source of the bias in asymmetric timeliness of earnings.