Traditionally, conditional conservatism is estimated using the Basu (1997) model, which captures the fundamental spirit of conservatism -- the asymmetric sensitivity to losses relative to gains. However, this model also leads to anomalies in the earnings-return relationship in the recent sample periods. I propose and find that a modification to the Basu model -- relating change in earnings (as against earning levels) to stock level -- corrects the earnings-returns anomaly while preserving the essential features of asymmetric sensitivity to losses versus gains. Equally importantly, I find that the modified Basu model reveals lower-system conservatism than previously reported, suggesting that the modified model has substantive implications for the level of conservatism. I provide evidence on the validity of this measure in two ways. First, I confirm that it is associated with known determinants of conservatism such as discretionary accrual behavior in predictable ways. Second, I show that the modified model corrects for the econometric problem of ”the loss effect” in the Basu model, as shown by Patatoukas and Thomas (2011).
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。