臺大管理論叢 NTU Management Review VOL.29 NO.1

223 NTU Management Review Vol. 29 No. 1 Apr. 2019 Table 5 reports the empirical results from Reg. (1) based on the unbalanced-panel fixed effect model (denoted as the IM model). The coefficients of X t and X t3 are respectively 1.386 ( t = 8.45) and 0.387 ( t = 6.21), which is positive and statistically significant at the 1% level. This result is consistent with the findings documented by Collins et al. (1994) and provides evidence supporting the use of the CKSS model as a benchmark model to examine the hypotheses. Importantly, the coefficient of IM t *X t is -10.060 ( t = -5.07), which is negative and statistically significant at the 1% level and supports the first hypothesis. The informativeness of current earnings definitely decreases for firms which recognize asset impairment. Meanwhile, the coefficient of IM t *X t3 is 2.041 ( t = 2.23), which is positive and statistically significant at the 5% level. This means that the informativeness of future earnings is enhanced for firms with recognized asset impairment. The second hypothesis gains empirical support in the analysis. As conjectured, the recognition of asset impairment presents distinctive informativeness patterns for current and future earnings, i.e., the informativeness of current earnings (future earnings) is decreased (increased) when firms recognize asset impairment. Reg. (2) (denoted as the REV model) considers reversal of asset impairment. Table 5 shows that the coefficients of IM t *X t and IM t *X t3 are respectively -11.350 ( t = -4.79) and 2.871 ( t = 3.07), and both statistically significant at the 1% level. This suggest that the informativeness of current earnings (future earnings) is decreased (increased) for firms recognizing impairment losses without reversals in the following year. However, the coefficient of REV t *IM t *X t3 is -16.300 ( t = -1.53), which is negative and statistically insignificant. Thus, the third hypothesis does not gain empirical support in the analysis. It is interesting to note that the combined coefficient of IM t *X t3 and REV t *IM t *X t3 ( β 8 + β 13 ) is -13.429 ( t = -1.23), which represents the entire informativeness of future earnings for firms with asset impairment reversals, and is negative and statistically insignificant. This result suggests that impairment reversals weaken the positive association between informativeness of future earnings and initial asset impairment recognition, which lead to no significant difference in earnings informativeness of impairment reversals firms and that of firms without impairment. The coefficient of REV t *IM t *X t is 35.498 ( t = 1.67), which is positive and marginally statistically significant. The combined coefficient of IM t *X t and REV t *IM t *X t ( β 7 + β 12 ) is 24.148 ( t = 1.12), which represents the entire