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

Earnings Informativeness of Long-Lived Assets Impairment Recognized and Reversals 236 5.4 Subsample Examinations In the initial analysis, this study uses a dummy variable for the reversing sub-sample in Reg. (2) to examine whether the informativeness of future earnings is mitigated for firms with asset impairment reversals. An alternative approach to test the third hypothesis, based on Reg. (1), but directly examining the “with” versus “without” impairment reversals subsample. This study reruns Reg. (1) using three subsamples: (1) 7,864 observations for the “Non-reversals vs. Non-impairment” sub-sample (875 non-reversing plus 6,989 without impairment observations); (2) 7,057 observations for the “Reversals vs. Non-impairment” sub-sample (68 reversing plus 6,989 without impairment observations); and (3) 943 observations for the “Reversals vs. Non-reversals” sub-sample (875 non-reversing plus 68 non-reversing observations). The empirical results are presented in Table 9. From the “Non-reversals vs. Non-impairment” model in Table 9, the coefficients of IM t *X t and IM t *X t3 are respectively -11.055 ( t = -4.70) and 2.696 ( t = 3.13), both statistically significant at the 1% level. The coefficients of IM t *X t and IM t *X t3 are respectively 29.782 ( t = 0.96) and -12.965 ( t = -1.30) in the “Reversals vs. Non- impairment” model, both statistically insignificant. The empirical result from “Reversals vs. Non-reversals” model (a firm reversed its impairment is denoted as one) is reported in the right-hand column of Table 9. The coefficients of IM t *X t3 and REV t *IM t *X t3 are respectively 10.959 ( t = 2.06) and -32.948 ( t = -2.25), both statistically significant at the 5% level. However, the combined coefficient of IM t *X t3 and REV t *IM t *X t3 is -21.989 ( t = -1.46), negative and statistically insignificant. 17 These additional results reveal that the informativeness of current earnings (future earnings) decreases (increases) when firms recognize the impairment of long-lived assets without reversals in the following year. Nevertheless, the informativeness of future earnings is mitigated for impairment reversal firms. This provides additional empirical support for the findings. 18 17 We use impairment reversals ratio, which is measured by the reversal amounts in year t +1 divided by the impairment recognized in year t , to replace the dummy variable for the firms with assets impairment reversals in the initial model and rerun “Reversals vs. Non-reversals” model. This setting allows this study to further examine the effects of assets impairment with different magnitude of reversals, compared with the non-reversals sample, on earnings informativeness. The untabulated results reveal that the coefficients of IM t *X t and IM t *X t3 are 11.819 ( t = 1.17) and 10.037 ( t = 1.90). The coefficients of REV t *IM t *X t and REV t *IM t *X t3 are 65.919 ( t = 1.43) and -71.232 ( t = -2.15). The results again reveal the informativeness of future earnings is mitigated for the impairment reversals firms. 18 We also match the 68 reversing samples with the same observations who recognized impairment yet without reversals using Reg. (3) and rerun the equation. The results do not qualitatively change the primary findings.