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

Earnings Informativeness of Long-Lived Assets Impairment Recognized and Reversals 220 share (X t ) in year t is respectively 0.263 (0.042) and 0.088 (0.081) in the entire sample. The mean (median) of total earnings per share (X t3 ) for three years in year t is 0.301 (0.218). The mean (median) of assets recognized as impaired (IM t ) is 0.002 (0.000) of the total assets in the entire sample. However, the mean (median) of impairment for firms without reversals (875 observations) and with reversals (68 observations) is respectively 1.53% (0.6%) and 2.11% (1.5%). The magnitude of recognized impairment for the reversing firms is larger than that of firms without reversal. The mean recognized impairment reversals is 1.27% (untabulated), suggesting that approximately 60% of the initially recognized impairment was reversed in the following year. Given the rather large standard deviation of our empirical variables, this study adopts White’s (1980) heteroskedastity consistent covariance matrix estimator to correct estimates of the coefficient covariances in the possible presence of heteroskedasticity in all regressions. Table 4 presents the correlations among the related variables. Asset impairment (IM t ) is found to be significantly negative-associated with past, current, and future earnings. Thus firms with significant asset impairment have negative earnings performance. We find the correlation between annual stock returns (R t ) and recognized asset impairment (IM t ) is negative and statistically significant, indicating that a recognized asset impairment results in negative stock returns. While most of the independent variables are highly correlated with each other, the variance inflation factors (VIF) of the pivotal interactive variables, i.e., IM t *X t , IM t *X t3 in Reg. (1) and REV t *IM t *X t , REV t *IM t *X t3 in Reg. (2) are 2.364, 1.177, 2.210, and 1.341, respectively, and do not suggest severe multicollinearity problems (Neter, Wasserman, and Kutner, 1989).