5.1. Time-series results using the Basu DT model
To test whether the Basu-based conservatism coefficients of our test companies differ between the years when they
overstate earnings (t¼ T), and those subsequent to the overstatements (t ¼Tþ t), we estimate a variant of model (3) using
only test firm data pooled across years. All years in the pre-overstatement period are deleted for this test.
Table 5 presents the regression estimation results. We find that the coefficient of RETit  NEGit is significantly positive
and, more importantly, the coefficient of POSTit  RETit  NEGit is positive and highly significant. These results indicate that
in the post-overstatement period, test companies report more conservatively than in the overstatement period, as earnings
become asymmetrically timelier in recognizing bad news.