Similar to GHN, this study investigates whether Basu-based measures of conservatism are able to detect situations in
which companies’ earnings are known to be significantly overstated, as revealed by subsequent restatements of earnings.
We examine whether Basu-based metrics detect changes in conservatism in companies’ income numbers surrounding
previous earnings overstatements. Our main findings can be summarized as follows. First, Basu’s DT conservatism metric
and the related C-Score metric reflect a lower conservatism level for restatement firms than for control firms in periods of
overstated earnings. Second, both Basu-based metrics reflect increases in conservatism of test companies’ earnings in
periods subsequent to overstatements. The increased conservatism is evident using both a time-series approach and a
cross-sectional approach. Third, increased conservatism is detectable only for test companies that improve corporate
governance subsequent to restatements. Test companies that do not improve corporate governance do not exhibit
increases in conservatism. Overall, these results are consistent with both Basu’s (1997) DT coefficient and Khan and Watts’
(2009) C-Score capturing predictable variation in conservatism around earnings overstatements.