Page 206 - 36-1

Does Litigation Experience Improve Audit Partners’ Audit Quality?

contrast, I estimate Model 1 using data for the bankrupt sample to obtain evidence of GCO
Type II errors. The coefficient for BEFORE is expected to be negative for the bankrupt
sample, suggesting that defendant partners are more likely to commit GCO Type II errors
before litigation than nondefendant partners.

A decrease in GCO Type I errors is observed when the coefficient for AFTER for the
nonbankrupt sample is significantly smaller than that for BEFORE. In contrast, a decrease
in Type II GCO errors is observed when the coefficient for AFTER for the bankrupt sample

is significantly larger than the coefficient for BEFORE. The learning hypothesis would be
supported if the coefficient for AFTER for the nonbankrupt sample is significantly smaller
than the coefficient for BEFORE, or is significantly larger for the bankrupt sample than
the coefficient for BEFORE. However, the contagious hypothesis would be supported if
the coefficient for AFTER for the nonbankrupt sample is not significantly smaller than

the coefficient for BEFORE (no reduction in GCO Type I errors), or is not significantly
larger than that for BEFORE for the bankrupt sample (no reduction in GCO Type II
errors). Finally, the conservative hypothesis would be supported if the coefficient for

AFTER is significantly larger than the coefficient for BEFORE for both the nonbankrupt
and bankrupt samples (i.e., increases in GCO Type I errors and decreases in GCO Type II
errors).

3. Findings

The empirical results for GCO Type I errors show that, compared to other partners,
defendant partners tend to issue GCOs for nonbankrupt firms both before and after being

sued. Moreover, the differences in the likelihood of committing GCO Type I errors
between the pre- and post-litigation periods do not reach statistical significance for the
defendant partners themselves. This suggests that defendant partners fail to improve their
subsequent audit quality and continue to commit GCO Type I errors. Nevertheless, the
contagion phenomenon is mitigated after the district courts rule in favor of the defendant

partners.
Regarding the results for GCO Type II errors before litigation, defendant partners do
not differ from other partners in their likelihood of issuing GCOs to soon-to-go-bankrupt
firms. However, after litigation, defendant partners are more likely to issue GCOs to soon-

198