To test the assertion from Maydew and Shackelford 2006 that separating audit
and tax service provision potentially leads to higher audit quality, we next examine
the relation between missing consensus earnings forecasts absent tax expense management
and third-to-fourth-quarter ETR changes for firms that do not pay tax fees
to their auditors. We venture to determine whether these companies’ ETRs
changed between the third and fourth quarters and whether the magnitudes of these
changes differed for firms that met or missed their consensus earnings forecasts
absent ETR changes. We add a dichotomous variable, Other, coded 1 for firms that
do not pay tax fees to their auditors and interact this variable with Miss. We
acknowledge that these firms likely invested in tax planning by purchasing these
services from lawyers or accountants other than their auditors or by staffing “inhouse”
tax departments. Although we cannot quantify these firms’ spending on tax
services because the Audit Analytics database only contains data on firms’ tax fees
paid to their auditors, our focus with the Other variable is to determine whether
these firms also decrease ETRs when they would otherwise miss consensus earnings
targets. Thus, we code Tax_Fees as $0 and Other as 1 for these firms. We only
use Other in years 2002 through 2004 because we cannot distinguish whether a
missing observation for Tax_Fees indicates that the firm did not pay its auditors for
tax services or if tax fees paid to auditors were included in other non-audit services
before these amounts were required to be reported separately in 2003.9 We fit (3)
using the full sample of 1,802 firm-year observations, which includes an additional
329 observations with values of $0 for Tax_Fees in 2002, 2003, and 2004 that are