Like DGM 2004, we include a dummy variable for firms missing their forecasts
absent tax expense management (
Miss
). If the DGM 2004 earnings management
explanation holds for our sample, the coefficient on the variable of interest,
Miss
,
will be negative, indicating that firms decrease their ETRs more between the third
and fourth quarters when they fall short of analysts’ targets (absent changes in
ETR) than when they reach or exceed these targets. Additionally, on the basis of
DGM 2004, we expect a positive coefficient for
Induced_Chg_ETR
and
Tax_Owed
, and a negative coefficient for
ETRQ3
. Also following DGM 2004,
when estimating each of our models, we calculate Huber-White
t
-statistics by clustering
observations by firm.
We then expand on DGM 2004 by adding auditor-provided tax service fees
(
Tax_Fees
) and the interaction between
Tax_Fees
and
Miss
to investigate the relation
between investments in tax planning and changes in ETR between the third
and fourth quarters, both for firms that would meet or beat their consensus earnings
forecasts without third-to-fourth-quarter ETR changes and for those that would
otherwise miss.
7
We estimate (2) using a reduced sample of 1,473 firm-year