

17
臺大管理論叢
第
28
卷第
1
期
INDEP
The proportion of independent directors.
DUAL
An indicator variable that equals one if the CEO also serves as chairman of the
board and zero otherwise.
DEV
The ratio of cash flow rights to control rights.
SALES_R
Related party sales over total sales.
INST
The percentage of common stocks held by institutional investors.
IMR
The inverse Mills ratio generated from the first-stage Probit regression model (3).
The second set of control variables (
ROA, NOL, and ∆NOI
) captures a firm’s
profitability and the presence of net operating loss. More profitable firms tend to have
higher effective tax rates (Chen et al., 2010), but they may have more incentives to be
engaged in tax avoidance (Rego, 2003; Wilson, 2009). Firms with net operating loss may
entail tax loss carryback or carryforward which affects the effective tax rates. The third set
of control variables (
STD_ROA and ∆SALES
) captures performance volatility. It could be
more difficult for firms with higher performance volatility to manage tax saving plan
(Cazier, Rego, Tian, and Wilson, 2009). The fourth set of control variables (
CASH, PPE,
INTAN, RD, ADV, and SGA
) captures a firm’s asset mix and expenditures that could
impact its ETRs. In particular, we control for the level of firms’ cash holdings to account
for firms’ cash needs that might be necessary for certain types of tax avoidance (McGuire,
Omer, and Wang, 2012); the tax code typically allows corporations to take depreciation
expense on property, plant, and equipments over periods much shorter than their economic
lives. Thus, more capital-intensive firms are expected to have lower ETRs (Gupta and
Newberry, 1997); the tax code also grants corporations the differential book and tax
treatments of intangible assets and RandD expenditures. RandD intensive firms are
especially affected by RandD-encouraging tax credit (Chen et al., 2010; Grubert and
Slemrod, 1998). We also include firms’ advertising expense and SGandA expense since
firms can manage tax avoidance plans via discretional spending such as advertising
expenses and selling, general, and administrative expense (Dyreng et al., 2010). Finally,
we control the number of investees of a firm since firms with more investees can use
transfer pricing and income shifting among the parent and investees to reduce tax burden.
We also control for firms’ industry and year because firm-specific characteristics might
vary systematically by industry and economic environment over time (Rego, 2003).