24 /304
公司避稅與金字塔結構
24
5. Empirical Results
Table 3 presents the regression analysis of model (1). In columns (1), we report the
results of estimating equation (1) using
GAAP_ETR
as the dependent variable. We find
that the coefficient on
LAYER
is -0.011 (p-value < 0.01), which suggests that firms which
have more number of layers tend to be engaged in more tax avoidance activities. In
column (2), we report the results of re-estimating equation (1) using
CASH_ETR
as the
dependent variable. We find that the coefficient on
LYEAR
is -0.008 (p-value < 0.05), also
suggesting that firms which have more number of layers tend to be engaged in more tax
avoidance activities. Likewise, in columns (3), we report the results of estimating equation
(1) using
LT_ETR_3Y
as the dependent variable. We find that the coefficient on
LYEAR
is
-0.008 (p-value < 0.05); in columns (4), we report the results of estimating equation (1)
using
LT_ETR_5Y
as the dependent variable and the coefficient on LYEAR is -0.010
(p-value < 0.05).
In addition to these primary results, several of the estimated coefficients for the
control variables are statistically significant. For example, focusing on column (1), when
GAAP_ETR
is used as a measure of tax avoidance, the coefficients on
CASH, INTAN
, and
ln(NUM_INVESTEE)
are positive and significant, while the coefficients on
SIZE, NOL,
∆NOI, ∆SALES
, and
RD
are negative and significant. In column (2), when the
CASH_ETR
is used as a measure of tax avoidance, the coefficients on
INTAN
, and
ln(NUM_
INVESTEE)
are positive and significant, while the coefficients on
SIZE, MB, LEV, NOL,
∆NOI, STD_ROA, ∆SALES
, and
RD
are negative and significant.
In addition, following Balakrishnan, Blouin, and Guay (2012) and Armstrong,
Blouin, Jagolinzer, and Larcker (2015), we also use the industry-size-adjusted effective
tax rate by adjusting the firm's
GAAP_ETR (CASH_ETR)
with the mean
GAAP_ETR
(CASH_ETR)
of the firm's size and industry peers. The use of these industry-adjusted
measures is based on the premise that, all else equal, similar firms in terms of industry and
size should have similar tax avoidance opportunities. The industry-adjusted measures of
tax avoidance capture cross-sectional variation in firms’ tax avoidance after benchmarking
a given firm’s tax avoidance relative to that of similar-sized firms in the same industry.