26 /304
公司避稅與金字塔結構
26
Among firms with similar tax avoidance opportunities, firms with lower effective tax rate
can be considered to be engaged in more tax avoidance (Balakrishnan et al., 2012). Table
4 presents the results of our multivariate analyses using industry-adjusted effective tax
rates as dependent variables. Specifically, using
IND_GAAP_ETR
as a dependent variable,
we find that the coefficient on
LYEAR
is -0.011 (p-value < 0.01); using
IND_CASH_ETR
as a dependent variable, the coefficient on
LAYER
is -0.010 (p-value < 0.01). Firms which
have more number of layers tend to be engaged in more tax avoidance activities. We also
get the qualitatively similar results using
IND_LT_ETR_3Y
and
IND_LT_ETR_5Y
.
Overall, these results suggest that the controlling shareholders (along with managers) can
conduct tax avoidance activities by creating complex structures to obscure the underlying
intent and to avoid detection by the government. The increase in the number of layer
grants pyramidal firms more opportunities to take tax avoidance without being detected.
Table 5 presents the regression model (2). The definition of tax haven countries is
based on Durnev et al. (2017). In columns (1), (2), (3), and (4), we report the results of
estimating equation (2) using
GAAP_ETR, CASH_ETR, LT_ETR_3Y
, and
LT_ETR_5Y
as
the dependent variables respectively. In column (1) and (2), We find that the coefficient on
LAYER
is -0.019 (p-value < 0.001) and -0.022 (p-value < 0.01), which is consistent with
the results of model (1), suggesting that firms which have more number of layers tend to
be engaged in more tax avoidance activities. Moreover, the coefficient on
LAYEAR*TAXH1
is -0.024 (p-value < 0.05) and -0.044 (p-value < 0.01), suggesting that
the positive association between tax avoidance activities and having subsidiaries in tax
haven countries becomes more pronounced as the number of layers in corporate pyramids
increased. Likewise, in column (3) and (4), we find that the coefficient on LYEAR is
-0.024 (p-value < 0.001) and -0.030 (p-value < 0.001), which is consistent with the results
of model (1). Moreover, the coefficient on
LAYEAR*TAXH1
is -0.049 (p-value < 0.01) and
-0.047 (p-value < 0.01).
Table 6 presents the results using the definition of tax haven countries as in Dyreng
and Lindsey (2009). The results qualitatively lead to similar inferences. Overall, these
results suggest that the positive association between tax avoidance activities and having
subsidiaries in tax haven countries become more pronounced as the number of layers in
corporate pyramids increases.