4 /304
公司避稅與金字塔結構
4
countries
1
, firms are not required to disclose the structure of corporate pyramids; such
mandatory disclosures in Taiwan provide a natural choice to test our research questions. In
order to measure the “length” of layers, we identify all intermediate layers connecting the
parent company and the lowest-tire firms. That is, the holding company (i.e., the parent
company) at the apex of investment structure indirectly controls firms sitting on the lowest
tier of investment layer through intermediate companies. If firms have multiple chains in
investment structure, we will focus on the longest chain which has the largest number of
intermediate layers, and calculate the number of intermediate layers connecting the parent
company and the lowest-tier firms.
We then examine the association between tax avoidance and the length of corporate
pyramids. Following prior studies (Phillips, 2003; Rego, 2003; Dyreng, Hanlon, and
Maydew, 2010; Robinson, Sikes, and Weaver, 2010), our first measure of tax avoidance is
GAAP effective tax rate, which is designed to capture tax avoidance activities that directly
affect net income via the tax expense. Prior literature suggests that a lower effective rate
reflects higher incidences of tax avoidance (e.g., Gupta and Newberry, 1997; Rego, 2003).
Consistent with our expectation, we find that firms with more layers within the pyramidal
structure engage in more tax avoidance than do firms with fewer layers. Our results are the
same when measuring tax avoidance with cash effective tax rate and long-term cash
effective tax rate over three-year period (five-year period). Our results remain the same
when controlling variables documented in the literature do affect tax avoidance activities,
such as firm size, market-to-book ratio, financial leverage, and etc.
In addition, we measure tax-haven intensity using the number of investees in tax
haven divided by total number of investees and interact the intensity variable with the
number of layers. We find that the well-documented positive association between tax
avoidance activities and having investees in tax haven becomes more pronounced as the
number of layers increases. The results support the notion that the number of layers
facilitates tax avoidance activities.
The relation between layers and tax avoidance may have endogenous issues. In
particular, investment layers help facilitate tax avoidance activities and managers who
intend to engage into tax avoidance activities may tend to build more layers. To mitigate
1 For example, in the US, publicly traded firms are required to disclose their subsidiaries in Exhibit 21
of 10K. However, from Exhibit 21 sections of each firm’s annual 10-K report, the whole structure of
pyramidal ownership is not available as firms are not required to disclose the number of layers in
which each subsidiary is located.