Page 85 - 34-3

NTU Management Review Vol. 34 No. 3 Dec. 2024

Hypothesis 1: Other things equal, there exists a negative association between tax risk and
firm value.

Hypothesis 2: Other things equal, there exists a positive association between tax avoidance

and firm value.

Hypothesis 3: Other things equal, tax risk moderates the positive valuation of tax

avoidance.

2. Design

We compile the data from the Taiwan Economic Journal (TEJ), which provides

financial statement information of domestic listed companies in Taiwan. The empirical
model is based on data spanning the period from 2000 to 2019. After excluding TDRs,
F-shares, and financial and insurance companies, the initial sample consists of 28,158
observations. Additionally, we remove 4,218 observations with negative pre- tax income

and 9,163 observations that are missing data which is required to calculate our variables of
interest from the dataset, resulting in a final sample comprising 14,777 observations.
Hypothesis 1 posits that higher tax risk is associated with lower investor evaluations
of the company. To test this hypothesis, the paper constructs a regression model with

firm value as the dependent variable, measured using Tobin’s Q (TOBINSQ). The primary
independent variable is tax risk, proxied by the standard deviation of the five-year cash
effective tax rate (SD_CETR5). We estimate the following regression model using ordinary
least squares (OLS):

TOBINSQ = β + β SD_CETR5 + ρ CONTROL + IND + YEAR + ε , (1)
0
i
1
it
it
it
it
t
where i denotes the firm and t denotes the year. CONTROL represents the control variables,
including return on assets (ROA), five-year standard deviation of ROA (VOL_ROA5),
the interaction term of ROA and its standard deviation, sales growth rate (SGR), capital
expenditures (CAPEX), research and development expenditures (RD), net operating loss
(NOL), firm size (SIZE), leverage (LEVERAGE), foreign income (FOREIGN), advertising

75