12 / 372
審計人員之產業專精與客戶租稅規避：中國實證研究
12
Control Variables
Big10
= 1 when the audit firm is the top 10 based on revenue according to the CICPA’s top 100
audit firm list, otherwise 0.
Soe
= 1 for state-owned firms and zero otherwise.
Size
= the natural log of total assets
Roa
= the ratio of net income over total assets
Lev
= the ratio of total net income over total assets
CFO
= cash flow from operations scaled by the beginning total assets
4. Data and Discussion of Empirical Results
4.1 Sample Selection Process
We use all A-share listed companies on the Shanghai and Shenzhen Stock Exchange as
our initial sample, which consists of 10,272 firm-years for fiscal years 2008-2012. This study
excludes some observations based on the following sample selection criteria: (1) exclude
financial companies; (2) firm-years with negative or missing values of income tax expense;
(3) firm-years with negative or missing values of equity; (4) industry-years with auditors less
than five and auditor-years with clients less than five;
9
(5) firm-years with missing values of
audit fee or audit tenure; (6) firm-years with missing values of control variables. The final
sample usable for the study is 7,692 firm-years, with complete information on all variables.
This study collects the financial information from the China Stock Market and Accounting
Research (CSMAR) database. This study hand-collects audit fee, audit tenure, and tax rate
from annual financial reports of listed companies. Table 2 reports the sample selection
process:
Table 2 Sample Selection Process
Initial sample for fiscal year 2008-2012
10272
Less: the financial companies
(226)
Less: firm-years with negative or missing values of income tax expense
(149)
Less: firm-years with negative or missing values of equity
(964)
Less: industry-years with auditors less than five and auditor-years with clients less than five
(196)
Less: firm-years with missing values of audit fee or audit tenure
(103)
Less: firm-years with missing values of control variables
(842)
Equals: final sample
7692
9 Because if the industry-years and auditor-years are too few, the measurement error of auditor industry
expertise will be large.