1

Trade Liberalization and Corporate Income Tax Avoidance

Yao Lu Xinzheng Shi*

Tsinghua University

December 2016

Abstract

We identify the effect of trade liberalization on corporate income tax

avoidance in a sample of Chinese manufacturing firms, using China’s entry

into the World Trade Organization (WTO). We find that firms engage in

more tax avoidance in industries with larger tariff reductions. We also find

that firms with better corporate governance engage in less tax avoidance than

their counterparts. Further analysis shows that firms with a lack of cash or a

high demand for cash before WTO entry tend to engage in more tax

avoidance after WTO entry. Our study also provides evidence that

manipulating costs is one way that firms avoid corporate income tax.

Keywords: Trade liberalization; tax avoidance; WTO entry

JEL code: D22; F61; F63; H26

* Yao Lu is an Associate Professor in the Department of Finance in the School of Economics and Management, Tsinghua University; Xinzheng Shi is an Associate Professor in the Department of Economics in the School of Economics and Management, Tsinghua University. We wish to thank Rui Wang for providing excellent research assistance, and Hongbin Li and the participants in the CCER Summer Institute 2016 for helpful comments. Corresponding author: Xinzheng Shi (shixzh@sem.tsinghua.edu.cn).

2

1. Introduction

After the Second World War, many developing countries chose import

substitution as a strategy for industrialization. However, in the past three

decades, many countries have begun to favor global economic integration,

particularly trade liberalization, as a development strategy. Previous research

has documented that, faced with a more competitive environment after trade

liberalization, 1 firms invest more in technological development, 2 which

improves their productivity3 and therefore the likelihood of their survival.4

Collectively, this leads to rapid growth in the whole economy.5

However, one possible behavioral response of firms to trade

liberalization, i.e., corporate income tax avoidance, has seldom been studied.6

Corporate income tax avoidance is a commonly observed behavior among firms.

Schneider and Ernste (2000) point out that tax avoidance is widespread in

developing countries and estimate that the tax avoidance rate is above 50% in

many low-income countries. In China, the National Auditing Office uncovered

RMB11.89 billion (roughly 1.6 billion dollars) in tax avoidance in 2003, based

on a nationwide investigation of 788 randomly selected companies in 17

provinces and cities (Asian Wall Street Journal, A2, September 20, 2004). In

this paper, we investigate the effect of trade liberalization on corporate income

tax avoidance by manufacturing firms in China, taking advantage of China’s

1 Levinsohn, 1993; Krishna and Mitra, 1998; Brandt et al., 2012; Lu and Yu, 2015. 2 Rodrik, 1988; Brandt and Thun, 2010. 3 Topalova and Khandelwal, 2011; Pavcnik, 2002; Amiti and Konings, 2007; Tybout and Westbrook, 1995; Krishna and Mitra, 1998; Brandt, et al., 2012; Ederington and McCalman, 2007; Trefler, 2004; Schor, 2004; Lileeva and Trefler, 2010; Harrison, 1994. 4 Baggs, 2005. 5 See Nunn and Trefler (2010) and Brandt and Thun (2010). However, some researchers find that the benefits of trade liberalization are not distributed equally among different groups. For example, Topalova (2007) finds that trade liberalization leads to an increase in the poverty rate and poverty gap in rural districts where industries more exposed to liberalization are concentrated. 6 “Corporate income tax avoidance”, “corporate tax avoidance” and “tax avoidance” are used interchangeably in this paper. Following the literature, we do not distinguish between tax avoidance (legal) and tax evasion (illegal) but call both of them tax avoidance.

3

entry into the WTO.

The effect of trade liberalization on tax avoidance is not straightforward

from a theoretical perspective. Previous studies (Levinsohn, 1993; Krishna and

Mitra, 1998; Brandt et al., 2012; and Lu and Yu, 2015) have shown that a

reduction in import tariffs due to trade liberalization increases the level of

competition in the domestic market. As Shleifer (2004) notes, the cost of ethical

behavior is high under heightened competitive pressure. In addition, Slemrod

(2004) finds that firms whose profit performance decreases due to greater

competition may resort to noncompliance so that they have more cash to invest

and thus improve their prospects. Trade liberalization may therefore lead to

higher incentives for corporate income tax avoidance. However, the increased

competitiveness in product markets fostered by trade liberalization may increase

the real value of the same amount of money. That is, one dollar gained or lost

means more for firms in a more competitive environment. As Slemrod (2004)

also notes, people normally value a given amount of punishment higher than the

equivalent amount of tax savings. Thus, trade liberalization may increase the

real cost of being caught, making firms more conservative and leading to a

lower incentive for corporate tax avoidance. Therefore, the theoretical

prediction of the effects of trade liberalization on corporate tax avoidance is not

clear, and empirical studies are needed.7

A common difficulty in studying tax avoidance is how to measure it. In

the literature, book income is usually used as a proxy for firms’ true accounting

profits, and therefore the gap between book and taxable income (scaled by total

assets) is used as a measure of tax avoidance (Desai, 2003, 2005; Desai and

Dharmapala, 2006). However, this approach works only for publicly listed firms,

as book income is usually not available for non-listed firms. In this paper, we

7 Trade liberalization could affect firms’ tax avoidance through other channels such as reducing input tariffs, increasing excess to foreign markets, and changing domestic legal environments. However, robustness checks in our paper show that these possible channels do not drive the main results.

4

follow Cai and Liu (2009) in calculating an imputed profit based on the national

income account. That is, we calculate the imputed profit by subtracting

intermediate inputs from gross output. The gap between imputed profit and

reported profit is our measurement of tax avoidance. However, the imputed

profit can be legitimately different from the true accounting profit, due to

differences in revenue and expense recognition rules between the national

income account and the General Accepted Accounting Principles (GAAP) with

which accounting profit is calculated.8 Fortunately, the panel feature of Chinese

manufacturing firms allows us to control for any time-invariant systematic

difference between these two accounting systems, which may mitigate this

problem. In addition, we conduct several robustness checks, described later, to

justify this measurement.

After China entered the WTO, import tariffs were reduced to a uniquely

low level, which means that reductions were correspondingly larger for

industries with higher tariff levels before WTO entry. Combining before–after

variation and the different levels of tariffs before WTO entry across industries,

we use a difference in difference (DID) strategy to identify the effect of trade

liberalization on tax avoidance, using data from Chinese manufacturing firms.

We find that WTO entry induced firms in industries with a tariff one percentage

point higher in 2001 to engage in RMB76,000 (roughly $12,000) more tax

avoidance, accounting for 5% of the corporate income tax paid in 2001.

Considering that the average tariff decreased by roughly five percentage points

from pre- to post-WTO entry, the induced aggregate corporate income tax

avoidance would account for 25% of corporate income tax paid in 2001.

We conduct several tests to investigate the validity of our identification

strategy. We first check whether the pre-existing time trends of firms’ tax

avoidance are the same for different industries. We then check whether firms

8 For example, asset depreciation rules can be different; gross output in the current year is not equal to revenue in the same year.

5

adjusted their tax avoidance behavior in anticipation of WTO entry. We also

investigate whether the reform of state-owned enterprises (SOEs) and regulation

changes for foreign direct investment (FDI) that occurred in the same period

have a contaminating effect on our main results. WTO entry may change law

implementation, and we check whether this change affects our results. WTO

entry decreases not only output tariffs but also input tariffs, and enlarges firms’

exposure to international markets. We further investigate whether our main

results are affected by these two factors. As processing traders are not affected

by the tariff change, we estimate the same effects for processing traders as a

placebo test. Finally, we check whether sample attrition affects our results. All

of the results of these tests justify our identification strategy.

We then conduct several tests to justify the validity of our tax avoidance

measurement. We first identify a subsample of industrial firms that are publicly

listed and for which information on the book-tax income gap is therefore

available. For this subsample, we find a significantly positive correlation

between the imputed-reported profits gap and book-tax income gap. We then

estimate the effect of WTO entry on the book-tax income gap using this

subsample, and find that WTO entry significantly enlarges the book-tax income

gap, confirming our paper’s main finding. Second, we follow Cai and Liu (2009)

to investigate how WTO entry affects the response of reported profits to

imputed profits. We find that WTO entry weakens the response, which confirms

our finding that WTO entry induces more corporate income tax avoidance.

Third, we investigate whether firms over-report output because of declining

sales income after WTO entry, which may over-estimate imputed profit and

therefore our measurement of corporate tax avoidance. We find no evidence that

firms over-report output. These results provide support for our tax avoidance

measurement.

In addition, we conduct other robustness checks to examine whether our

main finding is robust to different outcomes and explanatory variables. The

6

results show that our main finding is robust to different tests.

After establishing a causal relation between trade liberalization and tax

avoidance, we investigate the heterogeneous effects in terms of firms’ corporate

governance. We find that firms with weak corporate governance (i.e., SOEs)

engage in more tax avoidance than other types of firms in response to the

reduction in tariffs after WTO entry.

We then examine the possible mechanisms through which trade

liberalization affects firms’ corporate tax avoidance behavior. We find that firms

that have higher leverage, lower profitability, or faster growth before WTO

entry engage in more tax avoidance after WTO entry. This finding suggests that

being short of cash or a higher demand for cash could be the reason firms

engage in more tax avoidance in a more competitive environment. Finally, we

investigate how firms engage in tax avoidance by estimating the effects of WTO

entry on costs per unit of sales income (including administrative costs, sales

costs, financial costs, wages, and benefits). The costs for firms in industries with

larger tariff reductions increase more, suggesting that manipulating costs could

be a way to avoid corporate income tax.

Our paper makes several contributions to the literature. First, previous

research has documented that integration with the global economy is an

important factor driving economic growth in developing countries. However,

very few studies investigate firms’ tax avoidance behaviors in response to the

trade liberalization. Our paper is the first to study this issue in a developing

country. 9 Besides the extension of the literature on the effects of trade

liberalization, focusing on developing countries is of particular importance. Tax

administration in these countries is far less effective than in developed countries

such as the U.S., induced tax avoidance is more likely to take illegal forms,

suggesting that the social benefits of economic growth may not be as large as

9 The only other paper studying the same issue is Chen and Lin (2013), using data on publicly listed firms from the U.S. They find that more tariff reductions are related to more tax avoidance, and the effect is greater for financially constrained firms and smaller for firms with better corporate governance.

7

expected. Policymakers need to exert more effort to increase the efficiency of

tax administration to international standards during the process of integration

with the global economy. Second, we also investigate the mechanism through

which trade liberalization affects tax avoidance and one possible approach used

by firms to engage in tax avoidance. Although the results can only be

considered suggestive, they provide useful insight into firms’ tax avoidance

behavior. Third, we find that better corporate governance can mitigate firms’ tax

avoidance behavior, providing additional evidence supporting predictions by

Desai and Dharmapala (2006). Lastly, using China’s entry into the WTO as a

natural experiment, we can credibly identify a causal relation between trade

liberalization and firms’ corporate tax avoidance.

Our paper is also linked to the public finance literature that examines

various determinants affecting tax avoidance activities; see, e.g., Johnson et al.

(1997, 1998), Slemrod (2004), Desai and Dharmapala (2006, 2009), Crocker

and Slemrod (2005), Hanlon and Slemrod (2009), Kim et al. (2011), and Beck

et al. (2014). In particular, Fisman and Wei (2004) identify evidence of

pervasive tariff evasion in China, and Cai and Liu (2009) investigate the

relationship between competition and corporate tax avoidance among Chinese

manufacturing firms. A most recent study by Chen et al. (2016) find that a

substantial fraction of Chinese firms’ response to fiscal incentive for R&D

investment is due to tax evasion.

The rest of this paper is organized as follows: Section 2 introduces the

institutional background; Section 3 describes the data and empirical strategies;

Section 4 presents the main results; Section 5 shows robustness checks and

heterogeneous effects; Section 6 investigates the mechanisms; Section 7 studies

how firms engage in tax avoidance; and Section 8 offers our conclusions.

2. Background

2.1. The Corporate Income Tax System in China

8

The evolution of China’s corporate tax system can be divided into four

periods: (1) 1949-1978; (2) 1979-1994; (3) 1995-2007; and (4) 2008-present.

The central planning system dominated the first period (1949-1978),

when the majority of Chinese industrial firms were state-owned and all

surpluses were sent to the government. Therefore, as described in Wu (2003),

there was no corporate income tax in this period.

The Chinese government introduced an income tax system in 1979. In

the 1980s, large SOEs were subject to an income tax rate of 55%. These firms

also divided after-tax profits between themselves and the government according

to a given formula. The tax rates were 10-55% for small SOEs and 35% for

private firms.

In the 1980s the Chinese government introduced SOE reforms that

focused on “delegating decision power and giving incentives.” These reforms

continued through the 1990s, when a modern corporation system focusing on

corporatization and governance was emphasized. During this process, many

small and medium-sized SOEs were privatized and others were transformed into

corporations. As a part of these reforms, the Chinese government enacted the

Corporate Income Tax Code in 1994, which overhauled corporate taxation.

One feature of the corporate income tax schedule in the 1994-2007

period was that domestic and foreign firms were subject to different tax rates.

With the exception of small firms with taxable incomes of less than RMB30,000,

which paid 18% corporate income tax, all domestic firms paid 33% income tax.

However, most foreign firms paid only 15% corporate income tax. In addition,

various exemptions, tax credits, and reductions were applied to most foreign

firms, meaning that they generally enjoyed much more favorable tax treatment.

The different treatment of foreign firms and domestic firms lasted until

2007. A new Corporate Income Tax Code was announced at the end of 2007.

Since 2008, the corporate tax rate for both domestic and foreign firms has been

set at 25%. The government phased in the increase for tax rates on foreign firms

9

over the following five years.

Tax collection agencies in China were reformed in 1994. Before 1994,

all taxes were collected by provincial tax collectors, and the central government

and provincial governments divided the tax revenue according to a given

formula. In the 1994 reform, taxes were classified into central taxes and local

taxes. Under the supervision of the State Administration of Taxation, a National

Taxation Bureau and provincial taxation bureaus were established to separately

collect central taxes and provincial taxes. Today, corporate income tax is

classified as a central tax and is therefore collected by the National Taxation

Bureau. The growth in corporate income tax, like the growth in the economy,

has been impressive since 1994; total corporate income tax revenue was

RMB93 billion in 1998 and increased to RMB878 billion in 2007 (China

Statistical Yearbook, 2008).

2.2. China’s Entry into the WTO

Before the economic reform in 1978, Chinese trade took place within a

central-planning framework. The State Planning Commission made plans for

almost all of China’s export and import activities, and a few foreign trade

companies controlled by the Ministry of Foreign Trade were responsible for

carrying out export and import plans. In 1977, Chinese trade accounted for only

0.6% of world trade by volume (Lardy, 1994).

Although China began its economic reform at the beginning of the

1980s, trade liberalization did not start at the same time. On the contrary, the

government raised import tariffs on most commodities in the early years of

reform. The average tariff rate was 43% in 1985. This high tariff level was

maintained until 1992 (Lardy, 2002).

Starting in 1986, as part of China’s efforts to join the WTO, the Chinese

government implemented several tariff cuts. The average import tariff rate was

reduced from 43% in 1992 to 17% in 1997. There was little change in tariff after

10

1997 until China joined the WTO at the end of 2001(Lu and Yu, 2015).

According to the agreement, the Chinese government promised to reduce the

average tariff level to under 10% by 2005. Figure 1 shows the changes in the

average tariff rates over time.

The average value of import tariffs hides the variation in tariffs across

industries. In 2001, the variation in tariffs over two-digit industries was large,

equal to 0.10 (compared with the mean, 0.15). Additionally, the different levels

of import tariffs in different industries before WTO entry meant there were

different reductions in tariffs in different industries after WTO entry. Figure 2

shows the correlations between industry import tariffs in 2001 and the tariff

reductions from 2001 to 2002, 2003, and 2004. Figure 2 also shows the

correlation between industry import tariffs in 2001 and the 2002-2004 average

industry import tariffs. We can see from these graphs that industries with higher

tariffs in 2001 had larger reductions in tariffs after China entered the WTO.

3. Data and Empirical Strategies

3.1. Data

In this study, we mainly use two types of data: manufacturing firm data

and import tariffs.

The manufacturing firm data come from the National Bureau of

Statistics of China (NBS). All SOEs and non-SOEs with sales above RMB5

million (roughly $769,000) are required to file a report on their production

activities and accounting and financial information with the NBS every year.

This is the most comprehensive firm-level dataset in China, and it is used to

calculate matrices in the national income account and major statistics published

in the China Statistical Yearbooks. The data collected by this survey have been

widely used by researchers, e.g., Lu, Lu, and Tao (2010); Brandt, Van

Biesebroeck, and Zhang (2012); Brandt et al. (2012); Yu (2014); Lu and Yu

(2015).

11

The NBS issues an identification code for each firm such that we can

construct a firm panel. A four-digit Chinese Industrial Classification (CIC) code

is also assigned to each firm. However, the classification system for the industry

code was changed in 2003, from GB/T 4754-1994 in 1995-2002 to GB/T

4754-2002 after 2002. To achieve consistency in industry codes for the whole

period, we convert the industry codes to GB/T 4754-1994.

Although this dataset includes rich information, some samples are still

noisy, largely because of misreporting by firms. Following the literature (Cai

and Liu, 2009; Yu, 2014; Lu and Yu, 2015), we delete any observations for

which one of the following is true: (i) the value of fixed assets is below RMB10

million; (ii) the value of total sales is below RMB5 million; (iii) the number of

employees is less than 30; (iv) total assets are smaller than liquid assets; (v)

total assets are smaller than fixed assets; (vi) total assets are smaller than the net

value of fixed assets; or (vii) accumulated depreciation is smaller than current

depreciation. After this procedure, we have 308,179 observations, including

103,505 unique firms. All of the monetary values are deflated to 1998 values.

We use data from 1999 to 2004: three years before WTO entry (1999-2001) and

three years after WTO entry (2002-2004).

An important variable in our study is tax avoidance. One widely used

measure of tax avoidance is the book-tax gap (Desai, 2003, 2005; Desai and

Dharmapala, 2006). However, the manufacturing firm dataset we use in this

study does not include information about book income. Therefore, following

Cai and Liu (2009), we use the gap between imputed profits and reported profits

as a proxy for tax avoidance. The reported profits, ��, are the profits reported

by firms when they file their annual reports with the NBS. The imputed profits

are calculated according to the national income accounting system as follows:

�� = � − �� − � − � �� − ���� − ��� (1)

Here, � is the gross output, �� is the intermediate inputs excluding

financial charges, � is the financial charges, � �� is the total wages paid,

12

���� is the current depreciation, and ��� is the value-added tax. All of

these variables are reported by firms. Therefore, the tax avoidance engaged in

by firms, ��, is measured as �� minus ��. We normalize the gap in this

paper using total assets.

One caveat to bear in mind is that the difference between imputed profits

and reported profits reflects not only the tax avoidance engaged in by firms but

also the systematic differences between the accounting system and the national

income account system, including different expense recognition rules, asset

depreciation rules, and tax credit and tax loss carry-over rules. In our analysis,

we include firm fixed effects to control for time-invariant systematic differences,

and we also conduct robustness checks to investigate the validity of our

measurement.

The import tariff data are from the World Integrated Trade Solution

(WITS) provided by the World Bank. Import tariffs are set at the six-digit level

of the Harmonized System (HS) product classification. To merge the data with

the manufacturing firm categories, which have only CIC four-digit industry

codes, we map the HS six-digit codes onto the CIC four-digit codes. As each

CIC four-digit industry includes several six-digit industries, each of which has

an import tariff, we construct an average tariff for each four-digit industry as

follows:

� ������ =�

�∑ � ������

���� . (2)

Here, � ������ is the average tariff level for a four-digit industry k in year t;

� ������ is the tariff level in a six-digit industry j in year t. Using the same

method, tariffs in three-digit and two-digit industries can be calculated. Using a

simple average, rather than a weighted average (using imports as weight), can

avoid bias due to the endogenous response of imports to tariffs. However, our

results are robust when imports are used as weights when calculating four-digit

tariffs, as shown in the robustness section.

13

Table 1 shows the summary statistics for the main variables. Panel A of

Table 1 shows the summary statistics for firm characteristics. On average, the

reported profits by firms are RMB10 million, while the imputed profits are

RMB17 million. The firms on average underreport profits by 7 million. In our

data, 33% of firms are exporters and 8% of firms are processing traders. Among

all firms, 24% are SOEs and 14% are foreign invested firms. We can also see

that on average their sales income is RMB177 million and they have 764

employees. Panel B in Table 1 shows the average tariffs in two-, three-, and

four-digit industries, which are about 16%.

3.2. Empirical Strategy

Two variations are combined to identify the effect of trade liberalization

on tax avoidance. One is the tariff reduction from before WTO entry to after

WTO entry. Another results from the larger tariff reductions of industries with

higher tariff levels before WTO entry. Essentially, we exploit the DID strategy.

The baseline regression is estimated as follows:

����� = � + ��� �����" � ∗ �$%&� + ��'� + �� �� + (���. (3)

Here, ����� is the tax avoidance of firm i in the four-digit industry k in

year t.10 � �����" � is the tariff level in industry k in 2001. �$%&� is a

dummy variable indicating the post-WTO year, i.e., 2002-2004 in our study. ��

is the coefficient of interest. ��'� is a set of firm fixed effects. We also

include year fixed effects, �� ��, in the regression to control for different levels

of average tax avoidance in different years. (��� is an error term with a mean

equal to zero. Standard errors are calculated by clustering over industry.

One issue to consider is that industry tariffs in 2001 were not randomly

assigned; it is possible that industries with higher or lower tariffs in 2001 could

be systematically different. These pre-existing differences could affect firms’ tax

10 We normalize tax avoidance using firms’ total assets.

14

avoidance behavior differently after WTO entry, thus contaminating our

estimates. Fortunately, Lu and Yu (2015) show that the industry-level output

share of SOEs, average wage per worker, and export intensity have robustly

significant effects on industry-level tariffs in 2001. Therefore, to alleviate this

concern, we follow Gentzkow (2006) and Lu and Yu (2015) to control for the

interaction of the output share of SOEs, average wage per worker, and export

intensity in 2001 with �$%&� in the regression.

Several other assumptions need to hold to obtain consistent estimates

of Equation (3). First, we assume that tax avoidance in industries with higher

tariffs in 2001 would have followed the same trend as in industries with lower

tariffs in 2001 if there had been no WTO entry. We test this assumption using

data before WTO entry, i.e., data from 1999 to 2001. The details are shown in

Section 4.2.

Second, one possible issue is that China’s entry into the WTO at the

end of 2001 was expected, so firms might have adjusted their tax avoidance

behavior even before the actual tariff reductions. As a robustness check, we

include an additional control in the regression, � �����" � ∗

*+� �� � ,��$�� ��* �+&�-� , to investigate whether firms changed their

behavior in anticipation of WTO entry.

Third, if other policy changes in this period affected industries with

high versus low tariffs differently, then our estimates would be biased. In the

early 2000s, there were two important reforms: SOE reform and the relaxation

of FDI regulations. To address this concern, we add the SOE shares, i.e., the

number of SOEs divided by the number of firms in each industry, and FDI, i.e.,

the logarithm form of the number of foreign invested firms in each industry, to

the regression to check whether the main results hold.

Fourth, besides output tariffs, WTO entry reduced input tariffs as well,

making it possible for Chinese firms to use more imported intermediate inputs.

Furthermore, China’s trading partners might have also decreased their tariffs on

15

Chinese products, giving Chinese firms greater access to larger foreign markets.

Both could affect firms’ tax avoidance as well. To address this concern, we add

industry-level input tariffs and export intensity to the regression to check the

robustness of results.

Fifth, WTO entry may improve law implementation to different

degrees in different provinces. If more industries with lower tariffs were based

in provinces in which law implementation was more significantly improved by

WTO entry, then our estimates could be biased. To address this concern, we add

an index for law implementation at the provincial level to check whether our

results change.11

To further investigate the assumptions behind the identification method

used in our strategy, we also conduct a placebo test using processing traders. To

address potential selection bias due to sample attrition, we also estimate

Equation (3) using a balanced panel.

4. Empirical Results

4.1. Main Results

Table 2 shows the main results. In column 1, we control for industry

and province fixed effects. The coefficient of � �����" � ∗ �$%&� is 0.057 and

statistically significant at the 1% level. It shows that the higher the tariff level in

2001, the greater the increase in tax avoidance. In column 2, we add the

interactions of �$%&� with three industry-level variables in 2001, i.e., average

wage per worker, export intensity, and SOE shares, to the regression. The

coefficient of � �����" � ∗ �$%&� remains very similar at 0.059 and is still

significant at the 1% level. In columns 3 and 4, we replace the industry and

province fixed effects with firm fixed effects, and in column 4 we add the

interactions of �$%&� with the three variables in 2001. The coefficient of

� �����" � ∗ �$%&� decreases to 0.033 in column 3 and 0.031 in column 4.

11 This index is constructed by Fang and Wang (2006).

16

Both are significant at the 1% level. The similarity of these two coefficients also

suggests that industry tariffs in 2001 are not likely to be endogenous. Using the

specification in column 4, which we prefer, we can see that for firms in

industries with tariffs 10 percentage points higher in 2001 (roughly one standard

deviation), the increase in corporate tax avoidance induced by WTO entry is

RMB757,000 more, accounting for roughly 47% of corporate income tax paid

by an average firm in 2001.

4.2. Testing the Identification Assumptions

In this section, we conduct several tests to see whether the assumptions

behind our identification are valid.

Pre-existing Time Trend. We test whether the tax avoidance behavior of

firms in different industries follows the same pre-existing time trend. We use the

sample from before WTO entry, i.e., from 1999 to 2001. We first define a

dummy denoting 2000 and 2001, and then replace Post in Equation (3) with this

dummy and estimate the equation. We also define another dummy denoting

2001 and replace Post in Equation (3) with this dummy. We then estimate this

equation again. The results of these two exercises are shown in columns 1 and 2

in Table 3. We can see that no coefficient of the interactions of the newly

defined dummies and � �����" � is significant, suggesting that the

pre-existing time trends are the same for firms in different industries.

Behavior Change Before WTO Entry. One concern is that firms in

industries with higher tariffs in 2001 were more likely to change their tax

avoidance behavior if they expected tariff reductions after WTO entry, thus

biasing our estimates. To address this concern, we add � �����" � ∗

*+� �� � ,��$�� ��* �+&�-� to the regression.

*+� �� � ,��$�� ��* �+&�-� is a dummy variable indicating 2000, one

year before WTO entry. The results are shown in column 3 in Table 3. We can

see that the coefficient of � �����" � ∗ *+� �� � ,��$�� ��* �+&�-� is

17

not significant, suggesting that firms did not change their behavior before WTO

entry.

Other Policy Changes. In the early 2000s, there were two other reforms:

SOE reform and the relaxation of FDI regulation. If these two reforms affected

firms in different industries differently, then our estimates would be biased. To

address this concern, we add the share of SOEs among all firms in the industry

and the number of foreign invested firms in the industry to the regression. The

results are shown in column 4 in Table 3. We can see that no coefficients of

these two variables are significant, and that the coefficient of interest remains

positive and significant.

Input Tariffs and Larger Foreign Markets. Entry into the WTO

decreased not only output tariffs but also input tariffs, making it possible for

Chinese firms to use more imported intermediate inputs, which are usually

considered of better quality. Furthermore, China’s trading partners might have

also decreased their tariffs on Chinese products, giving Chinese firms greater

access to larger foreign markets. Having better inputs and gaining access to

larger foreign markets could have increased firms’ profitability and therefore

affected their tax avoidance behavior. To address this concern, we further add

industry-level input tariffs and export intensity to the regression. The results are

shown in column 5 in Table 3. We can see that the coefficient of interest is more

or less the same.

Change of Law Implementation. WTO entry might have had different

effects on law implementation in different provinces. If more industries with

smaller tariff reductions were located in provinces with larger improvements in

law implementation after WTO entry, then our estimates would be biased. To

address this concern, we add an index for law implementation at the provincial

level. The index was constructed by Fan and Wang (2006). The results are

shown in column 6 in Table 3. Adding this variable does not change the

coefficient of the interaction of tariffs in 2001 and the post dummy.

18

Processing Traders. A feature of the Chinese trading regime is that a

proportion of Chinese exporters are processing traders, i.e., firms that are

allowed to import intermediate materials free of tariffs but are required to export

all of their outputs. These firms should not be affected by China’s entry into the

WTO. We therefore use the sample of processing traders to estimate Equation

(3), and the results are shown in column 7 in Table 3. We can see that the

coefficient of � �����" � ∗ �$%&� is not significant.

Sample Attrition. We use an unbalanced panel of firms in the main

analysis. However, if firms were more likely to exit the market in industries

with higher tariff levels in 2001 and if these firms tended to engage in more (or

less) tax avoidance, our estimates would be biased. We test whether this concern

is valid by estimating Equation (3) using a balanced panel of firms. The results

are shown in column 8 in Table 3. We can see that the coefficient of

� �����" � ∗ �$%&� is still positive and significant, suggesting that sample

attrition is not a concern.

5. Robustness Checks

5.1. Justification of the Tax Avoidance Measurement

We use the gap between imputed profits and reported profits to measure

corporate tax avoidance. There are concerns about whether this method can

accurately measure tax avoidance. In this section, we provide several

justifications for using this measurement.

Correlation with Book-Tax Income Gap. A commonly used

measurement of corporate income tax avoidance is book-tax income gap (Desai,

2003, 2005; Desai and Dharmapala, 2006). Although we do not have

information on book income for industrial firms, we match industrial firms with

publicly listed firms and obtain a subsample of 325 firms between 1999 and

2004. For this small sample of firms, we have information on the book-tax

19

income gap.12 The summary statistics of these firms are presented in Appendix

Table A. We first regress the gap between imputed profits and reported profits

on the book-tax income gap. Table 4 shows the results: we can see that all the

coefficients are significantly positive and the magnitudes are large, between

0.357 and 0.591, which shows that these two measures are highly correlated. We

then replicate the regressions in Table 2, replacing the gap between imputed and

reported profits with book-tax income gap using this subsample. Table 5 shows

the results. We can see that the coefficients of Tariff2001*Post in all columns

are significantly positive, consistent with the results in Table 2. These results

suggest that the gap between imputed and reported profits is consistent with the

book-tax income gap in terms of measuring corporate tax avoidance. One caveat

we need to bear in mind is that the matched firms only account for a small part

of the whole sample, therefore the supporting evidence presented here can only

be considered as suggestive.

Response of Reported Profits to Imputed Profits. Following the approach

of Cai and Liu (2009), we estimate the effect of WTO entry on the response of

reported profits to imputed profits. If WTO entry reduced the response of

reported profits to imputed profits, we would have more evidence that trade

liberalization induces more tax avoidance. The results are shown in Table 6. We

can see that the coefficient of the triple interaction of the tariff in 2001,

post-WTO dummy, and imputed profits is negative and significant in all

columns. The results confirm our finding that WTO entry induces more tax

avoidance.

Over-reporting Output. One concern is that firms in industries facing

larger tariff cuts may be more likely to over-report outputs because of declining

sales income due to more intense competition after WTO entry. In other words,

imputed profits could be overestimated in industries with larger tariff cuts,

12 The list of publicly listed firms and information on the book-tax income gap come from RESSET (www.resset.cn).

20

making it hard to differentiate this phenomenon from higher corporate tax

avoidance. To see whether this issue may arise, we replicate regressions in Table

2 but replace the outcome variable with total output divided by sales income. If

firms were more likely to over-report output due to the decline of sales income

in industries with larger tariff cuts, we would expect to see a significantly

positive coefficient of Tariff2001*Post. However, none of the coefficients

shown in Table 7 are significant. This result shows that firms did not over-report

outputs due to the decline of sales income.

5.2. Other Measurements of Explanatory or Outcome Variables

In our main analysis, the four-digit industry-level tariffs are the

unweighted average of the six-digit industry-level tariffs. In this section, we

construct the four-digit industry-level tariffs using the weighted average of the

six-digit industry-level tariffs (1998 import values are used as the weight). For

further robustness checks, we also exploit the variation of two-digit industry

tariffs and three-digit industry tariffs, respectively. The results are shown in

columns 1-3 in Table 8. We can see that all of the coefficients of interest are

positive and significant, consistent with the main results.

In addition, we also use effective tax rate, which is the ratio of corporate

income tax paid by firms to pre-tax profits, to measure tax avoidance, and

investigate how WTO entry affects effective tax rate. The results are shown in

column 4 in Table 8. Although the coefficient of � �����" � ∗ �$%&� is not

precisely estimated, it is negative, providing some evidence that trade

liberalization induces more tax avoidance.

5.3. Heterogeneous Effects in Terms of Corporate Governance

Desai and Dharmapala (2006) emphasize the importance of corporate

governance in determining tax avoidance. They argue that firms with better

corporate governance tend to engage in less tax avoidance. As we do not have a

21

direct measurement of firms’ corporate governance, in this section we

investigate whether the effect of trade liberalization on tax avoidance differs for

firms with different types of ownership, i.e., SOEs, non-SOE domestic firms,

Hong Kong-Macau-Taiwan invested firms, and foreign invested firms. SOEs are

usually considered to have weak corporate governance, while foreign invested

firms have better corporate governance, especially in terms of financial

transparency (Bushman, Piotroski and Smith, 2004).

Table 9 shows the heterogeneous effects. In this table, columns 1-4 are

for SOEs, non-SOE domestic firms, Hong Kong-Macau-Taiwan invested firms,

and foreign invested firms, respectively. We can see that the coefficient of

� �����" � ∗ �$%&� is positive and significant in column 1 (SOEs), while the

coefficient is not significant in the other columns.

Our results show that firms with weak corporate governance respond to

WTO entry by engaging in more tax avoidance, reinforcing the prediction of

Desai and Dharmapala (2006).

6. Channels through which Trade Liberalization Affects Tax Avoidance

Firms experiencing reduced profit performance due to the increased

competition induced by WTO entry tend to rely on noncompliance so that they

have more cash to invest and thus improve their prospects (Slemrod, 2004).

Therefore, firms short of cash or with a higher demand for cash tend to engage

in more tax avoidance. In this section, following Almeida, Campello, and

Weisbach (2004), we use firms’ profitability and leverage before WTO entry to

measure whether they are short of cash. Lower profitability or higher leverage

means a higher degree of cash shortage. We use firms’ growth rate before WTO

entry to measure their demand for cash. The higher the growth rate, the higher

the demand for cash.

Table 10 shows the results. Columns 1 and 2 are for firms with high and

low profitability, respectively. We can see that the coefficient of � �����" � ∗

22

�$%&� is significantly positive for firms with low profitability (column 2).

Columns 3 and 4 are for firms with high and low leverage, respectively. We can

see that the coefficient of � �����" � ∗ �$%&� is significant and positive for

firms with high leverage (column 3). The combination of these two results

suggests that firms short of cash tend to engage in more tax avoidance after

WTO entry.

Columns 5 and 6 are for firms with high and low growth rates before

WTO entry. We use the growth rate of firms to measure their demand for cash.

The higher the growth rate, the higher the demand for cash. We see that the

coefficients in columns 5 and 6 are significantly positive and that the coefficient

for firms with a higher growth rate is larger. This finding suggests that firms

with a higher demand for cash did engage in more tax avoidance after WTO

entry.

7. How Did Firms Engage in Tax Avoidance?

There are several ways firms can avoid corporate income tax, among

which under-recording sales revenues, over-recording costs, and transfer pricing

with affiliated firms are the most common. As we do not have information on

transfer pricing, our focus in this section is whether firms under-report sales

revenues or over-report costs after WTO entry. In particular, we investigate the

effect of WTO entry on administrative costs, sales costs, financial costs, wages,

and benefits, all of which are scaled by sales income.

Indeed, WTO entry significantly increases costs per unit of sales income.

The results are shown in Table 11. We can see that the coefficients in all five

columns are positive and significant at the 10% level or more. This result shows

that firms in industries with higher tariffs in 2001 spent more for each unit of

sales income, which is evidence that over-reporting costs could be a way that

firms avoid corporate income tax.

23

8. Conclusions

In this study, we investigate the effect of trade liberalization on

manufacturing firms’ tax avoidance, using China’s entry into the WTO as a

natural experiment. Comparing firms in industries with different tariffs in 2001

and therefore different reductions between the pre- and post-WTO entry periods,

we find that WTO entry induced firms in industries with tariffs 10 percentage

points higher to engage in tax avoidance totaling RMB757,000, accounting for

47% of the corporate income tax paid by an average firm that year.

Furthermore, we find that firms with better corporate governance (i.e.,

non-SOEs or foreign invested firms) tended to engage in less corporate tax

avoidance in response to WTO entry. We then investigate the possible channels

by which WTO entry affects firms’ corporate tax avoidance. We find that firms

short of cash or with a higher demand for cash engage in more tax avoidance

when facing a more competitive environment after WTO entry. We also find

evidence that firms avoid corporate income tax by over-reporting costs.

Trade liberalization has been considered an important factor in China’s

rapid economic growth in the past three decades. Our findings suggest that the

social benefits may be undermined by illegal activities induced by this trend.

Government policymakers should therefore work to improve tax administrative

efficiency during integration with the global economy.

24

References

Almeida, Heitor, Murillo Campello, and Michael S. Weisbach, 2004. “The Cash

Flow Sensitivity of Cash.” The Journal of Finance 59: 1777-1804.

Amiti, Mary, and Jozef Konings, 2007. “Trade Liberalization, Intermediate

Inputs, and Productivity: Evidence from Indonesia.” The American Economic

Review 97: 1611-1638.

Baggs, Jen, 2005. “Firm Survival and Exit in Response to Trade Liberalization.”

Canadian Journal of Economics/Revue Canadienne d’Economique 38.4:

1364-1383.

Beck, Thorsten, Chen Lin, and Yue Ma, 2014. “Why Do Firms Evade Taxes?

The Role of Information Sharing and Financial Sector Outreach.” Journal of

Finance 69.2: 763-817.

Brandt, Loren, and Eric Thun, 2010. “The Fight for the Middle: Upgrading,

Competition, and Industrial Development in China.” World Development 38.11:

1555-1574.

Brandt, Loren, Johannes Van Biesebroeck, Luhang Wang, and Yifan Zhang,

2012. “WTO Accession and Performance of Chinese Manufacturing Firms.”

Working paper.

Bushman, Robert M., Joseph D. Piotroski, Abbie J. Smith, 2004. “What

Determines Corporate Transparency?” Journal of Accounting Research 42.2:

207-252.

Cai, Hongbin, and Qiao Liu, 2009. “Competition and Corporate Tax Avoidance:

Evidence from Chinese Industrial Firms.” Economic Journal 119.537: 764-795.

Chen, Tao, and Chen Lin, 2013. “Trade Liberalization, Financial Constraints,

and Corporate Tax Avoidance.” Working Paper.

Chen, Zhao, Zhikuo Liu, Juan Carlos Suarez Serrato, and Daniel Yi Xu, 2016.

“Notching R&D Investment with Corporate Income Tax Cuts in China.”

Working Paper.

China Statistical Yearbook, National Bureau of Statistics Press, 2008.

25

Crocker, Keith J., and Joel Slemrod, 2005. “Corporate Tax Evasion with

Agency Costs.” Journal of Public Economics 89.9: 1593-1610.

Desai, Mihir A., 2003. “The Divergence between Book Income and Tax

Income.” Tax Policy and the Economy 17: 169-208.

Desai, Mihir A., 2005. “The Degradation of Reported Corporate Profits.”

Journal of Economic Perspectives 19.4: 171-192.

Desai, Mihir A., and Dhammika Dharmapala, 2006. “Corporate Tax Avoidance

and High-Powered Incentives.” Journal of Financial Economics 79: 145-179.

Desai, Mihir A., and Dhammika Dharmapala, 2009. “Corporate Tax Avoidance

and Firm Value.” Review of Economics and Statistics 91: 537-546.

Ederington, Josh, and Phillip McCalman, 2007. “The Impact of Trade

Liberalization on Productivity Within and Across Industries: Theory and

Evidence.” Working Paper.

Fisman, Raymond, and Shangjin Wei, 2004. “Tax Rates and Tax Evasion.”

Journal of Political Economy 112: 471-496.

Gentzkow, Matthew, 2006. “Television and Voter Turnout.” Quarterly Journal

of Economics 121.3: 931-972.

Hanlon, Michelle, and Joel Slemrod, 2009. “What Does Tax Aggressiveness

Signal? Evidence from Stock Price Reactions to News about Tax Shelter

Involvement.” Journal of Public Economics 93.1: 126-141.

Harrison, Ann E., 1994. “Productivity, Imperfect Competition and Trade

Reform: Theory and Evidence.” Journal of International Economics 36.1:

53-73.

Johnson, Simon, Daniel Kaufmann, and Andrei Shleifer, 1997. “The Unofficial

Economy in Transition.” Brookings Papers on Economic Activity 2: 159-221.

Johnson, Simon, Daniel Kaufmann, and Pablo Zoido-Lobaton, 1998.

“Regulatory Discretion and the Unofficial Economy.” The American Economic

Review 88: 387-392.

Kim, Jeong-Bon, Yinghua Li, and Liandong Zhang, 2011. “Corporate Tax

26

Avoidance and Stock Price Crash Risk: Firm-level Analysis.” Journal of

Financial Economics 100.3: 639-662.

Krishna, Pravin, and Devashish Mitra, 1998. “Trade Liberalization, Market

Discipline and Productivity Growth: New Evidence from India.” Journal of

Development Economics 56.2: 447-462.

Lardy, Nicholas R., 1994. China in the World Economy. Peterson Institute Press:

All Books.

Lardy, Nicholas R., 2002. Integrating China into the Global Economy.

Washington: Brookings Institution Press.

Levinsohn, James, 1993. “Testing the Imports-as-Market-Discipline

Hypothesis.” Journal of International Economics 35.1: 1-22.

Lileeva, Alla, and Daniel Trefler, 2010. “Improved Access to Foreign Markets

Raises Plant-level Productivity…For Some Plants.” Quarterly Journal of

Economics 125.3: 1051-1099.

Nunn, Nathan, and Daniel Trefler, 2010. “The Structure of Tariffs and

Long-Term Growth.” American Economic Journal: Macroeconomics 2.4:

158-194.

Pavcnik, Nina, 2002. “Trade Liberalization, Exit, and Productivity

Improvement: Evidence from Chilean Plants,” Review of Economic Studies 69:

245-276.

Rodrik, Dani, 1988. “Closing the Technology Gap: Does Trade Liberalization

Really Help?” NBER Working Paper.

Schneider, Friedrich, and Dominik Ernste, 2000. “Shadow Economies: Size,

Causes and Consequences.” Journal of Economic Literature 38: 77-114.

Schor, Adriana, 2004. “Heterogeneous Productivity Response to Tariff

Reduction. Evidence from Brazilian Manufacturing Firms.” Journal of

Development Economics 75.2: 373-396.

Shleifer, Andrei, 2004. “Does Competition Destroy Ethical Behavior?” The

American Economic Review 94: 414-418.

27

Slemrod, Joel, 2004. “The Economics of Corporate Tax Selfishness.” National

Tax Journal 57.4: 877-899.

Topalova, Petia, 2007. “Trade Liberalization, Poverty and Inequality: Evidence

from Indian Districts.” NBER Working Paper.

Topalova, Petia, and Amit Khandelwal, 2011. “Trade Liberalization and Firm

Productivity: The Case of India.” Review of Economics and Statistics 93.3:

995-1009.

Trefler, Daniel, 2004. “The Long and Short of the Canada-US Free Trade

Agreement.” American Economic Review 94.4: 870-895.

Tybout, James R., and M. Daniel Westbrook, 1995. “Trade Liberalization and

the Dimensions of Efficiency Change in Mexican Manufacturing Industries.”

Journal of International Economics 39.1: 53-78.

Wu, Jinglian, 2003. Contemporary Chinese Economic Reforms. Shanghai:

Shanghai Far-East Publisher.

28

Figure 1. Tariffs in Four-digit Industries (1999-2004)

Note: The average tariff in each year is the mean value of import tariffs for 476 China

Industry Classification (CIC) system four-digit industries. China entered the WTO at the end

of 2001, which is denoted by the vertical line in the figure.

.1.1

2.1

4.1

6.1

8T

ari

ff

1999 2000 2001 2002 2003 2004Year

29

Figure 2. Correlation between Tariffs in 2001 and Tariffs after WTO Entry

Note: The dots in the graphs are for CIC four-digit industries. The X-axis in each graph shows the average

tariff in 2001. The Y-axis in the top two graphs and the bottom-left graph is for tariff reductions from 2001

to 2002, 2003, and 2004, respectively. The Y-axis in the bottom-right graph is for tariff reductions from

2001 to the average tariff level over 2002 to 2004.

0.2

.4.6

Tari

ff R

eduction (

2001-2

002)

0 .2 .4 .6 .8Tariff in 2001

2001-2002

0.2

.4.6

Tari

ff R

eduction (

2001-2

003)

0 .2 .4 .6 .8Tariff in 2001

2001-20030

.2.4

.6T

ari

ff R

eduction (

2001-2

004)

0 .2 .4 .6 .8Tariff in 2001

2001-2004

0.2

.4.6

Tari

ff R

eduction (

2001-A

vera

ge(0

2-0

4))

0 .2 .4 .6 .8Tariff in 2001

2001-Average(2002-2004)

30

Table 1. Summary Statistics

Mean S.D.

Panel A. Firm characteristics

Reported profit (1,000 yuan) 10,079.356 231,839.186

Imputed profit (1,000 yuan) 16,893.610 233,166.300

Total assets (1,000 yuan) 244,199.425 1,208,199.082

Exporter 0.333 0.471

Processing trader 0.080 0.272

Export (1,000 yuan) 29,285.834 260,377.499

Effective income tax rate 0.179 15.137

Sales (1,000 yuan) 176,624.208 937,413.619

SOE 0.235 0.424

FDI 0.136 0.342

Administrative expenditures (1,000 yuan) 10,701.175 59,786.793

Sales expenditures (1,000 yuan) 6,506.970 44,024.018

Financial costs (1,000 yuan) 3,549.309 20,237.050

Total wages (1,000 yuan) 9,359.691 43,457.963

Total employment 763.770 2,573.373

OBS 281,522 281,522

Panel B. Tariff

42 two-digit industries 0.154 0.101

188 three-digit industries 0.159 0.105

476 four-digit industries 0.159 0.109

Note: Reported profits are the profits firms reported when they filed annual reports to the NBS. Imputed profits are equal to output minus intermediate inputs, financial charges, total wages paid, current depreciation, and value added tax. Effective income tax rate is the ratio of corporate income tax paid by firms to pre-tax profits. SOEs are defined as firms with more than 50% state-owned shares. FDI firms are defined as firms with more than 25% foreign-owned shares.

31

Table 2. Effects of Trade Liberalization on Firms’ Tax Avoidance

(1) (2) (3) (4)

Dependent variable: (Imputed Profits minus Reported Profits)/Total Assets

Tariff2001*Post 0.057 0.059 0.033 0.031

(0.012)*** (0.012)*** (0.011)*** (0.011)***

Average wage in 2001*Post

-0.001

-0.000

(0.000)***

(0.000)

Export intensity in 2001*Post

-0.026

-0.006

(0.009)***

(0.009)

SOE share in 2001*Post

-0.001

-0.008

(0.008)

(0.007)

Industry and province fixed effects Yes Yes No No

Firm fixed effects No No Yes Yes

Year fixed effects Yes Yes Yes Yes

Observations 281,522 281,522 281,522 281,522

R-squared 0.06 0.06 0.64 0.64

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%. Average wage in 2001 is the average wage per worker for each industry in 2001; export intensity in 2001 is the ratio of exports to sales in each industry in 2001; SOE output share in 2001 is the share of outputs produced by SOEs for each industry in 2001.

32

Table 3. Pre-assumption Tests (Dependent Variable: (Imputed Profits Minus Reported Profits)/Total Assets)

(1) (2) (3) (4) (5) (6) (7) (8)

Pre-trend

(1999-2001) Next year Additional Controls

Processing Traders

Balanced Panel

Tariff2001*Post

0.026 0.030 0.024 0.024 0.049 0.039

(0.011)** (0.011)*** (0.012)** (0.012)** (0.140) (0.012)***

Tariff2001*One year before

-0.016

(0.010)

Tariff2001*Dummy for 2000 and 2001 0.002

(0.014)

Tariff2001*Dummy for 2001

-0.010

(0.014)

SOE

-0.016 -0.019 -0.018

(0.011) (0.011)* (0.011)

Ln(FDI)

0.001 -0.001 -0.001

(0.001) (0.002) (0.002)

Input tariff

-0.035 -0.035

(0.042) (0.042)

Exports/Sales

0.093 0.093

(0.057) (0.058)

Index for law implementation

-0.001

(0.001)

Variable in 2001*Post Yes Yes Yes Yes Yes Yes Yes Yes

Firm fixed effects Yes Yes Yes Yes Yes Yes Yes Yes

Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes

Observations 123,795 123,795 281,522 281,522 218,592 218561 17520 108447

R-squared 0.71 0.71 0.64 0.64 0.65 0.65 0.64 0.48

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%; variables in 2001 include the average wage per worker for each industry in 2001, the ratio of exports to sales in each industry in 2001 and the share of outputs produced by SOEs for each industry in 2001. Index for law implementation is a provincial level variable indicating how well laws are implemented; a larger value indicates better law implementation.

33

Table 4. Correlation between Measures of Corporate Income Tax

(1) (2) (3)

(Imputed Profits minus Reported Profits)/Total

Assets

Book-tax income gap/Total assets 0.357 0.411 0.591

(0.113)*** (0.118)*** (0.117)***

Industry and province fixed effects No Yes No

Firm fixed effects No No Yes

Year fixed effects No Yes Yes

Observations 699 699 699

R-squared 0.01 0.41 0.64

Robust standard errors are in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%.

34

Table 5. Using Publicly Listed Firms

(1) (2) (3) (4)

Dependent Variable: Book-tax income gap/Total assets

Tariff2001*Post 0.078 0.091 0.067 0.070

(0.027)*** (0.026)*** (0.027)** (0.027)**

Average wage in 2001*Post

0.001

0.000

(0.001)

(0.001)

Export intensity in 2001*Post

-0.004

0.003

(0.021)

(0.024)

SOE share in 2001*Post

0.019

0.008

(0.020)

(0.019)

Industry and province fixed effects

Yes Yes No No

Firm fixed effects No No Yes Yes

Year fixed effects Yes Yes Yes Yes

Observations 699 699 699 699

R-squared 0.30 0.30 0.47 0.47

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%. The sample used in this table is drawn from publicly listed firms. Average wage in 2001 is the average wage per worker for each industry in 2001; export intensity in 2001 is the ratio of exports to sales in each industry in 2001; SOE output share in 2001 is the share of outputs produced by SOEs for each industry in 2001.

35

Table 6. Effects on the Responses of Reported Profits to Imputed Profits

(1) (2) (3) (4)

Dependent Variable: Reported Profits Scaled by Total Assets

Tariff2001*Post*Imputed profits scaled by total assets -0.096 -0.104 -0.073 -0.083

(0.049)* (0.050)** (0.036)** (0.036)**

Tariff2001*Post -0.003 0.003 -0.020 -0.014

(0.006) (0.006) (0.007)*** (0.006)**

Imputed profits scaled by total assets 0.143 0.142 0.064 0.063

(0.014)*** (0.013)*** (0.010)*** (0.010)***

Imputed profits scaled by total assets*Post 0.033 0.035 0.038 0.041

(0.014)** (0.014)** (0.010)*** (0.010)***

Tariff2001*Imputed profits scaled by total assets -0.044 -0.039 -0.023 -0.017

(0.048) (0.048) (0.043) (0.042)

Variable in 2001*Post Yes Yes Yes Yes

Firm fixed effects Yes Yes Yes Yes

Year fixed effects Yes Yes Yes Yes

Observations 281,522 281,522 281,522 281,522

R-squared 0.23 0.23 0.76 0.76

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%; variables in 2001 include the average wage per worker for each industry in 2001, the ratio of export to sales in each industry in 2001, and the share of outputs produced by SOEs for each industry in 2001.

36

Table 7. Effects of Trade Liberalization on Firms’ Output Relative to Sales

(1) (2) (3) (4)

Dependent variable: (Imputed Profits minus Reported Profits)/Total Assets

Tariff2001*Post -0.036 -0.024 0.010 0.018

(0.026) (0.026) (0.026) (0.026)

Average wage in 2001*Post

0.002

0.001

(0.001)***

(0.001)**

Export intensity in 2001*Post

-0.017

-0.042

(0.019)

(0.024)*

SOE share in 2001*Post

0.007

-0.004

(0.016)

(0.018)

Industry and province fixed effects Yes Yes

Firm fixed effects

Yes Yes

Year fixed effects Yes Yes Yes Yes

Observations 281,515 281,515 281,515 281,515

R-squared 0.02 0.02 0.56 0.56

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%. Average wage in 2001 is the average wage per worker for each industry in 2001; export intensity in 2001 is the ratio of exports to sales in each industry in 2001; SOE output share in 2001 is the share of outputs produced by SOEs for each industry in 2001.

37

Table 8. Other Robustness Tests

(1) (2) (3) (5)

Imputed Profits minus Reported

Profits/Total assets Effective Tax Rate

Tariff2001_New*Post 0.024

(0.010)**

Tariff2001_2digit*Post

0.046

(0.015)***

Tariff2001_3digit*Post

0.026

(0.010)**

Tariff2001*Post

-0.097

(0.127)

Variable in 2001*Post Yes Yes Yes Yes

Firm fixed effects Yes Yes Yes Yes

Year fixed effects Yes Yes Yes Yes

Observations 281,522 281,522 281,522 275,104

R-squared 0.64 0.64 0.64 0.33

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%; Tariff2001_New is at the four-digit industry level and is calculated as the weighted average of tariffs in the HS six-digit industry level using imports as weights. Variables in 2001 include the average wage per worker for each industry in 2001, the ratio of export to sales in each industry in 2001, and the share of outputs produced by SOEs for each industry in 2001.

38

Table 9. Heterogeneous Effects in Terms of Corporate Governance

(1) (2) (3) (4)

Dependent variable: (Imputed Profits minus Reported Profits)/Total Assets

SOE Domestic Non-SOE HK-Macau-Taiwan FDI

Tariff2001*Post 0.052 0.015 0.031 0.022

(0.014)*** (0.014) (0.033) (0.031)

Variable in 2001*Post Yes Yes Yes Yes

Firm fixed effects Yes Yes Yes Yes

Year fixed effects Yes Yes Yes Yes

Observations 66,661 156,937 35,347 188,251

R-squared 0.54 0.55 0.50 0.56

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%. Variables in 2001 include the average wage per worker for each industry in 2001, the ratio of export to sales in each industry in 2001, and the share of outputs produced by SOEs for each industry in 2001.

39

Table 10. Heterogeneous Effects

(1) (2)

(3) (4)

(5) (6)

Dependent variable: (Imputed Profits minus Reported Profits)/Total Assets

Profitability prior to WTO entry

Leverage prior to WTO entry

Sales growth prior to WTO entry

Top 50

percentile Bottom 50 percentile

Top 50 percentile

Bottom 50 percentile

Top 50 percentile

Bottom 50 percentile

Tariff2001*Post 0.012 0.050

0.042 0.013

0.047 0.029

(0.016) (0.012)***

(0.015)*** (0.014)

(0.017)*** (0.012)**

Variable in 2001*Post

Yes Yes

Yes Yes

Yes Yes

Firm fixed effects Yes Yes

Yes Yes

Yes Yes

Year fixed effects Yes Yes

Yes Yes

Yes Yes

Observations 120,910 103,500

109,742 114,653

76300 73541

R-squared 0.56 0.51 0.57 0.54 0.53 0.48

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%. Variables in 2001 include the average wage per worker for each industry in 2001, the ratio of exports to sales in each industry in 2001, and the share of outputs produced by SOEs for each industry in 2001.

40

Table 11. Suggestive Evidence for Tax Avoidance Channels

(1) (2) (3) (4) (5)

Ratio of

Administrative Expenditures

Sales Expenditures

Financial Costs

Wages Benefits

Tariff2001*Post 0.013 0.007 0.016 0.024 0.004

(0.007)* (0.003)** (0.005)*** (0.005)*** (0.002)**

Variable in 2001*Post

Yes Yes Yes Yes Yes

Firm fixed effects Yes Yes Yes Yes Yes

Year fixed effects Yes Yes Yes Yes Yes

Observations 281,472 281,436 281,515 281,515 281341

R-squared 0.72 0.81 0.68 0.74 0.45

Robust standard errors in parentheses are calculated by clustering over four-digit industry; * significant at 10%; ** significant at 5%; *** significant at 1%. Variables in 2001 include the average wage per worker for each industry in 2001, the ratio of exports to sales in each industry in 2001, and the share of outputs produced by SOEs for each industry in 2001.

41

Table A. Summary Statistics of Publicly Listed Industrial Firms

Mean S.D.

Panel A. Firm characteristics

Book-tax income gap/Total assets 0.003 0.033

Reported profit (1,000 yuan) 97,481.815 249,141.064

Imputed profit (1,000 yuan) 158,694.200 627,524.400

Total assets (1,000 yuan) 1,908,515.927 2,701,504.283

Exporter 0.571 0.495

Processing trader 0.008 0.088

Export (1,000 yuan) 108,796.604 352,569.848

Effective income tax rate 0.165 0.342

Sales (1,000 yuan) 1,228,955.401 2,327,948.040

SOE 0.321 0.467

FDI 0.035 0.183

Administrative expenditures (1,000 yuan) 72,780.540 170,093.573

Sales expenditures (1,000 yuan) 53,764.699 153,295.582

Financial costs (1,000 yuan) 19,427.010 41,365.234

Total wages (1,000 yuan) 51,402.294 105,109.469

Total employment 2,688.077 3,478.291

OBS 699

Number of firms 325

Number of 4-digit industries 162