China and WTO liberalization: Imports, tariffs and non-tariff barriers

Michele Imbruno

PII: S1043-951X(16)30014-1
DOI: doi: 10.1016/j.chieco.2016.02.001
Reference: CHIECO 919

To appear in: China Economic Review

Received date: 23 May 2015

Revised date: 9 February 2016
Accepted date: 9 February 2016

Please cite this article as: Imbruno, M., China and WTO liberalization: Imports, tariffs
and non-tariff barriers, China Economic Review (2016), doi: 10.1016/j.chieco.2016.02.001

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 ACCEPTED MANUSCRIPT

China and WTO liberalization:

Imports, Tariffs and Non-tariff barriers
Michele Imbruno1
1 Present address: IMT Institute for Advanced Studies Lucca, Piazza San Francesco, 19, 55100, Lucca, Italy.

T
Nottingham Centre for Research on Globalisation and Economic Policy

IP
University of Nottingham
University Park, Nottingham, NG7 2RD, United Kingdom.

CR
E-mail address: michele.imbruno@imtlucca.it

US
Abstract

N
This paper explores the effectiveness of different trade policy instruments on product-level
Chinese imports over the period of 2000-2006. More specifically, in addition to the declines
MA
in tariffs, we investigate the impact on imports of the gradual removal of non-tariff barriers
(NTBs) as agreed within the WTO’s accession protocol in 2001 (such as import quotas,
licenses and tendering requirements). We document that while manufacturing imports mainly
increase because of tariff cuts, agricultural imports grow due to the elimination of import
ED

licenses. However, we provide evidence that quota elimination is associated with a

redistribution of imports along a larger range of countries, whereas tendering liberalization is
connected to an import reallocation from OECD to non-OECD countries. Finally, we also
find that NTB protection is complementary to tariff protection and discriminating against
PT

foreign-owned manufacturing firms.

CE

Highlights
We investigate the effectiveness of both tariff and non-tariff barriers (NTBs) on product-level imports in China.
AC

While manufacturing imports mainly increase because of tariff reductions, agricultural imports grow due to the
elimination of licenses.
Both tariff and license liberalizations are associated with larger number of countries (firms) and average imports per
country (firm).
Quota elimination implies a redistribution of imports along a larger range of countries, whereas tendering liberalization
entails a reallocation from OECD to non-OECD countries.
NTB protection is found to be complementary to tariff protection and discriminating against foreign-owned manufacturing
firms.

JEL classification: F13, F14, F15.

Keywords: WTO liberalization, tariffs, non-tariff barriers, imports, China.

1
 ACCEPTED MANUSCRIPT

1. Introduction
Until the late 1970s, China played a marginal role in world trade, since both imports and
exports occurred under a context of a planned economy: for example in 1978, the Ministry of
Foreign Trade determined that only twelve state-owned firms could engage in international

T
trade. In the 1980s, China began a process of phasing out the quantitative planning of trade

IP
flows and to reform its trade regime, through introducing instead some limits on the trading
rights and using more some conventional policy instruments, such as tariffs and quotas.

CR
Indeed, import tariff rates initially increased, reaching an average duty rate of around 56 % in
1982, and then fell to 43 % in 1985, remaining quite stable for the rest of the decade. Quota or

US
license restrictions were still present for almost half of the imports over the same period.
However, during the 1990s, China drastically decreased all these trade barriers, so that in

N
2001 the average tariff rate was around 15 %; the share of imports under quota/license
MA
regulation fell to 8.5 %; and the number of firms with the right to trade abroad increased to
35,000. Moreover, China’s import openness was actually even greater once the privileges,
introduced for export-processing and foreign-owned firms, were accounted for. As an
ED

example, tariff exemptions were introduced in 1987 for the imports of raw
materials/parts/components used by the firm for the production of exported goods, as well as
PT

for the imports of capital goods by wholly foreign-owned firms (e.g. about 40 % of imports in
2000 were duty-free).
CE

To a large extent the unilateral measures of trade liberalization adopted by China over the
1990s were undertaken as part of the process in its World Trade Organization (WTO) entry,
AC

which officially took place on December 11, 2001. Indeed, China agreed to several conditions
for its WTO entry that were more stringent compared to other entrants, to not only obtain
credibility for its commitments to economic openness through trade with the rest of the world,
but also because it believed that those conditions were necessary for the country to benefit
from economic globalization.1 Ianchovichina and Martin (2006) explore how China’s WTO
accession might influence both China and the rest of the world, by accounting for the export
processing regime within the Global Trade Analysis Project model. They first estimate that
China would enjoy the largest benefits, due mainly to its own trade liberalization policies.
Then, they also find that while the developed economies (especially North America and
European Union countries) would gain – thanks to larger export opportunities towards China
and a decrease in the costs of imports from China – some developing countries in East Asia

1
See Branstetter and Lardy (2008), for further details.

2
 ACCEPTED MANUSCRIPT

are expected to lose – as China will become their most important competitor in third
countries.

Through signing the WTO accession protocol, China started to apply the most-favored-nation
tariff rates to all WTO members, and amongst other commitments, to reduce tariffs and on

T
yearly-basis to remove all other non-tariff barriers (NTBs) to imports, such as quotas, licenses

IP
and tendering requirements, by the end of 2004. According to the WTO’s China Trade Policy

CR
Review (2006) – CTPR (2006) henceforth – the average applied tariff rate in China fell to
12.2% after it joined the WTO and still further to 9.7% by 2005. While the phasing out of
import quotas, tendering requirements and licenses, as arranged by the protocol, was

US
successful.

N
This paper aims to explore the evolution of Chinese imports over the period of 2000-2006 and
to examine the role played by these different components of trade liberalization, i.e. tariffs
MA
cuts and the gradual removal of non-tariff barriers (such as quotas, licenses and tendering
requirements) as scheduled by China’s WTO accession protocol. Moreover, we also attempt
ED

to understand whether there is any relationship between tariffs and non-tariff barriers, since
one of the main concerns is that the two policy instruments might exhibit a link of perfect
substitutability, keeping a constant level of actual protection. More specifically, the current
PT

work attempts to address the following research questions: i) Is the elimination of non-tariff
trade barriers (NTBs) really effective on Chinese imports? Is there any dissimilarity amongst
CE

the different NTB reforms? ii) Is there any relationship between NTBs and tariffs? Do they
complement or substitute each other?
AC

While most of the literature analyzing the impact of trade reforms within an economy focuses
on import tariffs, several studies have highlighted the relevance of other trade restrictions. In
general, they find that non-tariff barriers are on average more trade restrictive than tariffs and
in many countries the former contribute relatively more to the overall level of trade
restrictiveness than the latter (WTO, 2012). For example, using data from 6-digit product
lines across 78 countries, Kee, et al. (2009) estimate that the ad-valorem tariff equivalent
(AVE)2 of NTBs3 was 12% on average (45% if the average is computed only over product
lines subject to NTBs) and document that in 55% of product lines subject to NTBs, the AVE
of NTB was even higher than the tariff rate. Moreover, their results show that NTBs on

2
AVE is the level of an ad-valorem tariff that would have an equally trade-restricting effect as the NTB.
3
Their measure of NTBs includes: price control measures, quantity restrictions, monopolistic measures, and
technical regulations.

3
 ACCEPTED MANUSCRIPT

average added an additional 87% to the level of trade restrictiveness imposed by tariffs, and in
34 countries NTBs contributed relatively more than tariffs to the overall level of trade
protection. Using trade data of 4-digit product lines from the EU and 14 other G-20 countries
over the period 2007-2010, Henn and Mcdonald (2011) show that imports decreased by 5%,
because of border restrictions applied during the financial crisis, and by 7%, as result of

T
IP
behind-the-border measures. When they further disaggregate the protectionist dummies
according to the measure type, they document that both traditional and non-traditional non-

CR
tariff measures were more effective than tariffs: such as quota, import ban, competitive
devaluation, trade defence measures, licensing requirements and discriminatory purchasing,

US
amongst import restrictions, as well as bailout and domestic subsidies, amongst behind-the-
border barriers.

N
The empirical literature on the linkage between tariffs and NTBs is quite scant and provides
MA
mixed results. For instance, Ray and Marvel (1984) highlight that the extent of NTBs can
either increase as a result of the main multilateral trade negotiations aimed at declining tariffs
(substitutability hypothesis), or increase to further reinforce the trade protection from tariffs in
ED

textile and agricultural sectors (complementarity hypothesis). More recent evidences have
supported either the latter hypothesis (Kee at al., 2009) or the former one (Dean et al., 2009).
PT

Limão and Tovar (2011) have developed a theoretical model showing that a government gains
from tariff constraint commitments through international agreements – even if this implies the
CE

use of less efficient NTBs – when its bargaining power position is weak relative to domestic
special interest groups. Using product-level data from Turkey, they find that NTBs increase,
AC

following tariff binding or bound tariff reduction, and the latter is more likely to occur when
the special interest groups are relatively stronger in their lobbying activities.

Our baseline results show that ordinary imports value’s growth in China is generally due to
both tariff and license liberalization, since the value of imports decreases following tendering
liberalization, or is unaffected by quota elimination. In particular, while the license effect
concerns the agricultural products, both tariff and tendering effects concern manufacturing
products.

A deeper analysis within product highlights that, thanks to both tariff declines and the
removal of licenses, China is able to import more from each country, as well as to increase
further the range of source countries, especially from the OECD area, i.e. China can access
more and better country-varieties. Reverse effects have been found by the elimination of

4
 ACCEPTED MANUSCRIPT

tendering requirements, which would suggest a potential reallocation of imports from high-
quality to low-quality varieties. Finally, quota abolition leads to a fall in import intensity per
country on the one hand, and an increase in the number of source countries from both OECD
and non-OECD areas equally, on the other hand, which would suggest that imports are
redistributed along a larger range of country-varieties.

T
IP
Furthermore, tariff liberalization implies an increase in both the number of importing firms

CR
with different ownership and average imports per firm for manufacturing products. Similar
effects have been found in agricultural products through the elimination of import licenses.
Conversely, in manufacturing products, the removal of licenses and the other traditional

US
NTBs, i.e. quotas and tendering requirements, would lead to an increase in the number of
foreign-owned importers at the expenses of the number of domestic and privately-owned
N
importers, in addition to a decline in the average import intensity per firm. Moreover, while
MA
tariff reduction allows more producers to import directly and more intermediaries to start
importing, the abolition of NTBs induces the entry of new production firms into the import
market and the exit of trade intermediaries.
ED

Finally, we rule out the hypothesis of substitutability between tariffs and NTB barriers as well
as the endogeneity of trade policies, as argued by Limão and Tovar (2011), which could
PT

potentially explain the controversial findings regarding NTB liberalization. First, we show
that tariff and non-tariff policies move in the same direction at a product level in China, since
CE

we find that applied tariff reductions are associated with NTBs’ eliminations. Second,
following Topalova and Khandelwal (2011) and Bas and Strauss-Kahn (2015), we provide
AC

evidence that a change in each trade reform in China (about tariff, import quotas, licenses and
tendering requirements, alternatively) is unlikely to be affected by the initial product-level
characteristics, which confirms that China’s willingness to be integrated with World Trading
System goes beyond the specific interest groups.

To the best of our knowledge, a similar study on China has been conducted by Bao and Qiu
(2010) who examine the impact of technical barriers to trade (TBTs) on Chinese imports at
the 2-digit sector-country level over the period 1998-2006, by controlling for tariffs, import
licenses and quotas. They find mixed results across sectors, periods, trade policy measure and
methodologies used. Unlike their work, we focus on product-level linkages between imports,
tariffs and traditional NTBs (i.e. import quotas, licenses and tendering requirements) only, but
at a more disaggregate level (6-digit) and using information about NTBs from the Protocol.

5
 ACCEPTED MANUSCRIPT

Furthermore, we also provide additional empirically stylized facts about the import evolution
within products (in terms of both intensive and extensive margins of country imports and firm
imports), as well as some evidence of a potential linkage between tariffs and the traditional
NTBs.

T
The rest of paper is organized as follows. Section 2 provides an overview about the WTO

IP
liberalization in China. Section 3 describes the data available through some descriptive

CR
statistics, while Section 4 empirically studies the impact of both tariffs and non-tariff barriers
(NTBs) on product-level imports value, as well as the potential linkage between the two kind
of trade policy instruments. Section 5 concludes.

N US
MA
2. China and WTO liberalization
This section describes China’s process of trade liberalization after the entry to the WTO.
More specifically, it provides some information about tariff and non-tariff barriers and their
ED

evolution over time.

PT

2.1. Import tariffs4

CE

Under the terms for its accession into the WTO, China bound 100% of its tariffs at ad
valorem rates and agreed to charge the MFN duty rate to all WTO members, except for El
Salvador. The CTPR (2006) documents that the simple average bound rate was 12.4% in 2002
AC

and decreased to 10% in 2005, while the applied MFN tariff rate tended to narrowly follow
the bound rate: the simple rate was 12.2% in 2002 and fell to 9.7% in 2005. 5 Agricultural
products exhibit on average a larger duty rate associated with a higher variability, compared
to manufacturing products.

China also offers some preferential tariff rates to some specific countries under its bilateral
and regional trade agreements (i.e. to the Special Administrative Regions of Hong Kong and
Macao,6 members of the Bangkok Agreement,7 members of ASEAN8 and Pakistan9), and to

4
Source: WTO (2006) China Trade Policy Review – WT/TPR/S/161.
5
Very few lines (less than 1%) are subject to applied tariffs at non-ad valorem rates, i.e. specific or compound
rates.
6
In 2005, the number of preferential tariff lines and the average tariff rate were: 1,061 and 8.2% for Hong Kong;
502 and 8.8% for Macao.

6
 ACCEPTED MANUSCRIPT

the least developed countries for some products under unilateral special agreements. 10 General
tariff rates are applied to El Salvador, the territories of some EU member states, and the WTO
non-members not involved in any reciprocal trade agreement with China.11

Preferential treatments (such as preferential tariffs) are reserved to border trade under certain

T
circumstances and reduced & zero tariffs are available for imports by special economic zones

IP
or for imports under processing trade regime (i.e. if imported goods are processed and

CR
exported).

US
2.2. Other import restrictions12

N
2.2.1. Trading rights, state trading and import prohibitions
In the period before the WTO accession, the right to trade was restricted to about 35,000
MA
Chinese firms with certain specific requirements (such as the minimum registered capital
requirement of CNY 5 million), and foreign-owned firms could only trade if the products
ED

were used in their own production or for exports.

Within the WTO protocol, China agreed to gradually reduce all these restrictions, so that,
PT

within three years, all enterprises located within country had the right to trade in all goods,13
except for those goods subject to state trading and those subject to trade prohibitions for
CE

health and environmental safety reasons.

AC

7
Bangkok agreement partners are: Bangladesh, India, Korea, Lao PDR, and Sri Lanka. In 2005, the number of
preferential tariff lines and the average tariff rate were 749 and 9.5%.
8
ASEAN’s partners in 2005 were: Brunei Darussalam, Cambodia, Indonesia, Lao PDR, Malaysia, Myanmar,
Singapore, Thailand and Viet Nam. In particular, under the Early Harvest Programme, China signed separate
preferential agreements with each member. In 2005, the number of preferential tariff lines for each member state
ranged between 358 and 536, while the average tariff rate ranged between 8.8% and 9.2%.
9
In 2005, the number of preferential tariff lines and the average tariff rate are 748 and 9.5%.
10
In 2005, they concerned about 182 tariff lines with an average tariff rate of 9.5% and 39 countries
(Afghanistan, Angola, Bangladesh, Benin, Bhutan, Burundi, Cape Verde, Cambodia, Central African Republic,
Comoros, Democratic Republic of Congo, Djibouti, East Timor, Eritrea, Ethiopia, Guinea, Guinea-Bissau, Lao
PDR, Lesotho, Liberia, Madagascar, Maldives, Mali, Mauritania, Mozambique, Myanmar, Nepal, Niger,
Rwanda, Samoa, Sierra Leone, Somalia, Sudan, Tanzania, Togo, Uganda, Vanuatu, Yemen, and Zambia).
11
In 2005, general tariff rates were applied to: Afghanistan, Andorra, Aruba, Bahamas, Bermuda, Bhutan,
Bonaire Islands, Canary Islands, Cayman Islands, Ceuta, Comorin, Curacao, El Salvador, French Guyana,
Gambier Islands, Gibraltar, Guadeloupe, Liberia, Marquesas Islands, Marshall Islands, Martinique, Melilla,
Montserrat Island, Nauru, Norfolk Island, New Caledonia, Palau, Palestine, Réunion, Sabah, Saint Martins
Island, Sao Tome and Principe, the Republic of San Marino, Seychelles, Society Islands, Tuamotu Islands, Tubai
Islands, Turks and Caicos Islands, Tuvalu, Vatican City, British Virgin Islands, and Western Sahara.
12
Sources: WTO (2006) China Trade Policy Review, WT/TPR/S/161; and WTO (2001) Protocol on the
accession of China, WT/L/432.
13
Trading rights concession to all firms does not imply that all importers can distribute goods within China.
Distribution services’ liberalization depends on some specific commitments of China in the GATS.

7
 ACCEPTED MANUSCRIPT

The 8-digit HS products that can be imported only by state-owned firms are listed in the
Annex 2A of the protocol and are related to: grain, vegetable oil, sugar, tobacco, crude oil and
processed oil, chemical fertilizer, and cotton. The 8-digit HS products subject to import
prohibitions have to be periodically notified by the authorities. According to the CTPR
(2006), in 2002, 32 products were fully subjected to import prohibitions and 405 products

T
IP
were partially subject to import prohibition. While the former decreased over time (to 30 in
2005), the latter increased (to 458 in 2005). In addition to these general import prohibitions,

CR
some products were subjected to import prohibitions only under the processing trade (they
account for 143 product lines in 2005, of which 86 are fully prohibited).14

US
The Annex 2B of the protocol displays all 8-digit HS products to be liberalized within three
years after the accession, so that all firms will be free to trade them. These products include
N
natural rubber, timber, plywood, wool, acrylic, and steel. To implement the WTO obligations
MA
about the trading rights’ extension, China amended the Foreign Trade Law of 1994, which
entered in force on July 1, 2004. Under this new regulation, any firm or individual willing to
trade internationally has to prepare some registration documents for identification purposes
ED

only with Ministry of Commerce, which in turn must complete the whole procedure within
five days.
PT

China agreed to allow the prices of all traded goods/services to be determined by market
forces, except for some specific goods/services listed in Annex 4 of the protocol, which may
CE

be subjected to price control. The 8-digit HS product lines under price control or government
guidance pricing concern tobacco, edible salt, natural gas, pharmaceuticals, vegetable and
AC

processed oil, fertilizer, silkworm cocoons, and cotton.

2.2.2. Non-tariff measures: Import quotas, tendering and licensing

Several changes occurred to non-tariff import restrictions, in terms of new legislation of
import licenses administration and the phasing out of products subject to import quotas,
tendering requirements15 or licensing, as established by China’s protocol of the WTO
accession.

14
The list of these products is not within the WTO protocol and cannot be found on the WTO website.
Tendering is a procurement system for machinery and electronic products. Tendering organizations arrange
15

procurement tenders for machinery and equipment needed by national ministries for major technical innovation
projects, public works and private enterprises. The bidding process was based on direct negotiation and was non-

8
 ACCEPTED MANUSCRIPT

In December 2001, China agreed to implement the gradual abolition of all import quotas and
import tendering requirements by the end of 2004,16 alongside the elimination of import
licensing for some 8-digit HS products upon accession (as described in Annex 3 of the
protocol). On January 1, 2005, all quota and tendering restrictions were removed, and three

T
import licenses regimes were notified by China to the WTO, including import licenses,

IP
automatic import licenses and tariff rate quotas for imports,17 whose complete details are
annually published by the Ministry of Commerce. Import licenses were required only for

CR
products in accordance with international conventions. In particular, firms must apply first for
an import permit to a specific entity according to the product, providing all documents and

US
certificates required. Once this permit is approved, the license is automatically granted by the
Ministry of Commerce within three working days. Products are subject to automatic import

N
licenses, if they are not subject to import restrictions but need import monitoring. Firms must
apply to authorized entities for them, by providing some documents, and if the application is
MA
correct, then the license is immediately granted by the Ministry of Commerce. While the
number of product lines partially or fully under import licenses decreased from 214 in 2002 to
ED

82 in 200518 (which include mainly chemicals and chemical products), the number of product
lines subject to automatic import licensing rose from 1,191 in 2002 to 1,205 in 2005, since the
goods listed under import tendering requirements until 2004 basically moved within list of
PT

goods requiring import monitoring (which comprises base metals, transport equipment,
machinery, plastics and rubber, textiles, and precision instruments).
CE

Finally, in order to restrict the quantity of imports, China introduced some tariff rate quotas19
AC

for wheat, maize, rice, soybean oil, palm oil, rape oil, sugar, wool tops, cotton, and chemical
fertilizers.20

transparent. Consequently, tender organizations and other interest parties may distort imports in favouring
domestic producers over foreign ones or discriminating against producers from certain countries. See the U.S.
International Trade Commission (1999) and the WTO (2006) China Trade Policy Review for more details.
16
The phasing-out period can be different across products. It can be upon accession, or alternatively in the year
2002, 2003, 2004 and 2005.
17
Import quotas were abolished in the end of 2004. According to the data provided by the authorities, two
chemical products that were subject to import licenses appear to be subject to import quotas in 2005.
Nevertheless, the authorities claim that they actually are just subject to licenses only.
18
In the end of 2004, two lines (sodium cyanide and platen screen press printing machinery) remained in the list
of products subject to licensing, although they were expected to be removed (upon accession and in 2002,
respectively).
19
It is a trade regime where quantities imported within a quota threshold are charged by lower import duty rates,
than those outside.
20
The list of these products in 2005 cannot be found on the WTO website.

9
 ACCEPTED MANUSCRIPT

2.2.3. Other measures

Other measures that can affect imports include: contingency measures, such as anti-dumping;
standards and other technical requirements, such as technical barriers to trade (TBT) and
sanitary and phytosanitary (SPS) measure; and government procurement.

T
Applications for anti-dumping investigations are generally made to the Ministry of

IP
Commerce. If there is any evidence of a causal link between dumping and injury to the

CR
domestic industry, the investigation can be initiated, and both the exporting country and the
applicant must be notified. Provisional anti-dumping measures can be adopted 60 days after
the decision to initiate the investigation has been published. Between 2002 and 2004, China

US
initiated 79 anti-dumping investigations (of which, 52 ended up with final measures and 7
with no measures). These investigations mostly frequently included imports of chemicals,
N
plastics and rubbers from Japan, Republic of Korea and the United States.21
MA
In China, there are four kind of standards (national, sectoral, local and enterprise standards),
which can be voluntary or mandatory. According to the analysis of the existing 21,000
ED

national standards by the Standardization Administration of China in 2004, only 44% of these
were still applicable through their relevance and alignment with international standards, while
about 44% required some revision and the remaining 12% had to be abolished. In addition,
PT

animals, plants and all related products need permits to be traded and are subject to quarantine
inspection for health and safety reasons.
CE

Without considering procurement by state-owned firms, public utilities and public organs of
AC

defense, Government procurement in China accounted for about 1.4% of the GDP in 2003.
According to the law on Government procurement, introduced in January 2003, government
entities can procure goods and services by using the following alternative methods: public
tendering, invitation for tendering, competitive negotiation, request for quotations, and single-
source procurement. In early 2002, China started the process to accede to the WTO agreement
on Government procurement to make the procurement system more transparent and open to
foreign competition.

Finally, there are some non-import protection measures in China, which could generate
similar effects as those arising from the import restrictions, such as the ‘pure export
subsidies’, i.e. fiscal advantages conditional on minimum export share requirements. Defever

21
However, over the same period, the other WTO members also initiated 152 anti-dumping investigations
against imports from China, and final measures were taken for 99 cases.

10
 ACCEPTED MANUSCRIPT

and Riaño (2012) show that these subsidies in China provide protection for domestic firms
and lead to a lower degree of market competition and consumers’ welfare, unlike the standard
export subsidies.

T
IP
3. Data and descriptive statistics

CR
3.1. Data sources and management
The analysis conducted within this work uses a balanced panel of 4,504 6-digit HS product

US
lines22 over the period of 2000-2006. Data are drawn from three main sources: the import data
are from the database of Chinese Customs Trade Statistics (CCTS), managed by the General

N
Administration of Customs of China; all tariff data are from the World Integrated Trade
MA
Solution (WITS) database developed by the World Bank; and all non-tariff trade barriers
(NTBs) are from China’s Protocol of the WTO accession available on the WTO’s website.
The CCTS database contains all monthly trade transactions between China and the rest of
ED

world. However, we focus only on imports. For each import transaction at the firm-product-
country level, the database records information about total value, FOB unit value (in US
PT

dollars), quantity and trade regime, also providing information about the firms, such as
ownership. Since our focus is to explore the effect of tariff and NTBs reductions on imports at
CE

the product-level, we dropped all observations related to a trade regime other than ordinary
trade regime (i.e. 42.9 % of all observations that account for 57.3% of the total value of
imports over the whole period) and collapsed the data to yearly information at the 6-digit HS
AC

product level. Then the import data were merged with product-level data regarding tariffs and
NTBs, and all observations where tariff data is missing were dropped (namely, 145
observations). Finally, we restricted our sample to those product lines present in all seven
years, losing 11% of observations.

3.2. Imports
Table 1 shows that 4,504 products are observed throughout the entire period (2000-2006).
The average imports value was about 18.2 USD millions in 2000, which increased threefold

22
Original information about all data (imports, tariffs and NTBs) is actually available at the 8-digit HS level.
However, we are forced to make a 6-digit HS level analysis, as some data (till 2001) are classified by HS1996
taxonomy and some are classified by HS2002 taxonomy (from 2002 onwards) and the concordance table
available is at the 6-digit level. Therefore, data based on HS2002 have been converted to HS1996.

11
 ACCEPTED MANUSCRIPT

by 2006. In 2000, products were on average sourced from 13 countries (9 of which were
OECD countries) by about 100 firms, of which 48% were state-owned firms (SOEs), 38%
were foreign-owned firms (FORs), and only 1% were domestic-private owned firms
(PRIVs).23 About 37% of firms importing a given product were trade intermediaries, while

T
about 63% were producers. When we compare these figures with those ones related the end of

IP
our sample period, we first notice that China generally became able to import a product from
many more countries and more from each country. In particular, the number of non-OECD

CR
countries increased relatively more than of OECD countries. Secondly, more Chinese firms
became able to import a given product with higher firm-level intensity. Specifically, while the

US
number of both private-domestic owned importers (especially) and foreign-owned importers
increased, the number of state-owned and other importers actually decreased. Finally, the

N
number of both producers and intermediaries rose over time such that the intermediary share
of importers declined.
MA
These patterns mainly concern the manufacturing sector, which accounts for about 93% of the
6-digit product lines in our sample. The agricultural products were on average imported by
ED

fewer firms and from a smaller number of countries (i.e. 25 and 9 respectively in 2000), but
with a greater intensity per country and per firm. Moreover, while the intensive margin of
PT

agricultural imports increased over time at both the country-level and firm-level, the related
extensive margin remained almost stable.
CE
AC

23
The remaining share (14%) refers to “other firms”.

12
 ACCEPTED MANUSCRIPT

Table 1: Imports value in China

PT
All goods Agricultural goods Manufacturing goods

2000 2006 2000 2006 2000 2006

RI
Std. Std. Std. Std. Std. Std.

SC
Mean Dev. Mean Dev. Mean Dev. Mean Dev. Mean Dev. Mean Dev.

NU
value of total imports 18.20 94.51 58.85 403.21 17.09 133.50 43.08 410.86 18.29 90.76 60.09 402.62

MA
average value of imports per country 1.25 8.20 2.88 22.30 2.60 22.16 6.55 59.55 1.14 5.82 2.59 16.05
number of countries 13 9 18 13 9 7 9 8 13 9 19 13
number of OECD countries 9 6 11 7 5 4 5 4 9 6 12 7

ED
number of non-OECD countries 4 4 7 7 4 4 5 5 4 4 8 7

PT
average value of imports per firm 0.37 3.27 0.65 5.44 0.53 3.37 0.77 5.37 0.35 3.26 0.64 5.45
number of firms 100 223 219 575 25 31 30 37 106 230 234 595
number of state-owned firms 48 95 CE 41 88 15 20 7 9 50 98 43 91
number of foreign-owned firms 38 103 117 359 5 8 8 12 40 107 126 372
AC
number of domestic private -owned firms 1 2 57 133 1 1 14 20 1 3 61 137
number of other firms 14 31 4 10 5 6 1 2 15 32 5 10

number of producers 63 156 158 452 14 17 16 20 67 161 169 467

number of intermediaries 37 71 62 130 11 15 14 19 39 73 65 134

Number of 6-digit product lines 4504 4504 328 328 4176 4176
Notes: Values are expressed in USD millions. Source: Author’s calculations using data on import transactions under ordinary trade regime from the database of Chinese Customs Trade Statistics (CCTS) managed by
the General Administration of Customs of China.

13
 ACCEPTED MANUSCRIPT

3.3. Trade barriers: tariff and non-tariff barriers

In Table 2, we see that in 2000, the average MFN applied tariff was around 16.9% and
dropped to 9.8% in 2006. Additionally, some products were subject to other import
restrictions, according to China’s WTO protocol. In the first year under analysis, 0.8% of

T
products could be imported only by state-owned firms and 0.7% of products remained under

IP
price control, while about 4.4% of products were under trading rights’ restrictions and must

CR
be liberalized within three years after the WTO accession. Moreover, 4.8% of goods were
subjected to the traditional non-tariff barriers (NBTs). More specifically, 2.8% were
subject to import quotas and 1.6% were subject to import tendering requirements, whereas

US
about 0.4% needed licenses only to be imported. In the final year of analysis, all NTB
restrictions were supposed to disappear. The Figure 1 displays the gradual trade liberalization
N
over time in terms of all import barriers as planned in the China’s accession protocol, as well
MA
as the declining trend of MFN applied tariff.
ED

Figure 1: Trade barriers across time

PT
.2

.04
CE
.15

.03
tariff rate

AC
.1

.02
.05

.01
0

2000 2001 2002 2003 2004 2005 2006

year

tariff trading rights restrictions

quota tendering
license

Source: Author’s calculations using data on MFN applied tariffs from World Bank’s WITS database, and data on non-tariff
trade barriers (NTBs) from China’s Protocol of the WTO accession available on the WTO’s website – i.e. on products
subject to trading rights restrictions, import quotas, tendering requirements and licenses.

14
 ACCEPTED MANUSCRIPT

Table 2: Applied tariffs and other import restrictions in China according to WTO Accession Protocol

All goods Agricultural goods Manufacturing goods

PT
2000 2006 2000 2006 2000 2006

RI
Std. Std. Std. Std. Std. Std.

SC
Mean Dev. Mean Dev. Mean Dev. Mean Dev. Mean Dev. Mean Dev.

NU
tariff 0.169 0.115 0.098 0.068 0.244 0.201 0.131 0.089 0.163 0.103 0.095 0.066

MA
Other import restrictions 0.096 0.295 0.012 0.111 0.043 0.202 0.04 0.195 0.1 0.3 0.01 0.101

state importing 0.008 0.089 0.008 0.089 0.037 0.188 0.037 0.188 0.006 0.076 0.006 0.076

ED
trading rights restrictions 0.044 0.204 0 0 0 0 0 0 0.047 0.212 0 0
quota 0.028 0.166 0 0 0 0 0 0 0.03 0.172 0 0

PT
tendering 0.016 0.127 0 0 0 0 0 0 0.018 0.132 0 0
license 0.004 0.066 0 0 0.027 0.164 0 0 0.003 0.051 0 0
price control 0.007 0.081 CE
0.007 0.081 0.021 0.145 0.021 0.145 0.006 0.074 0.006 0.074
AC

Number of 6-digit product lines 4504 4504 328 328 4176 4176

Notes: Both tariff rates and the shares of products lines subject to each non-tariff restriction are expressed in decimals. They need to be multiplied by 100, to know the tariff rate and the share of products under
non-tariff restrictions in percentage. For example, in 2000 all products on average exhibit a tariff rate around 16.9 %, while 9.6 % of product lines are subject to other import restrictions. Source: Author’s
calculations using data on MFN applied tariffs from World Bank’s WITS database, and data on non-tariff trade barriers (NTBs) from China’s Protocol of the WTO accession available on the WTO’s website – i.e.
on products subject to state importing, trading rights restrictions, import quotas, tendering requirements, licenses and price controls.

15
 ACCEPTED MANUSCRIPT

Again, these average figures more closely reflect the conditions within the manufacturing
sector, where the tariff rate declined from 16.3% to 9.5% and the share of products under
traditional NTBs was 5.1% in 2000 (3.0% under quota; 1.8% under tendering and 0.3% under
licenses). Agricultural products display higher tariff protection but in general are less
protected by traditional NTBs. The tariff rate in agricultural products was on average 24.4%

T
IP
in 2000 and fell to 13.1% in 2006, while the share of products under license was around 2.7%
in 2000. However, it is worth noting that compared to the manufacturing sector, the

CR
agricultural sector on average exhibits a higher share of product lines subject to state-
importing (3.7% versus 0.6%) and price control (2.7% versus 0.3%).

US
In Table 3, we note that the correlation across import restrictions is generally quite low (in
2000, it ranged between -0.08 and 0.31). State importing status seems to be positively
N
correlated with quotas, licenses and price control status, i.e. products that can be imported
MA
only by state-owned firms are more likely to be subject to a different kind of NTBs.
Meanwhile trading rights restrictions are positively correlated with quotas status: i.e.
products under trading rights’ limitations are also likely to be subject to other non-tariff
ED

barriers, such us quotas. Similarly, products under price control also seem to be restricted by
other non-tariff barriers, such as quotas and licenses.
PT
CE

Table 3: Correlation between import barriers

AC

state trading rights price

tariff importing restrictions quota tendering license control

tariff 1
state importing 0.3117* 1
trading rights restrictions -0.0790* -0.0191 1
quota 0.2538* 0.1956* 0.0689* 1
tendering -0.0312* 0.0276 -0.0276 -0.022 1
license 0.3075* 0.2940* -0.0142 -0.0114 -0.0086 1
price control 0.1462* 0.2992* -0.0175 0.0520* -0.0106 0.1998* 1

Notes: Significance at * 5%. Source: Author’s calculations using data on MFN applied tariffs from World Bank’s WITS database, and data
on non-tariff trade barriers (NTBs) from China’s Protocol of the WTO accession available on the WTO’s website – i.e. on products subject
to state importing, trading rights restrictions, import quotas, tendering requirements, licenses, and price control.

16
 ACCEPTED MANUSCRIPT

Finally, there is no statically significant correlation between quotas, tendering and licenses,
since they are mutually exclusive measures. When we focus on tariffs, we notice that they
display a statistically significant correlation with all other import limitations. In particular,
tariffs are positively correlated with all of them, except for the import tendering status and
trading rights restrictions status, where the correlation appears to be negative and quite low. In

T
IP
other words, products subject to some non-tariff import restrictions are generally associated
with larger tariff rates, apart from products subject to trading rights limitations and import

CR
tendering, which exhibit relatively low tariff rates.

US
4. Econometric analysis
N
MA
4.1. Imports and trade barriers
The main purpose of this section is to explore the relationship between the value of imports
into China and both tariff and non-tariff trade liberalization at the 6-digit HS product level,
ED

using balanced panel data of 4,504 products over the period of 2000-2006 and the following
baseline specification:
PT

ln M pt  1 tariff pt1   2 quota pt1   3 license pt1   4 tendering pt1   p   t   pt ,

CE

where ln M pt is the log of total imports value for the 6-digit HS product line p at time t;

tariff pt1 is the simple average of MFN applied tariff rate of the product p in the year t-1;
AC

quota pt1 is a dummy variable taking value one if the product category p includes some of

257 8digit-HS products subject to the import quota in the year t-1; tendering pt is a dummy

assuming value one if the product category p includes some of 120 8digit-HS products subject
to import tendering in the year t-1; and license pt is a dummy variable taking value one if the

product category p includes some of 47 8digit-HS products subject to the import license only
in the year t-1. 24 We also include the 6-digit HS product fixed effects  p and time dummies

 t to capture time-invariant product characteristics and some common macroeconomic

24
All non-tariff barriers (NTB) variables are based on the information from the WTO accession protocol of
China. For this reason, the license dummy variable switches from 1 to 0 in a single year (2002). While for the
other two NTB dummy variables, we have more variability across products, as scheduled by the protocol: i.e.
quota dummy variable can switch from 1 to 0 in 2002 or 2003 or 2004 or 2005, and tendering dummy variable
can switch from 1 to 0 in 2002 or 2004. See Figure 1 again.

17
 ACCEPTED MANUSCRIPT

shocks to all products respectively, while  pt is the error term. We lag all explanatory

variables one period to account for possible lags in adjustment. These lags might occur as
most product lines that were subject to quotas, tendering and licenses, moved to automatic
import licensing system for monitoring purposes, which was not very clear and transparent

T
over the first years of the WTO liberalization. All coefficients are expected to be negative

IP
since the imports value should grow following a fall in trade barriers.

CR
Table 4 displays the baseline results. In column 1, we explore the import-tariff effect only.
This has the expected negative effect and is statistically significant. The parameter estimate
implies that a 10 percentage points fall in the tariff rate leads to an increase in the imports

US
value of about 7.1 %. In the second column, we regress the imports value on NTBs variables.
Here, we can notice that, in line with our expectations, products subject to license-
N
liberalization exhibit higher import growth (by about 72%). Conversely, products subject to
MA
tendering-liberalization appear to decrease their imports volumes (by about 32%), and the
effect of quota abolition turns out to be statistically insignificant. These findings remain
essentially the same when all trade policy variables are included at the same time in column 3,
ED

although the coefficients’ magnitude of both tariff and license slightly decreases, due to the
complementarity between these two policy instruments, highlighted in the correlation table
PT

(Table 3).
CE

In columns 4 and 5, we re-estimate our baseline regression separately for agricultural and
manufacturing products, respectively. First, it is worth noting that agricultural products are
subject to licenses only, where the positive effect of liberalization seems to be even stronger
AC

(i.e. the magnitude is three times larger than that one documented for the whole sample),
while the tariff cut effect turns out to be statistically insignificant. Whereas, manufacturing
products are subject to all NTBs, and only the negative effect of tendering liberalization turns
out be statistically significant. However, the positive impact of tariff liberalization on
manufacturing imports appears to be statistically significant and even larger compared with
the effect found for the whole sample. The coefficient estimates now suggest that the value of
manufacturing imports increases by 9.7 %, following a reduction in the tariff of 10 percentage
points.

18
 ACCEPTED MANUSCRIPT

Table 4: Product-level imports value, tariffs and non-tariff barriers (NTBs)

PT
Dependent variable: All goods Agricultural goods Manufacturing goods
log value of imports (ln Mt)

RI
(1) (2) (3) (4) (5) (6) (7) (8) (9)

SC
tarifft-1 -0.709*** -0.598** -0.717 0.110 -0.879*** -0.972***
(0.240) (0.240) (0.504) (0.473) (0.268) (0.272)

NU
quotat-1 0.0872 0.111 0.105 0.146
(0.111) (0.109) (0.111) (0.108)

MA
tenderingt-1 0.320*** 0.323*** 0.338*** 0.346***
(0.0722) (0.0724) (0.0722) (0.0726)
-0.784** -0.624* -2.101*** -2.138*** 0.0994 0.288

ED
licenset-1
(0.358) (0.365) (0.505) (0.532) (0.326) (0.353)

PT
Product fixed effects         
Time dummies         
CE
Observations 27024 27024 27024 1968 1968 1968 25056 25056 25056
AC
R-squared 0.193 0.193 0.194 0.073 0.095 0.095 0.21 0.21 0.211
Number of 6-digit product lines 4504 4504 4504 328 328 328 4176 4176 4176
Notes: Standard errors (in parentheses) have been corrected for clustering at the product level. Significance at: *** 1%, ** 5%, * 10%. Source: Author’s calculations using data from the database of Chinese Customs
Trade Statistics (CCTS) managed by the General Administration of Customs of China; World Bank’s WITS database; and China’s Protocol of the WTO accession available on the WTO’s website.

19
 ACCEPTED MANUSCRIPT

4.2. Country-analysis and Firm-analysis within products

Table 5 shows that almost 2/3 of product-level imports growth from tariff liberalization is
explained by an increase in the average imports value per country (i.e. country-intensive
margin), while the remaining part is due to an increase in the number of source countries (i.e.
country-extensive margin), more specifically the OECD countries. Therefore, it appears that a

T
IP
decline in tariffs allows China to reach new and better quality country-varieties, in addition to
an increase in import intensity per each country-variety. This is much more evident for

CR
manufacturing products, where the change in country-extensive margin accounts for 42% of
the import enhancement from tariff cuts, and the number of OECD countries increases

US
slightly more than the number of non-OECD countries; whereas agricultural products’
imports value seems to be statistically unaffected by tariff reduction in all their country-
margins.
N
MA
Similarly, imports growth from the liberalization of import licenses is due to an increase in
both country-intensive (by about 2/3) and country-extensive margins (by about 1/3), where
the number of OECD source countries has increased more than the number of non-OECD
ED

countries. However, this is marked for agricultural products only. The decline in
manufacturing imports from liberalization of tendering is due mainly to changes in the
PT

country-intensive margin rather than the extensive margin (85% versus 15%): in particular,
while the number of OECD countries from which imports are sourced decreased, the number
CE

of non-OECD countries increased (although the related negative coefficient is statistically

insignificant). This would seem to imply some reallocation of imports from high-quality
AC

varieties to low-quality varieties, which would also partially explain the decrease in the value
of imports (per country). For manufacturing products, the import country-margin now also
appears to be sensitive to quota elimination: while the number of source countries increases,
the average imports value per country decreases. Since the number of source countries
increased almost equally from both the OECD and the non-OECD area, it seems that the
removal of quotas simply allow China to extend and diversify the range of countries from
which it obtained imports.

20
 ACCEPTED MANUSCRIPT

Table 5: Country-margins of product-level imports value, tariffs and non-tariff barriers (NTBs)

PT
All goods Agricultural goods Manufacturing goods

RI
average average number of
value of number of number of value of number of non- average value number of number of
Dependent variables (log) imports per number of OECD non-OECD imports per number of OECD OECD of imports per number of OECD non-OECD

SC
at time t country countries countries countries country countries countries countries country countries countries countries
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

NU
-0.381* -0.217*** -0.331*** -0.121 0.117 -0.00646 -0.121 0.0581 -0.559** -0.413*** -0.533*** -0.306***
tarifft-1
(0.199) (0.0841) (0.0832) (0.0950) (0.381) (0.180) (0.172) (0.191) (0.233) (0.0858) (0.0853) (0.112)

MA
0.277*** -0.166*** -0.145*** -0.149*** 0.290*** -0.144*** -0.126*** -0.128***
quotat-1
(0.0977) (0.0367) (0.0353) (0.0437) (0.0971) (0.0364) (0.0352) (0.0435)

ED
0.285*** 0.0384* 0.0458** -0.0433 0.291*** 0.0547** 0.0603*** -0.0276
tenderingt-1
(0.0673) (0.0228) (0.0232) (0.0408) (0.0674) (0.0229) (0.0232) (0.0409)

-0.422 -0.202** -0.149 -0.100 -1.478*** -0.660*** -0.641*** -0.236* 0.294 -0.00620 0.0692 -0.157

PT
licenset-1
(0.337) (0.0932) (0.0908) (0.0948) (0.531) (0.147) (0.132) (0.136) (0.323) (0.0972) (0.0910) (0.124)

Product fixed effects    CE         

Time dummies            
AC

Observations 27024 27024 26739 25268 1968 1968 1863 1570 25056 25056 24876 23698

R-squared 0.138 0.145 0.077 0.168 0.09 0.048 0.03 0.068 0.146 0.174 0.097 0.186

Number of 6-digit product lines 4504 4504 4498 4468 328 328 325 311 4176 4176 4173 4157

Notes: Standard errors (in parentheses) have been corrected for clustering at the product level. Significance at: *** 1%, ** 5%, * 10%. Source: Author’s calculations using data from the database of Chinese
Customs Trade Statistics (CCTS) managed by the General Administration of Customs of China; World Bank’s WITS database; and China’s Protocol of the WTO accession available on the WTO’s website.

21
 ACCEPTED MANUSCRIPT

Table 6 displays that overall both number of firms (i.e. firm-extensive margin) and the
average imports value per firm (i.e. firm-intensive margin) contribute to product-level imports
growth from tariff liberalization by 41% and 59%, respectively. Therefore, there is a clear
evidence that a tariff cut induces more firms to start importing, in addition to a higher import
intensity per firm, although only state-owned firms are able to start importing. When we split

T
IP
the sample into agricultural and manufacturing products, we can see that the documented
overall effects reflect changes found in the latter set of products, where the number of

CR
domestic-private owned importers increased even more than that of both state-owned and
foreign-owned firms. Conversely, for agricultural products, we find almost no significant

US
linkage between all firm-margins of imports and tariff liberalization. In fact, only the number
of domestic-private-owned firms seems to be affected and in a negative manner.

N
By taking into account all the product lines, imports growth from removing licenses appears
MA
to be almost equally due to both changes in the intensive and extensive firm-margins,
although only the effect on the latter is statistically significant. The extensive margin mainly
concerns state-owned firms and foreign-owned firms. However, these linkages are relatively
ED

stronger for agricultural products, for which the effects on both the firm-intensive margin and
the number of domestic-private-owned firms are also statistically significant. Conversely, the
PT

effects are weaker for manufacturing products, for which we document a fall in the number of
domestic-private-owned firms and an increase in the number of foreign-owned firms that
CE

import.

Tendering liberalization negatively affects both the firm-intensive and firm-extensive margins
AC

of (manufacturing) imports, although we find some evidence that the number of foreign-
owned firms has increased. Quotas’ elimination leads to a decrease in the average imports
value per firm and some contrasting effects on the number of importers, according to their
ownership. While the number of both foreign- and state-owned firms that imported products
increases, the number of domestic-private ones decreases. Considering these findings
altogether, it would appear that firms still have some constraints or difficulties in their ability
to start importing products that have been liberalized from quota and tendering restrictions,
unless they are foreign-owned or state-owned firms.

22
 ACCEPTED MANUSCRIPT

Table 6 : Firm-margins of product-level imports value, tariffs and non-tariff barriers (NTBs)

All goods Agricultural goods Manufacturing goods

PT
RI
number
of number of number of
number number of domestic number of number of domestic average number of number of domestic

SC
Dependent average value of state- foreign- private - average value state- foreign- private - value of state- foreign- private -
variables (log) of imports per number of owned owned owned of imports per number of owned owned owned imports number of owned owned owned
at time t firm firms firms firms firms firm firms firms firms firms per firm firms firms firms firms
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)

NU
-0.352** -0.246** -0.223* -0.173 0.0882 0.278 -0.168 -0.162 -0.0860 0.495* -0.522** -0.450*** -0.441*** -0.445*** -0.529***
tarifft-1

MA
(0.177) (0.118) (0.126) (0.124) (0.169) (0.342) (0.191) (0.232) (0.195) (0.284) (0.210) (0.141) (0.150) (0.151) (0.198)

0.197** -0.0859 -0.129** -0.160*** 0.0669 0.201** -0.0551 -0.0975* -0.130** 0.116*
quotat-1
(0.0890) (0.0561) (0.0567) (0.0553) (0.0711) (0.0888) (0.0554) (0.0561) (0.0544) (0.0702)

ED
0.266*** 0.0570 0.00859 -0.193*** -0.0779 0.264*** 0.0818* 0.0345 -0.170*** -0.0422
tenderingt-1
(0.0644) (0.0468) (0.0441) (0.0556) (0.0712) (0.0645) (0.0470) (0.0443) (0.0559) (0.0714)

PT
-0.321 -0.303** -0.320** -0.359** 0.0687 -1.168** -0.970*** -1.121*** -0.567** -0.624** 0.273 0.0150 0.0818 -0.478*** 0.284*
licenset-1

Product fixed
(0.313) (0.133) (0.138) (0.160)
CE
(0.145) (0.544) (0.193) (0.181) (0.265) (0.307) (0.258) (0.143) (0.103) (0.163) (0.145)

              
effects
AC
Time dummies               

Observations 27024 27024 26398 25706 23075 1968 1968 1832 1626 1551 25056 25056 24566 24080 21524

R-squared 0.096 0.235 0.169 0.329 0.744 0.09 0.052 0.286 0.067 0.497 0.098 0.274 0.171 0.359 0.762
Number of 6-digit
4504 4504 4500 4480 4467 328 328 326 317 318 4176 4176 4174 4163 4149
product lines
Notes: Standard errors (in parentheses) have been corrected for clustering at the product level. Significance at: *** 1%, ** 5%, * 10%. Source: Author’s calculations using data from the database of Chinese Customs Trade Statistics (CCTS) managed by the
General Administration of Customs of China; World Bank’s WITS database, and China’s Protocol of the WTO accession available on the WTO’s website.

23
 ACCEPTED MANUSCRIPT

Table 7: Firm-margins of product-level imports value, tariffs and non-tariff barriers (NTBs): production versus trade intermediation

PT
All goods Agricultural goods Manufacturing goods

RI
SC
number of number of number of number of number of number of
Dependent variables (log) number production intermediary number of production intermediary number of production intermediary

NU
at time t of firms firms firms firms firms firms firms firms firms
(1) (2) (3) (4) (5) (6) (7) (8) (9)

MA
tarifft-1 -0.246** -0.0802 -0.577*** -0.168 0.133 -0.297 -0.450*** -0.405*** -0.811***
(0.118) (0.124) (0.146) (0.191) (0.230) (0.243) (0.141) (0.137) (0.172)

ED
quotat-1 -0.0859 -0.129** 0.0456 -0.0551 -0.0919* 0.0698
(0.0561) (0.0554) (0.0607) (0.0554) (0.0543) (0.0599)

PT
tenderingt-1 0.0570 -0.0879** 0.161*** 0.0818* -0.0598 0.178***
(0.0468) (0.0405) (0.0499) (0.0470) (0.0408) (0.0500)

licenset-1 -0.303** -0.426*** CE -0.0994 -0.970*** -0.943*** -0.708*** 0.0150 -0.259* 0.184**
(0.133) (0.128) (0.129) (0.193) (0.223) (0.201) (0.143) (0.148) (0.0934)
AC
Product fixed effects         
Time dummies         

Observations 27024 26737 26327 1968 1902 1845 25056 24835 24482
R-squared 0.235 0.287 0.083 0.052 0.046 0.037 0.274 0.329 0.099
Number of 6-digit product lines 4504 4503 4499 328 328 324 4176 4175 4175
Notes: Standard errors (in parentheses) have been corrected for clustering at the product level. Significance at: *** 1%, ** 5%, * 10%. Source: Author’s calculations using data from the database of Chinese Customs
Trade Statistics (CCTS) managed by the General Administration of Customs of China; World Bank’s WITS database, and China’s Protocol of the WTO accession available on the WTO’s website.

24
 ACCEPTED MANUSCRIPT

Finally, Table 7 shows the results regarding the impact of the different trade policy
instruments on the number of producers and intermediaries importing foreign products, and
we find that: i) tariff reduction within manufacturing products leads to an increase in the
number of both production and intermediary firms (especially the latter); ii) quota elimination
within manufacturing products would benefit importing producers only, while iii) tendering

T
IP
liberalization within manufacturing products is associated with a decrease in importing
intermediaries; and iv) license liberalization within agricultural products entails an increase in

CR
both firm groups (especially manufactures compared to intermediaries), whereas the same
policy within manufacturing products implies an increase in producers and a decrease in

US
intermediaries. Consequently, while the NTBs removal allows more producers to start
importing directly, at the expenses of trade intermediaries who are forced to stop import

N
activity (especially within manufacturing sector), tariff liberalization would lead both
production and intermediary firms to start importing.
MA
ED

4.3. Tariffs versus Non-Tariff Barriers: complements or substitutes?

This subsection examines more deeply the potential relationship between tariffs and NTBs,
i.e. if they are complements or substitutes, through the following specification:
PT

tariff pt   1 quota pt   2 license pt   3tendering pt   p   t   pt ,

CE

where each coefficient can be negative if the tariff policy and the related NTB policy exhibit a
substitutability linkage, or positive if the two trade policy instruments have some
AC

complementarity ties. The results are displayed in Table 8, where we note a certain
complementarity between the tariff and each NTB policy under analysis, which is relatively
stronger for licenses, especially within agricultural products in respect to manufacturing ones,
and relatively weaker for tendering. However, there is no evidence of potential substitutability
between tariffs and traditional NTBs.

By ruling out the hypothesis that tariffs have been adjusted towards higher protection within
products subject to NTB liberalization, it remains quite difficult to understand why product-
level import value per source country (or per firm) decreases following both quota and
tendering liberalization, and the product-level number of source countries (or importing firms)
falls as tendering requirements have been removed. A potential reason could be that
traditional NTBs (quotas, licenses, etc.) have been basically replaced by the new more

25
 ACCEPTED MANUSCRIPT

stringent NTBs, such as TBTs, SPSs, antidumping measures, etc. Anderson and Schmitt
(2003) theoretically show that tariff liberalization increases the probability of using quota
restrictions, and when quotas are also removed, the adoption of antidumping policies become
more intense. Feinberg and Reynolds (2007) explore whether tariff liberalization (through
changes in bound tariffs) implies an increase in the adoption of antidumping measures

T
IP
(substitutability hypothesis), using data from about 24 countries during the period of 1996-
2003, and they confirm this hypothesis only for developing countries. Similar findings have

CR
been documented by Moore and Zanardi (2011), using data on applied tariffs from about 36
countries during the period of 1991-2002.

N US
Table 8: Tariffs and non-tariff barriers (NTBs)

Agricultural Manufacturing
MA
All goods
goods goods

Dependent variable:
tariff (1) (2) (3) (4) (5) (6)
ED

quota 0.0471*** 0.0482*** 0.0503***

PT

(0.0112) (0.0111) (0.0111)

tendering 0.00284 0.00507 0.00718*
(0.00430) (0.00427) (0.00425)
CE

license 0.287*** 0.288*** 0.360*** 0.209***

(0.0615) (0.0615) (0.113) (0.0512)
AC

Product fixed effects      

Time dummies      

Observations 31528 31528 31528 31528 2296 29232

R-squared 0.383 0.376 0.414 0.422 0.377 0.452

Number of 6-digit
product lines
4504 4504 4504 4504 328 4176
Notes: Standard errors (in parentheses) have been corrected for clustering at the product level. Significance at: *** 1%, ** 5%, *
10%. Source: Author’s calculations using data from World Bank’s WITS database, and China’s Protocol of the WTO accession
available on the WTO’s website.

26
 ACCEPTED MANUSCRIPT

4.4. Endogeneity of trade policy

A strand of trade theory literature highlights that trade policy reforms might be endogenous,
and consequently, we could have potential problems of reverse-causality, when analyzing the
impact of trade liberalization on imports. More specifically, it has been argued that high levels
of import penetration can lead domestic interest groups to intensify their lobbying for activity

T
for protection, so that estimates of trade liberalization effects on imports might be biased

IP
downwards. Using data from about 300 industries in the 1983 US manufacturing sector,

CR
Trefler (1993) documents that the impact of the NTB liberalization on imports was indeed
much larger when trade protection was modelled endogenously, in respect to the case where

US
trade protection was exogenously treated.

However, more recent studies have argued that trade policy reforms in China, especially

N
during its accession to the WTO, are far from being considered endogenous, since China’s
MA
willingness to become a market economy goes beyond the specific groups’ interests
(Branstetter and Lardy, 2008). Indeed, it has been documented an import tariff convergence
over time to a uniform level in all sectors in China during the period of 1995-2007 (Brandt, et
ED

al., 2012), as well as an insignificant correlation of (input) tariff changes between 2000 and
2006 with initial industry performance characteristics, such as value added, intermediate
PT

inputs, investments, etc. (Bas and Strauss-Kahn, 2015). In other words, several studies
provided evidence that changes in tariffs in China are exogenous, i.e. unlikely to be affected
CE

by the sectors’ performance or lobbying activities.

Following Bas and Strauss-Kahn (2015), we also check whether there is any correlation
AC

between a change in each trade policy variable during the period of 2000-2006 and the initial
characteristics at a more disaggregate level – i.e. at the 6-digit product level, instead of the 4-
digit industry level – through using however different explanatory variables according to the
available data. We include the import unit value (in log), the share of the OECD economies in
total source countries, the share of foreign-owned importers, and the share of trade
intermediaries in total importing firms, that are expected to negatively affect trade protection.
Indeed, if the foreign imported goods are highly expensive, they are more likely to be less
competitive than domestic ones, and therefore the latter would need less protection.
Moreover, export partners from OECD countries, foreign-owned importing firms, and trade
intermediaries might be strong lobbying groups in removing trade barriers for products that
they are mostly engaged with. Finally, we also account for the China-specific import elasticity
of substitution (computed by Broda, et al., 2006), and whether products were subjected to

27
 ACCEPTED MANUSCRIPT

state trading and/or trading rights restrictions in the beginning of the period, which are
expected to positively influence trade protection. Indeed, both homogeneous goods, i.e.
products that exhibit a high elasticity of substitution, and product groups including varieties
that were not freely traded, are more likely to be protected with other measures.

T
IP
Table 9 - Endogeneity of trade policies

CR
Dependent variables Δ tariff Δ quota Δ tendering Δ license

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

import unit value (in log) 5.93e-06 -0.00334 -0.0132*** 0.000507

US
(0.000981) (0.00513) (0.00436) (0.000386)
foreign share of import firms 0.00248 0.0555 0.0269*** 0.000246
(0.0124) (0.0341) (0.00880) (0.00772)
OECD share of source countries 0.0116 -0.00246 0.00285 -0.00315

intermediary share of import firms

(0.0131)
-0.00846 N (0.0233)
-0.00760
(0.00529)
-0.00654
(0.0111)
0.0116
MA
(0.0117) (0.0289) (0.0107) (0.00852)
state importing -0.270*** -0.259** -0.0915 -0.270
(0.0773) (0.120) (0.0833) (0.166)
trading rights restrictions -0.0120 -0.159 6.88e-05 -0.000584
(0.0414) (0.111) (0.00967) (0.00147)
elasticity of import substitution -0.000879 -0.0203 -0.00131 -0.000287
ED

(0.00474) (0.0153) (0.00256) (0.000272)

constant -0.00651 0.0143 0.0665*** -0.00604
(0.0169) (0.0455) (0.0228) (0.0118)
PT

2-digit Industry fixed effects    

CE

Observations 4,453 4,453 4,453 4,453

R-squared 0.406 0.237 0.105 0.370
Notes: Dependent variables are changes in trade policies related to the entire sample period (2000-2006). Explanatory variables refer to the
initial year of the sample period (2000). Standard errors (in parentheses) have been corrected for clustering at the 2-digit industry level.
Significance at: *** 1%, ** 5%, * 10%. Source: Author’s calculations using data from the database of Chinese Customs Trade Statistics
AC

(CCTS) managed by the General Administration of Customs of China; China’s Protocol of the WTO accession available on the WTO’s
website, and Broda, et al. (2006).

Table 9 shows that changes in tariffs, quotas and licenses do not exhibit any statistically
significant correlation with all explanatory variables, except for state-trading. It seems that
products subject to state-trading are associated with more intense tariff and quota
liberalizations, compared with other products. This is surprisingly in contrast with our
expectations, since products subject to state-trading are expected to be slower in the process
of international liberalization. This can be considered a further signal that trade integration
policies are not impeded by domestic public interests. However, it is worth noting that the
latter variable is time-invariant in our dataset and is therefore captured by the product fixed
effects in all our regressions displayed in the previous sections. Finally, tendering
liberalization appears to be uncorrelated with the majority of initial product characteristics –

28
 ACCEPTED MANUSCRIPT

including state-trading status – except for initial import unit value (proxying the price of
foreign products) and the presence of foreign-owned importers. More specifically, reductions
in tendering restrictions are associated with higher foreign imported products prices and lower
presence of foreign-owned importers, which suggest that tendering protection can be stronger
for products that are domestically less competitive and/or imported by foreign-owned firms.

T
IP
Thus, while we confirm a strong exogeneity of tariff, quota and license policies in China in
line with former studies, we are aware that tendering policy might be weakly exogenous.

CR
US
5. Conclusion
This paper attempts to explore the imports value evolution in China over the period of 2000-
N
2006 through investigating the relevant role played by non-tariff barriers removal, as
MA
scheduled within China’s WTO accession protocol, in addition to tariff liberalization. The
main findings show that the growth of imports under the ordinary trade regime in China is due
mainly to both tariff cuts and license barriers removal. The value of imports was found to be
ED

declining as tendering requirements were removed and unaffected by the abolition of quotas.
These results differ slightly between agricultural and manufacturing products. The
PT

agricultural imports value increased primarily due to the removal of import licenses, while
tariff reductions were found to have no statistical significant effect. No agricultural products
CE

appear to be subject to any other traditional NTBs (i.e. quotas or tendering). In contrast, the
manufacturing imports value changed principally due to increases resulted from tariff
AC

liberalization, given that the removal of tendering requirements had the surprising effects of
lowering import volumes, while the abolition of quotas and licenses had no significant effect.

The country-analysis within products also shows that the change in imports from tariff and
tendering liberalization (in manufacturing sector) as well as from license liberalization (in
agricultural sector) occurs because of changes in both the country-intensive and extensive
margins. China imported from many more countries (especially OECD countries) and
imported more from each country due to tariff and license liberalization. Furthermore, a
decrease in both the number of OECD source countries and the average imports per country
from the removal of import tendering requirements would tend to suggest some potential
import reallocation from high-quality to low-quality country-varieties. The opposite changes
that were found for both the extensive country-margins of product imports, from the
elimination of quotas, associated with a decline in the intensive country-margin, would

29
 ACCEPTED MANUSCRIPT

suggest a redistribution of imports along a larger range of countries. In other words, while it
seems there was no discrimination from quota restrictions between developed and developing
source countries, we find some evidence of potential discrimination from tendering
restrictions of non-OECD countries in favor of OECD countries. The reason might be that,
amongst foreign countries, only the developed were able to be considered suitable, according

T
IP
to the stringent tendering requirements.

CR
The firm-analysis within products documents that imports’ increase from tariff reduction in
the manufacturing sector is also due to changes in both the firm-extensive and intensive
margins, i.e. both number of importing firms (especially private-domestic ones compared to

US
state-owned and foreign-owned firms) and average imports per firm increased. For
agricultural products, we find a decrease in the number of private-domestic owned firms,
N
linked to a tariff cut. Imports growth from license liberalization, found in agricultural
MA
products, appears to be connected to increases in both the number of importing firms (of
different ownership) and the import intensity per firm. We also find some evidence of an
increase in the number of foreign-owned importers, at the expenses of the number of Chinese
ED

private-owned importers for manufacturing products, due to the removal of import licenses.
Declines in manufacturing imports, linked to tendering liberalization, occur through both
PT

firm-intensive and extensive margins, although the number of foreign-owned firms actually
increases. We again notice some kind of reallocation across firms, following the abolition of
CE

quotas at the expenses of private-domestic owned firms, given that a decline in the average
imports per firm is associated with a lower number of private-domestic owned firms and a
AC

larger number of other firm groups. These findings would suggest a potential discrimination
of foreign-owned manufacturing firms from all NTB restrictions (quotas, licenses and
tendering requirements), and since these firms are found to be on average more productive
than the other firms, we can confirm that import permits were not efficiently allocated in
China – like export quotas (Khandelwal, et al., 2013) – implying misallocation of resources.
Furthermore, while tariff reduction allows more producers as well as more intermediaries to
import, NTB liberalization seems to reallocate imports from intermediary firms to direct
importers.

Finally, we exclude any hypothesis of substitutability between tariffs and the traditional non-
tariff barriers, since products subject to NTB liberalization, through the elimination of quotas,
licenses and tendering, turn out to be associated with larger tariff cuts. We also provide some

30
 ACCEPTED MANUSCRIPT

evidence that China’s trade policy reforms are unlikely to be influenced by product-level
characteristics and lobbying activities.

To conclude, of the trade policy instruments that we have studied, tariffs seem to be best
placed to explain the product-level growth in Chinese imports of manufacturing goods over

T
the period of 2000-2006. For agricultural products, the elimination of import licenses plays a

IP
more important role. The insignificant effect of quota removal and the significant negative

CR
effect of the abolition of tendering requirements are more surprising. An explanation for these
results could be that the traditional NTBs have been replaced by new forms of NTBs, such as,
antidumping, technical barriers to trade (TBT) and sanitary and phytosanitary (SPS)

US
measures. We leave further exploration of these new trade policy instruments for future
research.
N
MA
Acknowledgments
ED

The author would like to thank Fabrice Defever, Richard Kneller and Alejandro Riaño for
their helpful comments and suggestions, as well as Zheng Wang for sharing some data. All
remaining errors are his own.
PT
CE

References

Anderson, S. P., & Schmitt, N. (2003). Nontariff Barriers and Trade Liberalization. Economic
AC

Inquiry, 41, 80-97.

Bao, X., & Qiu, L. D. (2010). Do Technical Barriers to Trade Promote or Restrict Trade?
Evidence from China. Asia-Pacific Journal of Accounting & Economics, 17, 253–280.
Bas, M., & Strauss-Kahn, V. (2015). Input-trade liberalization, export prices and quality
upgrading. Journal of International Economics, 95, 250-262.
Brandt, L., Van Biesebroeck, J., Wang, L., & Zhang, Y. (2012). WTO Accession and
Performance of Chinese Manufacturing Firms. CEPR Discussion Papers 9166,
C.E.P.R. Discussion Papers.
Branstetter, L., & Lardy, N. R. (2008). China’s Embrace of Globalization. In L. Brandt, & T.
G. Rawski (Eds.), China’s Economic Transition: Origins, Mechanisms, and
Consequences (pp. 829-826). New York: Cambridge University Press.
Broda, C., Greenfield, J., & Weinstein, D. (2006). From Groundnuts to Globalization: A
Structural Estimate of Trade and Growth. NBER Working Paper No. 12512.

31
 ACCEPTED MANUSCRIPT

Dean J. M., Signoret, J. E., Feinberg, R. M., Ludema, R. D., & Ferrantino, M. J. (2009).
Estimating the Price Effects of Non-Tariff Barriers. The B.E. Journal of Economic
Analysis & Policy, 9, 1-41.
Defever F., & Riaño, A. (2012). China’s Pure Exporter Subsidies. CEP Discussion Papers
dp1182, Centre for Economic Performance, LSE.
Feinberg, R. M., & Reynolds, K. M. (2007). Tariff Liberalisation and Increase Administrative

T
Protection: Is There a Quid Pro Quo? The World Economy, 30, 948-961.

IP
Henn, C., & Mcdonald, B. (2011). Protectionist Responses to the Crisis: Damage Observed in
Product-Level Trade. Washington, D.C., International Monetary Fund, IMF Working

CR
Paper 11/139.
Ianchovichina, E., & Martin, W. (2006). Trade Impacts of China's World Trade Organization
Accession. Asian Economic Policy Review, 1, 45-65.

US
Kee H., Nicita, A., & Olarreaga, M. (2009). Estimating Trade Restrictiveness Indices.
Economic Journal, 119, 172-199.

N
Khandelwal, A. K., Schott, P. K., & Wei, S.-J. (2013). Trade Liberalization and Embedded
Institutional Reform: Evidence from Chinese Exporters. American Economic Review,
MA
103, 2169-95.
Limão, N., & Tovar, P. (2011). Policy choice: Theory and evidence from commitment via
international trade agreements. Journal of International Economics, 85,186-205.
ED

Moore, M. O., & Zanardi, M. (2011). Trade Liberalization and Antidumping: Is There a
Substitution Effect? Review of Development Economics, 15, 601-619.
Ray, E. J., & Marvel, R. (1984). The Pattern of Protection in the Industrialized World. Review
PT

of Economics and Statistics, 66, 452-58.

Topalova, P., & Khandelwal, A. (2011). Trade Liberalization and Firm Productivity: The
Case of India. The Review of Economics and Statistics, 93, 995-1009.
CE

Trefler, D. (1993). Trade liberalization and the theory of endogenous protection: an

econometric study of US import policy. Journal of political Economy, 101, 138-160.
AC

U.S. International Trade Commission (1999). Assessment of the Economic Effects on the
United States of China’s Accession to WTO.
WTO (2006). China Trade Policy Review, WT/TPR/S/161.
WTO (2012). World Trade Report 2012: Trade in natural resources, Geneva, WTO.

32