Trade Liberalization, Tariff Overhang and Antidumping Filings in

Trade Liberalization, Tariff Overhang and
Antidumping Filings in Developing Countries+
Thomas J. Prusa*
Rutgers University and NBER
Yue Li**
University of International Business
and Economics–Beijing and World Bank
April 2009
Preliminary Draft
Abstract
It has been argued that antidumping (AD) facilitates trade liberalization by serving
as a release valve for protectionist pressure. We focus on a group of prominent ADusing developing countries and examine the validity of the pressure release
hypothesis. The fact that these new AD users generally began using AD after
significant trade liberalization supports the hypothesis. Yet, if pressure exists to
reverse liberalization, it is not because products that were subject to AD scrutiny
were treated poorly in the liberalization programme. For nearly all major new
users products subject to AD investigations do not have unusually low tariffs nor
have they experienced unusually large tariff reductions. Our main point is that
that AD is not the only policy option. For all countries studied the tariff bindings
are weak in the sense that the bound levels exceed the applied levels by a
considerable margin. Because of this, developing countries can raise their applied
tariffs without running afoul of WTO rules. This is not just a theoretical possibility
– we show that for some countries 8 out of 10 HS6 tariffs have been increased in
recent years. Countries could certainly increase applied rates for industries that
are particularly susceptible to injury from trade. By doing so AD’s vague and
economically illogical rules could be reformed without placing other WTO trade
liberalization initiatives at risk. The reality is, however, that industries prefer AD
because AD duties are often far larger than the bound rates. With the cumulation
provision industries can effectively use AD protection against all major import
sources. Because AD is not needed as pressure release valve and because it
involves high duties against most import sources, AD is likely more damaging to the
world trading system than the pressure release valve hypothesis suggests.
Paper is VERY preliminary. Many cites to related literature are missing. This initial draft is just
to get our ideas down on paper and is not ready for circulation.
* Department of Economics, Rutgers University, New Brunswick, NJ 08901. Email:
[email protected]
** World Bank, Email: [email protected]
+
0
Dorothy: Oh, will you help me? Can you help me?
Glinda: You don't need to be helped any longer. You've always
had the power to go back to Kansas.
The Wizard of Oz
Introduction
Antidumping is a statute that often perplexes economists. If another country opts
to sell goods to another at a low price, the importing country gains. Why would the
importing country want to counteract its partner’s generosity? Viner (1923) argued
that the importing country might want impose antidumping duties if the partner is
engaged in predatory dumping. Unfortunately, Viner’s explanation is not a
satisfactory justification for modern antidumping. When the rules and procedures
governing the calculation of antidumping duties are studied, it becomes clear that
the law punishes far more behavior than just predatory dumping. 1 Firms making
large profits on their exports are nevertheless subject to antidumping duties.
Simply put, the legal definition of “unfair” bears no resemblance to the economics
definition.
Political economists have a different interpretation of antidumping. Antidumping
(AD) is really not about punishing specific behavior, but rather serves a larger
purpose. The “big picture” is that all countries benefit from greater trade
Shin (1998) finds few U.S. antidumping cases would raise suspicions under USDOJ guidelines for
assessing economically unfair behavior. Competitive conditions in the industries using antidumping
could rarely are compatible with a profitable predatory pricing strategy.
1
1
liberalization. But, some politically important industries do not want trade
liberalization. Their opposition could make it impossible for a government to accept
a new trade agreement. These industries may not be efficient enough to export
even if their partners’ tariff barriers were reduced. These industries may in fact be
worried that their inefficiencies will be exposed if they must compete with foreign
firms. Rather than letting these industries’ objections put overall trade
liberalization at risk, policymakers use AD to allow these industries to insulate
themselves from foreign competition. AD facilitates trade liberalization by acting
as a release valve for protectionist pressure. It is an insurance policy for industries
worried they will suffer with more liberalization. That AD often results in the
imposition of large duties on exactly the type of the economically fair trade that the
WTO promotes is moot. That the rhetoric of AD has little connection to reality is a
small price to pay. Without a pressure release valve the WTO will not be able to
achieve new and welfare increasing agreements.
In this paper we question the validity of the pressure release hypothesis. We do
not disagree with the idea – AD definitely does offer protection to domestic
industries and by doing so mitigates (but does not completely eliminate) their
opposition to trade negotiations and agreements. Rather, we challenge the implicit
assumption that AD has to be the safety valve. Is there no other way for WTO
members to mollify potential obstructionists to new trade agreements? Is there
really a need for a law that defines “unfair behavior” in a way that is inconsistent
with the most basic lessons of microeconomics?
2
Clearly the answer is “no”. Most obviously the GATT/WTO has a safeguard
provision that allows countries to increase tariffs on imports that seriously injure a
domestic industry. With safeguards there is no need to distort the definition of
unfair. Safeguards have some drawbacks. They can only be imposed if imports
have increased, restrict the policy response, and can only be imposed for three
years. If these conditions are violated, partners can respond (retaliate) by raising
their own tariffs. Safeguard’s built-in compensation rules make it more difficult for
countries to simply “undo” negotiated liberalization by imposing new duties for
indefinite periods of time. This virtue may also explain safeguard’s unpopularity:
over the past decade AD investigations outnumber safeguard cases by a 20 to 1
ratio. Apparently industries needing a pressure release valve feel that limitations
on the safeguard provisions limit its usefulness.
In this paper we argue that many governments can mitigate pressure to reject trade
agreements by simply selectively raising the tariffs for the handful of key industries
(and/or products) that balk at trade liberalization. It may sound odd to say that the
way to promote liberalization is to selectively raise tariffs but once we get beyond
AD’s rhetoric, for all intents and purposes that is what is already being done. The
reality is that the new users driving current AD usage all have weak tariff bindings
or “tariff overhang”. By this we mean that the bound levels exceed the applied
levels by a considerable margin. Weak bindings characterize all the developing
countries studied in this paper and we believe most (all?) developing countries.
3
Because of the weak bindings, developing countries can raise their applied tariffs
without running afoul of WTO rules.
As a result of tariff overhang, the “necessity” for trade negotiators to ignore the
absurd nature of modern AD rules disappears. If countries need discretion to raise
tariffs to protect vital national interests and politically important industries, the
ability already exists in the negotiated tariff bindings. The charade that firms
accused of dumping are violating sound economic principles can end. We are not
arguing that there is no need for AD. Rather, the point is that for most WTO
members the flexibility to “release pressure” already exists. If we recognize this
leeway, trade negotiators can get started on rewriting the AD statutes in ways that
makes economic sense.
We acknowledge that phenomenon of tariff overhang does not hold for the major
developed AD users – the U.S, the EU, Australia, Canada. Thus these countries
may still need a pressure release valve. Yet, we also point out that for these
countries tariffs on manufactured products are already very small. Since
manufactured goods account for the bulk of their AD disputes the question emerges
what pressure needs to be relieved for these users? Can it stem from current
negotiations if developed countries are discussing moving from a 4% to 3% ad
valorem tariff? We submit that for most developed countries AD is not a result of
any new or forthcoming liberalization but rather simply a discretionary exception to
WTO commitments. The trading system would be more legitimate if it simply
acknowledged that countries need the flexibility to have exceptions for some small
4
percent of their imports. The WTO could decide on the maximum number of line
items (or perhaps import share) where discretion is available and then let countries
decide how to allocate the waiver (a scarce resource). The farce that AD is anything
but discretionary protection should end.
The rest of the paper is organized as follows. We will begin by reviewing the
emergence of the developing countries as the new prominent users of AD. The
proliferation of AD has shifted the focus from why traditional users need AD to why
new users need AD. After this general discussion we narrow our focus to nine
active AD users – Argentina, Brazil, Colombia, India, Mexico, Peru, South Africa,
Turkey, and Venezuela. These nine countries were chosen primarily for two
reasons. First, these nine are among the heaviest new users of AD. They account
for more than one-third of all AD actions worldwide over the past decade. The
second reason is more mundane. Our analysis requires good tariff coverage and
this requirement eliminates some other important new users where such data
simply does not exist for a sufficient number of years.
As we show, having an AD statute is not the same as using the statute. Often times
AD statutes are enacted but not used. A trigger is needed. We then document that
trade liberalization preceded these countries’ initial use of AD. The aggregate data
supports the view AD seems to be serving the “pressure release” role.
We then take a closer look at the applied and bound rates at the HS6 level and find
that the pressure release story might not be so obvious. If pressure exists to reverse
5
liberalization, it is not because products that were subject to AD scrutiny were
treated “poorly” in the liberalization programme. Generally speaking, the products
that were subject to AD scrutiny do not have unusually low tariffs nor have they
experienced unusually large tariff reductions. We find economically small
differences between applied tariffs for products that were subject to AD scrutiny
and those that do not. We also do not find that products that were subject to AD
scrutiny have experienced larger than average tariff reductions.2
We then present evidence in support of our main point: AD is not the only policy
option. For all countries studied there is a substantial tariff overhang. Because of
weak bindings, developing countries can raise their applied tariffs without running
afoul of WTO rules. One concern is that countries may be loath to exercise this
discretion. Doing so may show they are “weak” and can be manipulated by domestic
industries. While this may be true, it is not mean that countries do not exercise
discretion. We show that all of the countries in our dataset have raised their
applied tariffs for some products. For five of our nine countries more than half of
the HS6 lines have experienced increases in their applied rates. For three countries
(Mexico, Argentina, and Brazil) more than 70% of their HS6 lines have experienced
tariff hikes.
Developing countries could certainly increase applied rates for industries that are
particularly susceptible to injury from trade. There is no need for the new users to
South Africa and India are exceptions; for these countries products subject to AD did experience
bigger tariff reductions than other products. This is consistent with the findings of Bown and Tovar
(2008).
2
6
have to resort to AD in order to diffuse political pressure for protection. Because
they have another option new users have little to gain by embracing the current AD
rules and instead should advocate reform. By acknowledging the inherent
flexibility in their own tariff schedules, AD’s vague and economically illogical rules
can be reformed without placing other WTO trade liberalization initiatives at risk.
The reality is, however, that industries prefer AD because antidumping duties are
often far larger than the bound rates. With the cumulation provision industries can
effectively use AD protection against all major import sources.
Motivation – The Emergence of New Users
As discussed in Prusa (2001), until the mid-1980s, more than 99 percent of all AD
cases were filed by just four “traditional” AD users – the U.S., the EU, Australia,
and Canada. Beginning in the mid-1980s, however, more and more cases began to
be filed by countries that did not historically use AD (Figure 1). By the early-1990s,
the share of worldwide AD disputes accounted for by the traditional users fell below
one-half and now stands about 30%.
7
Figure 1 – Percent of AD Cases Filed By Traditional Users
100%
90%
% AD Cases by Traditional Users
80%
70%
60%
50%
40%
30%
20%
10%
0%
Developing countries account for the bulk of the new AD activity. From 1995 to
1999, developing countries filed 559 cases compared to the world total of 1029 cases.
One key difference between AD activity by new users and traditional users is the
intensity of use. As discussed in Finger, Ng, and Wangchuk (2002) and Prusa
(2005) per dollar of imports AD usage by new users is far higher than by the
traditional users. Brazil’s AD intensity is five times that of the U.S., India’s seven
times, and South Africa and Argentina’s twenty times. The intensive use of AD by
the new users provides compelling evidence for anyone one is concerned that AD
offers discretionary protection.
Why did the traditional users dominate the AD landscape prior to the mid-1980s?
It is not the case that they were the only countries with an AD statute. Table 1
8
reports information from Zanardi (2004). In order of their year of implementing an
AD statute, we list all GATT/WTO members with AD statutes through 2002. We
also report the number of AD disputes initiated by each country over the 1981-2001
period. As shown, many countries adopted AD prior to 1980, but only the four
traditional users regularly initiated AD disputes. Having an AD statute on the
books is a necessary but not sufficient condition for a country to apply AD against
its trading partners.
Table 1 – Year AD law Implemented and No. of AD disputes initiated, 1981-2001
Country
Canada
Australia
South Africa
United States
Japan
France
New Zealand
United Kingdom
Germany
Greece
Norway
Malawi
Zambia
Zimbabwe
Cyprus
Finland
Nigeria
Antigua Barbuda
Barbados
Jamaica
Malaysia
Uganda
Dominica
Grenada
South Korea
Saint Lucia
Portugal
Belgium
EEC
Ireland
Italy
Luxembourg
Yr Impl.
1904
1906
1914
1916
1920
1921
1921
1921
1951
1954
1954
1955
1955
1955
1956
1958
1958
1959
1959
1959
1959
1959
1960
1960
1963
1964
1966
1968
1968
1968
1968
1968
# cases
478
829
230
856
10
EEC
85
EEC
EEC
0 / EEC
0
0
0
0
0
16 / EEC
0
0
0
1
15
0
0
0
66
0
0 / EEC
EEC
784
EEC
EEC
EEC
Country
Yr Impl. # cases
Netherlands
1968
EEC
Austria
1971 9 / EEC
Argentina
1972
201
Uruguay
1980
3
Spain
1982 1 / EEC
Pakistan
1983
0
Taiwan
1984
73
India
1985
192
Singapore
1985
2
Chile
1986
17
Mexico
1986
180
Brazil
1987
143
Iceland
1987
0
Turkey
1989
94
Colombia
1990
27
Cuba
1990
0
Ecuador
1991
1
Israel
1991
25
Peru
1991
36
Bolivia
1992
0
Romania
1992
0
Trinidad Tobago 1992
8
Venezuela
1992
31
Bulgaria
1993
0
Slovenia
1993
1
Hungary
1994
0
Philippines
1994
19
Senegal
1994
0
Thailand
1994
7
Tunisia
1994
0
El Salvador
1995
0
Honduras
1995
0
Source: Zanardi (2004), Tables 1 & 2
9
Country
Indonesia
Nicaragua
Costa Rica
Guatemala
Panama
Paraguay
China
Czech Rep.
Morocco
Poland
Slovak Rep.
Uzbekistan
Cameroon
Egypt
Fiji
Kazakhstan
Kyrgyzstan
Lithuania
Russia
Albania
Belarus
Croatia
Ukraine
Latvia
Moldova
Saudi Arabia
Dominican Rep.
Armenia
Denmark
Sweden
Yr Impl. # cases
1995
33
1995
3
1996
6
1996
1
1996
2
1996
2
1997
22
1997
3
1997
0
1997
28
1997
0
1997
0
1998
0
1998
25
1998
0
1998
0
1998
0
1998
14
1998
2
1999
0
1999
0
1999
0
1999
3
2000
0
2000
0
2000
0
2001
0
2002
0
??
EEC
??
13 / EEC
In Table 2 we report information on AD usage by the nine key new users that are
the focus of this study. As shown, there is often a lag between the time a country
implements and the time a country starts using AD. In the case of Argentina, for
example, 19 years passed between its enactment of an AD statute and its first
dispute. Since the initial case, however, the Argentines have not been bashful users
having initiated 201 cases from 1991-2001. India had a seven year gap. South
Africa’s gap (80 years) is a bit misleading. There are reports of South Africa using
AD during the 1950s and 1960s. Unfortunately, the only consistently reliable
statistics on worldwide AD usage begins post-Tokyo Round. But, by the time the
GATT began to monitor AD use, international boycotts and trade sanctions over
apartheid isolated South Africa making AD moot. Thus, according to official WTO
reports, South Africa first began using AD in 1994. Since 1994 South Africa has
been one of the world’s most frequent users of antidumping.
Table 2 – AD Adoption and Implementation, Major New Users
Implementation of
Year AD First
AD Law
Initiated
Argentina
1972
1991
Brazil
1987
1988
Colombia
1990
1991
India
1985
1992
Mexico
1986
1987
Peru
1991
1994
South Africa
1914
1994
Turkey
1989
1993
Venezuela
1992
1993
Notes: Data from Zanardi (2004) and WTO reports
10
Time Gap
19
1
1
7
1
3
80
4
1
No. AD cases
initiated, 19812001
201
143
27
192
180
36
230
94
31
What leads a country to become a user of AD? Often times what spurs AD activity
is trade liberalization. The case of India is particularly instructive. India
underwent a major tariff reform in the early 1990s. As shown in Figure 2 India’s
average effective ad valorem tariff rate dropped from nearly 80 percent in 1990 to
about 30 percent in 1997.3 As seen in Table 2 and as discussed in Bown and
Tovar (2008) India subsequently became a frequent user of AD.
Figure 2 – India, Applied Tariff Rate
80
79
70
AD starts being used
Ad Valorem Applied Tariff Rate
60
57
50
40
32
30
29
31
28
20
16
10
0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
The same pattern is seen for other countries studied in this paper. To see this we
gathered data on the applied tariff rates applied for a period of time before each
country started using AD. As is well known, outside of a small set of OECD
Unlike data used later in the paper, this figure is based on India’s average overall tariff rate. We
do not have line item tariff data for India prior to 1995.
3
11
countries getting consistent data for on tariff coverage is difficult prior to 1996. We
were not able to get annual tariff rates for all the countries but rather only could
find “period averages”, where the period generally spans three years. Even then we
only were able to obtain data for seven countries.4 Table 3 presents the countries'
simple average applied tariff rates prior to the year when the first AD petition was
initiated.
Table 3 – Simple Average Applied Ad Valorem Tariff Prior to First AD Use
Argentina
Brazil
Colombia
India
Mexico
Turkey
Venezuela
2 Periods Prior
27.0%
n.a.
33.6%
98.8%
n.a.
26.6%
32.9%
1 Period Prior
20.5%
55.6%
30.4%
79.2%
30.9%
22.7%
30.6%
Period when AD
First Used
10.6%
30.0%
11.8%
53.0%
11.3%
9.0%
16.4%
It is clear that for all of the countries, applied tariff rates experienced a consistent
and sharp decline prior to the first AD case.5 For Brazil, for example, the applied
tariff fell from 55% to 30% during the period immediately prior to its starting to use
AD; for Mexico, the applied tariff fell from 31% to about 10% during the period prior
to its starting to use AD. For countries where we were able to assemble longer time
spans, the reductions are similarly impressive – Argentina, Colombia and Turkey’s
applied rates fell by about two-thirds and India and Venezuela’s applied rates fell
by about one-half over the two periods preceding the start of AD use.
Our main source of tariff data is the WTO. WTO tariff data does not cover years prior to 1995 so
we use UNCTAD tariff data for this analysis.
5 To confirm robustness, we find similar trend results also for weighted average applied tariffs.
4
12
Tariff Levels and AD Activity
The aggregate results provide some support for the pressure release hypothesis.
Countries that undertake large tariff reduction programmes apparently need to
start using AD. The above analysis, however, is based on aggregate tariff data and
thus tells us little about what is happening for the products actually subject to AD
investigations. It may be the case that tariffs are generally falling, but why do
some sectors seek AD protection while others do not? Do the injured sectors have
lower tariffs than other sectors? Have the injured sectors experienced greater than
typical tariff reductions?
To get at these questions, we need to look at highly disaggregated tariff data.
Following the Uruguay Round a uniform or harmonized system (HS) of tariffs was
adopted by WTO members.6 The WTO reports each member country’s tariff
schedules at the six-digit level (HS6).7 Tariffs for the vast majority of products are
recorded as ad valorem rates.8 There are three types of tariffs reported. First we
have the bound tariff rate which is the maximum tariff that can be imposed under
the WTO agreement. Second, there is the applied rate which reflects the official
tariff rate applied in a given year. In no cases can the applied rate be greater than
the bound rate but a country can choose to assess an applied rate below the bound
Prior to the Uruguay Round countries could use different systems to classify products and levy
tariffs.
7 Only the first six digits of the system are harmonized. Countries can choose to add subcategories if
they wish. For instance, the HS6 classification might define “10cm ball bearing”; a country could
add more digits (e.g., HS8) and distinguish “10cm ball bearing, carbon” and “10cm ball bearing,
aluminum”. Often the additional granularity is added only for certain products. For the majority of
countries and products the tariffs are defined at the HS6 level (i.e., there are no subcodes for most
HS6 lines).
8 We exclude products with specific or other tariff schemes from our analysis.
6
13
rate. Third, there is the effective rate. The effective rate is often the same as the
applied tariff but can vary from country to country due to preferential trade
agreements, customs unions, or other non-MFN arrangements. Throughout the
paper we assume that all imports are assessed at the effective rate.
Fortuitously, for most countries and in most AD disputes, AD duties on specific
products are imposed at the 6-digit tariff line level. In some cases countries apply
tariffs at a finer level (e.g., HS8 or HS10) and occasionally at a coarser level (e.g.,
HS4) but these are the exceptions for the countries studied in this paper. The fact
that AD activity and tariff schedules are classified with the same system allows us
to analyze the pre-existing level of protection for cases under investigation. Using
data reported and compiled by Bown (2005) we can identify all pertinent
information regarding the products being investigated: the year of investigation,
which country(ies) were involved, the HS6 codes studied, etc.
In order to understand what is happening at the product level, we first ask whether
products subject to AD investigations have less tariff protection than those that do
not seek AD protection. Table 4 compares the mean difference in effective ad
valorem tariff rates between products subject to AD protection and those not. More
specifically, we conduct the following regression on this subset of data:
݂݂݁௜௧ = ߙ + ߚܵ‫ݐ݆ܾܿ݁ݑ‬௜௧ + ߛܵ‫ݎ݋݅ݎ݆ܾܲݑ‬௜௧ + ߝ௜௧ ,
where
14
i
= observation distinguished by both HS6 product and exporting country
effit
= effective tariff rate, product/partner i, time t
Subjectit
= dummy for product/partner subject to at least one AD dispute at
time t
SubjPriorit = dummy if product subject to an AD dispute in some prior year
The regression is run for each country separately. The estimated value of β
represents average differences in effective rates between products subject to AD
investigation and those not.
We note that except for Brazil and Turkey the differences are statistically
significant. However, the direction of the difference and the relative economic
importance vary across countries. Interestingly, for all countries but Mexico and
Brazil products seeking AD protection have higher effective tariff rates than those
that do not (Table 4). This suggests a possible endogeneity issue: industries who
ultimately seek AD protection have been able to procure higher tariff rates in the
past.
15
Table 4 – Effective Tariff Rate, Products with and without AD Filings (HS6)
AD Filing
AD Filing in prior years
Constant
Argentina
1.698***
(0.094)
-1.143***
(0.155)
14.016***
(0.006)
Brazil
-0.109
(0.146)
-2.177***
(0.250)
16.467***
(0.006)
Colombia
0.921***
(0.216)
-0.770**
(0.332)
12.299***
(0.005)
India
1.193***
(0.258)
-1.260***
(0.461)
34.852***
(0.016)
Mexico
-1.533***
(0.085)
-2.374***
(0.125)
15.239***
(0.008)
Peru
4.171***
(0.088)
3.920***
(0.156)
13.347***
(0.004)
F
188.117
38.313
11.709
13.934
340.404
1443.687
r2
0.007
0.06
0.002
0.061
0.015
0.061
Numb of Obs
1,390,858
2,098,327
1,331,685
1,637,952
1,649,691
978,081
Note: Standard Errors in parentheses; * = significant at 10%, ** = significant at 5%, *** = significant at 1%
Turkey
0.232
(0.214)
2.275***
(0.408)
6.496***
(0.010)
Venezuela
3.830***
(0.190)
1.246***
(0.306)
13.368***
(0.006)
S. Africa
6.166***
(0.249)
2.528***
(0.319)
10.948***
(0.010)
16.182
0.007
1,106,116
212.151
0.005
1,205,806
339.92
0.007
2,143,404
Table 5 – Change in effective tariff rates (t-3 - t-1), Products with and without AD Filings (HS6)
AD Filing
AD Filing in prior years
Constant
Argentina
0.002
(0.058)
0.338***
(0.091)
-0.067***
(0.004)
Brazil
0.368***
(0.092)
0.712***
(0.152)
1.420***
(0.004)
Colombia
India
0.004
-0.514***
(0.072)
(0.118)
-0.018
0.912***
(0.097)
(0.211)
-0.048*** 2.803***
(0.002)
(0.008)
Mexico
Peru
Turkey Venezuela
0.021
-0.072 0.246**
0.240***
(0.098)
(0.050) (0.107)
(0.082)
-0.237*** 0.083
0.087
-0.158
(0.035)
(0.065) (0.186)
(0.123)
-0.270*** 0.881*** 0.668***
0.340***
(0.003)
(0.002) (0.005)
(0.003)
F
6.904
19.151
0.018
18.058
22.683
1.832
2.794
r2
0.041
0.114
0.047
0.052
0.054
0.049
0.008
N
993,470 1,604,603 976,569 1,228,464 1,051,856 677,133 703,892
Note: Standard Errors in parentheses; * = significant at 10%, ** = significant at 5%, *** = significant at 1%
5.084
0.022
861,290
S. Africa
-1.282***
(0.072)
0.272***
(0.092)
0.524***
(0.003)
163.612
0.025
1,667,092
16
It is not clear in our minds, however, that the differences are important. The tariff
rates for HS6 products that were subject to AD scrutiny and those that were not
subject to an AD investigation are depicted in Figure 3. In most cases the difference
in effective rates is fairly small and probably not economically significant. In most
cases the difference is less than two percentage points. We conclude that for most
countries the overall level of protection does not differ in any remarkable way for
AD and non-AD using products. Said differently, there is not a clear connection
between effective tariff rates and AD filings at product level.
Figure 3 – Effective Tariff Rates by HS Code and AD status
40
36.0
34.9
35
Ad Valorem Effective Tariff Rate (%)
HS codes without AD
HS Codes with AD
30
25
20
17.5
15.7
15
17.2
16.5 16.4
17.1
15.2
14.0
13.7
13.2
13.4
13.3
12.3
10.9
10
6.5 6.7
5
0
Argentina
Brazil
Colombia
India
Mexico
Peru
Turkey
Venezuela
S. Africa
17
In Table 5 we compare the mean change in effective ad valorem tariff rates between
products subject to AD protection and those not. Specifically, we conduct the
following regression on this subset of data:
݂݂݁௜௧ିଷ − ݂݂݁௜௧ିଵ = ߙ + ߚܵ‫ݐ݆ܾܿ݁ݑ‬௜௧ + ߛܵ‫ݎ݋݅ݎ݆ܾܲݑ‬௜௧ + ߝ௜௧ .
The idea here is to see if product that had the biggest reduction in tariffs were more
likely to result in AD investigations. The results are mixed. For India and South
Africa both have a large negative point estimate and a statistically significant
difference. For these two countries there is evidence that AD is used to recover or
undo the concessions granted in the liberalization programme. We note that our
findings are consistent with those reported by Bown and Tovar (2008) in their study
of India’s AD activity. For all other countries, however, the difference is either
statistically insignificant, of the wrong sign, or of very small magnitude.
Tariff Overhang
At this point we have learned that there is a striking relationship between trade
liberalization and when developing countries start to use AD. When we looked at
the pattern of tariff concessions actually granted, however, it is not clear that
products that ultimately have sought AD protection have any different effective
rates or were subject to greater tariff reductions than other products.
18
Whatever the exact motivation for why certain products seek AD protection, the fact
is that most countries already have the ability to relieve political pressure against
the liberalization programme without having to resort to AD. For all countries
studied, many products have weak tariff bindings. That is, the current bound tariff
is larger than the current effective tariff. As shown in Table 6, positive tariff
overhang is common for most of these new AD users. For six of the countries –
Argentina, Brazil, Colombia, Mexico, Peru, and Venezuela – less than 10 percent of
all HS6 lines by trade partner are strictly binding. For South Africa, about 30
percent of its line items have zero overhang. Only India and Turkey have most of
their line items strictly bound, but even in their cases about 40 percent of the lines
are weakly bound.
Table 6 – Fraction of Observations with Zero Overhang, 1999
Argentina
Brazil
Colombia
India
Mexico
Peru
South Africa
Turkey
Venezuela
HS6*Partner with Zero Overhang
Share (%)
No. Obs.
0.65
644
2.65
3,271
0.02
18
56.1
57,433
8.53
9,065
0.01
7
30.67
36,519
57.63
57,952
1.09
935
Total No. of
HS6*Partner
99,347
123,431
88,779
102,372
106,238
75,237
119,078
100,556
86,129
Moreover, often the tariff overhang is quite large which means there is a large gap
between the bound and effective tariff rates. In Figure 4 we present box and stem
plots of the overhang for each country using the tariff schedules for one
representative year (1999). To make the figure more readable, we trim the plots so
19
all line items with overhang greater than 50 percentage points are dropped. The
figure show that for six countries – Argentina, Brazil, Colombia, Mexico, Peru, and
Venezuela – the potential for discretionary tariff is large. Not only are most line
items weakly bound but the gap is large. In addition, even the three countries
(India, Turkey, and South Africa) that have a significant fraction of their lines
strictly bound, there remain many line items that are not bound. It is possible that
the weak bindings are associated with many of the same products that have sought
AD protection. Therefore, it is possible that even for these countries discretionary
increases in effective tariffs could mitigate the political pressure against trade
liberalization.
0
bound rate - effective rate, simple average
10
20
30
40
50
Figure 4 – Distribution of Tariff Overhang, 1999 (Trimming at Overhang ≥ 50)
IND
PER
TUR
VEN
ARG
20
BRZ
COL
MEX
ZAF
Tariff Overhang and AD Activity
To get an idea of the scope of discretionary protection available to products seeking
AD protection, we run a regression similar to those described above but now have
the tariff overhang as the dependent variable:
ܱ‫ݎ݁ݒ‬ℎܽ݊݃௜௧ = ܾ݊݀௜௧ − ݂݂݁௜௧ = ߙ + ߚܵ‫ݐ݆ܾܿ݁ݑ‬௜௧ + ߛܵ‫ݎ݋݅ݎ݆ܾܲݑ‬௜௧ + ߝ௜௧ .
21
Table 7 –Tariff Overhang, Products with and without AD Filings (HS6)
AD Filing
AD Filing in prior years
Constant
Argentina
Brazil
Colombia
-0.331*** -1.716*** 4.811***
(0.106)
(0.132)
(0.592)
2.511***
0.25
10.014***
(0.175)
(0.226)
(0.910)
18.344*** 12.308*** 27.749***
(0.024)
(0.024)
(0.057)
India
-0.103
(0.292)
-0.523
(0.522)
7.598***
(0.070)
Mexico
Peru
1.278*** -3.994***
(0.079)
(0.095)
2.256*** -3.681***
(0.116)
(0.167)
19.685*** 14.872***
(0.028)
(0.017)
Turkey
0.999***
(0.335)
0.919
(0.640)
7.688***
(0.050)
F
537.844 3501.355
37.931
385.737 1790.067 3958.803 26.669
r2
0.006
0.029
0
0.004
0.017
0.054
0
Numb of Obs
1,390,858 2,098,327 1,331,685 1,637,952 1,649,691 978,081 1,106,116
Note: Standard Errors in parentheses; * = significant at 10%, ** = significant at 5%, *** = significant at 1%
Venezuela
-4.188***
(0.328)
-1.745***
(0.527)
20.741***
(0.040)
S. Africa
-0.708**
(0.302)
9.442***
(0.387)
8.581***
(0.052)
114.948
0.001
1,205,806
212.836
0.002
2,143,404
Table 8 – Binding Tariff Overhang (Normalized), Products with and without AD Filings (HS6)
AD Filing
AD Filing in prior years
Constant
Argentina
Brazil
Colombia
India
-0.690*** -0.362*** -0.569*** -0.078***
(0.054)
(0.051)
(0.114)
(0.016)
-0.587***
-0.109
0.117
-0.017
(0.097)
(0.089)
(0.175)
(0.028)
1.933***
1.560***
3.139***
0.378***
(0.003)
(0.002)
(0.003)
(0.001)
Mexico
0.215***
(0.051)
0.301***
(0.075)
2.183***
(0.005)
Peru
Turkey Venezuela
-0.568*** -0.508*** -1.134***
(0.028)
(0.155)
(0.068)
-0.757*** -1.313*** -0.773***
(0.049)
(0.280)
(0.109)
1.561*** 2.262***
2.394***
(0.001)
(0.007)
(0.002)
F
100.905
25.942
12.598
12.887
17.07
329.466 16.552
r2
0.001
0.004
0.002
0.008
0.007
0.059
0.017
Numb of Obs
1,309,377 2,012,243 1,297,112 1,605,660 1,512,601 973,448 916,404
Note: Standard Errors in parentheses; * = significant at 10%, ** = significant at 5%, *** = significant at 1%
167.404
0.001
1,185,311
S. Africa
0.152***
(0.054)
2.182***
(0.069)
1.006***
(0.003)
511.019
0.003
1,284,648
22
The results are presented in Table 7. The constant term reflects the mean overhang
for those products not seeking AD protection. Six of the countries average double
digit overhangs. The three countries with single digit average overhangs all have a
sizeable number of line items with strict bindings which implies those that are not
strictly binding have sizeable overhangs (confirming what we saw in Figure 4). If
we look at the β parameter, we see that the overhang for products that sought AD
protection is consistently smaller than for other products. While the difference in
most countries is not large, it does raise the possibility that governments might be
somewhat more limited in using discretionary tariff increases to release
protectionist pressure. They may have flexibility for sensitive products, but they
may not have enough leeway to satiate domestic protectionist demands.
23
Figure 5 – Weak Bindings, by HS Code and AD status
35
32.6
30
HS codes without AD
27.7
Tariff Overhang (Binding - Effective)
HS Codes with AD
25
21.0
20.7
19.7
20
18.3 18.0
16.6
14.9
15
12.3
10.9
10.6
10
8.7
8.6
7.9
7.7
7.6 7.5
5
0
Argentina
Brazil
Colombia
India
Mexico
Peru
Turkey
Venezuela S. Africa
There are at least two reasons why this might not be a significant problem. First,
for most countries the difference in overhang is less than one or two percent. The
similarity between the overhang values is depicted in Figure 5. Only Peru and
Venezuela have considerably smaller overhang for AD products. Second, the
smaller overhang might reflect that the products seeking AD protection had higher
than normal effective tariffs. That is, the industries that seek AD protection just
have been able to garner a lot of protection. In attempt to control for this we
“normalize” the overhang by the size of the effective tariff,
24
ܾ݊݀௜௧ −݂݂݁௜௧
݂݂݁௜௧
= ߙ + ߚܵ‫ݐ݆ܾܿ݁ݑ‬௜௧ + ߛܵ‫ݎ݋݅ݎ݆ܾܲݑ‬௜௧ + ߝ௜௧ .
The results are shown Table 8. As shown, for all nine countries line items that have
sought AD protection have a lower normalized overhang. The magnitude of the
difference is also sizeable. The results suggest that that products with low tariff
overhang is more likely to face AD petitions. This might mean that simple
discretionary tariff increases might not be sufficient to relieve the protectionist
pressure. (This will be an issue we will discuss further below.)
Are Countries Willing to Adjust Effective Tariffs?
Another reason why discretionary increases tariffs may be insufficient to diffuse
political pressure against trade liberalization is the possibility that countries are
loath to adjust their tariff schedules. For instance, governments may feel that
adjusting tariffs may signal that they are weak. Once the government accedes to
one or two requests for tariff increases, it will be over-run with such requests. AD
has the “virtue” of being misunderstood by most outsiders. While in practice AD
essentially is serves the same function as discretionary increases in effective tariffs,
AD obfuscates the process and hence does not create an avalanche of protection
requests. AD allows governments to commit to their announced effective tariff
schedules while still granting large tariffs for politically important industries.
25
The reality, however, is that governments often do adjust their effective tariffs, fare
more often than to cause worry that a few more line item increases would be
awkward. To see this we reviewed each country’s effective tariff rates over time.
We narrowed our focus to just the years following each country’s initial use of AD.
While the results are not sensitive to the time period we look at, we thought it
would be most likely the case that adjustments to the effective rate would be less
likely given the availability of raising protection via AD.
To insure that we are observing actual tariff changes we restricted our tariff
database to only those HS6 codes that represent a single line item product (i.e., no
sub-codes). By doing this we can insure that the observed changes in the tariff rate
does not simply reflect product shifting at the HS8 or HS10 level.
26
Table 9 – Share of HS6 products with at least one tariff increase in 15 years after initial AD dispute
South Africa
Argentina
Brazil
Colombia
Mexico
Peru
Venezuela
Turkey
India
Year of Initial AD
Disputes
1994
1991
1988
1991
1987
1994
1993
1993
1992
No. Products
with Tariff Increases
314
3,563
3,415
2,637
2,300
148
1,021
204
2,356
No. of
Products
4,702
4,631
4,253
4,672
3,237
4,431
4,384
2,580
4,716
Percent of HS6
Products with Tariff
Increases
6.7%
76.9%
80.3%
56.4%
71.1%
3.3%
23.3%
7.9%
50.0%
27
Table 9 gives our findings. Most countries have increased the effective tariff rate on
the majority of their products since the time they began using AD. For instance, in
the case of Argentina and Brazil, about 8 out of 10 HS6 codes experienced at least
one increase in the effective tariff rate. Colombia and India have raised the effective
rate on about half the line items and Venezuela for about one out of four products.
Only three countries has raised the effective rate on a small number of products.
But, even for these countries, the observation that 3-10% of line items have had
their effective rate raised is impressive.9 We note that this is still a higher
incidence of protection increases than we observe via AD.
The Magnitude of Protection
We want to return to the question of whether discretionary tariff increases would be
sufficient to release the political pressure against liberalization. We think the
answer depend largely on how one conceives the extent of the pressure that needs to
be released.
If one believes that the government needs some discretion to “undo” the
liberalization that has injured key products and industries, then the answer is a
resounding “yes”; the flexibility is already in the tariff schedules. The overhang is
greater than the tariff concessions that were recently granted.
This is especially true for South Africa given that almost one-third of its line items have zero
overhang. Thus, for the product where discretionary protection is an option, South Africa has raised
its effective rates about 10% of the time.
9
28
To see this we calculated the size of the overhang as compared to the change in the
effective rate:
‫ ݕݐ݈ܾ݅݅݅ݔ݈݁ܨ ݂݋ ݊݅݃ݎܽܯ‬୧୲ = ܱ‫ݎ݁ݒ‬ℎܽ݊݃௜௧ − ∆‫݂݂݅ݎ݂݂ܽܶܧ‬௜
= ܾ݊݀௜௧ − ݂݂݁௜ − ሼ݂݂݁௜௧ିଷ − ݂݂݁௜௧ିଵ ሽ .
and then run a regression
݉ܽ‫ݕݐ݈ܾ݅݅݅ݔ݈݂݁ ݂݋ ݊݅݃ݎ‬௜௧ = ߙ + ߚܵ‫ݐ݆ܾܿ݁ݑ‬௜௧ + ߛܵ‫ݎ݋݅ݎ݆ܾܲݑ‬௜௧ + ߝ௜௧ .
In Table 10 we present the results. For all countries the margin of flexibility is far
larger than the recent tariff concessions that were granted. For seven of the nine
countries the margin is well into the double digits (e.g., 15-20 percentage points); for
India and Turkey the margin is about 7 percentage points. While smaller, the fact
remains that all of the countries have more than enough flexibility in their
schedules to be able to “undo” recent concessions. For the products that sought AD
protection we see some mixed results. For three of the nine countries that margin
of flexibility is larger than for other products but for the other six countries the
margin is a bit smaller. In all cases, however, we believe that the difference in AD
using products and those not using AD are economically insignificant. Our bottomline is that these countries do not have to resort to the economically dubious AD
protection.
29
Table 10 – Difference between Tariff Overhang and Change in Effective Tariff (t-3 to t-1)
Argentina
AD Filing
AD Filing in prior years
Constant
Venezuela
S. Africa
-0.095
-2.276***
0.472
0.323
1.745*** -3.613*** -1.297*** -3.583***
(0.109)
(0.149)
(0.675)
(0.312)
(0.284)
(0.094)
(0.387)
(0.350)
2.111***
-0.217
9.648*** -1.232** 2.994*** -3.682***
0.895
-1.666***
(0.180)
(0.250)
(0.921)
(0.555)
(0.117)
(0.162)
(0.659)
(0.531)
18.261*** 15.104*** 26.802*** 7.093*** 19.134*** 16.643*** 7.479*** 21.701***
(0.008)
(0.007)
(0.017)
(0.021)
(0.008)
(0.005)
(0.018)
(0.012)
Brazil
Colombia
India
Mexico
Peru
-0.173
(0.305)
9.240***
(0.391)
9.032***
(0.014)
F
69.119
117.933
55.192
2.907
346.949 1007.641
r2
0.006
0.036
0
0.01
0.006
0.034
N
1,092,817 1,728,034 1,065,348 1,330,836 1,171,735 752,370
Note: Standard Errors in parentheses; * = significant at 10%, ** = significant at 5%, *** = significant at 1%
Turkey
6.511
0
804,448
57.706
0.001
947,419
279.627
0.001
1,786,170
30
Table 11 – Difference between AD duty and Tariff Overhang (Counties with Ad Valorem ADD)
Argentina
Brazil
Peru
Venezuela
Turkey
No. Products
(HS6)
217
100
60
46
122
No. Cases
103
81
34
20
35
No. Investigations
65
53
13
14
19
Mean
106.59
29.16
42.22
20.71
37.13
Median
39.26
19.55
13.49
20.03
30.93
SD
237.09
46.66
128.61
33.51
23.92
No. Investigations
43
30
9
13
7
Mean
106.27
40.70
47.20
20.13
19.44
Median
38.65
27.37
13.49
20.22
4.70
SD
264.81
57.30
141.47
33.26
28.29
Number of tariff lines = 1 (effective rate)
Argentina
Brazil
Peru
Venezuela
Turkey
No. Products
(HS6)
137
50
49
38
12
No. Cases
63
41
28
17
12
Table 12 – Difference between AD duty and Change in effective tariff rates (t-3 - t-1),
(Counties with Ad Valorem ADD; No tariff lines=1)
Argentina
Brazil
Peru
Venezuela
Turkey
No. Products
(HS6)
No. Cases
No.
Investigations
94
23
43
32
8
41
17
23
13
8
21
13
8
10
3
Mean
108.31
80.84
67.53
50.10
41.74
Median
55.26
45.50
31.49
55.01
28.90
SD
282.85
71.01
150.87
30.70
28.51
31
However, these results do not imply that countries can use their tariff schedules to
grant the same level of protection as AD. If one believes that the government needs
some discretion to act just like AD, then the answer to the question “is tariff
flexibility enough” is a resounding “no”; on average, AD duties are far larger than
the flexibility in the tariff schedules.
To investigate this we compared the ad valorem AD duties with the tariff overhang.
Since only five countries in our sample report AD duties as ad valorem we must
limit ourselves to looking at just Argentina, Brazil, Peru, Venezuela, and Turkey.
In Table 11 we report some statistics on the HS6 codes subject to investigation for
each country. For Argentina, for instance, there were 216 HS6 codes that we
subject to AD investigations. Those codes were identified from 65 AD investigations
and 63 AD cases.10 If each investigation involved just one exporting country, then
the number of cases and investigation would be identical. However, some
investigations involve more than one country which means a country will report
more cases than actual investigations. With cumulation, the cases are almost
always a part of the same investigation.
For Argentina the average AD duty for these products was about 106 percentage
points greater than the overhang. Given that Argentina has an average overhang
of about 18 percentage points (a big number), it becomes clear that AD results in
prohibitive tariffs. Similar, albeit not quite so dramatic differences are found for
If each investigation involved just one exporting country, then the number of cases and
investigation would be identical. However, some investigations involve more than one country which
means a country will report more cases than actual investigations.
10
32
all of the countries. Typically, we see that AD results in protection that is about
20-40 percentage points larger than the already considerable overhang gap.
In the bottom panel of Table 11 we repeat the analysis, but this time limiting
ourselves to just those HS6 products that have just one tariff line associated with
them. Again, the idea here is to eliminate the possibility of measurement error due
to product shifting at finer levels of the tariff schedule. With one exception
(Turkey), the results very closely mirror the results from the full sample. Namely,
AD duties are far, far larger than the tariff overhang gap. For Turkey, when we
limit ourselves to just HS6 codes with a single line item, the number of observations
falls dramatically, from 122 to 12. With the much smaller sample we see that AD
duties are still considerably larger than the overhang but now the difference is just
19 percentage points rather than 38 percentage points.
In Table 12 we explore the extent to which AD duties “undo” recent tariff
reductions. As seen, the AD duties are 40-100 percentage points higher than any
recent tariff change. Thus, if AD is acting as a pressure release valve, it must be
releasing pressure not associated with recent liberalization.
Given that we have shown that the overhang would more than compensate for any
recent tariff reductions, the results makes on question what the meaning of
“pressure release” is. Apparently, if one thinks that AD is a pressure release
mechanism, “return to autarky” is what is needed to release pressure.
33
Conclusion and Future Research Direction
In the paper we have explored the extent to which AD is needed to release political
pressure for protection stemming from recent trade liberalization programmes. We
find that the new users first use of AD is generally preceded by significant tariff
reductions. This relationship is apparently the empirical basis for the pressure
release hypothesis.
However, once we explore the idea further we find little clear support in the tariff
schedules for the political necessity to use AD. With a couple of exceptions,
products/industries and that have sought AD protection have not experienced
unusually large reductions in tariffs nor do they have unusually low tariffs.
Moreover, if countries need to selectively “undo” the recent tariff concessions to
release the protectionist pressure the power already exists within their tariff
schedules. For all the countries studied in the paper – countries that account for
the bulk of AD use by new users over the past decade – the applied tariffs are only
weakly bound. The tariff overhang for most products far exceeds any recent tariff
reductions.
In the Wizard of Oz Dorothy searches for some way to get back where she belongs –
Kansas. Similarly, industries often struggle to find a way to get back to the time
when they had high tariffs to protect them from foreign competition. In Dorothy’s
case, only at the end of the story is it revealed that she did not need the wizard to
get her back – she had the power all along. In the case of developing countries they
do not need AD to “undo” the injurious effects of recent tariff liberalizations.
34
Rather, they have had the power all along in their tariff schedules. Developing
countries already have the power to relieve the pressure in their tariff schedules –
they just need to learn to use it.
The concern that such discretionary manipulation of their applied schedules would
result in a widespread abandoning of their concessions seems improbable. First, it
already is the case that countries often adjust their applied schedules. For all
countries studied, the incidence of increased effective tariffs rates far exceeds the
incidence of AD duties. Second, we are not suggesting countries violate the bound
rates in the WTO agreement. Adjusting the effective rates is well within their
rights.
From our view, the big advantage of our proposal is that it might allow the new
users to realize that they are not wed to the current AD rules. If they pursue AD
reform they are not handcuffing their ability to respond to domestic demands for
relief. Reforming AD would be valuable because as it is currently formulated it
promotes trade policy that runs counter to other WTO principles. Modern AD is not
about stopping predatory pricing, it is about granting protection to politically
powerful industries. That AD as practiced is not what the its rhetoric is has been
long understood by trade insiders and negotiators. Modern AD is corrosive to WTO
negotiations because all members know that whatever tariff concessions are
negotiated, they can be easily reversed by claiming unfair trade.
35
Data Description
Antidumping data:
“Global Antidumping Database”: provides AD initiation date, reporting country
name, investigated country name, dumping decisions and associated dates,
Harmonized System (HS) product code for the product under investigation
the date when the first AD case was initiated in a country is also found in
this database. For further detailed information, see (Bown 2005);
Zanardi (2004) provides the date when a country first adopted AD law.
Tariff Rate data:
World Trade Organization Tariff Data: reports effective ad valorem tariff rates and
bound ad valorem tariff rates categorized by 6-digit HS product code. Year
coverage varies over countries.
UNCTAD Tariff Data: provides applied ad valorem tariff rates for highly
aggregated industries. Data covers 1984 to 1993 for most countries.
Value of Imports data:
United Nation Comtrade database: bilateral imports data categorized by 6-digit HS
product code.
References
Bagwell, K. and Staiger, R. W. “Enforcement, Private Political Pressure, and the
General Agreement on Tariffs and Trade/World Trade Organization Escape
Clause.” The Journal of Legal Studies, 2005, 34, pp. 471-513.
Blonigen B.A., Bown, C., 2003. “Antidumping and Retaliation Threats”, Journal of
International Economics, 60: 249-273.
Blonigen, B.A., Prusa., T.J., 2003. “Antidumping”, in Kwan Choi, E., Harrigan, J.
(eds.), Handbook of International Trade. Oxford, U.K. and Cambridge, MA:
Blackwell Publishers.
Blonigen, Bruce and Prusa, Thomas J. “Antidumping,” E. K. Choi and J. Harrigan,
Handbook of International Trade. Blackwell publishing, 2003.
Bown, C., 2005. “Global Antidumping Database”, World Bank Policy Research
Paper No. 3737.
DeVault, J., 1996. “The Welfare Effects of US Antidumping Duties”, Open
Economies Review, 7: 19-33.
Egger, P., Nelson, D.R., 2006. “How Bad Is Antidumping?: Evidence from Panel
Data”, University of Nottingham Research Paper No. 2007/17.
36
Feinberg, R., Reynolds, K., 2005. “Tariff Liberalization and Increased
Administrative Protection: Is There a Quid Pro Quo?”, forthcoming in the
World Economy.
Feinberg, R., Reynolds, K., 2006. “The Spread of Antidumping Regimes and the Role
of Retaliation in Filings”, Southern Economic Journal, 72: 877-890.
Finger, J. M. and Nogues, J. (2005). Safeguards and Antidumping in Latin
American Trade Liberalization: Fighting Fire with Fire, Macmillan Palgrave
(World Bank), Houndmills, Basingstoke, Hampshire.
Finger, J.M., Ng, F., Wangchuk S., 2002. “Antidumping as a Safeguard Policy”, in
Stern, R. (ed.), Issues and Options for US-Japan Trade Policies, Ann Arbor:
University of Michigan Press.
Gallaway, M.C., Blonigen, B.A., Flynn, J.E., 1999. “Welfare Costs of the US
Antidumping and Countervailing Duty Laws”, Journal of International
Economics, 49: 211-244.
Mankiw, G.N., Swagel, P.L., 2005. “Antidumping: The Third Rail of Trade Policy”,
Foreign Affairs, 84: 107-119.
Messerlin, P.A., 1990. “Anti-Dumping Regulations or Pro-Cartel Law? The EC
Chemical Cases”, The World Economy, 13: 465-492.
Moore, M.O., Zanardi, M., 2006. “Does Antidumping Use Contribute to Trade
Liberalization? An Empirical Analysis”, CentER working paper 61.
Nelson, D. (2006). The political economy of antidumping: A survey, European
Journal of Political Economy 22(3): 554-590.
Prusa, T.J., 2001. “On the Spread and Impact of Antidumping”, Canadian Journal
of Economics, 34: 591-611.
Prusa, T.J., “Antidumping: A Growing Problem in International Trade,” The World
Economy. May 2005, 28(5), 683-700.
Prusa, T.J., Skeath, S., 2002. “The Economic and Strategic Motives for Antidumping
Filings”, Weltwirtschaftliches Archiv, 138: 398-413.
Prusa, T.J., Skeath, S., 2005. “Modern Commercial Policy: Managed Trade or
Retaliation?”, in K. Choi and J. Hartigan (ed.), Handbook of International
Trade, vol. 2, Blackwell.
Shin, H. J. (1998). Possible instances of predatory pricing in recent u.s.
antidumping cases, in R. Lawrence (ed.), Brookings Trade Forum, Brookings
Institution, pp. 81-97.
Vandenbussche, H. and M. Zanardi (2005),”The Global Chilling Effects of
Antidumping Proliferation”, CEPR discussion paper no. 5597, London.
Vandenbussche, H. and M. Zanardi (2008), “What Explains the Proliferation of
Antidumping Laws?”, Economic Policy, January
Viner, J. (1923). Dumping: A Problem in International Trade, University of Chicago
Press, Chicago.
Wooldridge, Jeffrey M. Econometric Analysis of Cross Section and Panel Data.
Cambridge, Massachusetts: The MIT Press, 2002.
Zanardi, M., 2004. “Antidumping: What are the Numbers to Discuss at Doha?”, The
World Economy, 27: 403-433.
37
Zanardi, M., 2006. “Antidumping: a Problem in International Trade”, European
Journal of Political Economy, 22: 591-617.
38