WHAT IS GOING ON WITH CONTEMPORARY PROTECTIONISM IN LATIN AMERICA? AN
OVERVIEW*
Alejandro D. Jacobo** and Ileana R. Jalile§
ABSTRACT
This paper updates the stock of discriminatory policy instruments in Latin American countries and
explores the determinants of tariff and non-tariff measures. Using pre and post-2008 crisis trade and
protection data, the level of tariff barriers and the Anti Dumping (AD) initiations are explained by macro
and micro variables. The study finds that Intra-industry trade (IIT) is an important source of revenue for
governments. The results indicate that the crisis did not increase protectionism in countries where
powerful exporters demand cheap imported inputs, but it did where this lobby is not powerfully enough to
overcome the need to raise public revenues. It also finds that governments are enthusiastic to favor their
exporters by reducing tariffs on inputs used by (upstream) home exporters in order to enhance their
competitive position with foreign users. Concerning non-tariff barriers determinants, all the countries in
the sample show a positive relationship between AD initiations and the tariff level. This could indicate
that tariff and non-tariff instruments as protectionist measures are both complementary. Finally, an
appreciation of the currency against the currency of their trading partners makes an AD initiation more
likely to occur in some countries of the sample. However, the crisis has not reinforced the relationship
between an appreciation/devaluation on the probability of an initiation of an AD procedure.
JEL Classification: F13
Keywords: Trade policy, tariffs, non tariff barriers, protectionism, antidumping, intra industry trade
1. Introduction
Different authors and studies have carefully documented new measures that discriminate against
foreign products activated in world trade since the 2008 global financial crisis. The 7th Global
Trade Alert Report (hereinafter GTA-7), for example, illustrate that Latin American
*
This paper is part of a larger effort made by Global Trade Alert (GTA) and the Latin American Trade
Network (LATN) for the examination of contemporary protectionism. Financial support from the Center
for Economic Policy Research is gratefully acknowledged by the authors.
**
Universidad Nacional de Córdoba and Universidad Católica Argentina.
§
Universidad Nacional de Córdoba.
governments did seek to use protectionist policy instruments to respond to the crisis and that
unilateral discriminatory measures mushroomed after the outbreak of the recession.1
To cope with the global crisis, major economies implemented a trading scheme and subsidies,
cheap access to credit and other tax deductions and exemptions for exporters helped the
recovery in world trade (Evenett, 2010; Tussie, 2012). However, due to different reasons —
mainly due to the lack of resources— others economies were unable to generate these stimulus
packages and they use tariff and non-tariff measures as protectionist instruments.
As to Latin America Countries (LAC), while some of them used tariff measures to protect one
or more sectors affected by the global crisis, other economies started to assemble a trade policy
pattern notably characterized by major movements in non-tariffs barriers as well (Dalle and
Lavopa, 2010). The emergency tools used by LAC are such important in manner and magnitude
that deserves a study monitoring what is going on with Latin America protectionism.
The aim of this study is twofold. First, to provide a brief and updated description of the stock of
discriminatory policy instruments in LAC. This will help us to identify the protectionist (tariff
and non-tariff) measures imposed by the region on different trading partners. Second, to develop
an empirical model that explores the determinants of these policy instruments. The use of pre
and post crisis trade and protection data will allow us to search possible variations in the
determinants of trade policy responses.
The paper proceeds as follows. Section two provides an overview on recent trade barriers
involving LAC. Section three develops a model in which the presence of discriminatory policies
in a particular sector from a specific country depends on macro and microeconomic
determinants. Section four presents the estimation and the results. Section five concludes.
2. Protectionist policy instruments in LAC: An overview
1
Managed Exports and the Recovery of World Trade: The 7th Global Trade Alert Report (2010).
This section provides a regional perspective on trade barriers involving Latin American
countries according to the World Integrate Trade Solution (WITS), Temporary Trade Barriers
(TTB) and Global Trade Alert (GTA) databases.2 It reviews the policy instruments and
identifies those countries using more as well as those suffering most protectionist policies. We
update the last review of protectionist measures enacted for Latin America provided by the
GTA-7.3
WITS database retrieves information on trade and tariffs compiled by international
organizations: The United Nation Commodity Trade (UN-COMTRADE) database contains
information on exports and imports by detailed commodity and partner country; the United
Nations Trade Analysis Information System (UN-TRAINS) database shows information on
imports, tariffs, para-tariffs and non-tariff measures; and the World Trade Organization’s
(WTO) Integrated Data Base (IDB) and WTO’s Consolidated Tariff Schedule Data Base (CTS)
which contains information on imports by commodity and partner country, Most Favored
Nation (MFN) Applied Rate, Preferential Tariffs (when available), Bound Tariffs (BND) as well
as other indicators.
The Temporary Trade Barriers Database (TTBD) website hosts newly collected and freely
available detailed data on more than thirty different national governments’ use of policies such
as AD, global safeguards (SG), China-specific transitional safeguard (CSG) measures, and
countervailing duties (CVD).
GTA database includes trade barriers (e. gr. tariffs and AD measures) as well as others measures
related to trade policy (e. gr. foreign investment related measures). GTA classifies these
measures into three categories: (1) green measures (which involve liberalization on a nondiscriminatory basis or an increase in transparency), (2) amber measures (that involve
discrimination against foreign commercial interests and include already applied measures and
measures announced or under consideration which, if implemented, would discriminate those
2
These databases are available through the following links: http://wits.worldbank.org/wits/,
http://econ.worldbank.org/ttbd/ and http://www.globaltradealert.org.
3
We partially follow Rozemberg and Gayá (2010) for comparison purposes.
interests), and (3) red measures (that are already in force and certainly discriminate against
foreign commercial interests).
About trade barriers, we present information on protectionist measures imposed by 10 LAC
using data available from GTA database. These countries are: Argentina (ARG), Bolivia (BOL),
Brazil (BRA), Chile (CHL), Colombia (COL), Ecuador (ECU), Mexico (MEX), Paraguay
(PRY), Peru (PER) and Venezuela (VEN). Although GTA database considers 27 Latin America
economies, not all the countries have started to be monitored at the same time and/or have
implemented measures, and therefore were not included in our analysis.4 Thus, to make a
comparison with the last GTA-7 and to avoid distortions by introducing countries that
previously were not considered, we finally analyze only 10 economies. The stock of trade
restriction comprises barriers from November 2008 (when GTA database started its job and
began to list measures) through 03/02/2012 (when the GTA database was downloaded for this
study).
Figure 1 distinguishes green, amber or red measures implemented by each country. As shown in
the figure, Argentina leads the ranking with the application of red and amber measures (127 in
total), followed by Brazil (63) and Mexico (13). Argentina also exhibits the highest ratio in the
relationship between protectionist (red and amber) to green measures (12.7). According to the
last GTA-7, in Argentina the use of red and amber measures increased by 84%.
4
This situation could generate a bias against those first monitored. For example, Uruguay applied
measures before the GTA-7 was published. However, these measures were not considered in this report
which means that Uruguay was not monitored.
Figure 2
Number of Measures Implemented by LAC, according GTA classification
130
120
116
110
100
90
80
70
60
60
54
50
40
30
20
11 10
10
9
12
11
1
8
5 4
4 3 5
2 2
6
9
2 2
5
0 1
3
0 0
1 0 1
1 0 1
CRI
CHL
0
ARG
BRA
MEX
VEN
PER
Source: Authors' own estimation based on GTA database.
COL
ECU
PRY
BOL
Red
Amber
Green
Figure 2 shows the stock of red and amber measures implemented in LAC by type of measure.
Trade defense measures (AD, countervailing duties (CVD) and safeguard) represent 30% of all
red and amber measures, followed by non-tariff measures (28%) and tariff measures (15%).
This situation exhibits some differences from the previous GTA-7: Non-tariff measures notably
increased from 8.9 to 28% while tariff measures augmented from 12.9% to 15%.5
“Other Measures” includes, inter alia, the following ones: Consumption subsidy, Import subsidy,
Competitive devaluation, Sanitary and Phytosantiary Measure, State-controlled company, Technical
Barrier to Trade, Local content requirement, Trade finance and Export subsidy.
5
Figure 4
Red and Amber Measures Implemented in LAC,
by Type of Measure
Trade defence measure
(AD, CVD, safeguard)
31%
Other
14%
Export taxes or
restriction
4%
Bail out / state aid
measure
4%
Investment measure
4%
Non tariff barrier (not
otherwise specified)
28%
Tariff measure
15%
Source: Authors' own estimation based on GTA database.
Each measure individually imposed by one of the LAC generally affects various jurisdictions
and sectors. Table 1 presents the ranking of jurisdictions affected by Red measures imposed by
the LAC. China is more affected by these measures than the other countries (6.5% of total),
followed by India, Korea and Indonesia (2.7% of total in each case). This situation shows some
differences from the last GTA-7, since the United States, Germany, Brazil, France and Spain are
not affected as before by red measures implemented by LAC.6
6
It is worth to mention that Table 1 assumes that when a measure referenced in GTA database contains
more than a type of measure affecting different countries, then each of these measures will
simultaneously affect these countries.
Table 1
Red Measures Imposed by Latin
America, by Affected Country
Red
Affected Jurisdictions
Measures
China
195
India
84
Indonesia
80
Republic of Korea
79
Thailand
78
Malaysia
77
USA
75
Singapore
72
Hong Kong
69
Viet Nam
69
Philippines
65
Germany
63
Pakistan
62
Brazil
57
Others
1863
Total
2988
Source: Authors' own estimates based
on GTA database.
All selected countries in the region imposed red measures to other LAC. In Table 2 we can
observe the number of red measures implemented among the countries under analysis.
Table 2
Number of Red Measures Imposed among Selected LAC
ARG
BOL
2
BRA
9
2
Implementing Jurisdiction
CHL COL ECU MEX
1
1
1
2
Affected Jurisdiction
ARG
BOL
14
BRA
43
2
1
CHL
35
5
1
COL
30
1
ECU
27
1
1
MEX
21
2
8
1
PRY
30
4
PER
20
1
4
1
VEN
17
1
1
1
Source: Authors' own estimates based on GTA database.
2
1
1
1
1
1
1
1
2
1
PRY
4
4
5
2
2
PER
1
VEN
2
1
2
2
3
2
4
1
2
Thus, Latin America does not provide a better treatment to regional partners, but quite the
opposite. Argentina is, by far, the most active user of measures that discriminate against
commercial interest of other LAC. Those countries involved in economic integration processes
are also affected by measures that restrict intra-zone trade (e. gr. MERCOSUR).
With respect to jurisdictions that implement protectionist measures, the Russian Federation and
Argentina leads the world ranking with 130 and 116 initiatives respectively, followed by the
UK, Germany, China, India and Brazil, as shown in Table 3.
Table 3
Red Measures Imposed, by jurisdiction
Implementing
Red Measures
Jurisdictions
Implemented
Russian Federation
130
Argentina
116
UK
64
Germany
60
China
59
India
57
Brazil
54
France
54
Spain
50
Italy
49
Austria
47
Hungary
47
Greece
46
Ireland
45
Netherlands
45
Others
1206
Total
2129
Source: Authors' own estimates based on GTA
database.
Table 4 presents information on the evolution of tariff barriers in LAC in recent years. As
shown, the countries have not passively used this kind of trade policy measure. In most of the
countries we observe an upward tendency in the level of the Applied Tariff (t) after the global
financial crisis. As we also observe, there is a great policy space for the countries to further
increase their tariff and remain within the bounds of GATT-WTO commitments indicated by
the Bound Tariff (tbnd).7
Table 4
Evolution of Tariff Barriers in LAC
Countries
Argentina
Bolivia
Brazil
Chile
Colombia
Ecuador
Mexico
Paraguay
Years
Tariff Measures
2002
2003
2004
2005
2006
2007
Applied Tariff (t)
14.78
14.73
11.85
10.60
10.74
10.80
MFN Tariff (tMFN)
14.78
14.73
13.40
12.35
12.37
12.40
Bound Tariff (tBND)
31.64
31.72
31.86
31.95
31.93
31.96
7.76
7.19
6.48
9.27
8.63
2009
2010
9.83
9.76
11.41
11.48
11.59
13.43
31.97
31.96
31.85
6.20
6.18
7.49
9.58
8.53
8.46
8.46
10.23
11.78
Applied Tariff (t)
9.87
MFN Tariff (tMFN)
9.87
Bound Tariff (tBND)
39.97
39.97
39.98
39.98
39.98
39.98
39.97
39.97
Applied Tariff (t)
14.56
14.37
13.28
12.39
12.20
12.23
13.10
13.34
13.37
MFN Tariff (tMFN)
14.56
14.37
14.28
13.28
13.24
13.25
14.38
14.62
14.67
Bound Tariff (tBND)
30.67
30.75
30.71
30.80
30.87
30.90
30.96
30.94
30.94
4.89
4.86
2.23
1.96
1.39
5.97
4.86
6.00
6.00
5.99
5.99
5.99
5.99
5.99
25.06
25.05
25.05
25.06
25.06
25.06
25.06
11.44
11.87
11.33
10.81
10.73
10.71
11.20
12.40
12.82
12.77
12.81
12.79
12.73
12.75
38.15
n.a.
Applied Tariff (t)
6.99
MFN Tariff (tMFN)
6.99
Bound Tariff (tBND)
25.04
Applied Tariff (t)
12.45
MFN Tariff (tMFN)
12.45
Bound Tariff (tBND)
38.57
38.29
38.14
38.05
38.06
38.06
38.15
Applied Tariff (t)
12.02
11.60
11.82
9.82
10.03
9.73
8.36
9.37
MFN Tariff (tMFN)
12.02
12.20
12.58
12.44
12.51
11.70
10.16
10.16
Bound Tariff (tBND)
21.84
21.98
22.65
22.49
22.57
22.17
21.64
21.64
Applied Tariff (t)
15.54
18.34
10.21
9.20
8.01
7.35
MFN Tariff (tMFN)
18.38
18.34
17.50
14.72
14.48
Bound Tariff (tBND)
34.98
34.98
34.97
34.94
34.95
Applied Tariff (t)
13.41
13.42
9.20
8.38
7.19
8.01
8.33
7.98
7.97
MFN Tariff (tMFN)
13.41
13.42
12.50
11.50
10.28
11.06
11.04
11.04
11.07
Bound Tariff (tBND)
32.67
32.69
32.45
32.67
32.70
32.75
32.63
32.61
32.61
9.65
9.19
8.57
8.54
3.86
3.73
4.78
10.18
9.58
9.60
9.64
4.96
4.74
5.17
30.08
30.07
30.06
30.07
30.08
30.08
30.06
n.a.
n.a.
n.a.
Applied Tariff (t)
Peru
2008
MFN Tariff (tMFN)
n.a.
n.a.
Bound Tariff (tBND)
Bound Tariff (tBND)
Applied Tariff (t)
MFN Tariff (tMFN)
13.54
8.26
34.98
34.98
34.99
9.86
9.64
9.52
9.50
9.61
9.59
12.53
12.33
12.32
12.35
12.32
12.32
31.25
31.29
31.52
31.52
31.54
31.58
31.58
31.58
13.54
12.22
12.69
12.22
12.17
11.84
12.07
13.02
13.04
13.44
14.25
14.31
14.36
14.55
14.48
Bound Tariff (tBND)
34.72
34.56
Source: Authors' own estimates based on data from WITS database.
34.42
34.43
34.43
34.45
34.60
34.48
7
14.33
5.28
11.17
14.33
Venezuela
14.33
MFN Tariff (tMFN)
6.39
12.69
11.15
Uruguay
Applied Tariff (t)
n.a.
n.a.
n.a.
Bound Tariffs (tbnd) are specific commitments made by individual WTO member governments and
represent the maximum Most-Favored Nation tariff level that a country may levy for a commodity. The
Most-Favored Nation tariffs (tmfn) are what countries promise to impose on imports from other members
of the WTO, unless the country is part of a preferential trade agreement. WITS database uses the concept
of Effectively Applied Tariff which is defined as the lowest available tariff and will be used as the
Applied Tariff (t) if a preferential tariff exists. Otherwise, the MFN applied tariff will be used.
According to GTA database, trade defense measures are the most used non-tariff trade policy
instruments in LAC countries. Within these instruments, TTB database ranks AD at the top of
the list. Table 5 presents information on 6-digit HS products with AD initiations per year for
some LAC countries.8
Table 5
Antidumping Initiations in Selected LAC
Years
Argentina
Brazil
Colombia
Mexico
Peru
2002
57
12
n.a.
17
48
2003
1
4
n.a.
17
48
2004
30
8
2
6
46
2005
16
6
2
18
5
2006
15
15
66
7
7
2007
11
19
2
4
1
2008
42
53
9
1
7
2009
89
17
3
3
18
2010
41
46
4
3
n.a.
Source: Authors' own estimates based on TTB database using iInformation on 6 Digits HS
products.
3. On the political-economy variables of trade policy
There is a vast theoretical and empirical literature analyzing the determinants of trade protection
in the economy. In recent decades, however, this literature has moved towards the
“endogenous” trade policy determination and constitutes the core of the literature on the
political economy of trade policy (Gawande and Krishna, 2006).9
In line with this literature, the aim of this section is to analyze the determinants of trade policy
in LAC and to verify if countries have changed their behaviour as a consequence of the crisis.
8
Due to availability of the data, the number of countries selected in Table 5 is reduced to five LAC.
See Gawande and Krishna (2006) for a summary of the empirical literature on the determinants of trade
protection.
9
Considering the evidence related to trade measures used by LAC we have already summarized
in Section 2, we analyze two different policy instruments: Tariff Barriers and AD. We explore
the determinants of both protectionist measures.
For this purpose, we use 6-digit HS tariff, non-tariff and trade data provided by WITS and TTB
databases to make inferences on the influence of micro and macroeconomic variables in
determining the source of protectionism. The level of disaggregated data will allow us to take
into account sectoral and partner countries differences that influence on trade protectionism.
This strategy is not a novel one. Among other authors Olarreaga and Vaillant (2011), Gawande
et al. (2011) and Bown and Tobar (2011) have already analyzed the determinants of trade
policies using disaggregated data as we do.
However, in comparison with the existing literature, we will focus specifically on Latin
America Region and we will try to see if there is a change in the behaviour of LAC after the
crisis with newly available data.
Data and Methodology
Our empirical approach have analyzed the following LAC: Argentina (ARG), Bolivia (BOL),
Brazil (BRA), Chile (CHL), Ecuador (ECU), Mexico (MEX), Paraguay (PRY), Peru (PER),
Venezuela (VEN) and Uruguay (URY) over the period: 2002-2010.10
As in Gawande et al. (2011), we explore the determinants of trade policy responses by
estimating two equations. First, the Tariff Barrier Equation where the dependent variable is the
effectively applied bilateral tariff. Second, the AD Equation where the dependent variable is
AD initiation. In both equations we explain the presence and level of trade barriers in a 6-digit
10
The last available year on data on trade and tariff and not tariff barriers provided by the WITS and TTB
databases is 2010. Since these databases provide information on Uruguay, this country is now formally
introduced in the analysis.
HS product imported from a particular country in a given year. This disaggregation is required
because tariff and non-tariff barriers are determined at the product level.
With regard to the Tariff Barrier Equation, the determinants of tariff barriers have been
extensively discussed in the literature.11 As in Gawande et al. (2011) and Olarreaga and Vaillant
(2011), we include in our analysis macro and microeconomic determinants of the level of tariff
barriers. As we mentioned above, as dependent variable in this equation we will use the
Effectively Applied Tariff, which is defined as the lowest available tariff. If a preferential tariff
exists, it will be used as the effectively applied tariff. Otherwise, the MFN applied tariff will be
used. Our proposed specification for this equation is as follows:
t g,p,t= α1(tbndprfg,p,t)+ α2(iitg,p,t-1)+ α3(VSg) + α4(VS1g)+ +αg + αp+ αt + εg,p,t
(1)
where tg,p,t represents the level of the Effectively Applied Tariff on good g, imported from
partner p at time t; tbndprg,p,t is a composite measure of tbnd and tprf (tbnd is the bound rate
commitment at the WTO and tprf is the preferential tariff rate) and represents the value of this
variable on good g, imported from partner p at time t; iitg,p,t-1 is a measure of intra-industry
trade on good g, imported from partner p at time t-1; VSg and VS1g are measures of vertical
specialization on product g; αg is an HS six-digit fixed effect; αp is a partner fixed effect and αt
is a time fixed effect.
The influence of Institutions is measured by the coefficient associated to the bound rate tbndprf
(1). While applied rates are determined by each country, they are bounded above by their
bound rate commitment at the WTO. The latter rates are determined in multilateral negotiations
and they are exogenous in our model. Countries do not make commitments in terms of “applied
protection” but instead in terms of the “ceiling” above which they commit not to raise their
applied duty. However, if a country decide to sign a PTA the new effective bound on its tariff
rate would be the preferential tariff rate (tprf ). Following Gawande et al. (2010), we define a
composite measure where tbndprf = tprf whenever tprf is applicable, or tbndprf = tbnd
11
See Gawande and Krishna (2008) for a review of this literature.
otherwise. The coefficient is expected to be positive and small if the structure of GATT/WTO
incentives keep applied tariff in check.
The coefficient 2 captures the impact of Intra-Industry Trade (ITT) on the tariff barrier level.
The construction of an intra-industry trade index at product level would allow us to measure the
trade in similar but differentiated products. Currently, an important share of trade is ITT. WITS
database allow us to construct the following ITT measure at the 6-digit HS level: IIT = 1 −
|Imports - Exports| / (Imports + Exports). Krugman (1981) demonstrate the gains from trade in
the presence of product varieties. According to this we would expect that higher IIT would
imply less protectionist pressures. However, more sophisticated models indicate that the
presence of IIT does not necessarily imply a negative correlation between IIT and tariff. For
example, in models featuring domestic and foreign duopolies, Brander and Spencer (1984) show
that rents could be shifted from foreign to home firms through a strategic tariff policy. Then,
even though the optimal action for both countries is to reduce tariffs, the unilateral incentive is
for governments to use tariffs to play zero-sum games. If tariffs are strategic, then, a positive
correlation between IIT and rents implies that tariffs should be positively associated with ITT.
Another example could be Jørgensen and Schröder (2006) who show that an optimal tariff
exists, below which welfare is reduced because there are too few domestic varieties and beyond
which there are too many inefficiently-produced, costly domestic varieties. On the other hand, if
tariffs in the countries are strategic as a source of government revenue one may expect a
positive correlation between ITT and the dependent variable.12
The literature also points out that vertical specialization could have an impact on the tariff level.
Vertical Specialization could be defined as production arrangements in which firms make final
goods via multiple stages located in multiple countries, as an important aspect of overall input
trade. We introduce two measures of vertical specialization: VS and VS1.13 VS is the share of
imports in a sector that is used directly and indirectly in the country’s own exports (i.e.
embedded as intermediate inputs). VS1 is the proportion of a sector’s exports used as
12
Gawande et al., 2011.
Theories offer several explanations for vertical production networks, including cross-country and/or
cross-industry differences in trade costs, factor prices, and the technological separability of production.
While there is some evidence to support these theories, little work goes beyond documenting broad facts
to provide a theoretically and micro-level empirical analysis of the importance of these explanations
(Hanson et al., 2003).
13
intermediates by exporters in other countries.14 These two variables have been constructed in
Daudin et al. (2010) using trade and input-output data from the Global Trade Analysis Project
(GTAP) database. To construct the variables VS and VS1, Daudin et al. (2010) computes valueadded trade for 66 regions and 55 sectors in 1997, 2001 and 2004.15 The construction of VS and
VS1 for different years depends on the input-output matrices for each reporting country. Since
LAC do not systematically update input- output data, VS and VS1 were generated using the last
available input-output matrices. It implies that VS and VS1 at each 6-digit HS code will be
constant along the period. Input-output tables indicate that the same sector is the larger user of
imports by that sector. While a positive coefficient in VS may indicate that the exporters are not
powerful enough to overcome the need to raise revenues, a negative coefficient on VS1 can be
interpreted as a global supply chain against protectionism.
Among other macroeconomic determinants of policy trade responses that may vary across years
are the level of activity, unemployment and institutional variables. These determinants have
been taken into account in Olarrega and Vaillant (2011) using year fixed effects. We follow the
same strategy. Besides, any particular determinant of protection towards a partner that is timeinvariant (as for example distance, institutional similarity, or similarities in the comparative
advantage) is controlled using partner fixed effects.
The literature has also points out other microeconomic determinants of trade policy instruments
such as the concentration of the sectors, output or the extent to which workers are unionised.
Unfortunately, we do not have data about these variables at the disaggregated 6-digit HS, so we
assume that these variables remain constant during the period and we control them using
product fixed effects.
14
VS1 measures the intensity of two sources of anti-protectionist pressure. High tariffs on imports in a
sector undermine the competitiveness of the sector’s exports that intensively use those imports. Inputoutput tables indicate that the same sector is the larger user of imports by that sector. As consequence, the
first source of anti-protectionism is exporters of that sector who will lobby against tariffs that raise their
input costs and make them uncompetitive. The second source of anti-protectionism is foreign lobbying by
exporters in other countries who depended on the source country for supplying them with intermediate
inputs. Low or zero tariffs in the source country are desirable for keeping their input costs down
(Gawande et al., 2011).
15
We thank Guillaume Daudin for generously providing us the data. We use concordance tables for
matching 55 sectors from the GTAP to the 6-digit HS codification used in our empirical approach.
With regard to the AD Equation, we should note that the determinants of Non-Tariff Barriers
have also been extensively studied in the literature.16 Our proposed equation specification is as
follows:
ADg,p,t=α1(uvg,p,t-1)+α2(mg,p,t-1)+α3(tg,p,t)+α4(VSg)+α5(VS1g)+α6(RBERp,t)+αg+αp+αt+εg,,p,t (2)
where ADg,p,t is a dummy variable indicating the presence of an AD on good g, imported from
partner p at time t; uvg,p,t is the unit value of good g, imported from partner p at time t; mg,p,t is
the value of imports of good g, imported from partner p at time t; tg,p,t is the Effectively Applied
Tariff on good g at time t; VSg and VS1g are measures of vertical specialization on product g;
RBERp,t is the real exchange rate with respect to partner’s p currency at time t; αg is an HS sixdigit fixed effect; αp is a partner fixed effect, and αt is a time fixed effect.
Among the most important macroeconomic determinants that vary among partners, we include
Real Bilateral Exchange Rates (RBER) and MFN (or effectively applied tariff) rates. The first
measures the impact of bilateral competitiveness of each country vis-à-vis each of its trading
partners. On the other hand, the coefficient associated to tariff rate indicates the extent to which
AD and tariff rates act as complementary or substitute trade policy. We have special interest in
the RBER coefficient. Following Olarreaga and Vaillant (2011) the sign of this coefficient is
ambiguous. Feinberg (1989) suggests that it should be positive as the depreciation of the local
currency increase the probability of finding dumping, while Knetter and Prusa (2003) suggest
that the coefficient should be negative because a depreciation of the local currency difficults
injury on the economy.
As microeconomic determinants that affect trade policy responses we consider the Price and the
Value of Imports which vary across partners, years and sectors. The coefficient associated to
these variables would indicate the casual effect from the price and size of imports on the
determination of the presence of an AD. To sum up, we postulate that the propensity to initiate
16
Se Aggarwal (2004) for brief review of the literature. See also Knetter and Prusa (2003), Prusa and
Skeath (2002), and Sabry (2000).
an AD procedure would increase with larger imports (2 >0), and that higher unit prices are less
likely to lead to finding dumping or injury from subsidies (1 < 0).
We have also included in our econometric specification the vertical specialization (VS and VS1)
measures. As we previously mentioned, we expect that an increase in vertical specialization
reduce the protectionism in the reporting country whether local governments favour global
supply chains. This means that AD initiations should be inversely related with vertical
specialization measures. On the other hand, a positive coefficient on VS could be associated
with the fact that exporters in the reporting countries are not powerful enough for fight against
protectionism, while a positive coefficient on VS1 could indicate that exporters of partner
countries are not lobbying against protectionism on local governments.
We follow the same approach presented in the tariff equation and we use year fixed effects to
control for domestic macroeconomic determinants of policy trade responses that vary across
years (such as the level of activity, unemployment and institutional variables). Any particular
determinant of protection towards a partner that is time-invariant (as for example distance,
institutional similarity, or similarities in the comparative advantage) is controlled using partner
fixed effects. Again, in order to control for microeconomic determinants of trade policy
instruments (such as the concentration of the sector, output or the extent to which workers are
unionised), we assume these variables keep constant in the period and consequently we control
for them using product fixed effects.
4. Estimation and Results
Estimates from a baseline partner and year fixed effects model of applied bilateral tariffs that
represents the influence of the variables considered are presented in Table 6 (see Appendix).
In the model, the year fixed effect controls for any domestic macroeconomic change such as the
level of economic activity or unemployment in the reporter countries. The partner fixed effect
controls for any particular determinant of protection towards that partner that is time-invariant,
as for example distance, institutional similarity, or similarities in the comparative advantage.
The model performs well.
The coefficient of 0.25 on tbndprf for Argentina indicates that if bound rate (adjusted for PTA
agreements) increases 1 point, Argentina’s bilateral applied tariff increases 0.25 points. Peru
exhibits lowest coefficient (0.14) while Ecuador shows the highest (0.61). Despite the
availability of tariff policy space, one reason for the small coefficients in Argentina is that since
the majority of trade is carried on with MERCOSUR partners the competition with others is not
probably an issue. Another reason is that the agreement has accelerated the decline of inefficient
industries in Argentina, so the country does not face protectionist demands from those sectors.
However, the small coefficients in the line of the table are the rule and they do not necessarily
reveals the feature of belonging to a PTA or the existence of completely efficient industries in
all the cases. Rather, the low coefficients may indicate that WTO incentives kept applied tariffs
in check.17
The coefficient of 1.64 on IIT for Argentina indicates that a higher intra-industry trade is
associated with an increase in Argentinean tariffs. This is quite the opposite of the prediction
from intra trade models that emphasize the additional welfare gains from expanding the
varieties.18 Besides, the positive sign on IIT could indicate the dependence of Argentina on
tariffs as a source of revenue. Since much of the Argentinean trade is with PTA’s partners more
revenues means higher tariffs on non PTA partners, even if trade with them is two-way trade in
similar goods. The selected countries have positive coefficients revealing revenues
requirements. For Mexico, however, IIT has a negative coefficient that may indicate the
additional welfare from expanding the variety in differentiated products. In this case, the gains
from trade appear to overwhelm the need to use tariffs for revenues motives.19
While measures of VS does not dissuade the use of tariff in Argentina, Bolivia, Brazil, Chile,
Colombia, Paraguay and Peru, it does deter their use in Ecuador, Mexico, Venezuela and
Uruguay. Recall that the VS measure of vertical specialization is the share of imports in a sector
17
Gawande et al. (2011).
The results presented in Jørgensen and Schröder (2006) and Brander and Spencer (1984) could also
explain the positive correlation we have found.
19
As in Gawande et al. (2011), we are unable to discriminate among different theories.
18
that is used directly and indirectly in the country’s own exports (i.e. embedded as intermediate
inputs). So, while the exporters of countries included in the first club of nations are not powerful
enough to overcome the need to raise revenues, the importance of exporters in the other club of
countries results obvious.
The second vertical specialization measure (VS1) shows a negative coefficient across the table
(with the exception of Uruguay). This could be interpreted as a global supply chain against
protectionism. Recall that this measure is the proportion of a sector’s exports used as
intermediates by exporters in other countries. Thus, the coefficients suggest that the
governments are enthusiastic to advance their exporters’ interests by reducing tariffs on the
inputs used by (upstream) home exporters in order to enhance their competitive position with
foreign users. The negative coefficients may also be taken as evidence for the idea that exporters
in foreign countries may (politically) influence home tariffs since their competiveness depends
on the supply of cheap inputs from home producers.
Following Gawande et al. (2011), each variable is interacted with a post-crisis dummy to find
out whether the relationships observed in Table 6 remained unaltered through the crisis or were
fundamentally changed by it. The results are presented in Table 7 (see Appendix).
Consider the coefficient on the interaction term tbndprfxI2009. The positive and statistically
significant coefficient in most of the cases (with the exception of Mexico, Peru, Venezuela y
Uruguay) indicate that the majority of the countries did not lower their tariffs and feel the
pressure in the post-crisis period to raise them. In the case of Argentina, for example, the
coefficient on tbndprf increased by 0.026 in 2009 over a pre-crisis coefficient of 0.24, signaling
a readiness to increase tariffs up to the bound levels. For other countries, the change on
coefficients is still small, considering the magnitude of the crisis.
The coefficient on IITxI2009 for Argentina is negative. However, taken into account the overall
impact of IIT post- crisis on the level of the tariff (1.8771-1.2729), we may conclude that its
public finances depend on tariff revenues. The same conclusion may be applied for Brazil,
Colombia, Paraguay, Peru and Uruguay. In the case of Bolivia, Chile, Ecuador and Venezuela
the positive impact of IIT on the tariff level has been reinforced after the crisis. This situation
probably means that the need of revenue from foreign trade is even higher after the 2008
economic downturn. Finally, we observe that the overall impact of IIT on Mexico has changed
after the crisis. In fact, while before the crisis the theory emphasizing the additional welfare
gains from expanding the varieties was verified, after the crisis the theory of tariff revenue
dependency could be applied.
Consider the vertical specialization measures: VSxI2009 and VS1xI2009. The latter term shows
large negative coefficient for Ecuador, Mexico, Brazil and Bolivia (-17.82, -12.47, -15.87 and 13.82 in each case). In the post-crisis period, the export sectors in Ecuador, Mexico, Bolivia and
Brazil’s partner countries seem to have a strong influence on lowering their tariffs, specifically
on products that the partners import from those countries for intermediate use.20 This source of
anti-protectionism is also evident in Peru, Paraguay, Venezuela, Colombia and, to a lesser
extent in Argentina. In the case of Chile and Uruguay, VS is the main source of antiprotectionism after the crisis where domestic exporters are the prime movers in demanding
lower protection on imported goods they intensively use. Both sources of anti-protectionism are
used in Ecuador, Mexico and Venezuela.
Other studies include in the analysis other regressors in their empirical estimation that are not
introduced in our analysis. For example, some of them include variables that measure the
influence of intermediate use as well as other political economy variables that intent to
determine the propensity of protectionism. However, if we assume that these variables are
surely time invariant across products during the period analyzed, we can control for their effect
on the level of the tariff using product fixed effects. The good fixed effect would also controls
for any other time invariant 6-digit HS determinant of protection, as the political strength of
producers. When we control for these fixed effect some of the variables that we have considered
in our econometric estimation are dropped because they were constructed using data that remain
constant across the period (e. gr. VS and VS1).
20
A lower cost makes partners more competitive and, in turn, this situation increase the purchases from
Ecuatorian, Mexican, Bolivian and Brazilian suppliers and expand their exports.
Table 8 (in the Appendix) presents the result of the estimation including good fixed effects. As
we can observe, the coefficients associated with the institutional variable tbndprf do not
generally present considerable changes compared with the previous specification. On the other
hand, we can observe some changes on the overall impact of intra industry trade on
protectionism after the crisis. While a positive impact of ITT in the level of the tariff was the
rule in the previous specification (indicating that weak tax system in these countries relies
almost at all on revenue tariff), when considering product fixed effect this relation has changed
for some countries. On this new specification there is evidence that in Argentina, Colombia,
Paraguay and Uruguay the gains from trade in similar but differentiated products appear to
overwhelm the need to use tariff for revenues motives.
We look at the incidence of AD initiations using conditional logit models with partner, product
and year fixed effects. Previous studies of trade defense measures have restricted their samples
only to sectors in which these kind of measures have taken place. In our study, we compare 6digit HS commodities on which AD investigations occurred with the overwhelming number of
cases in which these investigations do not exist. Due to the lack of data, we only have results for
Argentina, Brazil, Mexico, Colombia and Peru.
Table 9 (in the Appendix) presents the results of estimating the AD equation. We observed that
all the countries in our sample show a positive relationship between AD initiations and the tariff
level. This could indicate that, as a protectionist measures, tariff and non-tariff measures are
both complementary. This relationship is reinforced after the crisis only in Argentina. This
situation may indicate that this country may have stepped up AD investigations after the crisis
as a complement to tariff barriers.
The coefficient on BRER is negative and statistically significant for Argentina, Brazil and
Colombia. This indicates that an appreciation of their currency against the currency of their
trading partners makes a AD initiation more likely to occur. When this variable is evaluated in
the post crisis period, we find that the coefficient has reduced in Argentina and Colombia and
remains the same in Brazil. Consequently, the crisis has not reinforced the relationship between
an appreciation/devaluation on the probability of an initiation of an AD procedure.
The coefficients on VS are positive for all countries except for Peru. This exception indicates
that only in Peru the exporters are powerful enough to lobby against the initiation of an AD on
imports of goods used directly and indirectly by Peruvian exporters. For the post crisis years,
Argentina is the only country where we can observe a change in the relationship between VS
and AD. Specifically, it could indicate that Argentinean exporters are now more powerful in
fighting against AD initiations over their imports.
The coefficient on VS1 is negative in the case Colombia, but positive for Peru, México and
Brazil. A negative sign on this variable indicates that government favors global supply chains
(Colombia) while a positive one could indicate that foreign exporters do not have political
influence on the local economic policy (Brazil, Mexico and Peru). Besides, we can observe that
the negative impact of VS1 in the propensity of initiating an AD is not reinforced after 2008 in
Colombia.
While in Table 9 we observe that the propensity to initiate an AD is positively related with the
level of the tariff effectively applied on a particular product, it is important to take into account
that some problems of endogeneity may emerge.21 Our strategy in this paper is to control for 6digit HS product fixed effect. We suppose that the endogeneity problem could arise due to a
non-observed variable that determines both AD initiations and the level of the tariff. Such nonobserved variable could be the political strength of domestic producers of each 6-digit HS
product. Therefore, controlling for product, year and partner fixed effects is our last estimation
and we present the results of this specification in Table 10 (in the Appendix).
For Argentina we observe that the most important determinant of the probability of an AD
initiations is the RBER. It means that the propensity to initiate a trade defense measure in
Argentina strongly depends on the level of appreciation of its currency against its partner’s
countries and that for years after 2008 this relationship has been reinforced. While prior to 2008
the relationship between the level of the tariff and the probability of initiate an AD procedure
was not statistically different for zero, after crisis we can observe a complementarity between
both measures of protectionism. With regard to Brazil, the Table 10 indicates that the propensity
21
Aggarwal (2004) introduces a lagged tariff as independent variable, while others, such as Trefler (1993)
uses a simultaneous equation model.
to initiate an AD depends on the level of the tariff and the RBER. It indicates that tariff and non
tariff barriers are complementary and that the propensity to initiate an AD in Brazil depends on
the level of appreciation of its currency against its partner’s countries. Besides, the impact of
these variables on the probability of initiating an AD remains the same after the crisis.
In Mexico AD initiations and tariff appear to be complementary and the lower prices for
product imported from partners are more likely to lead to AD initiation. Both effects are not
reinforced after the breakdown of the crisis. For Colombia, the variables included in our
analysis do not explain the presence of an AD. After the crisis, however, the level of the tariff
and unit values of imports could explain the initiation of this non tariff barrier. According to our
results, the negative coefficient related to the tariff indicates that sectors with high tariffs are
less likely to get protected through AD initiations. Surprisingly, the coefficient associated with
unit values is positive and statistically significant. This could be explained by the fact that
products subject to AD tend to be timer invariant and that AD are more likely to be applied on
goods for which domestic demand is growing and this implies higher prices from these
partners.22 Finally, in the case of Peru, the unit value of imports is the only variable that could
explain the presence of an AD. According to our estimation, the lower prices for product
imported from partners are more likely to lead to AD being imposed on those imports. This
effect is not reinforced after the crisis.
5. Concluding comments
The paper provides a brief and updated description of the stock of discriminatory policy
instruments in LAC and it also explores the determinants of the protectionist measures imposed
by these countries.
There is an increase in the use of discriminatory measures by LAC and two of them (Argentina
and Brazil) lead the ranking of countries implementing discriminatory measures worldwide.
22
Olarreaga and Vaillant (2011).
Moreover, some of these measures mutually affect LAC. Trade defense measures (AD, CVD
and safeguard) represent 30% of all red and amber measures, followed by non-tariff measures
(28%) and tariff measures (15%).
With regard to the determinants of the measures imposed by LAC, we have found that WTO
and PTA (Preferential Trade Agreements) incentives appear to have kept applied tariff in
control. In spite of all LAC has plenty of space to raise tariff, they did not strongly use this
policy space for greater protectionism. On the other hand, following the 2008 financial crisis
most of the countries feel the pressure to raise their tariff up to the bound levels (with the
exception of Mexico, Peru, Venezuela and Uruguay).
The study also finds that ITT is associated with an increase in tariffs in most of LAC (with the
exception of Chile and Mexico). This is quite the opposite of the prediction from IIT models
that highlight the additional welfare gains from expanding the varieties. It could indicate that
these countries strongly depend on tariff as a source of government revenue. After the crisis the
overall impact of IIT on tariff level is positive and reinforces the dependence on tariff revenues
in almost all LAC.
We have found a positive coefficient for the VS measure for Argentina, Bolivia, Brazil, Chile,
Colombia, Paraguay and Peru, indicating that exporters of these countries are not powerful
enough to overcome the need to raise revenues. Our regressions show that the crisis did not
change the relationship between the level of VS and the tariff (except for the case of Chile).
Thus, we observe some heterogeneity across LAC since exporters in Ecuador, Mexico,
Venezuela and Uruguay (and Chile after 2008) did were successful in demanding protectionism.
The negative coefficient associated with the VS1 measure of vertical specialization (the
proportion of a sector’s exports used as intermediates by exporters in other countries) for almost
all countries analyzed, in general, suggest that governments are enthusiastic to favor their
exporters by reducing tariffs on the inputs used by (upstream) home exporters in order to
enhance their competitive position with foreign users. The negative coefficient could also
support the idea that foreign exporters have influence in determining liberalizing trade policy in
LAC.
As to AD determinants, tariff and non-tariff protectionist measures are complementary. The
evidence for Argentina indicates that this country may have further increased AD investigations
after the crisis as a complement to tariff.
The coefficient on BRER is negative and significant for Argentina, Brazil and Colombia. This
indicates that an appreciation of their currency against the currency of their trading partners
makes an AD initiation more likely to occur. When this variable is evaluated in the post-crisis
period, we find that the coefficient has been reduced in Argentina and Colombia and it remains
the same in Brazil. Consequently, the crisis has not reinforced the relationship between an
appreciation/devaluation on the probability of an initiation of an AD procedure.
The coefficients on VS1 are negative in the case Colombia, but positive for Peru, Mexico and
Brazil. A negative sign on this variable indicates that government favors global supply chains
(Colombia) while a positive one could indicate that foreign exporters do not have political
influence on the local economic policy. Besides, we can observe that the negative impact of
VS1 in the propensity of initiating an AD was not reinforced after 2008 in Colombia.
References
Aggarwal, A. (2004). “Macro Economic Determinants of Antidumping: A Comparative
Analysis of Developed and Developing Countries, World Development, vol. 32 (6): 1043- 1057.
Baldwin, R. and S. Evenett (2009). The Collapse of Global Trade, Murky Protectionism, and
the Crisis: Recommendations for the G20, A VoxEU.org Publication, Center for Economic
Policy Research, London.
Bchir, M. and S. Jean and D. Laborde (2005). “Binding Overhang and Tariff-Cutting
Formulas”, CEPII Working Paper 18, October.
Bown, C. P. (2011). ‘Introduction’, in C.P. Bown (Ed.) The Great Recession and Import
Protection: The Role of Temporary Trade Barriers, London, CEPR and the World Bank.
Bown, C. P. (2010a). “Taking Stock of Antidumping, Safeguards, and Countervailing Duties,
1990-2009”, World Bank Policy Research Working Paper 5436.
Bown, C. P. (2010b). “Temporary Trade Barriers Database”, http://econ.worldbank.org/ttbd/.
Brander, J. and B. Spencer (1984). Tariff protection and imperfect competition. In Monopolistic
Competition and International Trade, H. Kierzkowski (Ed.), Oxford: Oxford University Press.
Daudin, G., C. Rifflart and D. Schweisguth (2011). “Who Produces for Whom in the World
Economy”, Canadian Journal of Economics, 44 (4): 1403-1437.
Eichengreen,
B.
and
K.
O’Rourke
(2009).
“A
Tale
of
Two
Depressions”,
http://www.voxeu.org/index.php?q=node/3421.
Eichengreen, B. and D. Irwin (2009). “The Slide to Protectionism in The Great Depression:
Who Succumbed and Why?”, NBER Working Paper No. 15142.
Evenett, S. and M. Wermelinger (2010). “A Snapshot of contemporary protectionism: How
important are the murkier forms of trade discrimination?”, Asia-Pacific Research and Training
Network on Trade Working Paper Series, No. 83, September.
Evenet, S. (2009). “What can be learned from crisis era protectionism? An initial assessment”,
CEPR Discussion Paper Series No. 7494.
Feinberg, R. (1989). “Exchange Rates and ‘Unfair Trade”, Review of Economics and Statistics,
71 (4): 704-707.
Gawande, K., B. Hoekman and Y. Cui (2011). “Determinants of Trade Policy Responses to the
2008 Financial Crisis”, World Bank Policy Research Working Paper number 5862.
Gawande, K. and P. Krishna (2008). “The Political Economy of Trade Policy: Empirical
Approaches”, in E. Choi and J. Harrigan Handbook of International Trade, Basil Blackwell:
213-250.
Grossman, G., and E. Helpman (1994). “Protection for Sale”, American Economic Review,
84(4): 833-850.
Jacobo, A. and I. Jalile (2012). “On the Macro and Micro Determinants of Protectionist Policy
Instruments in Latin America: An Exploratory Note”, GTA-CEPR-LATN Joint Conference on
Analyses of Contemporary Protectionism: Implications for Latin America, Lima, Perú,
manuscript.
Jørgensen, J. and P. Schröder (2006). “Tariffs and Firm-Level Heterogeneous Fixed Export
Costs,” Contributions to Economic Analysis & Policy 5(1): Article 18.
Kee, H., I. Neagu, and A. Nicita (2010). “Is Protectionism on the Rise? Assessing National
Trade Policies During the Crisis of 2008”, World Bank Working Paper No. 5274.
Kee, H., A. Nicita, and M. Olarreaga (2008). “Import Demand Elasticities and Trade
Distortions”, Review of Economics and Statistics, 90 (4): 666‐682.
Knetter, M. and Prusa T. (2003). ‘Macroeconomic Factors and Antidumping Filings: Evidence
from Four Countries’, Journal of International Economics, 61 (1): 1-17.
Krugman, P. (1981). “Intra-industry Specialization and the Gains from Trade”, Journal of
Political Economy, 89 (5): 959- 973.
Hanson, G., R. Mataloni and M. Slaughter (2003). “Vertical Production Networks in
Multinational Firms”, NBER Working Paper No. 9723.
Hummels, D., J. Ishii, and K. Yi (2001). “The Nature and Growth of Vertical Specialization”,
Journal of International Economics, 54 (1): 75-96.
Olarreaga, M. and M. Vaillant (2011). “Micro and Macro Determinants of Trade Temporary
Barriers: The Brazilian Case Over the Last Two Decades”, Departamento de Economía,
Universidad de la República, Documento number 07/11.
Prusa, T. and S. Skeath (2002). “The Economic and Strategic motives for Antidumping
Filings”, Weltwirstchaliftches Archives, 138 (3): 389- 413.
Rozemberg, R. and R. Gayá (2010). “Global Crisis and Trade Barriers in Latin America”, in S.
Evenett (Editor) Managed Exports and the Recovery of World Trade: The 7th GTA Report. A
Focus on Latin America, Center for Economic Policy Research.
Sabry, F. (2000). “An Analysis of the Decision to File, the Dumping Estimates, and the
Outcome of Antidumping Petition”, International Trade Journal, 14 (2): 109-145.
Takacs, W. (1981). ‘Pressures for Protectionism: An Empirical Analysis’, Economic Inquiry, 19
(4): pp. 687-93.
WTO (2009). Overview of Developments in the International Trading Environment,
WT/TPR/OV/12.
Appendix
0.3577
R2
***
***
***
0.4537
225305
0.0944
-0.5107 ***
0.1855
4.5029
0.0891
***
0.4189
520806
Yes
Yes
0.1001
-23.5130
0.1689
17.6481
0.0353
1.2547
0.0011
BRA
0.2867
***
***
***
***
0.963
391623
Yes
Yes
0.0095
-0.0266
0.0113
0.1203
0.0062
0.0174
0.0001
***
***
***
CHL
0.2270
***
0.3294
370479
Yes
Yes
0.1308
-19.4677
0.1634
0.7665
0.0629
2.8131
0.0007
***
***
***
COL
0.1190 ***
0.5234
289586
Yes
Yes
0.1905
-8.7163
0.0916
-5.5615
0.0891
1.5468
0.0014
0.0008
MEX
0.3006 ***
0.5614
631402
Yes
Yes
0.1363
*** -42.9247 ***
0.0753
*** -13.7267 ***
0.0483
*** -0.3437 ***
ECU
0.6062
***
***
***
0.4527
147476
Yes
Yes
0.2249
-21.3775 ***
0.2684
11.0866
0.1634
1.5567
0.0017
PRY
0.1838 ***
0.3448
270131
Yes
Yes
0.1412
-20.6054
0.2945
15.1719
0.0675
2.5795
0.0016
***
***
***
PER
0.1411 ***
Notes: (1) Dependent variable is applied tariff; (2) tBNDPRF is the bound rate augmented by preferential rate when applicable; (3) Standard errors in italics; (2) *** p<0.01;
(3) Data pooled across 2002-2010.
405806
Yes
Yes
0.1182
-10.4062
0.1151
10.7300
0.0519
1.3201
0.0010
***
0.0013
1.6420
BOL
0.1820 ***
ARG
0.2502 ***
N
Partner FE
Year FE
VS1
VS
ITT
tBNDPRF
Table 6
0.3659
192941
Yes
Yes
0.2216
-18.6735
0.1658
-15.3033
0.1756
1.1945
0.0018
***
***
***
VEN
0.2384
***
0.4767
192591
Yes
Yes
0.1792
1.0467
0.1344
-12.3440
0.1110
0.5585
0.0017
***
***
***
URY
0.2019
***
405806
Yes
Yes
0.3586
225305
Yes
Yes
0.4764
0.2109
***
***
***
***
***
*** -13.8269 ***
0.4219
8.8649
0.2542
6.6616
0.0011
0.0544
0.1064
2.9715
0.2086
2.4029
0.2865
***
***
***
***
***
0.0936
***
2.4124
0.2933
-1.7366
0.1294
-1.2729
0.0021
0.0255
0.1309
-10.9265
0.1273
11.0759
0.0573
0.5260
0.0010
***
0.0014
1.8771
BOL
0.1656 ***
ARG
0.2436 ***
520806
Yes
Yes
0.4270
0.2316
-15.8650
0.4024
23.2473
0.0840
-0.4936
0.0018
0.0655
0.1121
-20.0695
0.1890
12.5779
0.0392
1.3880
0.0011
BRA
0.2707
***
***
***
***
***
***
***
***
391623
Yes
Yes
0.9631
0.0225
0.0142
0.0270
-0.2902
0.0143
0.0458
0.0003
0.0059
0.0108
-0.0309
0.0128
0.1829
0.0071
0.0067
0.0001
***
***
***
***
***
CHL
0.2262
***
370479
Yes
Yes
0.3302
0.2938
1.5520
0.3714
0.6947
0.1378
-1.0562
0.0012
0.0222
0.1503
-19.8583
0.1878
0.5570
0.0717
3.0861
0.0008
***
***
***
***
***
***
***
COL
0.1132 ***
289586
Yes
Yes
0.5307
0.4636
-17.8188
0.2205
-4.8493
0.2165
2.9222
0.0030
0.0859
0.2109
-5.2596
0.1022
-4.6159
0.0983
0.9891
0.0015
-12.4735
MEX
0.3307 ***
0.1032
0.5272 ***
631402
Yes
Yes
0.5677
0.2992
*** -12.4735 ***
0.1680
*** -6.6868 ***
***
0.0014
*** -0.1147 ***
0.1570
*** -39.6396 ***
0.0868
*** -11.9952 ***
0.0556
*** -0.4970 ***
ECU
0.5923
***
***
***
147476
Yes
Yes
0.4530
0.5015
-1.0052
0.6060
4.6313
0.3619
-1.7024
0.0022
0.0030
0.2629
***
***
***
-21.1225 ***
0.3115
9.8869
0.1923
2.0249
0.0018
PRY
0.1829 ***
270131
Yes
Yes
0.3550
0.3200
-0.7927
0.6754
1.2924
0.1460
-0.6157
0.0026
-0.1665
0.1603
-20.2187
0.3362
14.7202
0.0781
2.6949
0.0017
***
***
***
***
***
***
***
PER
0.1794 ***
Notes: (1) Dependent variable is applied tariff; (2) tBNDPRF is the bound rate augmented by preferential rate when applicable; (3) Standard errors in italics; (2) *** p<0.01;
(3) Data pooled across 2002-2010.
N
Partner FE
Year FE
R-squared
VS1xI2009
VSxI2009
ITTxI2009
tBNDPRFxI2009
VS1
VS
ITT
tBNDPRF
Table 7
192941
Yes
Yes
0.3661
0.4520
-1.6839
0.3374
-2.5947
0.3578
0.6435
0.0022
-0.0008
0.2808
-18.0240
0.2108
-14.3017
0.2227
0.9515
0.0020
***
***
**
***
***
***
VEN
0.2385
***
192591
Yes
Yes
0.4769
0.4645
0.7554
0.3408
-2.4302
0.3072
-1.0154
0.0026
-0.0040
0.1969
0.9201
0.1473
-11.9119
0.1202
0.7137
0.0017
**
***
***
*
***
***
***
URY
0.2027
***
***
403587
Yes
Yes
Yes
0.3115
0.0548
225304
Yes
Yes
Yes
0.4330
0.1906
5.1084
***
0.0815
-0.4264
-1.3989 ***
0.0275
0.0009
***
-0.0366
0.0009
0.0687
0.0009
***
0.0008
0.0399
BOL
0.1940 ***
ARG
0.3738 ***
520806
Yes
Yes
Yes
0.3337
0.0381
-0.0038
0.0196
0.0068
0.0008
0.0758
0.0009
BRA
0.3587
***
***
391613
Yes
Yes
Yes
0.9629
0.0114
0.0295
0.0063
-0.0294
0.0002
0.0042
0.0001
***
***
***
CHL
0.2338
***
370481
Yes
Yes
Yes
0.2576
0.0579
-0.0937
0.0315
-0.2291
0.0005
0.0202
0.0005
**
***
***
COL
0.1609 ***
289585
Yes
Yes
Yes
0.4788
0.1261
4.2240
0.0606
-0.8652
0.0017
0.2056
0.0013
0.0006
MEX
0.3939 ***
***
631358
Yes
Yes
Yes
0.4693
0.0651
0.6951 ***
0.0365
*** -0.3247 ***
0.0009
*** -0.1093 ***
ECU
0.4568
***
147469
Yes
Yes
Yes
0.3959
0.1882
-0.8320
0.1046
-0.5004
0.0012
0.0155
0.0012
***
***
***
PRY
0.3013 ***
270131
Yes
Yes
Yes
0.2334
0.0591
0.5684
0.0342
-0.1066
0.0010
-0.0323
0.0007
***
***
***
PER
0.2088 ***
Notes: (1) Dependent variable is applied tariff; (2) tBNDPRF is the bound rate augmented by preferential rate when applicable; (3) Standard errors in italics; (2) *** p<0.01;
(3) Data pooled across 2002-2010.
N
Partner FE
Year FE
Product
R-squared
ITTxI2009
ITT
tBNDPRFxI2009
tBNDPRF
Table 8
192932
Yes
Yes
Yes
0.2828
0.1639
0.7110
0.1228
0.1800
0.0011
-0.0098
0.0012
***
*
***
VEN
0.2956
***
192592
Yes
Yes
Yes
0.4215
0.1065
0.0697
0.0604
-0.3888
0.0009
0.0058
0.0012
***
***
URY
0.3601
***
Tabla 9
t
ARG
0.0444
***
0.0186
VS
VS1
BRER
3.5078
BRA
0.1059
0.0178
***
6.6032
1.1251
2.8009
-0.1454
3.5772
1.2038
1.6331
-1.5047
***
***
-2.2312
MEX
0.0229
***
0.0042
***
3.7400
***
1.0817
***
4.3331
COL
0.1379
***
PER
0.3377
0.0197
0.0441
12.3015
***
-5.6619
1.5355
2.2921
0.0231
-0.0048
0.0658
0.0030
***
0.0000
0.0000
***
-64.5921
***
***
2.4430
-13.7930
***
7.8290
*
-0.3534
***
0.0000
***
-33.5240
Imports
0.0000
0.0000
0.0000
0.0000
Unit Values
-0.0134
-14.0099
0.0240
7.2951
21.4922
46.7286
15.9726
-0.0089
-0.0033
-0.1069
0.0899
79.4899
0.0462
0.0894
0.1309
0.8666
-5.8989
-13.2480
0.5898
txI2009
0.0523
***
0.0238
0.2891
0.0000
-178.3344
0.0000
VSxI2009
-5.4049
1.8031
17274.2700
4.7066
16.1521
VS1xI2009
0.2014
-1.5298
-4.7378
23.1958
1.9595
10344.7600
6.5183
6.0903
5.2855
ImportsxI2009
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0070
0.0000
0.0000
UnitValuesxI2009
-0.0146
-13.9624
-45.8221
143.9287
49282.1500
102.7704
60.4284
TCRBxI2009
1.2404
-0.1363
-0.5507
0.0025
0.4894
3340.1530
0.5213
0.0016
8014.6860
N
Partner FE
Year FE
148284
Yes
Yes
125851
Yes
Yes
159138
Yes
Yes
53763
Yes
Yes
80329
Yes
Yes
***
0.0559
***
***
2.0494
0.8672
***
***
6.8371
0.5875
0.0000
***
***
20.2175
***
5.5488
0.0000
***
-57.3029
*
-128.1398
81.2093
Notes: (1) Dependent variable is a binary variable indicating the presence of an AD initiations in a particualr HS 6 digit sector;
(2) Standard errors in italics; (3) *** p<0.01.
Tabla 10
BRA
0.1708 ***
t
ARG
-0.0008
BRER
-1.5824
0.5957
0.9227
Imports
0.0000
0.0001
0.0000
0.0000
Unit Values
-0.0052
-1.8700
0.0067
4.4753
***
-2.6676
PER
0.0525
COL
0.1780
0.0392
0.3083
0.1765
***
0.0207
-0.0053
-0.3007
0.0783
0.0163
0.3135
***
0.0000
0.0000
0.0000
0.0456
0.0303
MEX
0.1604 ***
0.0002
0.0001
-187.3616
-60.5449
25.4968
119.8994
32.2589
0.0000
-42.1161 ***
**
0.0358
-0.0409
-0.0397
-0.3475
0.0274
63.9757
0.0508
0.1889
0.1165
ImportsxI2009
0.0000
0.0000
0.0000
0.0001
0.0001
0.0000
0.0127
0.0000
UnitValuesxI2009
-0.0025
1.7309
-27.9655
TCRBxI2009
1.3088
N
Partner FE
Year FE
Product FE
txI2009
0.0563
***
0.0002
195.9998 **
0.0001
-74.6243
11457.6900
85.7151
116.4329
133.4613
0.2594
-0.5714
0.0024
-137.7459
0.5224
2865.8620
0.5071
0.0239
12943.3900
11831
Yes
Yes
Yes
5644
Yes
Yes
Yes
6911
Yes
Yes
Yes
5604
Yes
Yes
Yes
1487
Yes
Yes
Yes
0.0349
***
***
Notes: (1) Dependent variable is a binary variable indicating the presence of an AD initiations in a particualr HS 6 digit
sector; (2) Standard errors in italics; (3) *** p<0.01.