Content uploaded by Toan Nguyen
Author content
All content in this area was uploaded by Toan Nguyen on Apr 25, 2018
Content may be subject to copyright.
Trade Diversion as Firm Adjustment to Trade
Policy: Evidence from EU Antidumping Duties
on Vietnamese Footwear∗
Nguyen Trong Hoai
University of Economics Ho Chi Minh City
Nguyen Truong Toan
Vietnam Netherlands Program
Pham Hoang Van
Baylor University
December 2015
Abstract: This paper evaluates the impact of the 2006 European Union
antidumping action on Vietnamese footwear in three markets: imports
to the EU, footwear producers in Vietnam, and the trade diversionary
adjustment of Vietnamese firms in the U.S. market. We find that the
AD action reduced Vietnamese footwear imports to the EU by as much
as 65%. Given that the EU makes up almost two-thirds of Vietnam’s
footwear exports and footwear is among the top four export industries
for Vietnam, this reduction is economically significant . Consistent
with predictions of our model, we also find evidence of trade
diversion by Vietnamese producers from the EU to the U.S. market.
Our difference-in-difference estimates of the AD actions on the value
of Vietnamese footwear imports to the U.S. ranged from 69-71% over
the period 2004-2007 and 69-72% in terms of quantity. These results
highlight the (perhaps unintended) indirect effects of trade policy in
third markets that can result from firms adjusting to trade barriers.
Our results are robust to triple difference specifications where we
adjust for trend-differences and a series of placebo specifications.
JEL Codes: F13, F15, L6
Keywords: Antidumping, Trade Diversion, Vietnam, Footwear
∗Contact: hoaianh@ueh.edu.vn, toan.nt@vnp.edu.vn, or van pham@baylor.edu. We thank
Nguyen Hoang Bao, Pham Thi Bich Ngoc, Dinh Cong Khai, Pham Khanh Nam, James Riedel,
and Le Cong Tru for helpful suggestions and comments. All errors are our own.
1 Introduction
The increasing integration of rich and poor countries has contributed to much of
the significant growth in world trade over the past few decades. For example,
Figure 1 shows imports to the U.S. over the 1990s and the first half of the new
century. Imports have increased by over 300% from 1996 to 2007. Most of this
growth has come from the expansion of trade with less-developed countries
(LDCs). Non-oil imports from LDCs surpassed those from rich countries in 1995,
and the gap between the two groups have continued to diverge ever since. The
growing imports from LDCs have, however, precipitated protectionist reactions
in many developed countries.
In this paper, we evaluate the effects of the European Union (EU)
antidumping duties in 2006 on footwear imports from Vietnam. As expected, the
duties reduced the quantity and value of imports to the EU. Interestingly,
however, we also find evidence that Vietnamese firms responded to the EU tariffs
by diverting footwear exports to the U.S. market. Our point estimates range from
69% to 72% increase in Vietnamese footwear exports to the U.S. from the period
2004 to 2007 as a result of the trade diversion due to the EU tariffs. These results
suggest that antidumping measures have unintended spillover effects into other
export markets which could exert pressure for further protectionist policies.
Antidumping duties (AD) were not common in the 1960s—about ten cases
per year on average (Schott 1994). This number increased rapidly in the 1970s and
1980s as more countries joined the GATT/WTO agreements. Accession to the
WTO took away a member country’s ability to impose discretionary tariffs in
general but allowed a country to impose duties if it could show evidence that a
partner country is dumping in its market (see Hansen & Prusa (1995) and
Blonigen & Prusa (2001) for discussions on the rise of AD as an instrument of
protection). Over 1,600 AD petitions were filed in the 1980s (Finger & Artis
CONSUMER GOODS IMPORTS (NOMINAL USD)
0.0E+00
5.0E+10
1.0E+11
1.5E+11
2.0E+11
2.5E+11
3.0E+11
3.5E+11
4.0E+11
4.5E+11
5.0E+11
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
CHINA
RICH
NONRICHNONOIL
WORLD
Figure 1. U.S. Consumer Goods Imports by Select Country Groups. Source:
World Development Indicators.
2
(1993)), and the use of AD measures by WTO members tripled from the beginning
of the 1980s to the end of the 1990s (Prusa (2005)). Since the world financial crisis
in 2008, developed countries have continued to increase the use of AD as
protection for domestic producers (Kee et al. (2013)) .
To understand the effects of the EU duties on Vietnamese footwear producers,
we measure the impact in three markets. First, we evaluate the direct effect of the
duties on the volume of footwear imports to the EU. Using Eurostat data, we
compare Vietnamese import volumes in the affected industry categories to those
of comparable countries before and after the imposition of the duties. Second, we
used data from the Enterprise Surveys collected by Vietnam’s General Statistical
Office in 2003, 2004, and 2006, 2007 to evaluate the effects on firm revenue, payroll
and employment. These survey years spanned before and after the investigation
period and imposition of the tax. We compare affected footwear firms to firms in
other sectors. Third, we consider the trade diversion effects of the EU duties by
comparing imports of Vietnamese footwear to the U.S. to other countries’ imports
to the U.S. using data from the International Trade Commision (USITC). In each
of these exercises, we apply a difference-in-difference methodology to the panel
data structure to estimate the effects of the EU duties. We discuss the
identification strategy, the choice of control groups, and the robustness exercises
in subsequent sections.
A number of studies have focused on the impact of AD in target countries.
Prusa (1994) finds that AD action pressures foreign firms to increase their prices in
the first stage of the process. Cuyvers & Dumont (2005) find that Asian exports to
the EU are significantly lower in both quantity and value after the petition is filed.
Brambilla et al. (2012) find that U.S. duties on Vietnamese catfish had significant
negative effects on incomes of Vietnamese catfish farmers. Our study extends this
literature but unlike these other papers, we evaluate the impact of this AD policy
3
on three markets: imports to the EU, footwear producers in Vietnam, and imports
to the U.S.. The second set of results are important for a small country like
Vietnam because the EU accounts for 65% of Vietnam’s footwear exports and
footwear is one of the most important industries in Vietnam. Finally, the third
result documents the non-trivial trade diversionary effects of trade policy, which
is often overlooked in discussions about the distortionary effects of AD as a tool
of protection. Other authors such as Prusa (2005) and Brenton (2001) have studied
the trade diversion effects of AD policy, but these studies typically focus on the
diversion of trade toward other non-named suppliers who now gain an
advantage in the subject market. Our study, instead, focuses on the diversion of
trade by suppliers named in the AD action toward non-named markets.
Our results on the diversion of trade toward non-named markets can be
compared with those in two papers that apply similar analysis to different data.
Bown & Crowley (2007) evaluate the effects of U.S. import restrictions on
Japanese exports of almost 5,000 products to 37 countries over the years
1992-2001. They find that the average U.S. antidumping duty on Japanese exports
leads to a 5% to 7% increase in Japanese exports of the same product to the
average non-U.S. market. The significantly smaller effect as compared to our
findings for Vietnamese footwear is not surprising for two reasons. First, their
result is an average treatment effect over 37 countries. Should affected suppliers
divert exports to other countries, they are only likely to do so to a small subset of
destination countries and negligibly to most countries in the sample. Our study
looks for evidence of trade diversion where diversion is most likely to take place:
from the EU to the U.S., the two largest markets for Vietnamese footwear exports.
Second, the Bown & Crowley (2007) result is an average treatment effect over
thousands of products. Japan is a developed, industrialized economy producing a
diverse range of products with different demand elasticities. Thus, we would
4
expect the trade distortionary effects to vary widely over the range of products.
Our results for Vietnamese footwear, a highly commoditized product, is likely to
be an example of a product whose trade is most distorted by protectionist policies.
It is also useful to compare our results to those in a second paper by Lu et al.
(2013) which exploits monthly export transaction data for Chinese firms. The
richness of firm-level characteristics in the dataset allows the authors to find
significant heterogeneity in the effects of U.S. antidumping action against Chinese
exporters over the period 2000-2006. U.S. antidumping action lowered export
volumes to the U.S. market by 32% to 67%, but the effects varied by product and
by whether the firm is a direct or indirect exporter, and single- or multi-product
firm. Lu et al. (2013) do not find that U.S. antidumping action cause Chinese
exporters to divert their exports to non-U.S. markets. This last result is the
average treatment effect over 346 product categories.
Our study complements these two other studies in an important way. By
studying the effects of one policy action on one specific product category, we can
uncover the potentially large distortionary effects of trade protection even where
the average effects over many industries and countries are muted. And because
pressure for protection from imports tends to emanate from specific industries,
our study sheds light on the potential propagation of trade policy across
countries. That is, trade barriers in one country can lead to trade barriers in third
countries because of the diversion of trade flows resulting from the original
policy.
The rest of this paper is organized as follows. Section 2 gives an overview and
background of Vietnamese footwear and the EU antidumping case. We introduce
a simple theoretical model in section 3 that generates predictions we can take to
the data. Section 4 describes the data and empirical strategy. Section 5 shows the
results. Concluding remarks are provided in section 6.
5
2 Background of Footwear and EU AD sanctions
Since the launch of market reforms in Vietnam in the late 1980s, Vietnam has
experienced sustained increases in income, accompanied by the rising importance
of exports. Real income increased three-fold from under 300 USD per capita in
1990 to almost 900 USD in 2010, while exports doubled from 36% to 72% of GDP.
Accompanying the growth in international trade, however, has been the
frequency of AD petitions against many of Vietnam’s key exports. Up until 2011,
Vietnam was the target of over 40 AD investigations, one-third of which resulted
in punitive duties. Most notable recent examples include the U.S. AD duties on
catfish in 2002 and shrimp in 2003.
The EU is one of Vietnam’s biggest export markets, reaching 17 billion USD in
2011. Trade between Vietnam and the EU has grown steadily, though Vietnamese
exporters have faced protectionist barriers in the form of AD action numerous
times in recent years. The EU targeted Vietnamese bicycles in 2004 and
Vietnamese footwear in July 2005.
In Vietnam, footwear ranked third behind rice and textile in non-oil export
value reaching 2.5 billion USD in 2004 (see Figure 2). The footwear industry
directly employs about 500,000 workers. Footwear enterprises are located around
the three major cities of Ha Noi, Ho Chi Minh, and Da nang. Vietnam produces a
variety of footwear products including leather shoes, athletic shoes. Labor is
cheap, and quality is sufficiently high for export to developed country markets
like the EU, the U.S., and Japan.
The EU is the biggest importer of Vietnamese footwear with nearly 1.77 bilion
U.S.D in 2004, making up 68% of all of Vietnam’s footwear exports. Vietnam is the
second biggest footwear exporter to the EU after China (followed by India,
Indonesia, and Tunisia) (see Figure 3).
On July 7, 2005, the EU launched AD investigations on Vietnamese and
6
Figure 2. Top five Vietnamese exports excluding oil in 2004 (million nominal
USD). Source: Vietnam General Statistics Office
Figure 3. Top five footwear exporters to EU (thousand nominal euro). Source:
Eurostat
7
Chinese leather shoes based on claims brought by the European Confederation of
the footwear industry (-CEC-) EC166/14 (2005). At that time, the CEC
represented 40% of EU suppliers. Its claim was based on Article 5 of European
Regulation EC384 (1995) last amended in EC461 (2004). The targeted products
were footwear containing leather as denoted by CN8 codes 64032000, 64033000,
64035111, 64035115, 64035119, 64035191, 64035195, 64035199, 64035911, 64035931,
64035935, 64035939, 64035991, 64035995, 64035999, 64039111, 64039113, 64039116,
64039118, 64039191, 64039193, 64039196, 64039198, 64039911, 64039931,
64039933,64039936, 64039938, 64039991, 64039993, 64039996, 64039998 and
64051000. The investigation period extended from 01 April 2004 to 31 March 2005.
Results were publicized in EC553 (2006). The resulting AD duty rates imposed on
the Vietnamese firms are shown in Table 1.
Table 1. EU AD on Vietnam and People Republic China Footwear.
Source: EC553/2006
Countries 7-Apr-2006 to
1-Jun- 2006
2-Jun-2006
to 13-Jul-2006
14-Jul-2006
to 14-Sept-2006
After
15-Sept-2006
Vietnam 4.2% 8,4% 12.6% 16.8%
China 4.8 % 9,7 % 14.5% 19.4%
As shown in Figure 4, imports of targeted products decreased significantly
from year 2005. In 2003, Vietnam’s export to the EU of targeted products grew
15% in value and 35.5% in quantity. After the launch of investigations, value was
stagnant in 2005, then contracted by 8.5% in 2006 and 13.8% in 2007. Quantity fell
by 5.5%, 12.78% and 13.05% in those same three years, respectively. The direct
effect of the AD action on targeted imports is consistent with the studies by
Brenton (2001) and Messerlin (1989). It is also similar to Prusa (2003) finding a
drop of 30% to 50% of AD products imported to the U.S.. Brenton (2001) finds an
average decrease of 20% in the first year after the beginning of an EU AD tax.
8
Figure 4. EU import value and quantity of Vietnamese footwear categories
covered and not covered by AD duties. Source:
http://epp.eurostat.ec.europa.eu/newxtweb/b/
3 A Theoretical Model
In this section, we construct a simple model to generate predictions on the effects
of the AD action on (1) exports to the target market, (2) production outcomes of
the targeted firms, and (3) third markets that see a diversion of trade by the
targeted firms away from the target market.
Consider a monopolist producing two goods that it sells to two export
markets denoted by inverse demand functions p1(q1)and p2(q2), where q1and q2
are quantities and p1and p2prices in the two respective markets. The two goods
are produced using the same technology so that cost is a function of total quantity
produced, C(q1+q2), where C(·)is increasing convex C0>0 and C00 >0. In
addition to production costs, the monopolist also faces an ad valorem tariff of rates
t1and t2to sell in each respective market.1
The monopolist chooses q1(and p1) and q2(and p2) to maximize its total
1Think of each tias being inclusive of transport and other per unit costs.
9
profit:
Π(q1,q2;t1,t2) = p2(q1)q1+p2(q2)q2−C(q1+q2) − t1q1−t2q2.
The two first order conditions are:
∂Π
∂q1
=p0
1(q1)q1+p1(q1) − C0(q1+q2) − t1=0 (1)
∂Π
∂q2
=p0
2(q2)q2+p2(q2) − C0(q1+q2) − t2=0 (2)
We assume the second order conditions are satisfied at the profit-maximizing
(q∗
1,q∗
2). That is, the Hessian matrix,
H=
∂2Π
∂q2
1
∂2Π
∂q2∂q1
∂2Π
∂q1∂q2
∂2Π
∂q2
2
is negative semi-definite:
∂2Π
∂q2
1
=p00
1(q1)q1+2p0(q1) − C00 <0 (3)
∂2Π
∂q2
2
=p00
2(q2)q2+2p0(q2) − C00 <0 (4)
|H|=∂2Π
∂q2
1
∂2Π
∂q2
2
−∂2Π
∂q2∂q12
=
(p00
1(q1)q1+2p0(q1) − C00) (p00
2(q2)q1+2p0(q2) − C00)−(C00)2>0. (5)
These second order conditions will be satisfied under common demand structures
such as constant price elasticity and linear demand and sufficiently convex cost.
Comparative statics for an increase in the import tariff, t1, gives us the
following predictions which we can test in subsequent sections of the paper.
Result 1 (Direct effects of an increase in import duty t1on market 1).
10
1. Quantity of exports to market 1 declines. ∂q1
∂t1=1
|H|
∂2Π
∂q2
2<0;
2. Price of exports to market 1 increases. ∂p1(q1)
∂t1=p0
1
∂q1
∂t1>0;
3. Value of Export to market 1 declines. ∂(p1(q1)q1)
∂t1=p1∂q1
∂t1(1
1+1)<0 since
the own-price demand elasticity, 1<−1 at the optimum.
Result 2 (Trade Diversion: Effects of increase in import duty t1on market 2).
4. Quantity of exports to market 2 increases. ∂q2
∂t1=1
|H|−∂2Π
∂q2
2=C00
|H|>0.
5. Price of exports to market 2 declines. ∂p2(q2)
∂t1=p0
2
∂q2
∂t1<0;
6. Value of Export to market 2 increases. ∂(p2(q2)q2)
∂t1=p2∂q2
∂t1(1
2+1)>0 since
the own-price demand elasticity, 2<−1 at the optimum.
7. Total exports to markets 1 and 2 decline.
∂(q1+q2)
∂t1=1
|H|(p00
2(q2)q2+2p0(q2)) <0. However, as the preceding
expression indicates, the extent of the decline depends on the price elasticity
of demand in market 2.
4 Empirical Methodology and Results
We estimate the effects of the EU’s AD action in three markets—Vietnamese
footwear imports to the EU, production outcomes of footwear firms in Vietnam,
and Vietnamese footwear imports to the U.S. In the subsequent sections, we
describe the difference-in-difference identification strategy for each of the three
analyses. We discuss our results and robustness specifications that include
validation and falsification exercises.
11
4.1 The Effect on Vietnamese Footwear Imports to the EU
We apply a difference-in-difference (DD) methodology to our three panel datasets
to evaluate the predictions of the model in the previous section. Our specification
takes the general form:
Yit =α+X
i
Ai+X
t
YEARt+β∗ddV Nit +it (6)
where Aiare country or industry fixed effects depending on the context, YEARt
are year fixed effects. The dummy variable ddVNit is our variable of interest. It
takes a value of one when the cross-sectional value is the treatment group and the
year is after the treatment year. The coefficient βis the DD estimate of the effect of
the AD action on our outcome variable Y. We assume the error term it is
independently identically distributed normal with mean zero.
We evaluate the effect of the EU AD on Vietnam footwear imports in the EU
using three outcome measures: value, quantity, and price, each in logs. For our
main specifications, we use India as the ‘control’ group. India is an appropriate
choice for the following reasons.
Figure 5 shows EU imports of Vietnamese footwear targeted by the AD action
along with imports of the same categories from Thailand, Indonesia, India, and
also with imports of Vietnamese apparel to the EU. First, as the figure shows, in
value, quantity and price, India closely tracks Vietnamese footwear for the period
1999-2004 . Thus, for at least six years pre-treatment, India is a country that
follows ‘parallel trends’ with Vietnam in terms of these three outcomes. This
feature makes India an attractive candidate as a control group for the
difference-in-difference analysis. Second, India is the country that the EU used as
the ‘comparable country’ to evaluate the costs of Vietnamese footwear production
as part of its dumping investigation.
12
Figure 5. Trend in import value, quantity, price of AD-category Footwear to the
EU.
13
Data for footwear imported to the EU were retrieved from the EUROSTAT
website (http://epp.eurostat.ec.europa.eu/newxtweb/). This database
contains all export and import data of EU members with intra- and extra-EU
partners. EU here is defined as the EU25 countries including: Belgium (BE),
Denmark (DK), France (FR), Germany (DE), Greece (EL), Ireland (IE), Italy (IT),
Luxembourg (LU), Netherlands (NL), Portugal (PT), Spain (ES) and United
Kingdom (UK) , Austria (AT), Finland (FI) and Sweden (SE), Cyprus (CY), Czech
Republic (CZ), Estonia (EE), Hungary (HU), Latvia (LV), Lithuania (LT), Malta
(MT), Poland (PL), Slovakia (SK) and Slovenia (SI). We use the 8-digit Combined
Nomenclature (CN8) data since the EU used CN8 codes to classify products
subjected to the AD action. Import data include value (in euro) and quantity (in
kilograms). Value does not include tariff, freight, insurances and other
surcharges. We calculate price by dividing value by quantity.
Since the EU AD action took place in 2005, we compare import value,
quantity and price for Vietnam with India before and after 2005. Table 2reports
the DD estimates of the effect of the AD action (coefficient on the ddVN variable)
for three time comparisons: 2004 with 2006, 2004 with 2007, and 2003, 2004 before
with 2006, 2007 after. The DD estimates for value and quantity are statistically
significant at the 1% level for each of these time period samples. Estimates of the
size of the effect of the EU action on value and quantity range from 50-65% and
70-97%. The AD action decreased the (before-tariff) price of Vietnamese footwear
imports in the EU but the estimates are not statistically significant.
The magnitude of our estimates are in line with Prusa (2003) who finds an
average drop in import value ranging from 50-70% over three years for AD
actions in the U.S. and the EU. Our estimates are larger than those found by
Brenton (2001) where imports to the EU decreased by 20% in the first year after an
EU AD tax. The difference is not surprising, however, since that study analyzed
14
the average effect over many countries. The sample included mostly AD action
against imports in industries from more industrialized countries such as Japan,
Korea, Taiwan. Higher value-added goods such as chemicals, machinery and
equipment will have lower demand elasticities as they are less commoditized
than footwear exports from Vietnam.
15
To check the robustness of our results we perform a number of additional
exercises. First, to address the possibility that Indian footwear imports and
Vietnamese footwear imports do not follow “parallel trends” after the treatment
year of 2005, we add Vietnamese apparel imports to the EU as an ‘additional
control group.’ This triple difference specification can accommodate the
diverging trends between Indian and Vietnamese footwear imports if the same
divergence in trends is common between Vietnamese apparel and Vietnamese
footwear imports to the EU. The results of the triple difference regression are
shown in table 3. The specification includes product and year fixed effects, a
Vietnam dummy, and two way interactions among the treatment and two control
groups. The coefficient of interest is on the three-way interaction term of the
Vietnam dummy, the dummy for sectors named by the AD action, and an
after-AD time dummy. The triple-difference estimates of the effect of the AD
action on Vietnamese footwear are negative and statistically significant for import
value and quantity consistent with the double-difference results. However, the
magnitude of the estimates are much larger. Absent evidence contrary to the
assumption of ‘parallel trends’ between Vietnamese and Indian footwear
post-treament2, the baseline double-difference specification serve as our main
results.
2Recall that the time series shown in Figure 5 supports the parallel trends assumption pre-
treatment.
16
Table 2. Difference in difference analysis for EU import of Vietnamese AD footwear with India as the control group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
VN 0.063 0.067 -0.004 0.048 0.044 0.004 0.048 0.044 0.004
(0.32) (0.33) (0.06) (0.31) (0.35) (0.09) (0.31) (0.35) (0.09)
ddVN -0.888*** -0.872*** -0.016 -0.702*** -0.682*** -0.020 -1.050*** -1.020*** -0.030
(0.17) (0.18) (0.05) (0.19) (0.20) (0.09) (0.21) (0.22) (0.07)
Constant 15.162*** 8.071*** 2.486*** 15.846*** 8.608*** 2.633*** 15.901*** 8.649*** 2.646***
(0.19) (0.21) (0.04) (0.14) (0.15) (0.02) (0.15) (0.16) (0.04)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.093 0.076 0.082 0.062 0.049 0.042 0.109 0.095 0.044
N 260 260 260 132 132 132 130 130 130
F 15.52 8.82 5.03 10.33 6.72 3.56 15.39 9.18 4.34
Standard errors are heteroskedastic robust * p<0.10, ** p<0.05, *** p<0.01
17
As a second robustness test, we use the import of Vietnamese apparel
categories as the control group instead of Indian footwear. Cursory inspection of
Figure 5 shows that Vietnamese apparel and footwear imports follow similar
trends pre-treatment making it a candidate control. Vietnamese apparel and
footwear are arguably similar technologically as well as in the markets in which
they operate. There is also no reason for us to believe that the AD action on
footwear had anything to do with the apparel imports. For these reasons,
Vietnamese apparel imports serve as a good alternative control group for our
robustness exercise. Results of this double-difference specification are shown in
table 4. The regressions include year and product fixed effects. The variable of
interest is the interaction between the dummy for named sectors and the after-AD
year dummy. The estimated effects of the AD action are negative and statistically
significant for value and quantity. The magnitude of the point estimates are
similar to those of our baseline estimates using Indian footwear as the control
group. The big difference is the result of a large positive statistically significant
increase in the price.
18
Table 3. Triple-difference analysis for EU import of Vietnamese AD footwear with Indian Footwear and Vietnamese
apparel imports as control groups.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
VN -1.726*** -1.114*** -0.424*** -1.714*** -1.085*** -0.398*** -1.719*** -1.071*** -0.407***
(0.11) (0.11) (0.05) (0.12) (0.12) (0.05) (0.12) (0.12) (0.05)
VNADFw 1.788*** 1.181*** 0.420*** 1.762*** 1.129*** 0.402*** 1.767*** 1.115*** 0.411***
(0.33) (0.34) (0.07) (0.33) (0.36) (0.10) (0.33) (0.36) (0.10)
VNAfterAD 0.385*** 0.606*** -0.331*** 0.283*** 0.528*** -0.380*** 0.508*** 0.630*** -0.325***
(0.09) (0.09) (0.06) (0.11) (0.11) (0.07) (0.11) (0.11) (0.07)
ADFwAfterAD 0.281*** 0.237** 0.070* 0.310*** 0.306** 0.021 0.216* 0.142 0.080*
(0.10) (0.11) (0.04) (0.11) (0.13) (0.05) (0.12) (0.13) (0.05)
VNADFwAfterAD -1.273*** -1.478*** 0.315*** -0.985*** -1.209*** 0.360*** -1.558*** -1.650*** 0.295***
(0.19) (0.20) (0.08) (0.21) (0.23) (0.11) (0.23) (0.24) (0.10)
Constant 13.433*** 6.376*** 2.569*** 13.583*** 6.364*** 2.637*** 13.737*** 6.623*** 2.599***
(0.06) (0.05) (0.03) (0.06) (0.06) (0.03) (0.06) (0.06) (0.03)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.257 0.126 0.158 0.255 0.107 0.171 0.256 0.122 0.191
N 3,735 3,646 3,646 1,876 1,823 1,823 1,855 1,810 1,810
F 55.03 33.35 29.98 43.00 20.11 27.35 48.90 27.30 33.70
Standard errors are heteroskedastic robust * p<0.10, ** p<0.05, *** p<0.01
VN takes value of 1 if country is Vietnam and 0 for others.
ADFw takes value of 1 if CN8 is listed in sanction list, otherwise it is 0.
AfterAD takes value of 1 for 2006 and after and 0 for 2004 or before
19
Finally, as a placebo test, we use the same regression specification as our
baseline using Indian footwear as the control group but include only footwear
categories that were not affected by the AD action. The DD estimates for the effect
on non-AD footwear, shown in table 5, are almost all not statistically significant
for value, quantity, and price regressions. Taking all these results in their totality,
we can say that the there is a robust finding that there is a large negative
statistically significant effect of the AD action on Vietnamese import value and
quantity. In the following sections, we estimate the effect of these actions on the
actual footwear firms in Vietnam and the spillover effects into the U.S. footwear
market.
20
Table 4. Difference in difference analysis for EU import of Vietnamese AD footwear with Vietnamese Apparel as the
control group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
ddADFw -0.975*** -1.251*** 0.398*** -0.682*** -0.950*** 0.390*** -1.321*** -1.525*** 0.391***
(0.18) (0.18) (0.07) (0.19) (0.20) (0.10) (0.22) (0.22) (0.09)
Constant 11.884*** 5.431*** 2.178*** 12.329*** 5.442*** 2.183*** 12.691*** 6.367*** 1.882***
(0.05) (0.05) (0.03) (0.05) (0.04) (0.03) (0.05) (0.05) (0.03)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.103 0.156 0.039 0.065 0.145 0.062 0.166 0.228 0.070
N 1,847 1,776 1,776 922 879 879 918 882 882
F 28.84 36.41 9.14 13.46 30.24 11.67 36.39 51.32 12.76
Standard errors are heteroskedastic robust * p<0.10, ** p<0.05, *** p<0.01
ddADFw takes a value of 1 if the CN8 code is listed in the sanction list and year >2005; otherwise it is 0.
21
4.2 The Effect on Vietnamese Footwear Firms
In this section we present estimates of the impact of the AD tax on Vietnamese
footwear firms revenue, payroll and employment. We analyze two time windows,
and use several different control groups in the DD regressions. We also perform
analysis for subsamples by firm size: ‘small’ denoting enterprises with fewer than
100 employees, ‘medium’ for enterprises between 100 and 1000 employees, and
‘large’ for firms with at least 1000 employees.
To evaluate the impact of the AD action on Vietnam footwear firms, we use
the Vietnam Enterprise Survey. This dataset is collected by Vietnam’s General
Statistics Office (GSO) and its provincial counterparts. We use panel data for two
years 2004 and 2006. The 2004 survey contained 91,755 observations of which 329
are footwear firms while the 2006 surveyed 131,347 enterprises, of which 369 were
footwear. Henceforth, we will use the word enterprise and firm interchangeably
when referring to an observation in this dataset. The data include a firm’s
revenue, labor information, capital information, etc. For robustness, we increase
the data sample by including two additional years 2003 and 2007.
One difficulty arises when we combine data from 2006, 2007 with 2003, 2004.
The industry classifications changed in 2006, so we had to concord the industry
codes of year 2006 and year 2007 with the codes for 2003 and 2004. We use
apparel products as our control group in some of the empirical specifications and
wood, papers, textile (WPT) as the control in some specifications when we
compare to our footwear treatment group. We use SITC 4-digit classifications for
footwear to indicate firms targeted by the AD action (the treatment). Revenue and
payroll are measured in millions of Vietnam Dong for one year; capital is
measured in millions of Vietnam Dong. We include firm and year fixed effects in
all regressions. We also control for time varying firm differences with capital and
labor used as well as female employees as a fraction of the total.
22
Table 5. Difference in difference analysis for EU import of Vietnamese non-AD footwear with India as the control group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
VN 1.216** 1.467*** -0.015 0.964 1.581*** -0.059 0.905 1.411*** -0.060
(0.50) (0.47) (0.08) (0.58) (0.47) (0.08) (0.57) (0.46) (0.08)
ddVN -0.167 -0.415 0.013 0.614* -0.188 0.181* -0.244 -0.594 0.015
(0.35) (0.30) (0.11) (0.35) (0.30) (0.10) (0.50) (0.40) (0.10)
Constant 12.713*** 5.867*** 2.414*** 12.874*** 6.050*** 2.324*** 12.828*** 5.928*** 2.438***
(0.24) (0.24) (0.04) (0.29) (0.27) (0.05) (0.27) (0.25) (0.05)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.101 0.135 0.007 0.132 0.192 0.038 0.045 0.111 0.040
N 361 351 351 172 167 167 184 177 177
F 7.90 10.08 0.90 10.01 4.97 2.15 1.69 3.34 2.16
Standard errors are heteroskedastic robust * p<0.10, ** p<0.05, *** p<0.01
23
Table 6 shows results for log revenue as the outcome variable using apparel
firms as the control group for the years 2004 and 2006. The DD estimates of the
effect of the EU action on revenue are consistently negative for all three groups of
firm sizes. However, none of these results are significant at the 10% level. Not
reported in these tables, we also find no consistent effects when we study the
2003-2007 period. Similarly, we find no statistically significant effect when we
choose the control group to be wood and wooden product, papers, and textiles
firms. We also find no robust effect of the AD duties on payroll or on employment
as reported in Tables 7 and 8.
In sum, the EU AD action seems to have no discernible effect on the footwear
firm’s overall activity. This is consistent with industry aggregate export data from
the GSO shown in Figure 6 where footwear export value seems to be unaffected
by the EU AD action.
In the next section we test for evidence of diversion of Vietnamese footwear
to the U.S. market. Expansion in the U.S. market might have made up for the
reduction in exports to the EU market dampening the contractionary effect on
Vietnamese firms.
24
Figure 6. Main Vietnamese export categories.
25
Table 6. The Effect of EU Antidumping Duties on Firm Revenue.
Small Medium Large
(1) (2) (3) (4) (5) (6)
ddFw -0.197 -0.135 -0.030 -0.106 -0.018 0.098
(0.20) (0.19) (0.13) (0.11) (0.10) (0.09)
logFemale -0.016 -0.059 -0.070
(0.17) (0.16) (0.37)
lnL 0.368*** 0.486*** 0.260**
(0.08) (0.07) (0.11)
lnK 0.033 0.191*** 0.158
(0.10) (0.07) (0.11)
Constant 7.024*** 5.646*** 9.167*** 4.997*** 11.426*** 8.014***
(0.04) (0.93) (0.02) (0.96) (0.03) (1.94)
Year FE Yes Yes Yes Yes Yes Yes
Firm FE Yes Yes Yes Yes Yes Yes
R-squared 0.005 0.100 0.088 0.302 0.169 0.323
N 1,388 1,376 1,343 1,329 455 454
F 0.50 4.39 11.93 21.10 7.61 6.86
The dependent variable is log revenue. Regressions are for two years 2004 and 2006.
ddFW is a dummy variable equal to 1 for a footwear firm in 2006.
*p < 0.10, **p < 0.05, ***p < 0.01
26
Table 7. The Effect of EU Antidumping Duties on Firm Payroll.
Small Medium Large
(1) (2) (3) (4) (5) (6)
ddFw 0.060 0.073 -0.084 -0.049 0.121 -0.097
(0.17) (0.16) (0.13) (0.12) (0.14) (0.12)
logFemale -0.079 -0.096 -0.180
(0.14) (0.17) (0.48)
lnK 0.048 0.069 0.072
(0.08) (0.08) (0.14)
lnL -0.366*** -0.472*** -0.652***
(0.07) (0.07) (0.14)
Constant 2.222*** 3.436*** 2.477*** 4.920*** 2.639*** 7.569***
(0.03) (0.75) (0.02) (1.01) (0.04) (2.51)
Year FE Yes Yes Yes Yes Yes Yes
Firm FE Yes Yes Yes Yes Yes Yes
R-squared 0.070 0.200 0.059 0.211 0.067 0.438
N 1,395 1,395 1,329 1,329 454 454
F 7.72 10.04 7.79 13.05 2.71 11.21
The dependent variable is log firm payroll. Regressions are for two years 2004 and 2006.
ddFW is a dummy variable equal to 1 for a footwear firm in 2006.
*p < 0.10, **p < 0.05, ***p < 0.01
27
Table 8. The Effect of EU Antidumping Duties on Firm Employment.
Small Medium Large
(1) (2) (3) (4) (5) (6)
ddFw 0.134 0.157 0.042 0.027 -0.103 -0.074
(0.17) (0.18) (0.08) (0.08) (0.07) (0.07)
logFemale -0.120 0.094 0.104
(0.15) (0.11) (0.22)
lnK 0.031 0.094** 0.143*
(0.09) (0.05) (0.08)
Constant 3.221*** 3.514*** 5.744*** 4.486*** 7.722*** 5.682***
(0.03) (0.83) (0.01) (0.65) (0.02) (1.43)
Year FE Yes Yes Yes Yes Yes Yes
Firm FE Yes Yes Yes Yes Yes Yes
R-squared 0.013 0.018 0.001 0.020 0.026 0.070
N 1,409 1,395 1,343 1,329 455 454
F 1.31 0.95 0.16 1.23 1.01 1.38
The dependent variable is log firm employment. Regressions are for two years 2004 and 2006.
ddFW is a dummy variable equal to 1 for a footwear firm in 2006.
*p < 0.10, **p < 0.05, ***p < 0.01
4.3 Trade Diversion: The Effects on U.S. Footwear Imports
Our theory predicts that Vietnamese footwear firms will adjust to the EU AD tax
by increasing its export to other existing markets. In this section we test that
prediction by looking at the change in Vietnamese footwear exports to the U.S.
caused by the EU action. The U.S. is the second largest market for Vietnamese
footwear after the EU making its market a likely candidate for Vietnamese firms
to divert its output.
We use U.S. import data provided by the US International Trade Commission
through its dataweb website (http://dataweb.usitc.gov/). Import and export
from all countries to and from the U.S. are available at the HTS10-digit level.
Values are given in thousand U.S. dollars, quantities are reported in units
depending on the line of product. For example, some lines of footwear are
28
Figure 7. Trend of export value, quantity, price of Vietnam Footwear and other
industries to the U.S..
29
measured in dozens while some are measured in kilograms. Price is the average
price calculated as value per unit quantity.
If trade is being diverted, it is reasonable that firms will engage in diversion
toward products in which they already have sales. It is unreasonable to believe
that firms will start up new products in the U.S. market in the short run just
because taxes are raised in the EU. Therefore, to test for evidence of the trade
diversion story, we include only lines with non-zero imports from Vietnam before
the AD action. Trade diversion by the firms need not be limited to the footwear
categories that were hit by the AD duties. The firm could react to higher barriers
in one set of products by increasing production in similar products. We allow for
this possibility by including in the analysis all footwear categories rather than
those just subject to AD duties in the EU.
We apply a difference-in-difference (DD) methodology to our three panel
datasets to evaluate the predictions of the model in the previous section. Again,
our specification takes the general form:
Yit =α+X
i
Ai+X
t
YEARt+β∗ddFWit +it (7)
where Aiare country or industry fixed effects depending on the context, YEARt
year fixed effects. The dummy variable ddFWit is our variable of interest. It takes
a value of one when the cross-sectional value is the treatment group and the year
is the after the treatment year. The coefficient βis the DD estimate of the effect of
the AD action on our outcome variable Y. We assume the error term is it is
independently identically distributed normal with mean zero.
The crucial assumption for proper identification in the DD methodology is
the constant difference between the treatment and control groups across time in
the absence of the treatment. After the U.S. and Vietnam entered into a bilateral
30
trade agreement (BTA) in 2001, Vietnam’s exports to the U.S. grew much faster
relative to other similar countries. Thus for the DD analysis of imports to the U.S.
market, it is inappropriate to use another country’s imports as the control group.
Since the BTA likely affected all Vietnamese goods similarly, we choose
Vietnamese exports in another sector as a control. Figure 7 shows the trend in
imports (value, quantity and price) from Vietnam to the U.S. for a number of
sectors. From a visual inspection of the time trends between 2003-2005, a few
sectors could plausibly serve as our control. We choose Vietnamese apparel as the
main control group in our DD estimates.
We later relax this ‘parallel trends’ assumption by adding a second control
group in a triple difference specification. In addition to Vietnamese apparel
exports to the U.S. as the main control group, we add Vietnamese exports of WPT
products to the U.S. and also exports in the control period of 2002 and 2003 as the
third difference. Finally, we report results from a number of placebo
specifications.
Our main results are shown in Table 9. There is strong evidence that the EU
AD action caused an increase in Vietnamese footwear exports to the U.S.. In the
2003-2007 subsamples as well as the 2004/2006 and 2004/2007 samples, the DD
estimates for the effects on import value and quantity are positive and significant.
These effects are economically significant. Comparing 2006 to 2004, the difference
in value and quantity were about 45% (0.31 log points). Comparing 2007 to 2004,
these differences were about 66%. The effect on price was not statistically
significant. This is not surprising as the demand for Vietnamese footwear is likely
to be highly elastic.
Table 10 shows the DD results using the WPT sectors as the control group.
Comparing 2006 to 2004, the difference in value and quantity were 36% and 33%
respectively. Comparing 2007 to 2003, these differences were 66% and 72% .
31
To address the possibility that the differences between treatment and control
groups are time-varying, we add a third difference to our analysis. For this DDD
analysis, we compare 2007 and 2004 imports of Vietnamese footwear and apparel
which we compare to the difference between the two sector imports for a control
time period of 2002 and 2003.3We also repeat the analysis with imports of wood,
paper and textile (WPT) as the control group. Results are shown in Table 11.
The DDD estimates are the coefficients on the three-way interaction term
FwTimeWindow. These estimates on value are positive and significant when we
use apparel and when we use WPT as control groups. The point estimates of the
effect of trade diversion to the U.S. market from 2004 to 2007 is about 100%. The
effect on quantity is about 66% while the effect on price is not statistically
significant. These results are consistent with our DD estimates.
Finally we perform a series of placebo tests to check the validity of our
results. Tables 12, 13, and 14 tests for effects of the AD action on Vietnamese
Chemical Products, Plastic, and Food and Beverage imports in the U.S.. We expect
the EU action to have no effect on imports of these products. The DD estimates
for these products show no statistically significant effects on value, quantity or
price for Chemical products, and similarly for Plastic and Food and beverage.
5 Concluding Remarks
In this paper, we evaluate the effect of the 2006 EU antidumping action on
Vietnamese footwear on three markets: imports to the EU, footwear producers in
Vietnam, and imports to the U.S.. Vietnamese footwear imports to the EU fell by
as much as 66% as a result of the AD duties. This represents a big share of the
Vietnamese footwear exports as the EU is the biggest destination for this industry.
3See Bell et al. (1999) for a discussion of this trend-adjustment of the DD estimate.
32
However, we find little discernible effect on Vietnamese footwear firms in terms
of revenue, employment and payroll.
The reason could be that output from Vietnamese footwear firms are diverted
to other export markets. Consistent with predictions of our theoretical model, we
find evidence of trade diversion by Vietnamese producers from the EU to the U.S.
market. Our difference-in-difference estimates of the AD actions of the value on
Vietnamese footwear imports to the U.S. ranged from 69-71% over the period
2004-2007 and 69-72% in terms of quantity. Our results are robust to triple
difference specifications where we adjust for trend-differences and a series of
placebo specifications.
The trade diversionary effects we document in this paper suggest another
non-trivial (unintended) consequence of protectionist trade policy. While the
trade diversion toward other export suppliers have been documented, trade
diversion by the targeted producers to other markets is often overlooked in
discussions about the distortionary effects of AD. This effect is important because
it suggests that protectionist trade policy in one end market could pressure other
end markets to enact similar policies. These policy spillovers could contribute to
the proliferation of AD action and other similar measures of protection.
33
References
B. Bell, et al. (1999). ‘Getting the unemployed back to work : the role of targeted
wage subsidies’. International Tax and Public Finance 6(3):339–360.
B. A. Blonigen & T. J. Prusa (2001). ‘Antidumping’. Tech. rep., National Bureau of
Economic Research.
C. Bown & M. Crowley (2007). ‘Trade deflection and trade depression’. Journal of
International Economics 72(1):176–201.
I. Brambilla, et al. (2012). ‘Adjusting to trade policy: Evidence from US
antidumping duties on Vietnamese catfish’. The Review of Economics and
Statistics 94(1):304–319.
P. Brenton (2001). ‘Anti-dumping policies in the EU and trade diversion’.
European Journal of Political Economy 17(3):593–607.
L. Cuyvers & M. Dumont (2005). ‘EU Anti-dumping Measures against ASEAN
Countries: Impact on Trade Flows*’. Asian Economic Journal 19(3):249–271.
EC166/14 (2005). ‘European Commission Regulation (EC) No C166/14/2006’.
Official Journal of the European Union .
EC384 (1995). ‘European Commission Regulation (EC) No 384/96’. Official Journal
of the European Union .
EC461 (2004). ‘European Commission Regulation (EC) No 461/2004’. Official
Journal of the European Union .
EC553 (2006). ‘European Commission Regulation (EC) No 553/2006’. Official
Journal of the European Union .
J. M. Finger & N. T. Artis (1993). Antidumping: How it works and who gets hurt.
University of Michigan Press.
W. L. Hansen & T. J. Prusa (1995). ‘The road most taken: The rise of Title VII
protection’. The World Economy 18(2):295–313.
H. L. Kee, et al. (2013). ‘Is protectionism on the rise? Assessing national trade
policies during the crisis of 2008’. The Review of Economics and Statistics
95(1):342–346.
Y. Lu, et al. (2013). ‘How do exporters respond to antidumping investigations?’.
Journal of International Economics 91(2):290–300.
P. A. Messerlin (1989). ‘The EC antidumping regulations: A first economic
appraisal, 1980-85’. Review of World Economics 125(3):563–587.
34
T. J. Prusa (1994). ‘Pricing behavior in the presence of antidumping law’. Journal
of Economic Integration pp. 260–289.
T. J. Prusa (2003). ‘On the spread and impact of anti-dumping’. Canadian Journal of
Economics/Revue canadienne d’e’conomique 34(3):591–611.
T. J. Prusa (2005). ‘Anti-dumping: A Growing Problem in International Trade’.
The World Economy 28(5):683–700.
J. J. Schott (1994). ‘Uruguay Round: An Assessment’. Peterson Institute Press: All
Books .
35
Table 9. Difference in difference analysis for US import of Vietnamese footwear with Vietnamese Apparel as the control
group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
ddFw 0.518*** 0.519*** -0.013 0.376** 0.365* 0.004 0.527*** 0.524** 0.010
(0.14) (0.16) (0.06) (0.18) (0.19) (0.07) (0.19) (0.22) (0.07)
Constant 11.777*** 8.206*** 3.583*** 11.650*** 8.177*** 3.479*** 11.244*** 8.133*** 3.564***
(0.05) (0.05) (0.02) (0.04) (0.04) (0.02) (0.04) (0.05) (0.02)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.048 0.017 0.036 0.026 0.013 0.010 0.073 0.036 0.022
N 4,472 4,460 4,460 2,187 2,183 2,183 2,280 2,274 2,274
F 28.28 9.45 21.92 10.00 5.02 4.36 31.12 14.98 10.01
Standard errors are heteroskedastic robust. * p<0.10, ** p<0.05, *** p<0.01
36
Table 10. Difference in difference analysis for US import of Vietnamese footwear with Vietnamese WPT as the control
group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
ddFw 0.470*** 0.467*** -0.002 0.316* 0.291 0.014 0.537*** 0.545** 0.003
(0.14) (0.15) (0.05) (0.18) (0.19) (0.07) (0.19) (0.21) (0.07)
Constant 11.480*** 8.299*** 3.172*** 11.388*** 8.122*** 2.994*** 11.400*** 7.939*** 3.020***
(0.04) (0.04) (0.02) (0.03) (0.03) (0.01) (0.04) (0.04) (0.02)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.048 0.016 0.030 0.029 0.015 0.007 0.066 0.030 0.022
N 5,761 5,625 5,625 2,804 2,740 2,740 2,950 2,876 2,876
F 32.67 9.86 21.92 14.68 7.54 3.93 33.36 14.73 11.54
Standard errors are heteroskedastic robust . * p<0.10, ** p<0.05, *** p<0.01
37
Table 11. Trend-adjusted Triple Difference Analysis comparing years 2004 and 2007 with an additional control of years
2002 and 2003.
Apparel control WPT control
logValue logQuant logPrice logValue logQuant logPrice
Time 0.902*** 0.828*** 0.074** 0.833*** 0.767*** 0.070**
(0.06) (0.07) (0.03) (0.06) (0.06) (0.03)
window 1.040*** 0.844*** 0.195*** 1.006*** 0.827*** 0.167***
(0.07) (0.08) (0.04) (0.07) (0.07) (0.03)
TimeWindow -0.474*** -0.589*** 0.115** -0.404*** -0.527*** 0.120***
(0.10) (0.11) (0.05) (0.09) (0.10) (0.05)
TimeFw -0.289* -0.133 -0.125* -0.220 -0.073 -0.121*
(0.17) (0.17) (0.07) (0.17) (0.17) (0.07)
FwWindow -0.059 0.120 -0.175*** -0.025 0.137 -0.147**
(0.19) (0.20) (0.07) (0.19) (0.20) (0.07)
FwTimeWindow 0.774*** 0.649** 0.102 0.705*** 0.587** 0.098
(0.25) (0.28) (0.10) (0.25) (0.27) (0.10)
Constant 10.417*** 7.213*** 3.212*** 10.146*** 7.323*** 2.823***
(0.05) (0.05) (0.02) (0.04) (0.05) (0.02)
Product FE Yes Yes Yes Yes Yes Yes
R-squared 0.170 0.099 0.056 0.161 0.089 0.045
N 4,225 4,212 4,212 5,373 5,241 5,241
F 76.26 47.72 21.98 85.36 49.65 20.57
Standard errors are heteroskedastic robust. * p<0.10, ** p<0.05, *** p<0.01
Time takes value of 1 if year is 2003 or 2007 and 0 for 2002, 2004 .
Fw takes value of 1 if hts2 is equal to 64, otherwise it is 0.
Window takes value of 1 for 2004 or 2007 and 0 for 2002 or 2003
38
Table 12. Difference in difference analysis for US import of Vietnamese Chemical Products with Vietnamese Apparel as the
control group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
ddChemicalPr 0.138 0.257 0.041 -0.160 1.105 -0.706 0.409 0.214 0.240
(0.39) (0.51) (0.20) (0.65) (0.99) (0.50) (0.50) (0.58) (0.16)
Constant 11.406*** 7.814*** 3.602*** 11.157*** 7.874*** 3.830*** 11.520*** 7.946*** 3.585***
(0.04) (0.05) (0.02) (0.05) (0.05) (0.02) (0.04) (0.04) (0.02)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.030 0.007 0.034 0.064 0.014 0.094 0.014 0.004 0.012
N 3,890 3,868 3,868 1,996 1,984 1,984 1,894 1,884 1,884
F 14.72 3.48 17.52 23.30 4.35 35.17 5.04 1.40 5.50
Standard errors are heteroskedastic robust. * p<0.10, ** p<0.05, *** p<0.01
39
Table 13. Difference in difference analysis for US import of Vietnamese Plastic with Vietnamese Apparel as the control
group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
ddPlastic 0.432** 0.461 -0.073 0.422 0.398 -0.126 0.224 0.413 -0.133
(0.18) (0.30) (0.16) (0.28) (0.51) (0.27) (0.18) (0.31) (0.20)
Constant 11.339*** 7.872*** 3.478*** 11.046*** 7.950*** 3.676*** 11.463*** 8.129*** 3.544***
(0.04) (0.05) (0.02) (0.04) (0.05) (0.02) (0.03) (0.04) (0.02)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.037 0.009 0.031 0.076 0.013 0.077 0.017 0.008 0.008
N 4,188 4,060 4,060 2,150 2,085 2,085 2,038 1,975 1,975
F 18.89 3.71 18.48 29.67 4.22 35.13 7.27 2.75 3.33
Standard errors are heteroskedastic robust. * p<0.10, ** p<0.05, *** p<0.01
40
Table 14. Difference in difference analysis for US import of Vietnamese Foodstuffs and Beverages with Vietnamese Apparel
as the control group.
2003, 2004 / 2006, 2007 2004/2006 2004/2007
logValue logQuant logPrice logValue logQuant logPrice logValue logQuant logPrice
ddFoBe 0.152 0.197 -0.045 0.283* 0.423*** -0.140** 0.008 -0.020 0.028
(0.13) (0.12) (0.05) (0.16) (0.16) (0.06) (0.15) (0.14) (0.06)
Constant 11.379*** 8.129*** 3.251*** 11.663*** 8.196*** 3.467*** 11.694*** 8.372*** 3.322***
(0.04) (0.04) (0.02) (0.05) (0.05) (0.02) (0.04) (0.04) (0.02)
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Product FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.036 0.010 0.036 0.082 0.022 0.095 0.013 0.004 0.011
N 4,358 4,358 4,358 2,234 2,234 2,234 2,124 2,124 2,124
F 21.90 7.00 22.19 41.47 15.53 41.22 5.53 1.52 5.57
Standard errors are heteroskedastic robust. * p<0.10, ** p<0.05, *** p<0.01
41
