Anti-dumping, Intra-industry Trade and Quality

Reversals

Jose L. Moraga-Gonzalez∗

Jean-Marie Viaene†

Revision: July 19, 2012

Abstract

We examine an export game where two firms (home and foreign), located in two differentcountries, produce vertically differentiated products. The foreign firm is the most efficient interms of R&D costs of quality development and the foreign country is relatively larger and en-dowed with a relatively higher income. The unique (risk-dominant) Nash equilibrium involvesintra-industry trade and the foreign producer manufactures a good of higher quality than thedomestic firm. For low enough transport costs, this equilibrium is characterized by unilateraldumping; otherwise, reciprocal dumping emerges. We show that the implementation of an-tidumping (AD) policy can change significantly the nature of the game and give rise to variousnew Nash equilibria. For some parameters, an AD policy leads to a quality reversal in theinternational market whereby the low-quality firm becomes leader. We show that such a pol-icy is desirable for the implementing country, though world welfare decreases. The paper alsoestablishes an equivalence result between a price undertaking and an anti-dumping duty.

JEL Classification: F12, F13Keywords: anti-dumping duty, domino dumping, intra-industry trade, price undertaking,

product quality, quality reversals

∗Corresponding author: VU University Amsterdam, Department of Economics and Business Administration, DeBoelelaan 1105, 1081 HV Amsterdam, The Netherlands. E-mail: j.l.moragagonzalez@vu.nl. Moraga is also affiliatedwith the University of Groningen, the Tinbergen Institute, the CEPR, and the Public-Private Sector Research Center(IESE, Barcelona).†Erasmus University Rotterdam, Tinbergen Institute and CESifo. Email: viaene@ese.eur.nl.

1

1 Introduction

Leapfrogging is central to the catching-up process of developing and transition economies. The

underlying idea is that latecomers may be able to bypass old vintages of technology and thereby

become highly competitive. In addition, quality reversals are important for making industry leader-

ship persistent, since the higher profits accruing from high-quality production provide firms with the

financial resources needed for the continued adoption of the latest technologies and cost-reducing

investments.

The cases of leapfrogging have been diverse, depending upon industries and countries, but

there is strong evidence against the outward-oriented argument by which the market mechanism

gives enough incentives for firms in emerging markets to reach the technology frontier (Brutton,

1998). Instead, the policy issue has been to design market protection schemes to induce learning,

knowledge accumulation and quality leadership. In doing so, governments have traditionally used

an array of non-market and market instruments, including trade and industrial policy as well as

antidumping policy.

A classic example is the battle between US and Japanese semiconductor manufacturers, which

dates back to 1960s when the American producers started to expand their operations in Japan.

The Japanese government employed quotas and tariffs to slow down the penetration of the Ameri-

can companies; in addition, it subsidized heavily the development of semiconductor manufacturing

technology and the expansion of its industry worldwide (Borrus et al., 1986). In the early 1980s,

American companies saw their market shares and profits fall, making it difficult to maintain cost

and quality standards. Towards the middle of the 1980s several US firms, including Micron Tech-

nology Inc., Intel and AMD, accused their Japanese competitors of dumping in the US market and

elsewhere (Hughes et al., 1997). Anti-dumping duties and retaliatory tariffs on Japanese imports

of electronic gear helped the American firms catch up in terms of learning-by-doing and quality

standards later in the 1980s. The battle was ended by a series of pacts between the US and the

Japanese administrations.1

The Chinese newsprint industry is another example illustrating the use of anti-dumping mea-

sures as an instrument to gain time for development and acquisition of advanced equipment. In

1997 Chinese newsprint makers made dumping allegations against overseas producers. As a result,

the Chinese government started to levy anti-dumping tariffs in 1999, for a period of five years.

The anti-dumping regime promoted rapid development in the domestic newsprint industry, and

industry production increased by 40% during the period (Lu, 2004). In 2004, the Chinese Ministry

of Commerce announced that it would continue to impose anti-dumping duties on US, Canada and

1Lee and Lim (2001) also stress the important role of the government in providing market protection of fixedduration in their study of the catching-up experience of the Korean industries, another classic example of leapfrogging.

2

South Korea’s newsprint imports.

For members of the World Trade Organization (WTO), market protection via tariff barriers is

nowadays limited by countries’ commitments to bind their customs duty rates during multilateral

trade negotiations. In contrast, anti-dumping measures whose application is governed by Article

VI of GATT are not limited.2

Table 1: Antidumping initiations by reporting country

Country 1995 2001 2008 1995-20101

World 1572 371 213 3752

Traditional 63 153 44 1220Users (40.1%)3 (41.2%) (20.7%) (32.5%)

Australia 5 23 6 212Canada 11 25 3 152

EU 33 28 19 414USA 14 77 16 442

New Users 62 169 143 1822(39.5%) (45.6%) (67.1%) (48.6%)

Argentina 27 28 19 277Brazil 5 17 23 184China 0 14 14 182India 6 79 55 613

South Korea 4 4 5 111Mexico 4 6 1 98

South Africa 16 6 3 212Turkey 0 15 12 145

Rest of the 32 49 20 710world (20.4%) (13.2%) (10.0%) (18.9%)

Notes: (1) June 2010; (2) Number of new initiations; (3) Share of group of users in total initiations in a particular year or period (in percentages).

Source: Authors’ own calculations based on WTO website.

Table 1 documents the worldwide upward trend in the use of anti-dumping laws and illustrates

that, while only a few developed countries –mainly Australia, Canada, the EU and the US– were

traditional users of anti-dumping action in the old days, anti-dumping is nowadays the trade policy

of choice of developing and transition countries as well. Of the 3752 anti-dumping initiations

reported by the WTO from January 1995 to June 2010, about 48.6% are issued by the so-called

new users. In 1995 new users initiated almost the same number of cases as traditional users but

initiated 45.6% of the new investigations in peak year 2001 and 67.1% in 2008, in the midst of the

financial crisis. By June 2010, the WTO members with the most investigations include India, US,

EU, Argentina, Australia and South Africa.

Remarkably, developing and emerging economies target most of their AD initiations against

firms from more developed countries. As Table 2 indicates, the percentage of initiations aimed at

2For example, the Chinese anti-dumping case in chloroprene rubber applies a 151% anti-dumping duty against USexports (Bown, 2010). This is low compared to the US anti-dumping case in vector supercomputers that imposed arecord anti-dumping duty of 454% against NEC (Maur and Messerlin, 1999).

3

firms from countries with a larger (2008) income per capita is 99% for India, the largest initiator

of AD measures, and range between 34% and 95% for other new users.

Table 2: Share of AD initiations against exporting countries with a higher GDP per capita1

Argentina Brazil China India South Korea Mexico South Africa Turkey

0.35 0.59 0.95 0.99 0.52 0.58 0.64 0.34

Notes: (1) GDP per capita in 2008; (2) Share of cumulative AD initiations targeted at exporting countries with a higher GDP per capita than

the reporting country; for varying periods that span between 1987 and 2009.

Sources: GDP per capita from the World Bank; authors’ own calculations based on the Global Antidumping Database (Bown, 2010).

The aim of this paper is to provide a theoretical model to help explain these observations. For

this purpose, we introduce a novel model of intra-industry trade with product quality differentia-

tion. We first examine the strategic incentives of oligopolistic firms to dump exports in developing

economies. We then study the incentives of the domestic firm to file a petition for antidumping

action and, in turn, of the domestic government to impose a price undertaking (PU). Our main re-

sult is that an anti-dumping intervention can lead to a quality reversal. In that case, domestic firm

incentives are aligned with the government incentives and antidumping actions are then expected

to arise. To the best of our knowledge, this is the first paper identifying trade conditions under

which an anti-dumping action is welfare improving for the intervening country.

Bearing in mind that anti-dumping (AD) actions by a government sanction findings of dumping

by exporting firms into a particular country, a number of stylized facts have inspired our framework

of analysis:

• There is substantial empirical evidence that product quality matters, as globalization of the

international economy involves more trade with transition and developing countries. Because

of lower quality standards in these countries, local firms produce and export goods whose

quality is inferior to that of Western firms. As a result, a significant proportion of trade is

now characterized by different levels of quality. In a sample of 60 countries Hallack (2006)

shows that there are large differences in the quality of products that are exported. Using a

novel method to account for variation in trade balances induced by horizontal and vertical

differentiation, Hallack and Schott (2011) substantiate the importance of product quality by

tracing the evolution of manufacturing quality for the world’s top exporters. Greenaway et al.

(1994, 1995) show that over two thirds of all intra-industry trade in the UK involves trade of

vertically differentiated goods. From the numerous anti-dumping investigations, considerable

differences in the types of products made worldwide emerge. In particular, hearings and

public reports reveal that, besides prices, perceived quality differences are important in many

AD cases (USITC, 2001, 2002a, 2002b, 2003).

• The heterogeneity of countries involved in international trade suggests substantial differences

in consumer tastes and incomes across countries as well as cost asymmetries across firms.

4

• Petitions filed by a US industry against imports concern products which are usually classified

under 10 digit subheadings of the Harmonized Tariff Schedule of the United States. At this

level of disaggregation, sources of supply of this product in a domestic market are a few firms.

Even in large trading blocs like the US or the EU, it is common that a case concerns two

players, a local and a foreign producer. See, for example, EC (2002b, p. 25 and 48) and

USITC (2001, 2002b, 2008).

We analyze an international trade game between two firms located in two different countries

(developed vs. developing) that produce quality-differentiated products. Domestic and foreign

consumers have heterogenous preferences for quality. Markets are asymmetric in that they differ

in size and in the distribution of consumer tastes. The quality-differentiated good is supplied at

home and abroad by a local firm and by imports from the foreign producer. Markets are not totally

served in equilibrium, implying endogenous market sizes. Quality development is costly and firms

in developing countries face higher costs in order to produce a given quality level. The paper focuses

on situations where there is dumping by the high quality producing country. After the domestic

government either opts for free trade or AD policy, firms play a two-stage game.3 In the first stage,

firms select the qualities to be produced, and incur the fixed costs; in the second stage, firms engage

in an export and price competition game.

We show the existence of a unique (risk-dominant) free trade equilibrium that is characterized

by intra-industry trade. The foreign firm, which is more efficient, produces a good of higher

quality than the domestic firm. Since consumers across countries differ in their concern for quality,

unilateral dumping by the foreign firm into the domestic market arises in equilibrium. When

transport costs are sufficiently high, reciprocal dumping obtains.

Focusing on situations where dumping is unilateral, we study the effect of a PU on the trade

equilibrium and show that a PU may lead to radical changes in market structure. We first show

that a PU may result in the strategic exit of the foreign firm from the domestic market. When this

happens, we may expect the domestic firm to file a petition since, it if it were honored, this firm

would see its profits to increase. We show however that a domestic government who cares about

aggregate domestic welfare would not honor the petition.

Secondly, we show that a PU may lead to the exit of the domestic firm from the foreign market.

The rationale for doing so is that the overall level of prices in the markets would increase. Also in

this case, the domestic firm may wish to file a petition for antidumping action, which again would

not be honored by a welfarist domestic government.

3In our model the enactment and application of law are assumed simultaneous since there will be plain evidencethat the WTO dumping criteria are met. In practice, however, cases are less clear-cut, implying that more stagesin the game should be included to embed administrative costs and recognize the discrete time intervals between theenactment of law, the filing process and the application of law.

5

Finally, we show that a PU may result in a quality reversal, that is, a change in quality leadership

with the home firm becoming quality leader. In this case, imposing a PU is socially desirable for

the domestic country, though world welfare decreases. There are two such equilibria: one in which

there is intra-industry trade, another where the foreign firm producing low quality refrains from

exporting to the home market.

The rest of the paper is organized as follows. Next section discusses results of the model in light

of the related literature. Section 3 presents the details of our model. Section 4 solves for the free

trade equilibrium and establishes the conditions for dumping. The effects of a PU are examined in

Section 5. Section 6 shows that our results are not undermined if we allow the firms to choose the

level of the quality they produce. This section also establishes an equivalence result between AD

duties and price undertakings in our model. Section 7 concludes.

2 Related literature

Our paper is a contribution to the study of antidumping in oligopolistic industries. The litera-

ture on antidumping is extensive and the reader is referred to Feenstra (2003) and the survey of

Blonigen and Prusa (2003) for a detailed discussion. With the exception of Vandenbussche and

Wauthy (2001), the literature has dedicated little attention to the role of quality differences in the

determination of dumping and the desirability of AD policy. As we will show, this is an important

omission since AD policies modify firms’ incentives to select product quality which, in turn, affect

the extent of competition in the international market and hence firms’ profits and social welfare.

In some situations, we show that AD legislation might even lead to a quality reversal in the in-

ternational market. Various papers (see e.g. Ethier and Fisher, 1987; Fisher, 1992; Leidy and

Hoekman, 1990; Reitzes, 1993) have examined how AD protection gives firms incentives to alter

their price or output decisions vis-a-vis free trade in order to influence the AD outcome. This may

lead to higher or lower welfare depending on the existing market structure. Anderson et al. (1995)

examine a variant of the reciprocal dumping model of Brander and Krugman (1983) where two

governments can enact AD law or not. They find that welfare-maximizing governments impose

no law in equilibrium. Moreover, though an individual firm has an incentive to lobby for AD law,

consumer welfare increases and firm profits fall if laws are bilaterally enacted. Another branch in

the literature studies how AD policy influences the incentives of firms to collude (see e.g. Staiger

and Wolak, 1992; Veugelers and Vandenbussche, 1999).

The distinctive feature of our paper is the study of the effects of antidumping legislation in

an international market where firms produce vertically differentiated products. In this regard,

Vandenbussche and Wauthy (2001) is the paper most closely related to ours. They study a game

of one-way trade between a domestic and a foreign firm. Using the “lay” definition of dumping

6

(Weinstein, 1992) where the competing local price is used as a proxy for the “normal” value of

the good, they show that, relative to free trade, a PU gives the foreign firm incentives to be more

aggressive and become the quality leader in the domestic market. AD law leads in this case to

lower social welfare for the home country. In contrast, we use a model of intra-industry trade and

therefore allow for the more standard definition of dumping, the one put forward by the WTO (see

the WTO website).

Our paper is also related to the work explaining how product quality matters in international

trade. The monopoly problem is discussed in Mussa and Rosen (1978) and Krishna (1987), and

oligopoly versions of this model have received substantial attention in the international trade lit-

erature. More related to our model, Herguera et al. (2002) study optimal trade policy in a model

of one-way trade and Motta et al. (1997) analyze the sustainability of quality leadership once

countries open up to trade. Moraga-Gonzalez and Viaene (2005) study trade and industrial policy

in transition economies. Saggi and Sara (2008), extending the work of Horn (2006), analyze the

WTO’s national treatment clause in a two-country model where quality of goods and/or market

size are heterogeneous across countries. Our paper contributes to this work on quality leadership by

showing that the export game has a unique free-trade equilibrium characterized by intra-industry

trade and that firms’ cost asymmetries are crucial to sustain quality leadership.

Vandenbussche and Zanardi (2008) identify factors that explain why countries adopt and use

AD laws. Among the variables that are considered as potential drivers of the emergence of AD

laws, their empirical results suggest that past trade liberalization is a contributor to the likelihood

of a country’s adoption of AD law. Moreover, retaliatory motives are at the heart of proliferation.

Against this background which depicts a certain misuse of AD law by governments, this paper in-

troduces a new interpretation of domino dumping which is theoretically different from the standard

retaliatory motive associated to AD law enactment but which is empirically hard to distinguish

in the data. Once the home government enacts AD law, foreign rival firms which suffer from an-

tidumping enforcement, have incentives to react strategically by exiting the home market. As a

result of decreased competitive pressures in the home market, local firms which charge higher prices

may dump their products abroad and lead the foreign government to take antidumping actions.4

Though AD measures may have small effects on global trade, these trade effects may be large

for some AD-imposing countries. For example, Vandenbussche and Zanardi (2010) test the null

hypothesis of no-trade effect of AD measures at the aggregate level using a gravity model of bilateral

trade connecting new users of AD laws and their trade partners. Estimates for several countries

4A domino effect in trade policy arises when a protectionist measure in one country leads to another. For example,in Anderson (1992, 1993) domino dumping arises when exporting firms dump to obtain more export licences in thefuture when there is a positive probability of a future voluntary export restraint. Increased enforcement may noteliminate dumping and may even lead to a rise in dumping, toppling another domino. In our setting, though adomino effect may arise, antidumping enforcement is effective in the first place.

7

suggest chilling trade effects that are sufficiently large to offset the gain in trade volumes obtained

from various rounds of trade liberalization. In our setting, we show that while some Nash equilibria

are indeed characterized by a reduction in trade volumes, other equilibria with intra-industry trade

do not show such a pattern.

3 The Model

Two firms sell goods that are vertically differentiated. These two firms, located in two different

countries, home and foreign, produce goods for their own market and, eventually, for exports. The

firm located in the foreign (home) country is referred to as the foreign (home) firm and all foreign

variables are denoted by an asterisk “∗”. We index destination countries by i = 1, 2 where subscript

1 refers to the home country and subscript 2 to the foreign country. We assume there are transport

costs, denoted by r, associated to moving goods from country to country. For convenience, we

suppose transport costs are of the “iceberg” type.

Product quality may be one of two types: high-quality qh and low-quality q`, with qh > q`.5

We adopt the cost specification of pure vertical product differentiation models, where the costs of

quality mainly fall on fixed costs and involve only a small or no increase in unit variable costs (see

Shaked and Sutton, 1982, 1983). In particular, once the home (foreign) firm picks the quality of the

goods to be offered, it pays a fixed cost C(q) = cq2/2, q = {qh, q`} (C∗(q) = c∗q2/2) and produces

at a marginal cost which is normalized to zero.6 Fixed cost asymmetries across firms in different

countries may capture differences in the available production technologies as well as in the costs

of labor and capital. They are important in this paper because they help us pin down a unique

equilibrium in qualities. As we will see later, the size of these costs is not relevant and can then be

taken to be arbitrarily small, in particular, we let c→ 0, and c∗ → 0, with c∗/c = γ ∈ (0, 1).7

Assume there is a population of measure m (1) at the home (foreign) country, with 0 < m ≤ 1.

Consumers buy at most one unit and have preferences given by the following quasi-linear (indirect)

utility function: U = θq − p, if a unit of a good of quality q is bought at price p, and 0 otherwise.

Parameter θ is consumer specific and measures the utility a consumer derives from consuming a

unit of quality. Assume that θ is uniformly distributed over [0, λθ] at home, and over [0, θ] abroad,

with 0 < λ ≤ 1, θ > 0. Tirole (1988) shows that θ is the inverse of the marginal utility of income

so our assumption λ ≤ 1 implies that foreign consumers have higher incomes on average and

more sophisticated tastes. Our specification of demand thus captures income differences between

5In Section 6, we allow firms to choose quality from a continuum.6This normalization is without loss of generality provided that the main bulk of costs falls on fixed costs rather

than on variable costs. Adding small marginal costs of production makes computations cumbersome and obscuresthe presentation of the results substantially without adding further insights.

7The specification of the cost function could be more general without affecting results qualitatively. For example,Moraga-Gonzalez and Viaene (2005) use cost functions with a degree of homogeneity k ≥ 2. While larger k valuesaffect results quantitatively, they do not alter them qualitatively.

8

countries via λ and size differences between countries via parameter m. The assumption that λ is

less than 1 serves to focus the model on the interaction between firms located in developing and

developed countries.8

Let p` and ph be the prices for low- and high-quality products in the domestic country. Suppose,

for a moment, that ph > p`, i.e., high quality is sold at a higher price, an assumption that will be

verified later. Firms’ demand functions are obtained as follows. There is a consumer, denoted by

θ, who is indifferent between buying high quality or low quality. Using the utility function, it is

readily seen that θ = (ph − p`) /(qh − q`). Likewise, let θ denote the consumer indifferent between

acquiring the low-quality good or nothing at all, that is, θ = p`/q`. Then, the high-quality good

is demanded by those consumers such that θ ≤ θ ≤ θ and the low-quality variant is demanded by

those buyers such that θ ≤ θ < θ. As θ is uniformly distributed on [0, λθ], home demands for high-

and low-quality goods are:

Dl(.) = m

(ph − p`

λθ(qh − q`)− p`

λθq`

), Dh(.) = m

(1− ph − p`

λθ(qh − q`)

)(1)

Note that one of these demands is met by imports from the foreign firm. Proceeding in the same

way we can compute the foreign demands for high- and low-quality goods:

D∗` (.) =p∗h − p∗`θ(qh − q`)

−p∗`θq`

, D∗h(.) = 1−p∗h − p∗`θ(qh − q`)

, (2)

where p∗` and p∗h are foreign prices of low- and high-quality, p∗h > p∗` . As before, one of these

demands is met by imports from the home firm.

4 Free trade equilibrium

The assumptions of our model depict a situation in which a domestic firm, located in a smaller

and poorer country, considers, besides supplying its own market, to export to a larger and richer

country. It faces competition from a foreign firm, which is more efficient. Given this, the following

three questions arise:

• What is the pattern of trade that emerges in equilibrium once both countries open up to

trade?

• What are the product qualities that are produced by each firm in equilibrium?

• For which of the two firms, if any, is it optimal to dump its good in the international market?

We address these three issues in this section and focus on the market equilibrium under free

trade. Since there are two quality levels available for choice, qh and q`, this section studies an

8This sets the paper apart from Vandenbussche and Wauthy (2001), who study dumping into a developed country.

9

export game where the domestic and foreign firms can freely pick which of the two qualities to offer

and which prices to set in the two different countries.9

Risk-dominant Nash equilibrium

There are four relevant continuation games,10 each of them ensuing after the firms have picked

a quality-and-export strategy profile. The quality-and-export strategy set of a firm i is Si =

{{q`&E}, {qh&E}} for i = 1, 2 where E stands for export and means that the firm will export its

good to the other market. In what follows we compute the price equilibrium corresponding to each

quality-and-export strategy profile.

Suppose first that the two firms choose to produce different quality levels. Firms choose their

prices to maximize their world revenues. The firm selling high quality has revenues equal to

ph(1−r)Dh+p∗hD∗h while the corresponding revenues for the low-quality firm are p`D`+p

∗` (1−r)D∗` .

Taking the first order conditions (FOCs) and solving for the Nash equilibrium yields the following

prices in the domestic country:

p` =θλ(qh − q`)q`

4qh − q`, ph =

2θλqh(qh − q`)4qh − q`

In the foreign country we have p∗` = p`/λ, p∗h = ph/λ.

Suppose now that firms choose to produce the same quality levels. Since firms compete in a

Bertrand fashion, it is clear that competition will drive prices down to zero thereby eroding all

variable profits of the firms.

Given this information, we can now fold the game backwards and write down in Table 4 the

normal-form game at the choice-of-quality and export stage.

Table 3: Quality strategy profiles under free trade

Foreign firm

Home firm

qh&E q`&E

q`&E R` −cq2

`2 , Rh −

c∗q2h

2 − cq2`2 ,−

c∗q2`

2

qh&E − cq2h2 ,−

c∗q2h

2 Rh −cq2

h2 , R` −

c∗q2`

2

where R` and Rh, which are given by:

R` =θqh(qh − q`)q`(1 +mλ− r)

(4qh − q`)2;Rh =

4θq2h(qh − q`)(1 + λm(1− r))(4qh − q`)2

(3)

denote the revenues of a low- and a high-quality firm under free trade, respectively.

9To be sure, quality choices are endogenous here, though the quality levels are exogenously fixed. In Section 6 weexamine the case in which quality levels are also endogenous.

10Strategies for which a firm refrains from exporting are strictly dominated under free trade so they can be ignoredin this section. Later in Section 5, we show that, once AD action is in place, either the home or the foreign firmmay strategically choose to exit the export market, thereby relaxing price competition and gaining a competitiveadvantage.

10

Assuming product differentiation gives sufficient revenues to cover the (small) fixed costs, this

game has two Nash equilibria. In one equilibrium, {q`&E, qh&E}, the domestic firm produces the

low-quality product and in the alternative equilibrium, {qh&E, q`&E}, it is the home firm that

produces high quality. As discussed in detail in Motta et al. (1997) and in Moraga-Gonzalez

and Viaene (2005), selection of equilibria in this type of games requires to use the risk-dominance

criterion of Harsany-Selten. Applying this criterion to our game, we now show that as long as

firms’ costs differ, irrespective of whether they are large or small, one of the equilibria can readily

be refined away.11

Let G11 = R` + c2(q2h − q2` ) be the gains the domestic firm obtains by predicting correctly

that the foreign firm will select equilibrium 1, {q`&E, qh&E}. Likewise, G12 = Rh − c2(q2h − q2` )

denotes the gains the domestic firm derives by forecasting correctly that the foreign firm will select

equilibrium 2, {qh&E, q`&E}. Similarly, for the foreign firm we have G21 = Rh − c∗

2 (q2h − q2` )

and G22 = Rl + c∗

2 (q2h − q2` ). It is said that equilibrium 1 risk-dominates equilibrium 2 whenever

G11G21 > G12G22. It is straightforward to verify that this inequality holds whenever γ < 1. In

fact, we need to show that:(R` +

c

2(q2h − q2` )

)(Rh −

c∗

2(q2h − q2` )

)>(R` −

c

2(q2h − q2` )

)(Rl +

c∗

2(q2h − q2` )

)Define Z = c

2(q2h − q2` ). Then:

(R` + Z) (Rh − γZ) > (Rh − Z) (Rl + γZ) ⇐⇒ Z(Rh +Rl)(1− γ) > 0

The next proposition characterizes the assignment of qualities across firms under free trade.

Proposition 1 (Quality leadership) In the unique risk-dominant free trade equilibrium, the

foreign firm produces high quality and the domestic firm chooses low quality.

The quality leader is thus the most efficient firm in terms of development costs.12 This important

result indicates that the sustainability of quality leadership under free trade hinges on firms’ cost

asymmetries only and not on market size or income differences.

To assess welfare effects of antidumping action, we shall use the weighted sum of consumer

surplus and firms’ payoffs as a measure of social welfare in every country:

W = β1CS + β2R, βj ∈ {0, 1}, j = 1, 2 (4)

11See also Zhou et al. (2002), who show that there is only one Nash equilibrium provided cost differences acrossfirms are sufficiently large.

12In our paper quality leadership refers to the fact that one firm produces higher quality than the other and suchleadership is obtained via cost differences. By contrast, in Aoki and Prusa (1996) leadership refers to Stackelbergleadership.

11

With exogenous levels of quality, firms’ payoffs are equal to revenues. In Section 6 where we allow

firms to choose the level of the quality of their products, development costs are included as part

of welfare calculations. Weights β1 and β2 characterize policymaker’s preferences. If β1 = β2 = 1

we have the usual (neutral) definition of social welfare; if β1 = 1 and β2 = 0, the government cares

only about consumer surplus; finally when β1 = 0 and β2 = 1, only firm’s revenues matter for the

policy maker.13 Domestic consumer surplus is given by:

CS = m

∫ λθ

ph−p`qh−q`

(xqh − ph)dF (θ) +m

∫ ph−p`qh−q`

p`/q`

(xq` − p`)dF (θ)

where F (θ) is the cumulative distribution function of θ.

Let us define µ = qh/q`, with µ > 1 since qh > q`. Variable µ represents the quality gap

between firms’ variants and measures the degree of product differentiation. The next proposition

characterizes the free trade equilibrium:

Proposition 2 (Free trade equilibrium) There is a unique risk-dominant free trade equilibrium

of the export game. This equilibrium involves intra-industry trade and is characterized as

follows: (i) The domestic firm picks q` and the foreign firm picks qh; (ii) These goods sell at

prices:

p` =θλ(µ− 1)q`

4µ− 1, ph =

2θλqh(µ− 1)

4µ− 1(5)

p∗` = p`/λ, p∗h = ph/λ. (6)

in the two countries. (iii) Market demands for each firm are:

D` = mµ

4µ− 1; D∗` =

µ

4µ− 1;Dl +D∗l =

(1 +m)µ

(4µ− 1)(7)

and Dh = 2D`; D∗h = 2D∗` ; (iv) Firms’ revenues are given by (3) with R` and Rh being

domestic and foreign firm’s revenue respectively. The ratio of revenues is:

RhRl

= 4µ

[1 + λm(1− r)

1 + λm− r

](8)

(v) Consumer surplus in each country is:

CSFT =λmθqhµ(4µ+ 5)

2(4µ− 1)2, CS∗FT =

θqhµ(4µ+ 5)

2(4µ− 1)2= CSFT /λm (9)

13Note that AD actions seem to be in favor of firms. Dominant players have good lobbying power and consequentlythey find it easier to get governmental support. In contrast, there are practical problems in dealing with consumerinterests as it requires educating consumers about the role of antidumping.

12

(vi) Assuming β1 = β2 = 1, social welfare is:

WFT =θqh [2(µ− 1)(1− r) + λm(4µ− 1)(µ+ 2)]

2(4µ− 1)2,

W ∗FT =θqhµ(12µ− 3 + 8mλ(1− r)(µ− 1))

2(4µ− 1)2

(vii) The Grubel-Lloyd measure of two-way trade (in value) is:

GL = 100

{1−|p∗l (1− r)D∗l − p∗hD∗h|p∗l (1− r)D∗l + p∗hD

∗h

}= 100

{1− |(1− r)− 4λmµ|

1− r + 4λmµ

}(10)

This proposition describes the benchmark of our analysis. For any µ, demands are positive so

the market equilibrium involves intra-industry trade in vertically differentiated products, like in

the contributions of Falvey (1981) and Shaked and Sutton (1984). The Grubel-Lloyd index (10)

depends on the primitive parameters of the model: relative country size m, relative taste difference

λ, transport cost r and the gap between qualities qh and q`. Besides being a measure of product

differentiation, µ relates to the extent of price competition between firms since, by taking the ratio

of prices in (5) and (6), we obtain ph/pl = p∗h/p∗l = 2µ. The importance of producing high quality

is further stressed by the ratio of revenues in (8), which is equal to 4µ when r = 0. This is partly

explained by the discrepancy in hedonic prices, the hedonic price of high quality being twice that

of low quality (ph/qh = 2pl/ql).

Occurrence of dumping

Let us check whether dumping arises in this free trade equilibrium. The intra-industry trade

nature of the trade equilibrium is useful since it allows us to apply the traditional WTO definition of

dumping. The latter relies on three principles. First, a local firm may petition the government for

relief if dumped imports materially injure the competing domestic firm. In the market equilibrium

just described, domestic and foreign demands are such that the foreign firm’s market shareDh/(Dl+

Dh) is two thirds of the domestic market (see (7)). It is clear that this proportion is large enough

to justify injury. A second issue to substantiate a case for antidumping action is that the products

involved must be ’like’ products. The ’likeness’ means alike in all respects, or, alternatively, having

characteristics closely resembling those of the product under consideration, like in our model. Third,

a product is to be considered as being dumped if its export price to a particular country is less

than a ’normal’ value, standard used by the WTO, or less than a “fair value,” standard used by

the US government. There are different ways of calculating a product’s “normal” or “fair” value.

The standard definition of dumping, the one put forward by the WTO (see the WTO website), is

when the fob export price to a particular country is less than the price the firm normally charges

in its own market. In our framework, this definition of dumping amounts to comparing p` with

13

p∗` (1− r), and p∗h with ph(1− r). This yields the following result:14

Proposition 3 (Dumping) Under free trade high-quality products are always dumped; when λ >

1− r both types of product are dumped and dumping is reciprocal.

Our result above is in line with traditional treatments of dumping in that it shows the possibility

of reciprocal dumping based on transportation costs (Anderson et al. 1995; Brander and Krugman,

1983; Weinstein, 1992). However, dumping is unilateral when transportation costs are small or

zero. Even with positive transportation costs, only the high-quality product may be dumped but it

all depends on the cross-country difference in the distribution of tastes. A popular belief is that low-

quality goods are dumped in export markets because they usually command a lower price. In our

setting, the reverse occurs more often: it is the high-quality good that is dumped into the smaller

and poorer country. Dumping of high-quality products arises because cross-country differences in

the distribution of tastes provide the foreign firm with incentives to cut its export price relative to

the price it charges in its own market. The intuition is simple: higher income, here translated into

more sophisticated tastes, gives rise to higher prices in the foreign market for all goods.

5 Antidumping policy

Two AD policy instruments are commonly used by governments, namely, price undertakings and

antidumping duties. The former is more commonly used in the EU while the latter is observed

more frequently in the US. A PU is a binding commitment to raise export prices so that either the

dumping or the injury suffered from dumped imports by the domestic country is eliminated (GATT,

1991, p. 74). An antidumping duty equalizes the price that consumers in different countries pay

for the same good by means of a duty. The analysis that follows examines price undertakings,

antidumping duties being studied later in Section 6 where an equivalence result between these two

AD instruments is derived.

5.1 Quality and export strategies of firms

What are domestic government’s incentives to intervene? The answer to this question is obtained

by comparing domestic welfare attained under free trade with that under a PU. As the focus of

our analysis is on the new users of AD policy (cf. Table 1), we limit the number of issues by

considering only the case where high-quality products are dumped in the free trade equilibrium.

This corresponds to the parameter space where λ < 1 − r. For any non-negative value of r, we

14Another possibility is to consider the competing local price as a proxy for the normal value. Though this ischaracterized as the “lay” definition of dumping (Weinstein, 1992) the method has been used in theory (Vandenbusscheand Wauthy, 2001) and applied in a number of cases, among others in Mexico (Niels, 2004). This is a definition oflast resort, useful only when other methods of calculations are not possible.

14

can appropriately choose λ so that condition λ < 1 − r is satisfied. To make things simple, let us

assume r = 0 in which case λ can vary freely between 0 and 1.15

With high-quality products being dumped, a PU constrains the foreign firm to set an export

price that is equal to its local price, that is:

ph = p∗h (11)

We note that this constraint gives rise to an array of strategic responses by the firms at play.

For example, as we will show below, for some parameter values, the foreign firm prefers to exit

the home market. The question is then whether it prefers to continue to produce high quality or

instead prefers to deviate by producing low quality. In the latter case, we say that a quality reversal

occurs. For alternative parameter values, it is the domestic firm that exits the foreign market since

by doing so it raises prices in all markets, which ultimately benefits it. Therefore, to find out when

it is optimal for the domestic government to impose a PU, we need to solve the continuation game

played by the firms once the PU is in place. We solve the game in what follows.

Let E and NE denote an exporting and non-exporting strategy respectively. The strategy set

of a firm i is

Si = {{qh&E}, {qh&NE}, {q`&E}, {q`&NE}}, i = 1, 2

and the game played by the two firms is summarized in the following table:

Table 4: Quality and export strategy profiles of the two firms

Foreign firm

Home firm

qh&E qh&NE q`&E q`&NE

q`&E R(s11), R∗(s11) R(s12), R∗(s12) 0, 0 θλmq`4 , 0

q`&NE R(s21), R∗(s21) θλmq`4 , θqh4 0, θq`4

θλmq`4 , θq`4

qh&E 0, 0 θλmqh4 , 0 R(s33), R∗(s33) R(s34), R∗(s34)

qh&NE 0, θqh4θλmqh

4 , θqh4 R(s43), R∗(s43) θλmqh4 , θq`4

In each cell, R (R∗) is a shorthand notation for expressions representing the international rev-

enues of the home (foreign) firm under different strategies. Other cells report the simpler expressions

for either monopoly profits when one firm selects itself out of the export market, or for zero duopoly

profits when firms choose to produce the same quality level. To shorten the expressions, we do not

write the corresponding fixed costs of quality.

15An array of interesting situations arise when both governments decide to enact an AD policy. With r 6= 0, andreciprocal dumping, both governments may have an incentive to take antidumping actions. However, governmentshave opposite incentives in that actions by the home government are aimed at helping the home firm to gain qualityleadership while the foreign government wants to counter that.

15

Table 5: Trade properties of firm strategies

Concepts Strategies

Quality reversal s33, s34, s43, s44

Intra-industry trade s11, s33

Trade chilling effects s12, s21, s22, s24, s34, s42, s43, s44

Trade promotion s13, s31

Ambiguous trade effects s14, s23, s32, s41

In terms of their impact on international trade, as Table 5 indicates, only cells s11 and s33 allow

for intra-industry trade. Trade chilling effects arise when, for strategic reasons, one or two firms

decide not to export. Extreme chilling effects are observed in cells (s22, s24, s42, s44), where both

firms refrain from exporting and trade is zero. Trade promotion can take place as well, when firms

produce similar qualities and compete in a Bertrand fashion. Finally, some firm strategies have no

clear cut effect on trade with one flow increasing, the other decreasing.

Revenues in each cell of Table 4 are obtained from derivations similar to those for the free trade

equilibrium while imposing the price constraint (11). Let us start by looking closer at the pair of

strategies s11 = {q`&E, qh&E}, payoffs from other strategy pairs being obtained in a similar way.

Under strategy profile s11, the foreign firm produces and exports high quality while the domestic

firm produces and exports low quality. This is basically the same as under free trade but, because a

PU is in place, the foreign firm must set an export price that is equal to its local price to maximize

phDh+p∗hD∗h; the domestic firm has revenues equal to p`D`+p∗`D

∗` , where D`, Dh, D

∗` , D

∗h are given

in (1) and (2). Solving for the new Nash equilibrium in prices, we get:

p`(s11) =

θλ(1 +m)q`(µ− 1)

(m+ λ)(4µ− 1), p∗` (s

11) = p`(s11) (12)

p∗h(s11) =2θλ(1 +m)qh(µ− 1)

(m+ λ)(4µ− 1), ph(s11) = p∗h(s11) (13)

Demands become:

D`(s11) =

m(1 +m)µ

(m+ λ)(4µ− 1);D∗` (s

11) =(1 +m)λµ

(m+ λ)(4µ− 1)

D`(s11) +D∗` (s

11) =(1 +m)µ

(4µ− 1)(14)

Dh(s11) =2mµ

(4µ− 1)− m(1− λ)(2µ− 1)

(m+ λ)(4µ− 1);D∗h(s11) =

2µ

(4µ− 1)+m(1− λ)(2µ− 1)

(m+ λ)(4µ− 1)

Dh(s11) +D∗h(s11) =2(1 +m)µ

(4µ− 1)(15)

The revenues obtained by the firms in this case are:

R(s11) =θλ(1 +m)2µ(µ− 1)q`

(m+ λ)(4µ− 1)2; R∗(s11) =

4θλ(1 +m)2µ(µ− 1)qh(m+ λ)(4µ− 1)2

(16)

16

Comparing demand expressions (7) to (14) and (15), it is clear that trade volumes are not

modified by the imposition of a PU: world demand for low-quality products (D`+D∗` ) and for high-

quality products (Dh + D∗h) are similar under both free trade and PU. However, the distribution

of quantities changes: domestic firm’s local sales increase while its exports decrease by the same

volume. Likewise, exports of the foreign firm decrease while its local sales increase by the same

volume.16

Note that all demands are strictly positive except for possibly Dh(s11). The fact that Dh(s11)

can become zero for certain parameter configurations is of paramount importance in our setting,

since it means that a PU gives incentives to the high-quality foreign firm to exit the export market.

Therefore, suppose that firms play strategy s12 = {q`&E, qh&NE} instead, i.e., the foreign firm

exits the home market. The home market would be monopolized by the domestic firm and there

would only be competition in the foreign market. Equilibrium prices in the foreign market in this

situation are:

p∗` (s12) =

θ(µ− 1)q`(4µ− 1)

; pl(s12) =

θλq`2

; p∗h(s12) =2θqh(µ− 1)

(4µ− 1)

The revenues of the firms from the foreign market in that case turn out to be:

R(s12) =1

4mλθq` +

θµ(µ− 1)q`(4µ− 1)2

; R∗(s12) =4θq`µ

2(µ− 1)

(4µ− 1)2(17)

Consider now the pair of strategies s21 = {q`&NE, qh&E}. In this case the foreign firm must

maximize phDh+p∗hD∗h under the constraint that ph = p∗h. Likewise, the domestic firm has revenues

equal to p`D`, where D`, Dh and D∗h are given in (1) and (2). Taking the FOCs and solving for the

equilibrium prices yields:

p`(s21) =

θλ(1 +m)q`(µ− 1)

(m+ λ)(4µ− 1)− 3λ; p∗h(s21) =

2θλ(1 +m)qh(µ− 1)

(m+ λ)(4µ− 1)− 3λ= ph(s21)

Interestingly, the price of the foreign producer goes up and by implication so does the price of the

domestic producer. This implies that, at least potentially, the domestic firm can increase its profit

by exiting the foreign market. That this did not occur under free trade is due to the constraint the

foreign firm faces when maximizing profits. The revenues of the firms are given by:

R(s21) =θλm(1 +m)2µ(µ− 1)q`

[(m+ λ)(4µ− 1)− 3λ]2; R∗(s21) =

4θλ(1 +m)2qh(µ− 1) [(m+ λ)µ− λ]

[(m+ λ)(4µ− 1)− 3λ]2(18)

Suppose firms play strategy s33 = {qh&E, qh&E}, that is, there is a quality reversal. Both firms

16Notice that low-quality products would be dumped in this equilibrium if r > 0 since pl > (1− r)p∗l = (1− r)2pl.This observation reproduces again the notion of domino dumping: once a country enacts antidumping law, the rivalforeign firm that suffers from dumping may give an incentive to its own government to also take an antidumpingaction.

17

export to the distant markets. Prices are given by:

ph(s33) =θλq`(µ− 1)(4µ(λ+m) + (1− λ))

2(4µ− 1)(m+ λ), p∗h(s33) =

θq`(µ− 1)(4µ(λ+m)−m(1− λ))

2(4µ− 1)(m+ λ)

p∗` (s33) =

θλ(1 +m)(µ− 1)q`(4µ− 1)(m+ λ)

= p`(s33)

and the revenues of the firms are:

R(s33) =θλmq`(µ− 1) [4µ(m+ λ) + (1− λ)]2

4(m+ λ)2(4µ− 1)2+θq`(µ− 1) [4µ(m+ λ)−m(1− λ)]2

4(m+ λ)2(4µ− 1)2;

R∗(s33) =θλ(1 +m)2µ(µ− 1)q`

(m+ λ)(4µ− 1)2(19)

Suppose now firms play strategy s43 = {qh&NE, q`&E}. Prices would be given by:

ph(s43) =θλq`(µ− 1) [2(m+ λ)µ+ (1− 2λ)]

(m+ λ)(4µ− 1)− 3λ, p∗` (s

43) =θλ(2 +m)(µ− 1)q`

(m+ λ)(4µ− 1)− 3λ

and the revenue expressions by:

R(s43) =θλm(µ− 1)q` [2(m+ λ)µ+ (1− 2λ)]2

[(m+ λ)(4µ− 1)− 3λ]2; R∗(s43) =

θλ(2 +m)2q`(µ− 1) [(m+ λ)µ− λ]

[(m+ λ)(4µ− 1)− 3λ]2

Suppose finally that firms play the strategy s34 = {qh&E, q`&NE}. Prices would be given by

p∗` (s34) =

θ(µ− 1)q`4µ− 1

; p∗h(s34) =2θqh(µ− 1)

4µ− 1

and the revenues of the firms by:

R(s34) =1

4mλθq`µ+

4θµ2(µ− 1)q`(4µ− 1)2

; R∗(s34) =θq`µ(µ− 1)

(4µ− 1)2(20)

5.2 Nash equilibria

Having computed the profits of the firms under the different strategy profiles, we are now in a

position to analyze the game described in Table 4. The objective is to search for all Nash equilibria,

study the conditions for their existence and explore their trade properties. We note the first main

contribution of our paper now: imposing a PU dramatically changes the nature of the game, which

gives rise to new and interesting equilibria. In particular, we emphasize that imposing a PU may

either lead to a quality reversal, or to the exit of the foreign firm from the home market, or to the

exit of the home firm from the foreign market. The second main contribution is to show that when

the unique equilibrium that ensues after the domestic government imposes a PU involves a quality

reversal, then imposing a PU is indeed consistent with domestic welfare maximization.

18

5.2.1 Unchanged market structure

Let us start with situations in which the market structure prevailing under free trade is maintained

after imposing a PU. To verify whether the pair of strategies s11 = {q`&E, qh&E} continues to

be a Nash equilibrium after the policy, it is sufficient to compare R∗(s11) with R∗(s12) and R(s11)

with R(s21) since payoffs from other strategies are strictly dominated. Using (16), (17) and (18),

comparisons lead to the following:

R∗(s11) > R∗(s12) if and only if λ >1

2 +m

R(s11) > R(s21) if and only if (λ+m)(4µ− 1)(4µ− 7) + 9λ > 0.

When these two conditions are met, an equilibrium where firms play s11 exists. In order to

check whether the domestic country indeed wishes impose a PU, we compute the difference between

domestic welfare in this equilibrium, denoted W (s11), and that under free trade, denoted W (FT ).

The comparison gives:

2(m+ λ)2(4µ− 1)2

θ(1− λ)mµq`(µ− 1)

[W (s11)−W (FT )

]= −2m− λ [4(2m− 1)µ+ 5]− 3λ2(4µ− 1),

which is clearly negative because the first and third terms are negative and for the second term we

can prove that 4(2m− 1)µ+ 5 > 4(2m− 1) + 5 = 8m− 4 + 5 > 0.

The next proposition formalizes this result and the trade properties of this equilibrium.

Proposition 4 (Unchanged market structure). Suppose the parameters of the model satisfy

conditions:

λ− 1

2 +m> 0 (21)

(λ+m)(4µ− 1)(4µ− 7) + 9λ > 0 (22)

Then if the domestic government were to introduce a PU, the foreign (home) firm would continue

to produce high (low) quality and export it to the home (foreign) country. A PU would then result

in: (i) an increase in the (hedonic) prices of both variants in the domestic country; (ii) a decrease

in (hedonic) prices of both variants in the foreign country; (iii) a decrease in profits of both firms,

and (iv) a decrease in domestic consumer surplus. As a result, free trade is an equilibrium of the

game and high-quality products continue to be dumped into the domestic economy.

The PU works so as to level off the two markets and as a result domestic prices go up, foreign

prices go down, and firms profits decrease. Because of the price increase in the domestic market,

domestic consumer surplus decreases. These two facts together (profits and consumer surplus

decrease) explain the welfare result.

Notice that µ ≥ 7/4 suffices for (22) to be satisfied. When conditions (21) and (22) hold, the

19

domestic market continues to be attractive for the foreign firm, despite being constrained by the

domestic country’s policy. The foreign firm continues to export to the domestic country and the

equilibrium continues to exhibit intra-industry trade. It turns out that under such circumstances

it is in the domestic government interest to permit dumping and not to impose a PU.

Proposition 3 would point to there being a lot of antidumping cases in developing countries

against higher-quality products from the rest of the world. This is simply not the case in the data:

though the number of AD initiations is large, imports concerned represent a small fraction of trade

of developing countries. Proposition 4 offers an explanation for this outcome: an AD law would

simply be welfare reducing, which makes the tolerance of dumping socially desirable. The criterion

that the import product must be a “like-product” to the domestic petitioner’s product is another

very good reason for this. If product quality is different enough, as (22) suggests, it may be hard

to make the case of like-products.

5.2.2 Strategic exit of the foreign firm

Suppose that

λ <1

2 +m(23)

so that condition (21) fails and the pair of strategies s11 = {q`&E, qh&E} can no longer constitute

an equilibrium because if the home firm were to produce the low-quality good and export it abroad,

then the foreign firm would prefer to deviate by exiting the home market altogether in order to

no longer be subject to the PU. Given this, we next ask whether the pair of strategies s12 =

{q`&E, qh&NE} can constitute a Nash equilibrium.

We start by noting that, given that the foreign firm produces high quality and does not export

at all, the home firm faces a trade-off. On the one hand, it can choose to produce the low-quality

good and export it abroad, thereby obtaining the monopoly profits from the home market plus

the duopoly profits from the foreign market. On the other hand, it can choose to produce the

high-quality good, which would provide it with the corresponding monopoly profits at home only,

irrespective of whether it exports it abroad. The trade-off arises because the monopoly profits from

selling high quality in one market only can be higher than the profits obtained from selling low

quality in two markets. We note that s12 is a Nash equilibrium if the latter strategy provides the

home firm with a lower profit, that is, when R(s12) > θλmqh/4, which, after rearranging terms

gives:

λ <4µ

m(4µ− 1)2. (24)

This condition tells us that when the home country has a relatively low income and/or size, profits

from the foreign market are sufficiently large and the home firm prefers to produce low quality.

When both conditions (23) and (24) are met, an equilibrium where firms play s12 after the

20

domestic country imposes a PU exists. As before, in order to check whether imposing a PU is

consistent with domestic government incentives, we compute the difference SW (s12) − SW (FT ).

The next proposition gives the trade properties of this equilibrium and the welfare implications.

Proposition 5 (Strategic exit of foreign firm) Suppose the model parameters satisfy the con-

ditions:

λ < min

{1

2 +m,

4µ

m(4µ− 1)2

}(25)

Then if the domestic government were to introduce a PU, the foreign firm would continue to produce

high quality but stop exporting to the home market, while the domestic firm would produce and export

low quality. A PU would then result in: (i) an increase in the (hedonic) price of low quality in

the domestic economy (prices in the foreign market remain constant); (ii) a decrease in domestic

consumer surplus; (iii) an increase in profits of the home firm, and (iv) a decrease in domestic

welfare. As a result, an antidumping policy in the form of a PU can only be rationalized on the

basis of lobbying by the domestic firm (i.e., when β1 = 0, β2 = 1).

The intuition behind these results is as follows. Since the foreign firm exits the domestic market

so as to avoid the PU, prices in the foreign market become equal to the free trade prices. In the

domestic market, however, the home firm charges the monopoly price and therefore its profits

increase and domestic consumer surplus decreases. As it is usual in situations with market power,

the increase in profits does not offset the fall in consumer surplus and social welfare decreases. As

a result, a PU that causes the exit of the foreign firm from the domestic market can only be the

outcome of successful lobbying by the domestic firm.

To make sure that we remain in a subgame where only the domestic government may contem-

plate the imposition of a PU, we can impose an additional condition under which the domestic

firm does not dump its products when s12 is played. This condition requires that p`(s12) < p∗` (s

12),

which gives:17

λ <2(µ− 1)

(4µ− 1). (26)

We note that the region of parameters for which (25) and (26) jointly hold is non-empty (see section

5.3).

17This condition excludes what we call domino dumping, that is, dumping by the home firm abroad resulting fromthe strategic behavior of the foreign firm in response to the initial imposition of AD law. Domino dumping arisesbecause, once the home government enacts AD law, the foreign rival which is accused of dumping, may exit thehome market strategically. The latter move by the foreign firm changes the pricing behavior of the home firm in allmarkets and may lead to dumping by the home firm abroad. This would motivate the foreign government to take anantidumping action as well. Hence, domino dumping is theoretically different from the standard retaliatory motiveassociated with AD law, though it is difficult to disentangle the two in empirical work unless microdata are used.The above condition is imposed for simplicity, that is, in order to focus on whether the domestic government wishesto enact AD law in a region of parameters where the foreign government does not face dumping in the continuationgame.

21

5.2.3 Strategic exit of the home firm

The home firm may take advantage of the fact that a PU is in place by exiting the foreign market.

By strategically exiting the foreign market, the home firm induces the foreign firm to raise its price

in its local market, and, by the PU restriction, in the home market too. This may end up benefiting

the home firm. In fact, when

(λ+m)(4µ− 1)(4µ− 7) + 9λ < 0

condition (22) fails and the pair of strategies s11 = {q`&E, qh&E} is no longer an equilibrium

because, given that the foreign firm produces the high-quality good and exports it abroad, the

home firm prefers to deviate by exiting the foreign market altogether.

Given this, we next ask whether the pair of strategies s21 = {q`&NE, qh&E} can constitute a

Nash equilibrium. It is obvious that we need condition (22) to fail. A necessary condition for this

to happen is µ < 7/4, which suggests that only when the quality gap is sufficiently small the home

firm finds it advantageous to exit the foreign market. This is intuitive since when the quality gap is

large enough competition between the two products is not so fierce and the gains from weakening

competition by exiting are relatively low.

In addition, for the foreign firm not to deviate we require that R∗(s21) > θqh/4, which after

rearranging gives:

8λ(µ− 1)[1− 2µ+ 2(2 +m)(µm+ λ(µ− 1))]−m(4µ− 1)2 > 0.

Intuitively this condition requires that the quality gap is not too small for otherwise the foreign

firm would again prefer to exit the home market in order to avoid the PU constraint and obtain

the monopoly profits from its local market.

When both conditions are met, an equilibrium where firms play s21 after a PU is imposed by

the domestic country exists. As before, in order to check whether imposing a PU is consistent

with domestic government incentives, we compute the difference SW (s21) − SW (FT ). Our next

result discusses the trade properties of this equilibrium and the welfare implications of a PU for

the domestic country.

Proposition 6 (Strategic exit of home firm) Suppose the model parameters satisfy the condi-

tions:

(λ+m)(4µ− 1)(4µ− 7) + 9λ < 0 (27)

8λ(µ− 1)[1− 2µ+ 2(2 +m)(µm+ λ(µ− 1))]−m(4µ− 1)2 > 0 (28)

Then if the domestic government were to introduce a PU, the foreign firm would continue to produce

and export high quality, while the domestic firm would produce low quality and stop exporting to the

22

foreign market. A PU would then result in: (i) an increase in the (hedonic) prices of low and high

quality in the domestic economy; (ii) a decrease in domestic consumer surplus; (iii) an increase in

profits of the home firm if

λm(1 +m)2(4µ− 1)2 − [4λ(µ− 1) +m(4µ− 1)]2 (1 + λm) > 0

and (iv) a decrease in domestic welfare. As a result, an antidumping policy in the form of a PU

can only be rationalized on the basis of lobbying by the domestic firm (i.e., when β1 = 0, β2 = 1).

By exiting the foreign market the domestic firm relaxes competition in that market, which tends

to raise prices. By the PU restriction prices rise also in the domestic market, which tends to favor

the domestic firm. It turns out that profits of the home firm may or may not increase. Since the

price it charges domestically increases, profits tend to increase; however, exiting from the foreign

market altogether lowers its profits too. Domestic consumer surplus decreases and overall welfare

too. As a result, a PU that causes the exit of the home firm from the foreign market can only be

explained by successful lobbying by the domestic firm.

5.2.4 Quality reversal and intra-industry trade

Let us now consider the case in which a PU leads to a quality reversal in the marketplace, that is,

the shift of quality leadership from the foreign to the domestic firm. For this, we first study the

conditions under which the pair of strategies s33 = {qh&E, q`&E} is part of a Nash equilibrium.

As the foreign firm is subject to a PU, for s33 to be an equilibrium we require that the foreign firm

prefers to export its low-quality good to the home country rather than refraining from exporting

at all, that is:

R∗(s33) > R∗(s34) if and only if λ >1

2 +m.

Note that this condition is the same as condition (21). Likewise, given that the foreign firm exports

its good to the home country we require that the home firm prefers exporting its high-quality good

over not exporting it, that is:

R(s33) > R(s43).

Sufficient conditions for this inequality to hold are (see the Appendix):

(m+ λ)(4µ− 1)(1−√mλ)− 3λ > 0

m(1− λ) + 2(1− 2λ)√mλ < 0

Assume that (21) is met along with these two conditions. If the domestic government were

to impose a PU, there exists an equilibrium where the foreign firm would stay in the market

producing low quality and exporting it to the domestic country. In contrast the home firm would

produce and export the high-quality variant. Folding the game backwards, we now ask whether

23

the domestic country wants to deviate from the free trade equilibrium by indeed imposing a PU.

For this, we compute the difference between domestic welfare under s33 and that under free trade,

i.e. SW (s33)− SW (FT ). The following proposition summarizes our findings.

Proposition 7 (Quality reversal and intra-industry trade) Suppose the model parameters

satisfy the following three conditions: Condition (21),

(m+ λ)(4µ− 1)(1−√mλ)− 3λ > 0 (29)

m(1− λ) + 2(1− 2λ)√mλ < 0. (30)

Then if the domestic government were to introduce a PU, the home firm would produce high quality

and the foreign firm would produce low quality. A PU would then result in: (i) intra-industry

trade with no negative effect on world trade; (ii) an increase in the domestic (hedonic) prices of all

qualities;(iii) a decrease in domestic consumer surplus;(iv) a substantial increase in profits of the

home firm, and (v) an increase in the welfare of the domestic country. As a result, an antidumping

policy in the form of a PU can be rationalized on domestic welfare grounds in this case; world

welfare decreases though.

The intuition behind this result is as follows. The constraint on the foreign firm raises the

prices paid for low quality in the domestic and the foreign countries. Prices of high quality follow

suit. Domestic consumer surplus under s33 decreases for two reasons. First, home consumers pay

a higher price for all product variants. Second, less consumers demand the low-quality good since

hedonic price p`/q` increases. However, the reversal has a huge impact on domestic firm profits,

which more than offsets the negative impact on consumers and ultimately results in an increase in

welfare.

Figure 1 illustrates the percentage gain in (un-weighted) welfare resulting from the quality

reversal as a function of m and λ, compared to free trade. Parameters are θ = 1, µ = 6, q` = 0.05.

The light-shaded area gives the percentage gain for all possible (m, λ) combinations; the darker

area gives the percentage gain obtained when parameters are such that the equilibrium exists.

The figure first shows that for this equilibrium to exist, countries’ income differences have to be

relatively small; in addition, the figure reveals that the effect of a PU on welfare is very large.

Therefore, under the conditions in Proposition 7 the domestic government has a strong incentive

to impose a PU.

We can also verify if the domestic firm does not dump its products into the foreign country. For

this we compare ph(s33) and p∗h(s33) and obtain ph(s33)− p∗h(s33) = −(4µ− 1) < 0, which implies

that domino dumping does not arise in this case.

24

Figure 1: Percentage welfare gain under quality reversal equilibrium s33.

5.2.5 Quality reversal and exit of the foreign firm

Let us now consider the case where a PU results in the domestic firm gaining quality leadership

and to the exit of the foreign firm from the domestic market. For this, we study the conditions

under which the pair of strategies s34 = {qh&E, q`&NE} is part of a Nash equilibrium. Under this

strategy profile, the home firm produces high quality and exports it to the foreign country; the

foreign firm, by contrast, produces low quality to satisfy its local demand only.

Given that the foreign firm is not exporting its low-quality good to the home country, it is

obvious that the home firm does not have an incentive to deviate. To ensure that the foreign firm

does not deviate either we require that:

R∗(s34) > R∗(s33) if and only if λ <1

2 +m.

Note that this condition is the same as condition (23) and basically says that income differences

across countries should be small. Assume that it is met. Then, if the domestic firm were to impose

a PU, an equilibrium where the foreign firm chooses to produce low quality and sell it only locally

while the domestic firm chooses to produce high quality and export it abroad exists. Computing

the difference between domestic welfare under s34 and that under free trade, i.e. SW (s34) −

SW (FT ), we address the issue of whether the domestic country wants to deviate from the free

trade equilibrium by enacting AD law if firms are expected to play s34 thereafter. The following

proposition summarizes our findings.

Proposition 8 (Quality reversal and exit of the foreign firm) Suppose that condition (23)

holds. Then if the domestic government introduces a PU, there exists an equilibrium where the home

firm produces high quality and exports it to the foreign market, while the foreign firm produces low

quality and refrains from exporting it to the home market. A PU would then result in: (i) an

25

increase in the domestic (hedonic) price of high quality, while foreign prices remain constant;(ii)

a decrease in domestic consumer surplus;(iii) an increase in the profits of the home firm, and

(iv) in an increase in social welfare provided that 8(µ − 1) + λm(8µ − 11) > 0. As a result, an

antidumping policy in the form of a PU can be rationalized on welfare grounds in this case; world

welfare decreases though.

Since the home firm is the only seller in the domestic market, it is clear that the domestic price

of high quality is higher than under free trade. At the foreign market, since the foreign firm is not

constrained by the PU, prices are exactly the same as under free trade. As a result, it is clear that

the domestic firm obtains higher profits but consumers obtain a lower surplus. The welfare effect

is however ambiguous. While consumers lose, the home firm revenues increase a lot since after the

policy it becomes the high-quality producer.

We can additionally check if the domestic firm does dump its products into the foreign country.

For this we compare ph(s34) = θλqh/2 and p∗h(s34) and obtain:

ph(s34)− p∗h(s34) = θqh [λ− 4(µ− 1)/(4µ− 1)] ,

which is negative provided that λ < 4(µ−1(4µ−1) . Therefore, under this condition, the domestic firm does

not dump its products into the foreign country.

5.2.6 Quality reversal and exit of the home firm

Finally, we study the case where a PU results in the domestic firm gaining quality leadership but

exiting from the foreign market altogether. For this, we study the conditions under which the pair

of strategies s43 = {qh&NE, q`&E} is part of a Nash equilibrium. Under this strategy profile, the

home firm produces high quality but does not export it to the foreign country; the foreign firm, by

contrast, produces low quality to satisfy its local demand as well as the demand from the domestic

country.

We now argue that this cannot be an equilibrium because R(s43) < θµq`/4. In fact, we note

that

R(s43)− θµq`4

=θλ(2 +m)2q`(µ− 1) [(m+ λ)µ− λ]

[(m+ λ)(4µ− 1)− 3λ]2− θµq`

4< 0

if and only if

4µ2(λ+m)[4µ(λ+m)−2(4λ+m)−λ(2+m)2]+µ(4λ+m)2 +4λ(2+m)2[µ(2λ+m)−λ] > 0 (31)

We now observe that the second derivative of (31) with respect to m is equal to

2µ(4µ− 1)2 − 8λµ(µ− 1)(4 + 3m)− 8λ2(µ− 1)2 (32)

This expression decreases in m and in λ. If we set λ = m = 1 in (32) gives 2µ(37 + 8µ(2µ− 5))− 8,

26

which is positive for all µ. As a result, (32) is positive, which implies that the derivative of (31)

with respect to m increases in m. Let us now set m = 0 in the first derivative of (31) with respect

to m. This gives 8λ(µ− 1)(µ(4µ− 3)− 2λ(µ− 1)), which is always positive. From this we conclude

that (31) increases in m. We finally set m = 0 in (31) to obtain 16λ2(µ− 1)3 > 0. This concludes

the proof that R(s43) < θµq`/4.

5.2.7 Overview of equilibria

Figure 2 shows the regions of parameters for which the different equilibria discussed above exist.

There are two important points to be emphasized. First, unlike under free trade, for some pa-

rameters there is a unique Nash equilibrium. One way to interpret this result is in terms of AD

policy operating as an equilibrium selection mechanism. The second important point is that a

quality reversal may be the unique outcome following the imposition of a PU, which, as shown in

Proposition 7, is consistent with the domestic government incentives.

The size of the different regions depends on the magnitude of the quality gap µ. The left panel

of Figure 2 gives the existence regions for µ = 1.5. When λ and m are both relatively small, the

foreign firm exits the export market. This is because, given that the home market is not very

attractive, the foreign firm finds it profitable to focus its activity on its local market. By doing

so, it avoids the PU constraint and this increases the profits it can get from its local consumers.

There are two Nash equilibria in that case, s12 and s34. To select among them, we can again use

the Harsany-Selten criterion (cf. Section 4).

Let G11 = R(s12) − θλmqh4 + c

2(q2h − q2` ) be the gains the domestic firm obtains by predicting

correctly that the foreign firm will select equilibrium s12. Likewise, G12 = R(s34)− θλmq`4 − c

2(q2h−q2` )

denotes the gains the domestic firm derives by forecasting correctly that the foreign firm will

select equilibrium s34. Similarly, for the foreign firm we have G21 = R∗(s12) − c∗

2 (q2h − q2` ) and

G22 = R∗(s34) + c∗

2 (q2h − q2` ). Equilibrium s12 is said to risk-dominate equilibrium s34 whenever

G11G21 −G12G22 > 0. Using the expressions above we obtain

4(4µ− 1)2

θq2`µ(µ− 1)2(4µ+ 1)[G11G21 −G12G22] = 2cq`(1− γ)(1 + µ)− θλm.

This condition does not only depend on cost differences but on cost levels. Given our small invest-

ment costs assumption, s34 risk-dominates equilibrium s12.

When λ andm are moderately large, the foreign market is sufficiently profitable to be abandoned

by the foreign firm. In such a case we have a unique equilibrium that involves a quality reversal (s34

or s33). When λ is relatively high and m moderately large, there are two Nash equilibria, namely

s11 and s33. Under s11 the equilibrium is similar to free trade in that the market structure remains

unchanged. Under s33 there is a radical change since the home firm produces the high quality. We

can invoke the Harsany-Selten criterion to select among these two equilibria. However, in this case

27

we can only do it numerically. It turns out that s11 risk-dominates s33. Finally, when both λ and

m are large, we also have two equilibria (s21 and s33).

m m

λλ λλμ=5μ=1.5

S11: Unchanged market structure (Prop. 4)  S12: Strategic exit of foreign firm (Prop. 5)S21: Strategic exit of domestic firm (Prop. 6)   S33: Quality reversal and intra‐industry trade (Prop. 7)s34 : Quality reversal and exit of foreing firm (Prop.8)

Figure 2: Existence regions for the different types of equilibria.

The right panel gives the existence regions for µ = 5 instead. The implication of a rise in µ is

to reduce the number of possible equilibria with s34 being the unique equilibrium for relatively low

λ and moderate to large m. For these parameters, again, imposing a PU can be rationalized on

domestic welfare grounds.

6 Extensions

In this section we extend the model in two important ways. First, we allow firms to select product

quality from a continuum, so quality levels become endogenous. Second, we compare the incidence

of a PU with that of an AD duty and derive an equivalence result.

6.1 Endogenous quality

In the previous analysis we chose to work with exogenous quality levels in order to illustrate the

main insights in the simplest setting. The main purpose of this extension is to demonstrate that

our earlier results are not undermined when quality levels are chosen by the firms. Even though it

would be possible to replicate the entire analysis with endogenous quality levels, let us focus here

on two issues. First we illustrate how the free trade equilibrium is obtained when qualities can

be chosen from a continuum. Then, we show that for some parameters the domestic government

28

would find it desirable to impose a PU in order to induce a quality reversal.

In this section, we assume flexible production (Eaton and Schmitt, 1994), that is, once firms

invest in the necessary technology and organize their facilities to develop and produce one basic

product, they can produce various downgrades of this basic product at no cost. This idea is modelled

via the following specification of R&D costs: domestic firm’s costs of producing any two variants

q` and q∗` are C(q`, q∗` ) = cmax{q`, q∗` }2/2; likewise, foreign firm costs of producing variants qh and

q∗h are given by C∗(qh, q∗h) = c∗max{qh, q∗h}2/2, where c and c∗ are the usual development cost

parameters measuring R&D efficiency. This assumption serves to rule out quality strategy profiles

where a firm sells different qualities locally and abroad. Because of the presence of (non-trivial)

development costs, we shall now use the weighted sum of consumer surplus and firms’ profits π as

a measure of social welfare in every country: W = β1CS + β2π, βj ∈ {0, 1}, j = 1, 2.

The derivation of the free trade equilibrium follows the steps in section 4, with the additional

choice-of-quality stage. The first step is to show that non-exporting strategies are strictly dom-

inated. This implies that in equilibrium there is intra-industry trade. To select among the two

possible quality assignments, we use again the Harsany-Selten criterion. The next proposition gives

the result (for the proof see our working paper Moraga-Gonzalez and Viaene, 2007).

Proposition 9 (Endogenous quality gap) With endogenous qualities, the unique (risk-dominant)

free trade equilibrium of the export game involves intra-industry trade and is characterized as fol-

lows: (i) Goods sell at prices (5) and (6); (ii) Market demands for firms are (7); (iii) the quality

gap between the variants µ is the solution to:

4(4µ2 − 3µ+ 2)

µ2(4µ− 7)=c∗

c. (33)

(iv) The home firm produces a good of quality

q` =θ(1 +mλ)

c

µ2(4µ− 7)

(4µ− 1)3(34)

that is sold locally and exported to the foreign country; (v) the foreign firm produces a good of higher

quality

qh =4θ(1 +mλ)

c∗µ(4µ2 − 3µ+ 2)

(4µ− 1)3(35)

for its own market and for exports; (vi) Domestic firm’s profits are π = c∗q`qh/8, foreign firm’s

profits π∗ = 2cq`qh with π∗/π = 16c/c∗.

Observe that in this free trade equilibrium, variable µ is the unique solution to the third degree

polynomial in (33). It depends on firms’ development costs but is independent of λ and m. As

prices are similar to those of Proposition 2, Proposition 3 characterizes the conditions for dumping

in this context as well.

29

Let us now show that the strategies s34 continue to constitute an equilibrium after a PU is

imposed; recall that under s34, the home firm “leapfrogs” the foreign firm and produces a product

of endogenous higher quality; meanwhile the foreign firm finds it profitable to exit the domestic

market. Let us fix λ = 0.2, m = 0.5, c = 1, c∗ = 0.85 and θ = 100. Solving for the free trade

equilibrium, one obtains qh = 32.68 and q` = 5.51, with µ = 5.93. Firms’ profits are given by

π = 19.15 and π∗ = 360.42, with domestic social welfare W = 73.07.

Under the strategies s34, the domestic firm picks qh to maximize π = R(s34) − cq2h/2, and the

foreign firm chooses q` to maximize π∗ = R∗(s34)− c∗q2` /2, where R(s34) and R∗(s34) are given in

(20). Taking the FOCs with respect to quality levels and solving gives qh = 27.87 and q` = 5.57,

with µ = 5. Firms’ profits are given by π = 298.92 and π∗ = 17.68, with domestic social welfare

W = 347.17.

These results clearly show that imposing a PU is incentive compatible provided that the equi-

librium that ensues is s34. To see whether deviations are profitable, we ask whether any of the

firms wishes to pick another quality level and export strategy. Given the strategy of the foreign

firm, q`&NE, it is readily seen that the home firm does not wish to deviate because by doing so its

profits unambiguously fall. To see whether, given the strategy of the home firm, qh&E, the foreign

firm wishes to deviate, we first consider the case where it deviates to strategy {q`&E}. In this case,

it would obtain a deviating payoff equal to π∗d = R∗(s33)−c∗q2` /2, where the expression for R∗(s33)

is given in (19) and qh is fixed to 27.87 in the deviation. Maximizing profits with respect to q` gives

q` = 3.96 and a deviating profit equal to π∗d = 8.01. As a result, the deviation is not profitable.

Finally, we check whether the foreign firm wants to deviate to {qh&E}, in which case it jumps

over the quality of the home firm in the putative equilibrium. In that case, it would obtain a

deviating payoff equal to π∗d = R∗(s11) − c∗q2h/2, where R∗(s11) is given in (19) and q` would be

fixed to 27.87 in the deviation. Maximizing profits with respect to qh gives qh = 0 and a deviating

profit equal to π∗d = 0. As a result, the deviation is not profitable.

With these calculations we conclude that for the chosen parameters (which clearly fall in the

s34-labelled region of Figure 2), the equilibrium described by Proposition 8 remains true for the

cases when firms are allowed to choose their quality levels.

6.2 Equivalence results

The other popular instrument of AD legislation involves the imposition of a duty that equalizes the

price that consumers in different countries pay for the same good. Given that the price the foreign

firm charges locally is p∗h and the export price is ph, an antidumping policy in the form of a duty

involves a commitment by the domestic government to levy an ad valorem duty t that equalizes

30

the price that is paid by consumers in different countries:

ph(1 + t) = p∗h (36)

With an antidumping duty t, demands faced by the foreign and domestic firms are, respectively:

Dl(.) = m

(ph(1 + t)− p`λθ(qh − q`)

− p`

λθq`

), Dh(.) = m

(1− ph(1 + t)− p`

λθ(qh − q`)

)(37)

D∗` (.) =p∗h − p∗`θ(qh − q`)

−p∗`θq`

, D∗h(.) = 1−p∗h − p∗`θ(qh − q`)

(38)

Firms maximize profits given by:

π = plDl + p∗lD∗l −

cq2l2, π∗ = ph(1 + t)Dh + p∗hD

∗h −

c∗q2h2,

and anticipating the antidumping duty t given by:

t =p∗h − phph

. (39)

Plugging (39) in (37) and taking the FOC dπ∗/dph yields:

dπ∗

dph= m(1 + t)

(1−

p∗h − plλθ(qh − q`)

)= (1 + t)Dh(.)

The RHS of this FOC is simply the domestic demand for high quality. As long as Dh(.) is greater

than zero, the profits of the foreign firm are increasing in its export price ph. The optimal pricing

behavior of the foreign firm is then to set ph = p∗h, which implies that revenues from the duty

are zero, as has been shown earlier, for example in Feenstra (2003). In summary we obtain the

following equivalence result.

Proposition 10 An antidumping duty imposed by the domestic government results in an equal-

ization of international prices. Hence, antidumping duties and price undertakings are equivalent in

our model.

It is clear from Proposition 10 that, for rational firms, it does not matter whether a PU or an

AD duty is levied. What matters is the threat of an AD actoion. An important implication is that

antidumping duties could lead to product quality reversals under the same conditions as for price

undertakings.

Table 6: Types of antidumping measures1

AD measure Argentina S. Africa

PU2 69.1 0.0

AD duty3 30.9 100.0Notes: (1) percentages; (2) Price undertakings include cases where a duty is imposed if the price falls under a specific level; (3) Duty includes

specific and ad valorem duties. Source: Own calculations based on the Global Antidumping Database (Bown, 2010)

31

Proposition 10 would point out to there being an equal use of AD instruments. This is simply

not the case in the data as Table 6 shows. While Argentina has applied a PU in 69.1% of imposed

measures, South Africa relies exclusively on antidumping duties (Bown, 2010). What are the factors

that are conducive to this unequal use of AD instruments? For governments, the adoption of one

policy instrument hinges on the cost effectiveness of investigations and on the cost of implementing

these policies (Nakagawa, 2006). Very often, a PU is cheaper to adopt and monitor but does

not bring revenues when firms deviate from pricing rule (11).18 The exchange rate regime should

matter as well since the exchange rate pass-through of oligopolistic firms depends on the type of

product and on the underlying market structure (Dornbusch, 1987). With a high exchange rate

volatility, the timing of investigations matters and information acquisition becomes costly under a

PU. Moreover, the likelihood of deviations from pricing rule (11) is high in which case AD duties

would be collected.

7 Conclusions

We have presented a model of international trade where two firms located in two different coun-

tries produce quality-differentiated goods for their local markets and, eventually, for exports. An

important feature of our model is the existence of size and income differences across countries,

and of asymmetries in firms’ R&D costs. We have shown that, under free trade, the unique (risk-

dominant) Nash equilibrium involves intra-industry trade; in addition, the most efficient firm, the

foreign, is the quality leader in the international market. Since consumers in different countries

differ in their concern for quality, in equilibrium, unilateral dumping by the foreign firm into the

domestic country always takes place. With transport costs, reciprocal dumping may emerge. In

the context of dumping of high-quality products, we have studied the incentives of the domestic

firm to file a petition for an antidumping action and of the domestic government to honor it.

Our main result is that a PU may lead to a quality reversal in the international market. This

results in much greater profits for the home firm than under free trade, which leads to a greater

social welfare in spite of the fact that consumers lose. This gives the domestic government incentives

to impose a PU. A PU may also lead to the exit of the foreign firm, which favors the domestic

one and thereby helps explain why there is so much lobbying by domestic firms for antidumping

regulations. When AD duties are considered instead, we have derived an equivalence result between

AD duties and a PU.

Though the focus of the paper has been on unilateral dumping cases, we showed situations in

18Since there are two policy instruments for choice, a PU and an AD duty, it is possible to implement a formalpolicy game where policy-imposing governments can freely pick between the two instruments. There would be fourcontinuation games, each of them corresponding to a different instrument strategy profile. With an asymmetry inimplementation costs, we could compute a refined equilibrium in instruments like we did in Table 3.

32

which reciprocal dumping would be prevalent under free trade. Moreover, we encountered cases of

domino dumping, i.e. situations where once a country enacts antidumping law it triggers dumping

by its own firm on foreign markets. Further work could include strategic policy games between

governments with opposite incentives. The welfare implications of AD law in these cases are

still unknown. Lastly, key questions related to the equivalence between antidumping instruments

remain unanswered. A formal game between governments is one way to endogenize the choice of

AD instruments and to uncover when this equivalence fails.

Of course, an important issue is to establish the empirical relevance of our results. There is

plenty of empirical evidence supporting the main features of the model analyzed here (see e.g.

Anderton, 1999; Bils and Klenow, 2001; Crozet and Erkel-Rousse, 2004). Also, as mentioned

above, it is well-known that governments have traditionally used and continue to use an array a

trade and industrial policy instruments to support local producers. For members of the WTO,

market protection via tariff barriers is nowadays limited; in contrast, antidumping measures whose

application is governed by the WTO are not. The main hypothesis of this paper is that antidumping

is nowadays the trade policy of choice of developing and transition countries because it helps firms

located in these countries in their race for the market. We have also provided some examples in

which there presumably are quality upgrading effects of AD laws. As far as we know, there is no

empirical study about the product quality implications of AD policy. Combining the WTO data

on antidumping measures with recent databases like the “Analytical Database on International

Trade (BACI)” of CEPII (Paris),19 which explicitly accounts for intra-industry trade in vertically

differentiated products, may allow for the empirical testing of the various conclusions of the model.

19See http://www.cepii.fr/anglaisgraph/bdd/baci.htm.

33

8 ONLINE APPENDIX

Proof of Proposition 4. As explained in the main text, the conditions (21) and (22) follow from

the bilateral comparisons between R∗(s11) and R∗(s12) and between R(s11) and R(s21). Regarding

condition (22), we note that, as (4µ − 1) > 0 by assumption, a sufficient condition for R(s11) >

R(s21) is (4µ− 7) ≥ 0.

(i) Home prices. Compare domestic free trade prices in (5) with the same prices under strategies

s11, that is p`(s11) and ph(s11) in (12) and (13). Both prices under strategies s11 are higher if and

only if λ < 1. The same holds for domestic hedonic prices.

(ii) Foreign prices. If and only if λ < 1, foreign free trade prices in (6) are higher than the same

prices under strategies s11 in (12) and (13). Likewise for foreign hedonic prices.

(iii) Firm revenues. To verify whether free trade is preferable to s11 for the firms, it is sufficient

to compare R∗(FT ) in (3) with R∗(s11), and R(FT ) in (3) with R(s21). The revenue comparison

leads to R∗(FT ) > R∗(s11) and R(FT ) > R(s11) if and only if (1−λ)2 > 0, which is always satisfied

since λ < 1.

(iv) Domestic consumer surplus. Like for free trade, domestic consumer surplus under strategies

s11 is given by

CS(s11) =m

λθ

[x2

2qh − xph(s11)

]λθθ(s11)

+m

λθ

[x2

2q` − xpl(s11)

]θ(s11)

θ(s11)

where θ(s11) (taste parameter of the consumer indifferent between buying high or low quality) and

θ(s11) (taste parameter of the consumer indifferent between acquiring low quality or nothing at all)

are given by

θ(s11) =ph(s11)− pl(s11)

qh − q`=θλ(1 +m)(2µ− 1)

(m+ λ)(4µ− 1)

θ(s11) =pl(s

11)

q`=θλ(1 +m)(µ− 1)

(m+ λ)(4µ− 1)

Given this, we obtain the following expression for domestic consumer surplus:

CS(s11) =mλθqh

2(4µ− 1)2

{4µ2

[4− 4

(1 +m)

(m+ λ)+

(1 +m)2

(m+ λ)2

]+ µ

[−8 + 20

(1 +m)

(m+ λ)− 7

(1 +m)2

(m+ λ)2

]}

+mλθqh

2(4µ− 1)2

[1− 4

(1 +m)

(m+ λ)+ 3

(1 +m)2

(m+ λ)2

]A comparison of CS(s11) with CS(FT ) yields

CS(FT ) > CS(s11) iff 4µ(2m+ 3λ− 1) + (3 + 2m− λ) > 0

which is always satisfied under the condition (21). Intuitively, note that θ(s11) is larger than θ(FT )

34

and θ(s11) is larger than θ(FT ) if and only if (1 +m)/(m+ λ) > 1 which is always satisfied since

λ < 1. Hence with all prices being higher, with less consumers being served and less consumers

purchasing high quality, we obtain CS(s11) < CS(FT ).

Importantly, result (iii) together with result (iv) imply a domestic welfare loss following the

imposition of a PU by the home government. �

Proof of Proposition 5. As explained in the main text, the conditions λ < 12+m and λ <

4µm(4µ−1)2 follow from the bilateral comparisons between R∗(s11) and R∗(s12) and between R(s12)

and mθλqh/4, respectively. When these two conditions hold, then the pair of strategies s12 is a

Nash equilibrium.

(i) Home and foreign prices. Compare domestic free trade prices in (5) with the same prices

under strategies s12. This gives p`(s12) > p`(FT ) if and only if 2µ + 1 > 1, which is always the

case. Likewise for the domestic hedonic price of low quality. Foreign prices under s12 are identical

to those under FT .

(ii) Domestic consumer surplus. The computation of the domestic consumer surplus is simpler

than in Proposition 4 since the foreign firm does not export to the home country and hence only

low quality is locally available for consumption. Domestic consumer surplus under strategies s12 is

given by:

CS(s12) = m

∫ λθ

θ(s12)(xq` − p`(s12))dF (θ) =

θλmq`8

which is clearly lower than CS(FT ).

(iii) Firm revenues. Since the home firm becomes a monopolist in its own local market, it is

obvious that R(s12) > R(FT ).

(iv) Result (ii) together with result (iii) imply a loss in consumer surplus but a gain in domestic

firm’s revenues. Comparing welfare under s12 with welfare under FT gives:

SW (s12)− SW (FT ) = −θλm(4qh(qh − q`) + 3q2` )

8(4qh − q`)< 0.

As a result, an antidumping policy in the form of a PU can only be rationalized on the basis of

lobbying by the domestic firm (that is, when β1 = 0, β2 = 1).

Proof of Proposition 6. Comparing R(s21) with R(s11) gives the condition 16λ(µ − 1)2 +

m(4µ− 1)(4µ− 7) < 0. Comparing R∗(s21) with θqh/4 gives 16λ(1 +m2)(µ− 1) [λ(µ− 1) +mµ]−

[4λ(µ− 1) +m(4µ− 1)]2 > 0.

(i) Comparing p`(s21) with p`(FT ) gives the condition (4µ−1)(1−λ)+3λ > 0, which is always

satisfied. For the high-quality good the same condition arises, as well as for hedonic prices.

(ii) Because all prices are higher, it is obvious that domestic consumer surplus decreases.

35

(iii) Comparing R(s21) with RFT gives

[4λ(µ− 1) +m(4µ− 1)]2 (4µ− 1)2

θq`µ(µ− 1)

[R(s21)−RFT

]= λm(1 +m)2(4µ− 1)2 − [4λ(µ− 1) +m(4µ− 1)]2 (1 + λm) ,

from which we obtain the condition in the Proposition.

(iv) As usual with market power, an increase in prices raises firm profits but decreases consumer

surplus; the overall effect is negative.

Proof of Proposition 7. To verify whether s33 is a Nash equilibrium with a quality reversal,

it is sufficient to compare R∗(s33) with R∗(s34) and R(s33) with R(s43) since payoffs from other

strategies are strictly dominated. Using revenue expressions of Section 5.1, we obtain that R∗(s33) >

R∗(s34) if and only if λ > 1/(2 +m), which is condition (21). Regarding the comparison of R(s33)

with R(s43), it should be noted that the home firm is a non-exporter of high quality in s43. Therefore

a sufficient condition for R(s33) > R(s43) is that revenues from exports in s33 are larger than home

sales in s43. In other words R(s33) > R(s43) if p∗h(s33)D∗h(s33) > ph(s43)Dh(s43)where

p∗h(s33)D∗h(s33) =θ(µ− 1)q`

4

{4µ(m+ λ)−m(1− λ)

(m+ λ)(4µ− 1)

}2

ph(s43)Dh(s43) = λθm(µ− 1)ql

{2µ(m+ λ) + (1− 2λ)

(m+ λ)(4µ− 1)− 3λ

}2

.

After simplification we get

p∗h(s33)D∗h(s33)− ph(s43)Dh(s43) = 4µ(λ+m)[(m+ λ)(4µ− 1)(1−√mλ)− 3λ]

− (λ+m)(4µ− 1)[m(1− λ) + 2(1− 2λ)

√mλ]

+ 3λm(1− λ)

Upon observing this expression, it is readily seen that (29) and (30) suffice for R(s33) > R(s43).

(i) Intra-industry trade. Substituting prices ph(s33), p∗h(s33) and p∗l (s33) of Section 5.1 in demand

functions, we obtain:

Dh =4m(m+ λ)µ+m(1− λ)

2(m+ λ)(4µ− 1), D∗h =

4m(m+ λ)µ−m(1− λ)

2(m+ λ)(4µ− 1)

Dl =2mµ

(4µ− 1)+m(1− λ)− 2m(1 +m)µ

2(m+ λ)(4µ− 1), D∗l =

2µ

(4µ− 1)− m(1− λ) + 2λ(1 +m)µ

2(m+ λ)(4µ− 1)

The Grubel-Lloyd measure of two-way trade (in value) is:

GL = 100

{1−

∣∣p∗h(s33)D∗h − pl(s33)Dl

∣∣(p∗h(s33)D∗h + pl(s33)Dl)

}

Like (10), GL varies between 0 and 100 and is a (more complex) combination of the primitive

36

parameters of the model: m, λ, and µ. Regarding the sum of demands:

Dh +D∗h =2(1 +m)µ

(4µ− 1);Dl +D∗l =

(1 +m)µ

(4µ− 1)

which is similar to (7). Hence, though a quality reversal modifies bilateral trade flows, there is no

impact on total trade.

(ii) Home prices. Compare domestic free trade prices in (5) with the same prices under strategies

s33, that is ph(s33) and p`(s33). All domestic prices under strategies s33 are higher if and only if

λ < 1. Likewise for domestic hedonic prices.

(iii) Domestic consumer surplus. Like in the free trade case, domestic consumer surplus under

strategies s33 is given by

CS(s33) =m

λθ

[x2

2qh − xph(s33)

]λθθ(s33)

+m

λθ

[x2

2q` − xpl(s33)

]θ(s33)

θ(s33)

where

θ(s33) =ph(s33)− pl(s33)

qh − q`=

θλ

2(m+ λ)(4µ− 1)[(4µ− 1)(m+ λ)− (1 +m)]

θ(s33) =pl(s

33)

q`=θλ(1 +m)(µ− 1)

(m+ λ)(4µ− 1)

Comparing CS(s33) with CS(FT ) gives

CS(s33)− CS(FT ) =−λmθq`(1− λ)(µ− 1) [3 + 4µ(4m− 1) + λ(20µ− 3)]

8(λ+m)2(4µ− 1)2,

which is negative as long as

λ >4µ(1− 4m)− 3

20µ− 3

But since λ > 1/(2 +m), if we prove that

1

2 +m>

4µ(1− 4m)− 3

20µ− 3(40)

then inequality CS(s33) − CS(FT ) < 0 is always satisfied. To prove (40) notice that if m > 1/4

the inequality is true. Otherwise we need that

20µ− 3 > (2 +m)(4µ(1− 4m)− 3)

or 20µ− 3 > 8µ− 28µm− 6− 16µm2 − 3m

which, after simplification, gives 12µ+ 3 + 28µm+ 16µm2 + 3m > 0.

(iv) Revenues. Comparing R(s33) to R(FT ) we obtain that

4(m+ λ)(4µ− 1)2

θq`(µ− 1)

[R(s33)−R(FT )

]= m(1− λ)2 − 4µ[λ+m(3 + λ(3λ+m− 4))] + 16µ2(λ+m)(1 +mλ) (41)

37

We now prove that (41) is positive. First we take the derivative of (41) with respect to λ, which

gives

16µ2(1 +m2 + 2λm)− 4µ[1 +m(6λ+m− 4)]− 2m(1− λ) (42)

This expression is increasing in λ because its derivative with respect to λ is equal to 2m(1+4µ(4µ−

3)) > 0. Therefore, if (42) is positive at λ = 0 this means that (42) is always positive, which in

turns implies that R(s33) − R(FT ) increases in λ and since (41) evaluated at λ = 0 is equal to

m(1 + 4µ(4µ − 3)) > 0 the result R(s33) − R(FT ) > 0 follows. So it remains to show that (42) is

positive at λ = 0. Setting λ = 0 in (42) gives

16µ2(1 +m2)− 4µ[1 +m(m− 4)]− 2m (43)

Let us take the derivative of (43) with respect to m, which gives 8µ(2+m(4µ−1))−2 > 0. Therefore

(43) increases in m. Setting m = 0 in (43) gives 4µ(4µ − 1) > 0. As a result (43) is positive. The

proof is now complete.

(v) Welfare. Comparing W (s33) with W (FT ) gives

8 (λ+m)2 (4µ− 1)2

θq` (µ− 1)[W(s33)−W (FT )]

= m(1− λ)2(2m− λ)− 4µ[2λ2 + 2λm3 + λm(7− λ(2− λ)) +m2(6− 4λ(1− λ))

]+ 32µ2(λ+m)2(1 + λm)

(44)

We now notice that the second derivative of the RHS of (44) with respect to µ is equal to 64(λ+

m)2(1 + λm) > 0. Therefore the first derivative with respect to µ increases in µ. Setting µ = 1 in

the first derivative of the RHS of (44) with respect to µ gives

40m2 + 4λm [25 + 2m(2 + 7m)] + 8λ2(7 +m+ 14m2) + 60λ3m > 0,

from which we conclude that it is positive everywhere. As a result, (44) is increasing in µ. We now

set µ = 1 in the RHS of (44) to obtain

24λ2 + 24λm3 + 2m2 [5 + λ(6 + 25λ)] + λm [35 + λ(10 + 27λ)] > 0.

From this we conclude that the RHS of (44) is positive. Therefore, W(s33)> W (FT ).

World welfare, denoted WW , is the sum of welfare of the two countries. If we establish the

comparison WW (s33)−WW (FT ) we obtain:

WW (s33)−WW (FT ) = −θm(1− λ)2(µ− 1)(4µ+ 1)q`8(λ+m)(4µ− 1)2

which is clearly negative. �

Proof of Proposition 8. The conditions for a Nash equilibrium are explained in the main

38

text.

(i) It is straightforward to see that the prices in the foreign market are the same as under free

trade. The home price of high quality is much higher, since the home firm monopolizes the domestic

market.

(ii) Because the price is higher in the domestic market, domestic consumer surplus is lower than

under free trade;

(iii) Comparing profits yields

R(s34)−RFT =θq`µ

4(4µ− 1)2[4(µ− 1)(4µ− 1) + λm(5 + 4µ(4µ− 3))] > 0

(iv) Comparing welfare levels yields:

SW (s34)− SWFT =θq`µ

8(4µ− 1)[8(µ− 1) + λm(8µ− 11)] > 0

which, under the condition in the text, implies that welfare increases.

World welfare (WW ) is the sum of welfare of the two countries. If we establish the comparison

WW (s34)−WW (FT ) we obtain:

WW (s34)−WW (FT ) = −θλm(20µ− 11)q`8(4µ− 1)2

which is clearly negative. �

References

[1] Anderson, J.E. (1992), “Domino Dumping I: Competitive Exporters,” American Economic

Review 82(1), 65-83.

[2] Anderson, J.E. (1993), “Domino Dumping II:antidumping,” Journal of International Eco-

nomics 35(1-2), 133-150.

[3] Anderson, S., N. Schmitt and J.-F. Thisse (1995), “Who Benefits from Antidumping Legisla-

tion,” Journal of International Economics 38, 321-337.

[4] Anderton, B. (1999), “Innovation, Product Quality, Variety and Trade Performance: An Em-

pirical Analysis of Germany and the UK,” Oxford Economic Papers 51, 152-167.

[5] Aoki, R. and T.J. Prusa (1997), “Sequential versus Simultaneous Choice with Endogenous

Quality, ” International Journal of Industrial Organization 15-1, 103121.

[6] Blonigen, B.A. and T.J. Prusa (2003), ”Antidumping”, in E.K. Choi and J. Harrigan (Eds.),

Handbook of International Trade (Oxford, UK and Cambridge, MA: Blackwell Publishers),

Chapter 9.

39

[7] Bonroy, O. and C. Constantatos (2005), “Minimum Quality Standards and Equilibrium Selec-

tion with Asymmetric Firms,” mimeo.

[8] Borrus, M., L.D. Tyson, and J. Zysman (1986), “Creating Advantage: How Government

Policies Shape International Trade in the Semiconductor Industry,” in P. Krugman (Ed.),

Strategic Trade Policy and the New International Economics (Cambridge, MA: MIT Press).

[9] Bils, M. and P.J. Klenow (2001), “Quantifying Quality Growth,” American Economic Review

91(4), 1006-1030.

[10] Bown, C. P. (2010), “Global Antidumping Database,” Development Research Group, World

Bank.

[11] Brander, J. A. and P.R. Krugman (1983), ”A ‘Reciprocal Dumping’ Model in International

Trade,” Journal of International Economics 15, 313-23.

[12] Brutton, H. J. (1998), “A Reconsideration of Import Substitution,” Journal of Economic

Literature 36-2, 903-936.

[13] Cabrales, A., W. Garcia-Fontes and M. Motta (2000), ”Risk Dominance Selects the Leader:

An Experimental Analysis,” International Journal of Industrial Organization 18, 137-162.

[14] Crozet, M. and H. Erkel-Rousse (2004), “Trade Performances, Product Quality Perceptions,

and the Estimation of Trade Price Elasticities,” Review of International Economics 12, 108-

129.

[15] Dornbusch, R. (1987), ”Exchange Rates and Prices,” American Economic Review 77(1), 93-

106.

[16] Eaton, B.C. and N. Schmitt (1994), ”Flexible Manufacturing and Market Structure,” American

Economic Review, 84(4), 875-888.

[17] Ethier, W. J. and R. D. Ficher (1987), “The New Protectionism,” Journal of International

Economic Integration 2, 1-11.

[18] European Community-EC (2002a), Submission from the European Communities Concerning

the Agreement on Implementation of Article VI of GATT 1994, WTO Doha Development

Agenda Negotiations, Brussels.

[19] European Community-EC (2002b), Twentieth Annual Report from the Commission to the

European Parliament on the Community’s antidumping and Anti-Subsidy Activities, Brussels

(http://europa.eu.int/comm/trade/policy/dumping/reports.htm).

40

[20] Falvey, R. (1981), ”Commercial Policy and Intra-Industry Trade,” Journal of International

Economics 11, 495-511.

[21] Feenstra, R.C. (2004), Advanced International Trade: Theory and Evidence, Princeton: Prince-

ton University Press.

[22] Fisher, R. D. (1992), “Endogenous Probability of Protection and Firm Behavior,” Journal of

International Economics 32, 149-63.

[23] GATT (1991), Trade Policy Review, The European Communities, vol. 2 (GATT, Geneva).

[24] Greenaway, D., R. Hine, and C. Milner (1994), ”Country Specific Factors and the Pattern of

Horizontal and Vertical Intra Industry Trade in the U.K.,” Weltwirtschaftsliches Archiv 131,

77-100.

[25] Greenaway, D., R. Hine, and C. Milner (1995), ”Horizontal and Vertical Intra-Industry Trade:

A Cross Industry Analysis for the UK, ” Economic Journal 105, 1505-1518.

[26] Hallack, J. C. (2006), “Product Quality and the Direction of Trade,” Journal of International

Economics 68, 238-265.

[27] Hallack, J. C and P. Schott (2011), ”Estimating Cross-Country Differences in Product Quality,”

Quarterly Journal of Economics 126(1), 417-474.

[28] Harsanyi, J.C. and R. Selten (1988), A General Theory of Equilibrium in Games, Cambridge,

MA: MIT Press.

[29] Herguera, I., P. Kujal and E. Petrakis (2002), ”Tariffs, Quality Reversals and Exit in Vertically

Differentiated Industries,” Journal of International Economics, 58 (2), 467-492.

[30] Horn, H. (2006), “National Treatment in the GATT,’ The American Economic Review 96,

394-404.

[31] Hughes, J. S., S. Lenway and J. Rayburn (1997), “Stock Price Effects of U.S. Trade Pol-

icy Responses to Japanese Trading Practices in Semi-Conductors,” The Canadian Journal of

Economics 30-4, 922-42.

[32] Krishna, K. (1987), ”Tariffs versus Quotas with Endogenous Quality,” Journal of International

Economics 23, 97-122.

[33] Lee, K. and C. Lim (2001), “Technological Regimes, Catching-up and Leapfrogging: Findings

from the Korean Industries,” Research Policy 30, 459-483.

41

[34] Leidy, M. P. and B. M. Hoekman (1990), “Production Effects of Price and Cost-based an-

tidumping Laws under Flexible Exchange Rates,” Canadian Journal of Economics 23, 873-95.

[35] Lu, Z. (2004), “Learning to Use WTO Rules for Protection,” China Daily, July 6, 2004.

[36] Maur, J.-C. and P. A. Messerlin (1999), “Antidumping in Supercomputers or Supercomputing

in Antidumping? The Cray-Nec Case,” Institut d’Etudes Politiques de Paris, working paper.

[37] Moraga-Gonzalez, J.L. and J.-M. Viaene (2005), ”Trade Policy and Quality Leadership in

Transition Economies,” European Economic Review 49, 359-385.

[38] Moraga-Gonzalez, J.L. and J.-M. Viaene (2007), ”Anti-dumping, Intra-industry Trade and

Quality Reversals,” Tinbergen Institute DP # TI 2004-124/2, The Netherlands (downloadable

from http://www.tinbergen.nl/ moraga/ADpolicy.pdf).

[39] Motta, M., J.-F. Thisse and A. Cabrales (1997), “On the Persistence of Leadership or Leapfrog-

ging in International Trade,” International Economic Review 38(4), 809-824.

[40] Mussa, M. and S. Rosen (1978), “Monopoly and Product Quality,” Journal of Economic Theory

18, 301-317.

[41] Nakagawa, J. (2006), “Argentina,” in J. Nakagawa (Ed.), Antidumping Laws and Practices of

the New Users, (London: Cameron May).

[42] Niels, G. T. (2004), A Political Economy Assessment of Antidumping Policy in Mexico, Ph.

D. thesis, Erasmus University Rotterdam.

[43] Reitzes, J. D. (1993), “Antidumping Policy,” International Economic Review 34-4, 745-63.

[44] Saggi, K. and N. Sara (2008), “National Treatment at the WTO: The Roles of Product and

Country Heterogeneity,” International Economic Review 49(4), 1365-1394.

[45] Shaked, A. and J. Sutton (1982), “Relaxing Price Competition through Product Differentia-

tion,” Review of Economic Studies 49(1), 3-13.

[46] Shaked, A. and J. Sutton (1983), “Natural Oligopolies,” Econometrica 51-5, 1469-83.

[47] Shaked, A. and J. Sutton (1984), “Natural Oligopolies and International Trade,”in Monopo-

listic Competition and International Trade (ed. H. Kierzkowski), Oxford: Oxford University

Press.

[48] Staiger, R. and F. Wolak (1992), “The Effect of Domestic Antidumping Law in the Presence

of Foreign Monopoly,”Journal of International Economics 32, 265–287.

42

[49] Tirole, J. (1988), The Theory of Industrial Organization, Cambridge, MA: MIT Press.

[50] United States International Trade Commission - USITC (2001), “Canned Pineapple Fruit from

Thailand,” USITC Publication No. 3417, Washington DC.

[51] United States International Trade Commission - USITC (2002a), “Certain Folding Metal Ta-

bles and Chairs from China,” USITC Publication No. 3515, Washington DC.

[52] United States International Trade Commission - USITC (2002b), “Lawn and Garden Steel

Force Posts from China,” USITC Publication No. 3521, Washington DC.

[53] United States International Trade Commission - USITC (2003), “Certain Color Television

Receivers from China and Malaysia,” USITC Publication No. 3607, Washington DC.

[54] United States International Trade Commission - USITC (2008), “Matchbooks from India,”

USITC Publication No. 4054, Washington DC.

[55] Vandenbussche, H. and X. Wauthy (2001), “Inflicting Injury Through Product Quality: How

EU Antidumping Policy Disadvantages European Producers,” European Journal of Political

Economy 17, 101-16.

[56] Vanden Bussche, H. and M. Zanardi (2008), ’Antidumping,’ Economic Policy, 23 (Jan), 94-138

[57] Vanden Bussche, H. and M. Zanardi (2010), ’The Chilling Trade Effects of Antidumping

Proliferation,’ European Economic Review 54(6), 760-777.

[58] Veugelers, R. and H. Vandenbussche (1999), “European Antidumping Policy and the Prof-

itability of National and International Collusion,”European Economic Review 43(1), 1-28.

[59] Weinstein, D.E. (1992), ”Competition and Unilateral Dumping,” Journal of International

Economics 32, 379-387.

[60] World Trade Organization - WTO (2002) Annual Report 2002, Geneva.

43