A partnership with academia

Building knowledge for trade and development

Vi Digital Library - Text Preview

Nondiscrimination in GATT/WTO: Was There Anything to Begin with and is There Anything Left?

Discussion paper by T.N. Srinivasan, 2005

Download original document (English)

What: Non-discriminatory treatment is one of the main principles of the World Trade Organization (WTO). Under normal circumstances countries cannot discriminate between their trading partners: they must accord the same preferences to one country as they do to all their trading partners (Most Favoured Nation Principle [MFN]). Under WTO rules foreigners and locals also have to be treated equally in domestic markets (National Treatment [NT]). However, exceptions to non-discriminatory treatment are outlined in the WTO articles relating to customs unions and free trade areas, antidumping and safeguards. In this paper the author argues that these exceptions have become dominant over the past few years causing an erosion of the non-discrimination principle. In Section 2, non-discriminatory treatment, as mandated in the General Agreement on Tariffs and Trade (GATT), is outlined. Section 3 summarizes the exceptions to non-discrimination in the WTO. Section 4 examines the role of MFN and reciprocity as a means of self-enforcement of the GATT contract. In section 5 aspects of MFN and NT are analyzed with a simple theoretical model. Section 6 offers a conclusion on the role of non-discrimination in GATT/WTO. Who: For anyone dealing with the non-discrimination principle of the WTO. How: Can serve as an additional reading for anyone dealing with the non-discrimination principle of the WTO. Preliminary knowledge of the principles of non-discrimination and of economic theory would be an asset.

Nondiscrimination in GATT/WTO: was there
anything to begin with and is there
anything left?


Abstract : Nondiscriminatory treatment at its border of a like product coming
from another WTO signatory (Article I on Most Favored Nation Treatment), and
of domestic and foreign suppliers of like or similar products within its borders
(Article III on National Treatment) are widely held to be the fundamental
principles of GATT/WTO. Yet GATT included significant exceptions to
nondiscriminatory treatment, for example, in its articles relating to customs
unions and free trade areas, antidumping and safeguards. I argue that these
exceptions have become dominant over time so that not much nondiscrimination
remains in the global trading system. With the recent inclusion of services,
intellectual property and trade-related investment measures, traditional GATT
issues of tariff and non-tariff barriers at the border to market access have become
less important compared to regulatory barriers inside the border. It is an open
question whether nondiscrimination per se is a salient issue in thinking about
multilateral disciplines in these new areas.

1. Introduction

Robert Hudec was an unusual scholar. Trained in law, he was an excellent econ-

omist among lawyers, an outstanding lawyer among economists, and equally at

ease in engaging both groups in scholarly arguments. He has written extensively

and incisively on international trade law and legal aspects of the (non) institution

that laid out the rules of the international trade game, namely, the General

Agreement on Tariffs and Trade (GATT) and its successor institution, the World

Trade Organization (WTO).

The locus classicus for a critical assessment of the case against discrimination

in the use of trade measures from economic, legal and political perspectives is

Hudec’s (1988) analysis. There is not much I can hope to add to this classic and

will not attempt it in this paper. Many of his other writings on issues of discrimi-

nation, are reproduced in Hudec (1999a). His contributions to Bhagwati and

* Samuel C. Park, Jr Professor of Economics, Yale University, 27 Hillhouse Avenue, New Haven,

Connecticut USA 06511. E-mail : t.srinivasan@yale.edu. I thank Kyle Bagwell, Wilfred Ethier, Bernard
Hoekman, Philip Levy, and Robert Staiger for their helpful comments.

World Trade Review (2005), 4 : 1, 69–95 Printed in the United Kingdom
f T. N. Srinivasan doi :10.1017/S1474745605002144


Hudec (1996) are also relevant. In contrast to a typical footnote-festooned article

in law journals in which the ratio of footnotes (with a liberal sprinkling of Latin

words!) to a turgid text often exceeds one by a large margin, Bob’s articles are

fun to read and the average footnote ratio in them is far less than one! One of my

favorites is on GATT’s legal system, originally published in 1970 and reprinted

in Hudec (1999a). I would like to begin with it in my tribute to Hudec.

In this article, Hudec notes the contradiction in GATT between the complex and

carefully drafted, ‘with lawyerlike precision’, articles on substantive obligations

of the contracting parties, and enforcement procedures which ‘present a front of

ambiguity and uncertainty’, and ‘make no functional distinctions between breach

of legal obligations and other grievances’. He viewed the contradiction as ‘mani-

festation of a distinctive jurisprudence [that is] primarily the work of diplomats

rather than their lawyers’ (Hudec, 1999a: 17). A rationale for diplomats’ rather

than lawyers’ jurisprudence in GATT was the perception that handling disputes

over matters of international commerce was too important to be left to lawyers

and judges. Hudec traces the skepticism about lawyers to a report in 1932 of

the League of Nations’ Economic Committee. After quoting the following from

the report :

bodies composed of judges, who cannot be thoroughly well acquainted with all
the details of economic life, and who are rather inclined to rely on criteria of pure
law in judging cases in which situations of fact and technical considerations
are of predominant importance, do not always appear to operate in a way satis-
factory to the parties. Moreover, it appears... that the [International] Court itself
is of the opinion that judicial settlement is not always the best way of settling
disputes of an economic nature (League of Nations, 1932: 4)

he wryly remarks that: ‘The talk about the crippled intellect of judges and lawyers

makes interesting reading, but the rationale was not quite candid. ’ He then adds:

Most of the law’s subject matter is economic, and a very large part of it is just as
complicated as the subject matter of trade agreements. The real problem with
lawyers and judges had been their failure to understand the need for compromise
in these matters. The point (and the international lawyers of the day may well
have missed it) was that these legal obligations were not meant to be enforced to
the letter. (Hudec, 1999a: 21)

It is ironic that the diplomats’ jurisprudence, to use Hudec’s pithy description

of the GATT legal system and particularly its ‘political ’ dispute settlement

mechanism (DSM), which, according to Hudec (1999b), functioned well, has been

replaced in the WTO by an adversarial legal process.1 Although the persons now

1 Hudec (1999a: 17–41) discusses the negotiating history of the charter of the stillborn International

Trade Organization (ITO) of which GATT was to have been the Commercial Policy Chapter. The first

working draft of the ITO Charter submitted by the United States, ‘outlined a three step procedure for

dealing with legal questions. Complaints were to be investigated and ruled upon by the Executive Board,
an eighteen member executive committee charged with conducting the day-to-day business of the


constituting the Appellate Body of the WTO’s DSM cannot be characterized in

the same way as the League of Nations report characterized the bodies composed

of judges adjudicating economic disputes, the ‘criteria of pure law’ in judging

economic disputes, which the report was critical of, seems to be ascendant in

today’s WTO. The deliberate and constructive ambiguity of diplomats’ jurispru-

dence as incorporated in the GATT certainly opened up opportunities of legalistic

battles of textual interpretation. Despite this, the political process of GATT, in

which a defendant contracting party could veto the appointment of a panel for

adjudicating the complaint against it, and both defendants and plaintiffs could

veto the adoption of the findings of a panel if one were to be appointed, had strong

incentives to nudge the parties to arrive at a compromise.2 The WTO system, by

making the appointment of a panel in effect automatic and requiring a consensus

in the General Council for overturning the findings of the Appellate Body, has

eliminated these incentives. Largely at the successful insistence of the United States

(US), intellectual property protection was brought into the GATT, rather than

having it remain with the World Intellectual Property Organization. US pressure

was also behind ‘ legalizing’ the DSM of the WTO. This has encouraged the

demand for inclusion of labor standards and other issues into the WTO to take

advantage of its perceived strong enforcement capabilities through its DSM.

Another consequence of the ascendancy of pure law jurisprudence is the possi-

bility of judicial activism by the Panels and Appellate Body through interpretation

of ambiguous GATT texts. There is some evidence that this is already happening in

some of the decisions of the Appellate Body on some recent Article XX disputes,

ITO. Rulings of the Executive Board could be appealed to the Conference, the plenary body of all mem-
bers. Rulings of the Conference in turn, could be appealed to the International Court of Justice. This final

stage had one significant limitation. With only minor exceptions, appeals to the ICJ were to be allowed

only ‘‘ if the Conference consents’’ ’ (US Department of State, 1948). In the debates over the draft, some

objected to the ‘consent’ requirement on the grounds that the decisions of the Executive and the
Conference might well be ‘political ’ rather than objective. Others defended it essentially on the same

grounds as the League of Nations Committee against adjudication of economic disputes by lawyers.

Hudec describes the compromise that solved the ‘consent’ issue as follows: ‘It was agreed that legal
questions could be appealed to the ICJ as a matter of right. On the other hand, such appeals would have to

take the form of a request by the ITO itself for an advisory opinion. The parties to the dispute would not

appear as litigants before the court. In particular, the losing party would not be subject to a decree of the

Court. Enforcement would be in the hands of the ITO, where, presumably, men of sound ‘economic’
judgment could take account of the ‘economic facts’ (Hudec, 1999b: 26). In the GATT, there was no

provision for appeal to ICJ at all. Nor is there any in the WTO.

2 On the other hand, the veto feature also encouraged disputants to ignore the dispute settlement

process of GATT altogether and engage in threats and counter threats at an early stage, in effect drawing
‘lines in the sand’ and making a political compromise very difficult. In the WTO system, there is an

incentive for a regime in power to impose a trade measure to influence domestic politics, fully aware that

the measure would be found in violation of WTO rules by the DSM, but only after the desired outcome in
domestic politics has been achieved. Removing the offending measure then would have no political cost to

the imposing regime and inflict only a temporary loss to its trading partners. The recent imposition of tariff

on steel imports prices to the mid-term elections to the US Congress and its later removal once the DSM

ruled against it after the election is an example. I thank Wilfred Ethier for drawing my attention to these
aspects of dispute settlement in GATT and the WTO.

Nondiscrimination in GATT/WTO 71

for example, in the shrimp–turtles dispute. Judicial activism of domestic courts are

kept in check by legislatures which can amend the constitution or repeal the laws

interpreted too broadly by the judges. There is no analogue of the legislature in

the WTO to act as a check on its Appellate Body.

In the rest of the paper, I describe (in Section 2) nondiscriminatory treatment

as mandated in Articles I and III of GATT, followed (in Section 3) by exceptions

to nondiscrimination in other articles and their continuing consequences, which

together show that there is not much nondiscrimination left in the global trading

system. Section 4 is devoted to a brief description of the role of MFN and reci-

procity as means of self-enforcement of the GATT contract. In Section 5 I explore

aspects of MFN and NT with a simple analytical model. Section 6 concludes by

noting that, with tariff barriers on most goods becoming far less important in

restricting market access, and the growing importance of services in international

commerce, non-tariff barriers at the border and regulatory barriers inside the

border have become far more pressing concerns than the discriminatory treatment

that was the concern of Articles I and III.3

2. Nondiscrimination in GATT/WTO

Adverse discrimination

According to the Oxford Universal Dictionary of 1955, to discriminate is to ‘make

or constitute a difference in or between; to differentiate ’. Clearly, discrimination

in this sense has no normative connotation. The dictionary goes further and

defines, in a normative sense, ‘ to discriminate against ’ as ‘ to make an adverse

distinction with regard to’ (emphasis added) and interestingly, cites as an example,

the discrimination (presumably by the United Kingdom ‘against certain imports

from the United States ’. Read with its meaning in the same dictionary, it follows

that the use of the word ‘adverse ’ in this context means that the discriminator is

‘acting against or actively hostile’ to the subject of discrimination, and the effect of

discrimination is ‘unfavorable, injurious, calamitous’ to the subject.

Elimination of discriminatory treatment as a goal in GATT

The General Agreement on Tariffs and Trade (GATT) was a contract among its

signatories, called contracting parties.4 Since tariff policy is a choice of sovereign

3 Although formally the General Agreement on Trade in Services (GATS) concluded at the Uruguay

Round, includes the analogues of Articles I and III, namely, Articles II, XVI and XVII, the issues in services
trade are different and more complex than those in goods trade. Whether discriminatory treatment is

adequately addressed by these articles is an open question. See Mattoo (2000).

4 Not all signatories were nation states (e.g., Hong Kong). Their essential characteristic was that they
constituted separate customs jurisdiction. In order not to confuse a signatory with a nation state in GATT

documents, signatories are called ‘contracting parties’. The same documents, wherever something refers

to all the contracting parties collectively, the phrase ‘Contracting Parties’ is written in capital letters

instead of using ‘GATT’. The reason was to ensure that there was no hint of GATT being construed as an
international organization (Hudec 1999a: 37–38). In this paper, by GATT I mean the version called


entities, a voluntary contract among them to restrain its exercise has to be

mutually beneficial. Having signed (and ratified) the contract, the signatories

commit themselves to abide by the terms of the agreement and not to engage in

acts that either violate the terms or are not in conformity, even if they do not

violate, with the expectation of mutual gains each party has about the behavior of

others. In the context of GATT, an act by one contracting party has to be viewed

as discriminatory if and only if the act is ‘unfavorable, injurious or calamitous’ in

the sense of reducing the gains of one or more of the other contracting parties. Any

such agreement, if it is to be meaningful, has to include a definition of an adverse

discriminatory act and specify a remedial action. The determination of the

adverseness or otherwise of an act from the perspective of the agreement will

naturally depend on an identification of their interests that parties view as being

promoted (and protected) by the agreement. The objectives of GATT, as explicitly

stated by the original twenty-three contracting parties, are very broad: ‘raising

standards of living, ensuring full employment and a large and steadily growing

volume of real income and effective demand, developing the full use of the re-

sources of the world and expanding the production and exchange of goods’

(GATT, 1994: 486). The Marrakesh Agreement establishing the World Trade

Organization (WTO) reiterated these objectives and added the following: ‘ex-

panding the production of and trade in goods and services, while allowing for

the optimal use of the world’s resources in accordance with the objective of

sustainable development, seeking both to protect and preserve the environment

and to enhance the means for doing so in a manner consistent with their respective

needs and concerns at different levels of economic development’ (ibid: 6). In

agreeing to the GATT, the original contracting parties viewed themselves as

‘entering into reciprocal and mutually advantageous arrangements directed to

the substantial reduction of tariffs and other barriers to trade and to the elimin-

ation of discriminatory treatment in international commerce’ (ibid: 486, emphasis

added). Implicit in this is the presumption that reducing tariffs and other barriers

to trade, as well as eliminating discriminatory treatment, were essential to the

pursuit of the broad objectives of the contracting parties.

Principles of nondiscrimination in GATT architecture

The discriminatory treatment that the agreement seeks to eliminate or, more

precisely, the nondiscriminatory treatment it mandates, is encapsulated in two

fundamental principles. The first is the much celebrated ‘General Most Favored

Nation’, or MFN, treatment for short, mandated by the agreement’s very first

Article I. The second is the ‘National Treatment on Internal Taxation and

Regulation’, or NT for short, which constitutes Article III. MFN treatment is

about nondiscriminatory treatment by one party of all other contracting parties

GATT 1994 that was part of Annex 1A of the Uruguay Round Agreement. It is the original GATT of
October 1947 with all its subsequent modifications.

Nondiscrimination in GATT/WTO 73

with respect to customs duties and charges (i.e., ‘border’ measures) of any kind

on products it trades with them. NT forbids discrimination by a party in favor

of its domestic product with respect to domestic taxes and regulation (i.e., inside

the ‘border’ measures). It is trivially obvious that to take cognizance of dis-

crimination one has to limit the set of products such that any differentiation

between products within the set with respect to border or inside the border

measures would be deemed prima facie discriminatory. The text of GATT itself,

reports of its working parties, panels on disputes, as well as the Appellate Body

of WTO’s Dispute Settlement Mechanism, use several phrases such as ‘ like

products’, ‘similar products ’, ‘directly competitive products’, and ‘competitive

or substitutable products’ to mention a few, in an attempt to set such a limit.

Depending on the meaning attached to each of the phrases, the associated sets

could consist of a few or many products. To compound this problem, different

articles of GATT use different phrases among the above list, thereby opening the

possibility that such use was deliberate, that is the drafters of GATT in fact

intended the article to apply to a few or many products depending on the phrase


Several other articles of GATT refer to nondiscriminatory treatment. These

include, in particular, Article II.1, VIII.1, IX.1, and XIII.1. Without in any way

minimizing the significance of these articles in discussion of nondiscrimination

in GATT/WTO, in the rest of the paper I will focus largely on aspects of MFN

and NT covered by Articles I and III. GATT also included several exceptions

to nondiscrimination, which taken together seriously weaken, if not altogether

eliminate, the force of Articles I and III. I will return to these exceptions in Section 3


Relevance of nondiscrimination in GATT in liberalizing market access

The signatories to GATT apparently did not view nondiscriminatory treatment

as something to be of intrinsic value but only as an instrument for achieving

the broad objectives they hoped to achieve with the agreement. In other words,

they did not view nondiscrimination among contracting parties as epitomizing

a principle of justice or fairness in some well-defined sense. This being the

case, the argument sometimes advanced that such nondiscrimination in GATT

is unfair because it implies ‘equal treatment of unequals ’, whatever this means,

is irrelevant. It is clear that the signatories viewed discriminatory treatment in

5 There is a large body of case law involving the interpretation of these phrases in GATT/WTO

disputes (Bronckers andMcNelius, 2000; Horn andMavroidis, 2004; Hudec, 1999a: 365–367 and 2000;

Neven, 2000). From my probably very biased perspective of an economist, much of the discussion in the
panel reports and Appellate Body decisions appear to ignore the underlying economic issues in favor of

textual analysis of GATT, very much in the tradition of the Brahmin ‘Bhashyakars’ (or interpreters) in

Hinduism or of ancient rabbis of Talmud! Bob, who was clearly stung by the characterization of lawyers

and judges as incapable of economic analysis in the League of Nations Report, surely would have chided
me for my bias had he been with us!


international commerce, not as being unfair in some relevant sense (if any),

but as subverting the broad objectives that signatories sought to promote

through GATT.6 It raises the question of how precisely such subversion could


A somewhat superficial answer to this question can be seen in the fact that the

primary instrument of GATT for achieving the objectives is through liberalization

of the access of each contracting party to the markets of other contracting parties.

Any discrimination by one party in favor of a product originating in another party

in its market for the product through border measures is seen as affecting the

access of the parties not so favored to the same market.7 Article I prohibits such

discrimination and uses the phrase ‘ like products ’ to limit the set of products

within which prohibition applies. By the same token, an inside the border measure

favoring a domestic product in its market by a contracting party is seen as affecting

the access of all other contracting parties to the same market. Article III prohibits

such discrimination among products within a set that is specific to each relevant

paragraph of the article, if such discrimination is applied ‘so as to afford protec-

tion to the domestic product’ (GATT, 1994, Article III.l).

Analytical foundation for liberalization of market access

A deeper answer lies in the implicit analytical foundation for GATT’s emphasis

on market access and its liberalization through multilateral tariff negotiations8

covered by Article XXVIII bis. This foundation is the proposition, often called

the first theorem of neoclassical welfare economics, that a global competitive

market equilibrium under free trade, loosely speaking, maximizes global welfare

or, more precisely, is a Pareto optimum in the sense that any deviation from it,

if it raises welfare of someone somewhere, it would do so only at the expense of

someone else (Grandmont and McFadden, 1972). Thus, tariff or other barriers to

trade that prevent global market integration, ipso facto, prevent the equilibrium

in such segmented markets from being Pareto optimal. Given this proposition

6 Hudec (1999a: 227–250) provides very thoughtful analysis of the spuriousness and pernicious nature

of fairness as a basis for trade laws and trade actions in the context of US trade policy. He rightly points

out that ‘first, fairness is a matter that governments determine unilaterally: there are relatively few inter-

national agreements regulating the substance of such claims, and there is no recognized tribunal to
adjudicate them in common law fashion. Fairness is largely what a government wants to call unfair (sic).

Second, fairness claims call for unilateral concessions. No one pays to have unfair conduct corrected. To

say that certain conduct is unfair is to say that the guilty party must correct it for that reason alone. In

practice, this means that fairness claims can become a base for simply bullying others in search of greater
commercial advantage’ (Hudec, 1999a: 227). His short paper on the political morality of trade nego-

tiations (Hudec, 1999a: 215–225), though not about fairness, is also relevant since it raises the issue of

whether pursuing legitimate objectives relating to health and environment would be undermined by the
pursuit of the original objectives sought of GATT signatories through trade liberalization.

7 The relevant market from the perspective of Articles I and III, especially the latter, is not clearly

delineated in GATT or in case law (Neven, 2000).

8 In fact, the eight successful rounds of multilateral negotiations under the auspices of GATT covered
not only tariffs and other trade barriers but also rules governing trade.

Nondiscrimination in GATT/WTO 75

as the foundation,9 the total elimination of barriers to free trade through

successive reductions as the driving force of GATT/WTO makes eminent sense.

However, the validity of this proposition depends on the satisfaction of some

strong assumptions, including, in particular, that markets for goods, services

and factors are perfectly competitive and there are no production or consump-

tion externalities. Except for Article XVII on State Trading Enterprises, which

recognizes the possibility of such enterprises not acting ‘solely according with

commercial considerations’ (i.e., not behaving as competitive enterprises) and

also their being monopolistic (such as import monopoly, marketing boards, etc.),

GATT implicitly assumes that markets are by and large competitive. The ‘behind

the border’ regulatory issues of competition policies, that are meant to ensure

that markets remain competitive, and formulating multilateral disciplines for

them, did not figure in the agreements concluding any of the eight rounds of

GATT negotiations. The topic ‘Trade and Competition Policy’ was discussed in

the first ministerial meeting of the WTO in 1996 in Singapore, and a working

group was set up to study it. Subsequent attempts were made to include it in the

negotiating agenda for the ongoing Doha round. These did not succeed. The

General Council of the WTO decided on August 1, 2004 that no work towards

negotiation on these issues will take place within the WTO during the Doha


Externalities and GATT

Externalities in the precise sense of invalidating the Pareto optimality of a free

trade competitive market equilibrium (FTCME) if unaddressed are also not

explicit in GATT. However, the recognition in Article XX(b) that protecting

human and plant life could justify departures from nondiscrimination should be

viewed as, in fact, a recognition of failures either by producers or consumers of

internalizing the external consequences to human or plant life of their actions

in a FTCME. Another example is the agreement on Application of Sanitary and

Phytosanitary Measures (SPM) concluded in the Uruguay Round. It elaborates

rules for the application of the provisions of GATT, which relate to the use of

sanitary or phytosanitary measures, in particular the provision of Article XX(b)

and its chapeau.

Environmental externalities are well known. These are at the center of the

debates on global warming, for example. The decision on trade and environment

in the Uruguay Round to establish a committee on Trade and Environment open

to all members of the WTO was a response to the claims that trade liberalization

could conflict with environmental protection. It was also in part a recognition of

9 The analysis of Bagwell and Staiger (2002), which I discuss in Section 4, provides a related but

distinct foundation, namely that tariff liberalization under reciprocity provides a means of escape to a

mutually advantageous equilibrium from a non-cooperative one. This is consistent with the goals of the

signatories of GATT, namely to raise standards of living through reciprocal and mutually advantageous
tariff reduction.


the need for coordination of trade and environmental policies, without exceeding

the competence of the multilateral trading system, which is limited to trade policies

and those trade-related aspects of environmental policies which may result in

significant trade effects for the members of WTO. The committee’s terms of

reference, inter alia, are to address ‘ the relationship between the provisions of

the multilateral trading system and trade measures for environmental purposes,

including those pursuant to multilateral environmental agreements; the relation-

ship between environmental policies relevant to trade and environmental measures

with significant trade effects and the provisions of the multilateral trading system’

(GATT, 1994: 470–471). The committee has submitted several reports. At its

last meeting, the General Council of the WTO in July 2004 simply took note of

these reports without taking any further action.

Article XX(b), the agreement on SPM, and also Article III explicitly recognized

the possibility of a protectionist misuse of the exceptions to nondiscriminatory

treatment allowed under them and prohibited such misuse.10 Establishing such

misuse of allowed nondiscriminatory treatment by a contracting party involves,

in principle, showing that, first, there was discriminatory treatment; second,

there was protectionist intent or aim behind it ; and, third, that discriminatory

treatment had the effects of hurting other contracting parties. The record of

GATT/WTO jurisprudence on an ‘aim and effects ’ test is mixed (Hudec, 1999a:


For showing discriminatory treatment, it has to be shown, first, that the

products being treated differently are among ‘like’ products (or any other

phrase used in the relevant article for defining the set of products among which

discrimination is disallowed by it). Second, in the GATT architecture (at least

until the celebrated tuna/dolphin case) the only distinctions recognized were

based on qualities of products themselves or other characteristics that indirectly

affect product qualities, such as attributes of production processes or of the

producer. This being the case, even if the discrimination is based legitimately

on differences in products other than their qualities, it would violate Article III,

for example. Showing protectionist intent, obviously involves an examination

of policy makers’ objectives, and, in particular, whether alternative less pro-

tectionist means were available for achieving their objectives. Showing effects of

discriminatory treatment at the very least involves an analysis of the demand

and supply behavior in the market for the products. Needless to add that

establishing that nondiscriminatory treatment has, in fact, occurred is not


10 For example, in Section 5, I consider a model in which a domestic consumption tax which dis-

criminates in favor of a domestically produced import substitute, but which could be justified on the

grounds of differential externality effects (for example, environmental pollution) between domestic and

imported goods. The issue then becomes one of precluding the use of such a tax for purely protectionist
(i.e., to favor domestic substitute) purposes, while claiming it to be justified on environmental grounds.

Nondiscrimination in GATT/WTO 77

3. Exceptions to nondiscrimination in GATT/WTO

Relevant GATT Articles

There are several articles of GATT that allow exceptions to nondiscrimination. As

noted earlier, Article XX allows a contracting party to choose discriminatory

measures, which are not, applied in a manner which would constitute ‘a means

arbitrary and unjustifiable discrimination’ in order to achieve objectives listed

in the article. Article III allows discriminatory use of internal taxes, laws and

regulation as long as they are not ‘applied so to afford protection to domestic

production’. Article VI, and the agreement in the Uruguay Round on its imple-

mentation deal with anti-dumping measures, which are by definition discriminat-

ory, not only with respect to countries of origin of the product involved, but also,

in principle, between exporters from the same country who are found to have

dumped and those who are not. Article XIX on emergency action on imports

of particular products, and the agreements on safeguards in the Uruguay Round

relating to it, allow quantitative restrictions or quotas to be imposed.11 Although

Article XIII of GATT and Article 2, Paragraph 2, of the Safeguards Agreement

explicitly prohibit the discriminatory application of safeguards measures by

source, as Hudec (1988: 176–177) argued:

Where tariffs are concerned, there is a reasonably solid consensus about the
core meaning of the most-favored nation (MFN) concept_ There is no such
commonly accepted core meaning with regard to quantitative restrictions (QRs).
There is agreement about one basic requirement: if a QR is imposed on imports
of a given product, it must be imposed on imports of that product from all
countries. Although this is a necessary condition, it is not a sufficient condition.
Everyone agrees that the MFN principle requires something more in the case of
QRs. It requires that themanner of applying the QR also be ‘nondiscriminatory’.
Unfortunately, there has never been an agreed definition of exactly what this
added requirement contains.12

Other articles permitting the imposition of QRs and hence, subject to the

problem pointed out by Hudec, are the so-called balance of payments exceptions of

Articles XII.1 and XVIII.9. Article XIV permits exceptions to nondiscriminatory

treatment of QRs mandated in Article XIII in payments arrangements for current

11 Administered protection measures such as voluntary export restraint measures and, of course,

bilateral quotas such as those of the Multifibre Arrangements are, of course, discriminatory. However,

they involve implicit compensation to the injured since they get to keep the quota rents. Because of this
compensation feature, they make such measures less malign as compared to antidumping measures, which

are discriminatory but do not involve compensation. See Ethier (2002). My reading of Ethier’s analysis is

that some forms of discrimination combined with compensation for the discriminated against help in
precluding the use of worse forms of discrimination. However, I do not see this as an argument for

discrimination per se.

12 Ethier, in a private communication, points out that to the extent MFN contributed to lowering of

tariffs, it sowed seeds of its own destruction in the sense of making discriminatory non-tariff barriers more
attractive, since tariff preferences are not attractive at low levels of MFN tariffs.


transactions in order to restrict imports. Since by definition Customs Unions

and Free Trade Areas discriminate between their members and non-members with

respect to tariffs, by not ruling them out, but explicitly allowing their formation

subject to some admittedly stringent conditions laid out in Article XXIV, GATT

included another permitted exception to nondiscrimination. Article XXI allows

discriminatory treatment for national security reasons. Last, but not least, devel-

oping countries and least-developed countries have been exempted from almost

all articles of GATT that mandate nondiscriminatory treatment. Although the

provisions in GATT in favor of developing countries date back to 1955, Part IV on

Trade and Development incorporated in 1964 and the so-called Enabling Clause

of the Tokyo Round Agreement of 1979 contain the most important of the

exemptions from GATT disciplines that developing countries have been allowed.

These exemptions, under the rubric Special and Differential Treatment (SDT),

have become, in effect, entitlements that developing countries wish to extend

in the ongoing Doha Round.13 Thus, from its very inception, GATT’s ‘funda-

mental ’ principle of nondiscrimination was heavily diluted and compromised by


Have exceptions become the rule rather than proving it?14

I will confine myself to three of the exceptions to nondiscrimination among

the many discussed in Section A. The first relates to nondiscriminatory application

of QRs. This has been explicitly violated during the GATT era, although the

violations were not formally challenged. Agricultural trade has not been subject

to disciplines of GATT regarding the use of QRs, export subsidies and non-

discrimination almost since the signing of GATT in October 1947. Although the

Uruguay Round Agreement on Agriculture began a process of reintegrating agri-

cultural trade into GATT, it was fairly limited and far from complete.15 The ‘July

Package’ adopted by the General Council includes an agreement ‘to establish

detailed modalities ensuring the parallel elimination of all forms of export sub-

sidies and disciplines on all export measures with equivalent effect by a credible

end date. ’ The end date is yet to be agreed. However, tariff-quota schemes will

not be phased out. A complex tiered formula for overall reduction of aggregate

measure of support has been agreed to, though total elimination of trade-distorting

13 The trade preferences and related discriminations span most of the developing countries in the

WTO. The so-called G-90, consisting of 90 developing countries, is one such group.

14 Again, my Oxford Universal Dictionary defines: ‘The exception proves the rule’ as a legal maxim
meaning ‘Exception proves the rule in the cases not excepted.’

15 The attempt to convert various border measures into a single ad valorem tariff equivalent so that the

resulting tariff could be bound and reduced led to a so-called ‘dirty tariffication’ process by which tariff
equivalents were bound at absurdly high levels that were way above their applied levels. Reduction

commitments from such high levels made a mockery of the commitments – even at the end of the period of

implementation of these commitments, the bound levels would still be much higher than applied levels

so that there was virtually no liberalization of agricultural trade in the Uruguay Round Agreement on

Nondiscrimination in GATT/WTO 79

domestic support and of special and differential treatment are not contemplated.

Thus, significant discrimination will remain in agricultural trade even after

the conclusion of the Doha Round. Although QRs, such as ‘voluntary export

restraints ’ negotiated by the US and the European Union with Japan on its export

of automobiles, were eliminated as part of the Uruguay Round Agreement, the

possible use of QRs for addressing balance of payments (BOP) problems still exist,

since provisions of Articles XII and XVIII permitting such use have not been

repealed. In fact, India invoked Article XVIII when its QRs were challenged, even

though it did not have any BOP problem. Only after the WTO’s Appellate Body

ruled against it did India eliminate QRs in 2001.

Trade in textiles and apparel has been exempted from GATT rules since 1961;

the initial Short-Term Arrangement covering cotton textiles was quickly converted

to a Long-Term Arrangement in 1962, and 12 years later this was expanded into

the Multifibre Arrangement (MFA) in 1974, which covers trade in textiles made

from almost all natural and man-made fibers! The MFA has been a particularly

egregious exception to GATT rules: apart from being an outright violation of

MFN, it also permits the use of bilaterally negotiated trade quotas (in violation

of Article XI) on an item-by-item basis between each importer and exporter. One

cannot imagine a worse way of segmenting and heavily distorting markets. Under

the Agreement on Textiles and Clothing (ATC) in the Uruguay Round, MFA is

scheduled to expire on 1 January 2005, after being in force for nearly three

decades. However, the phasing-out of the quotas of MFA is back-loaded. Nearly

half the quotas of MFA will remain in place until the moment before its expiry.

Although the ten-year phase-out period was meant to enable the industry in

developed countries which would be uncompetitive in a free trade regime to

adjust, not much of an adjustment has taken place. The fear that a very competi-

tive China may capture the market after the phase-out of MFA has already

spawned attempts by a coalition of non-competitive producers in industrial and

developing countries to seek a delay of the phase-out and also to use the special

provision negotiated by members of the WTO in China’s agreement of accession

to the WTO to restrain its exports. Although Donald Evans, Commerce Secretary

in the US government, has opposed delaying the phase-out, it is not entirely out of

the realm of possibility that discriminatory treatment in some form will continue

to exist even after the phase-out of MFA.16

The exceptions from MFN for customs unions (CUs) and free trade areas

(FTAs) under Article XXIV of the GATT (and its updated version in the WTO),

under Part IV of the GATT, relating to economic development adopted in

1965, and under the enabling clause of the Tokyo Round of MTN in 1979,

‘Differential and More Favorable Treatment, Reciprocity and Fuller Participation

16 Ethier, in a private communication, points out that at least the injured DCs got compensated in part

for quota rents they kept in the MFA. This being the case, replacement of MFA on its phase-out by other
protective measures with no such compensatory features will hurt some DCs.


of Developing Countries ’, are serious, since any preferential and discriminatory

trade agreements, were they to be found consistent with these clauses, could last


Any proposed CU or FT agreement was required to be promptly notified to

the GATT for examination by a working party. In all, 98 agreements were notified

under Article XXIV during the life of the GATT from 1947 to the end of 1994,

including the most enduring of all, namely the European Economic Community

(EEC) and the European Free Trade Area (EFTA). A further 11 agreements were

notified by developing countries under the 1979 enabling clause. Working parties

were established to examine virtually all agreements. While 15 working parties

had not completed their examinations as of the end of 1994 and five did not

report for various reasons, out of the 69 that had submitted their repots, only

six explicitly acknowledged conformity with Article XXIV of the agreement, and

this six does not include the EC or the EFTA, and only two of the six are still

active! (WTO, 1995: 16). In fact, ‘no agreement was reached on the compatibility

of the Treaty of Rome [that established the EEC] with Article XXIV, and the

contracting parties agreed that because ‘there were a number of important

matters on which there was not at this time sufficient information_ to complete
the examination of the Rome Treaty_ this examination and the discussion of
the legal questions involved in it could not be usefully pursued at the present

time’. The examination of the EEC agreement was never taken up again’ (WTO,

1995: 11).17

Whether or not a CU or FTA that is consistent with Article XXIV would have

increased global welfare, it is abundantly clear that the procedures laid down for

examining such consistency have not worked. The WTO report rightly concludes


[w]hile the purpose of the Understanding on Article XXIV is to clarify certain
of the areas where the application of Article XXIV had given rise to controversy
in the past, and particularly as regards the external policy of customs unions, it
fell short of addressing most of the difficult issues of interpretation noted above.
For example, no consensus emerged in the Uruguay Round Negotiating Group
on GATT Articles concerning proposals made by several participants (notably
Japan), to clarify the substantially-all-trade requirement. It is evident, therefore,
that most of the problems that have plagued the working party process were not
solved in the Uruguay Round. (WTO, 1995: 20)

Article V of the GATS, which corresponds in many ways to Article XXIV on goods

trade, shares many of the unsolved problems of Article XXIV. In sum, the WTO

articles and procedures as they are now are unlikely to succeed in the future in

17 Philip Levy, in a private communication, points out that the European Coal and Steel Community,

the precursor of the EEC, was neither a CU nor an FTA. It clearly violated MFN. Its acceptance (or, more

precisely, its non-rejection for its violation of MFN) by contracting parties of the GATT is the ‘original
sin’ from which everything evolved subsequently!

Nondiscrimination in GATT/WTO 81

resolving, any more than GATT articles and procedures did, the tension, if not

outright contradiction, between discrimination which is an inherent feature of

CUs and FTAs and the fundamental principle of nondiscrimination.

WTO (2004: 68–70) reports that the number of preferential regional trade

agreements continue

to rise furthering the growth of preferential and discriminatory trade
relations_ The impasse experienced at Cancún and the resulting uncertainty
concerning the fate of the Doha Round has apparently precipitated the pursuit
of RTA partners with the announcement of negotiations of several new RTAs
and many more in the proposal stage_ in the Asian-Pacific region, where
the expected finalization of several bilateral RTAs will soon grant Mongolia the
status of the only WTO Member [among 148] not yet party to any RTA.

Developments in the global RTA landscape in 2003 show that all WTOMembers
are increasingly zealous in developing networks of preferential partners.
While the promotion of free trade at a preferential level may exert leverage for
openness and competitive liberalization in international trade relations thus
benefiting the multilateral process, this strategy carries certain inherent risks.
Such risks include the capacity of RTA partners to negotiate and administer
multiple agreements with the attendant risk of diminished attention to the
multilateral system; the creation of vested interests in FTA partners which will
resist dilution of preferential margins at the multilateral level ; and the impact
of these agreements on third countries, for example through trade and invest-
ment diversion. The Committee on Regional Trade Agreements (CRTA), the
body entrusted with verifying the compliance of RTAs with the relevant
WTO provisions, continued its examination of RTAs in 2003. However, the
CRTA made no further progress on its mandate of consistency assessment, due
to long-standing institutional, political and legal difficulties. Since the establish-
ment of the WTO, Members have been unable to reach consensus on the format,
and the substance, of the reports on any of the examinations entrusted to the

Of the original 23 contracting parties of GATT, nearly half were developing

countries (DCs), and many of them still are. As of 1 October 2004, an over-

whelming majority of the membership consists of DCs. However, DCs did not

participate effectively in the GATT and the first six founds of multilateral trade

negotiations under its auspices. In the seventh Tokyo Round, which concluded

in 1979, they participated in strength and with cohesion. Their experience in the

GATT18 up to the conclusion of the Tokyo Round could be interpreted in two

diametrically opposed ways. On the one hand, it could be said that from the

Havana conference on, the developing countries have been repeatedly frustrated

18 See Hudec (1987) and Srinivasan (1998) for an extended discussion of the experience of developing

countries in the GATT. On Special and Differential Treatment, see Hoekman (2004); Hoekman,
Michalopoulos, and Winters (2004); and Keck and Low (2004).


in getting the GATT to reflect their concerns. Tariffs and other barriers in

industrialized countries on their exports were reduced to a smaller extent than

those on exports of developed countries in each round of the MTNs. Products

in which they had a comparative advantage, such as textiles and apparel, were

taken out of the GATT disciplines altogether. Agriculture, a sector of great

interest to developing countries, largely remained outside the GATT framework.

‘Concessions’ granted to developing countries, such as the inclusion of Part IV

on trade and development and the Tokyo Round enabling clause on special and

differential treatment, were mostly rhetorical, and others, such as GSP, were

always heavily qualified and quantitatively small. In sum, the GATT was un-

friendly, if not actively hostile, to the interests of developing countries.

The other interpretation is that the developing countries, in their relentless

but misguided pursuit of the import-substitution strategy of development, in

effect opted out of the GATT until the Tokyo Round. The formal incorporation

at the Tokyo Round of their demands for a differential and more favorable

treatment, including not being required to reciprocate regarding any tariff

concessions’ by the developed countries, triply hurt them: once through the

direct costs of enabling them to continue their import-substitution strategies,

a second time by allowing the developed countries to get away with their own

GATT-inconsistent barriers (i.e., in textiles)19 against imports from developing

countries, and a third time by allowing the industrialized countries to keep

higher-than-average MFN tariffs on goods of export interest to developing

countries. Instead of demanding and receiving crumbs from the rich man’s table,

such as the Generalized System of Preferences (GSP) and a permanent status of

inferiority under the ‘special and differential ’ treatment clause, had they partici-

pated fully, vigorously, and on equal terms with the developed countries in the

GATT and had they adopted an outward-oriented development strategy, they

could have achieved far faster and better growth. The success of East Asia since

the mid-sixties, and China and India since the eighties, suggests that the second

interpretation is closer to the truth. Still, in the Doha Round DCs are insisting

on the continuation and extension of SDT. It is likely that their demands will

be conceded by the other members, thereby institutionalizing the inherently

discriminatory SDT.

Several distinct issues get confounded in the arguments for and against SDT.

First is the issue whether, within the same rule-based organization such as WTO,

the rules should be contingent on a country’s status as developed, developing

but not least developed, or least developed, given that the criteria for classification

of countries has nothing to do directly with the rules? The answer is no as long

as all members participate effectively in the making of rules.

19 Philip Levy correctly points out that these so-called ‘grey area’ measures, though they certainly

violate the spirit of GATT, do not ‘formally’ violate GATT rules because the violation was not formally
pronounced as such by a GATT body.

Nondiscrimination in GATT/WTO 83

Undoubtedly, the impact of a tariff on an economy and its trading partners will,

in general, depend both on the level of the tariff and also the market structure

(i.e., whether it deviates from the norm of pure competition). It is often the case

that the tariff structure (i.e., the vector of tariffs across commodities) in, say,

developed countries, is such that the tariffs and other barriers on commodities

that largely DCs export are higher (the problem of tariff peaks) than on other

commodities, and they are also higher on processed goods than on raw or semi-

processed material used in processing and which are again exported by DCs (the

problem of tariff escalation). However, such tariffs are almost always on an

MFN basis, so that their adverse impact on DCs is not from discriminatory treat-

ment of DCs in comparison to other exporters, which would have been a violation

of Article I. Certainly, there is a strong case for eliminating tariff peaks and tariff

escalation, but it is not a case for SDT. I have argued that such peaks and escal-

ation reflect, in large part, the fact that the DCs had ‘opted’ out of GATT for a

long time. In the GATT negotiating process, reciprocity (i.e., exchange of tariff

reductions by each part in return for similar reduction by others) is the norm.

Having gotten an exemption from having to reciprocate, the DCs lost any

bargaining leverage to get the other countries to lower peaks and avoid escalation

and have to depend essentially on their good will. Asking for continuation of SDT

is not of much help from this perspective.

In their very thoughtful analysis, Keck and Low (2004) distinguish the following

five arguments that have been advanced in favor of SDT:

. Special and differential treatment is an acquired political right.

. Developing countries should enjoy privileged access to the markets of their
trading partners, particularly the developed countries.

. Developing countries should have the right to restrict imports to a greater degree
than developed countries.

. Developing countries should be allowed additional freedom to subsidize exports.

. Developing countries should be allowed flexibility in respect of the application
of certain WTO rules, or to postpone the application of rules.

For arguments based on rights to be convincing, the rights have to emerge from

a normative framework, be coherent in the sense of the existence of an agency

whose duty it is to provide such rights, and have to have some universal values

underpinning them. None of these hold for SDT as a right. The normative frame-

work of GATT/WTO is the efficiency and Pareto optimality of free trade – SDT is,

by definition, in conflict with it. It cannot be the duty of the WTO to provide such

a right, since the WTO is simply an institutional framework for the conduct of

trade relations among its members related to agreements they have concluded

among themselves. Certainly, SDT has no intrinsic and universal value – whether

it even has an instrumental value in promoting development is arguable. The

second argument is, again, unconvincing. DCs should be interested in the removal

of all barriers to market access and not just relative reduction of barriers to


their export compared to those of others. Since trade restriction is rarely the

first best instrument for development, the third and fourth arguments are not


The fifth argument is more convincing in that it asks for flexibility in the appli-

cation of common rules rather than for different rules. It is reasonable to argue

that implementation of rules, and regulations and reallocation of resources in

response to trade liberalization do involve significant adjustment costs. Also,

the administrative structure and institutions (including legal institutions) of DCs

might not be adequate for implementing the rules and bringing about regulatory

changes quickly. For these reasons, giving DCs a longer time to bring their trade

policies and regulations in conformity with WTO rules and, above all, assisting

them with resources to ease the burden of adjustment would be appropriate while

formally repealing SDT.

It is evident that the exceptions to nondiscriminatory treatment that were

built into GATT are no longer exceptions but in fact the dominant feature of

the global trading system. It would not be surprising if a carefully estimated

proportion of world trade in goods and services as of 2004 that is free of any

discrimination turns out to be very small. Thus, discrimination is the rule and

nondiscrimination the exception in the WTO.

4. MFN and reciprocity as self-enforcement devices

In Section 2, I suggested that the analytical foundation for the focus of GATT

on progressive liberalization of market access with the end goal of barrier-free

movement of goods and services is the first welfare theorem of neoclassical

welfare economics, as extended to international trade, by Grandmont and

McFadden (1972). This extension shows that under certain assumptions, a free

trade competitive market equilibrium is Pareto optimal. An implication of

Pareto optimality is that the global resource allocation and trade flows associ-

ated with it are efficient. Hudec (1988) lucidly discusses how discriminatory

tariffs and quotas impose an efficiency loss. Both my argument and Hudec’s

in effect assume that the multilateral trade agreement GATT exists and focus

on its rules and roles. The statement that avoiding the repetition of the inter-

war experience with beggar-thy-neighbor trade policies, competitive devalu-

ation, etc. was the prime motivation for the negotiations that led to GATT,

though true, is not adequate. An analytical argument will not only explain

why the agreement was in the mutual interests of the parties, but also why

it has proved resistant to defections and remains stable over time, even though

20 The Doha Round is often called the Development Round. This is unfortunate since it leads to the

mistaken impression that once the Round is concluded to the satisfaction of all members of the WTO, the

problem of development will be solved. Unfortunately, the problem is far more complex, and the barriers

to trade and investment that developing countries face are not the dominant constraints on development.
The relevant constraints are largely in domestic political economy.

Nondiscrimination in GATT/WTO 85

it has no third party enforcement mechanism.21 In this section I will briefly note

two contributions (Bagwell and Staiger, 2002 and Ethier, 2004), which provide

such an argument and which also rationalize the pillars of MFN and reciprocity

in the GATT.

Bagwell and Staiger distinguish three approaches to determining the problem

that a trade agreement can be presumed to solve. In the first, which they call the

traditional economic approach, the problem is a terms-of-trade or other exter-

nality from unilateral (i.e., non-cooperative) tariff setting by national welfare

maximizing governments of countries which have market power. In the absence

of a trade agreement, the externality would lead to a trading equilibrium, which

is inferior, in the sense of being worse from every participant country’s perspective,

compared to an (cooperative) equilibrium associated with an appropriate trade

agreement. In other words, a trade agreement is a means of escape from a

Prisoner’s Dilemma.22 In the second, called the political economy approach,

the objective function of the government is generalized to include other con-

siderations, particularly political considerations, besides national welfare. The

generalization, while bringing in more realism, leads to the same conclusion as

the traditional approach, namely that a trade agreement is a means of escape

from a Prisoner’s Dilemma. The third, the commitment approach, is different

from the first two in that the inefficiency of the equilibrium without a trade

agreement arises, not from an externality, but from the inability of governments

to commit credibly to implementing contingent trade policies. Thus, a govern-

ment that announces a policy of reforms including trade liberalization may be

tempted to renege, if it is in its interests, once the private sector undertakes

the actions consistent with reform. Anticipating such reneging, the private sector

will not undertake the actions, and the announced reforms will fail. However,

if a trade agreement ties the government’s hands, and this tying is credible

because it is costly for the government to defect from the agreement, then credi-

bility of tying, in effect, means credibility of the government’s commitment to


In the first two approaches, since the source of the inefficiency of a trade

equilibrium in the absence of an agreement is the inefficiency from a terms-of-trade

21 Coneybeare (1987: 278–279) argues that multilateral agreements with MFN provisions are in-

herently unstable and would oscillate between multilateral arrangements and a reversion to bilateral and

other such contracts. He finds confirmation for his argument in international tariff history. The reason for

this oscillation is that ‘a large-numbers MFN system will break down, in the absence of an enforcement
mechanism against free-riders, but the ensuing contracting costs (and possibly predatory behavior) will

create pressure for a collective return to MFN norms. What is needed is some efficient combination of

multilateralism and bilateralism. Large numbers of bilateral negotiations may be time-consuming, but at
least they do not create the same incentive for unconditional defection that is induced by a large-numbers

MFN-public good game.’ The instability problem is avoided by the reciprocity requirement in the analysis

of Bagwell and Staiger.

22 Coneybeare (1987: part II) uses the title ‘Prisoner’s Dilemma Trade Wars’ in describing various
trade wars since the fourteenth century.


externality, the motivation for a trade agreement to avoid the inefficiency is

clear. However, in the commitment approach, the source of inefficiency is

basically domestic. As such, a trade agreement even if it contributes to its

amelioration, unless it is the only means for amelioration, it is not necessarily the

best means The attempt of Bagwell and Staiger (2002: 4, footnote 4) to link

the domestic credibility problem to a cross-country externality is not entirely


Turning to MFN, Hudec (1988: 177) points out that no part of the modern

MFN concept, which in any case was developed solely with respect to tariff

policy, has a very solid historical foundation. Besides the efficiency argument that

an MFN tariff policy will ensure that imports are procured from the cheapest

source, the other conventional argument in its favor is that, in its absence, any two

countries would be reluctant to agree to an exchange of concessions in a trade

agreement for fear that the benefits of concessions obtained for one party from

another will be eroded if, in the future, the second party offers even greater con-

cessions to a third party. This is the so-called ‘erosion of benefits or preferences ’

argument. Horn and Mavroidis (2001) provide a comprehensive survey of the

economic and legal aspects of MFN. Their description of the critical role played

by MFN, combined with reciprocity in the Bagwell–Staiger analysis, is concise

and illuminating.

Basically, in the GATT agreement, any renegotiation of tariffs under Article

XXVIII and XXVIII bis is subject to the reciprocity requirement that renego-

tiations ‘maintain a general level of reciprocal and mutually advantageous con-

cessions not less favorable to trade than that provided for in [the] agreement

prior to such negotiations’. Bagwell and Staiger show that, while MFN insures

against the conventional erosion of benefits of concessions in an agreement

through opportunistic deviation of a subset of parties at the expense of others, it is

not enough to eliminate fully such opportunism. However, MFN combined with

reciprocity requirement will lead to a renegotiation proof agreement that is pol-

itically efficient, so that any deviation from it can raise one party’s utility – which

allows for political considerations besides national welfare – only at the expense

of another’s. It should be noted that the politically efficient tariffs that the

renegotiation proof trade agreement sustains are not zero, but represent the

minimum that can be supported by the agreement. Thus, the Bagwell–Staiger

theory has nothing to say on whether the GATT style negotiation process con-

ditioned by MFN and reciprocity is an efficient means for achieving a barriers-free

global trading system.

In an extremely provocative and stimulating paper, Ethier (2004) argues

that what he dubs as the ‘Received Theory’ of multilateral agreements, based

solely on terms-of-trade externalities, is ‘ irrelevant to actual multilateral trade



TERMS OF TRADE’ (emphasis in the original). Moreover, in his view, more

Nondiscrimination in GATT/WTO 87

sophisticated versions of Received Theory such as, for example, that of Bagwell

and Staiger, have two implications:

(i) Small countries will never sign on to trade agreements. If they did, they
would surrender the use of trade policy for domestic objectives while receiving
absolutely nothing in return. (ii) Large countries will negotiate only trade agree-
ments that constrain terms-of-trade manipulation. Trade agreements that do
not do this would, for no reason, surrender the use of trade policy for domestic

In reality, we observe exactly the opposite. (Ethier, 2004: 304)

Thus, according to Ethier, ‘The Received Theory predicts what we do not

see and cannot explain what we do not see and cannot explain what we do see ’

(emphasis in the original). His own multicountry model deliberately does away

with market power induced terms-of-trade externalities by assuming that no

country has the ability to influence world prices. Governments have an objective

function that reflects political externalities ; these are basically two. First, ‘political

support ’ is ‘more sensitive to the direct effects of government actions than to the

indirect consequences [because] of bounded rationality [of the public] ’ ; second,

‘trade volumes influence political support independently, to some degree, from

their implications for factor reward’. Negotiations take place in a succession of

periods. Although he does not present a formal bargaining model – and the

government objective function is a reduced form – still the model ‘gives immediate

and transparent explanations of gradual liberalization, reciprocity, MFN, and

multilateralism’. He summarizes his argument as follows:

1. If governments, sensitive to political externalities, conduct trade policy as

they claim to be conducting it, they will not liberalize unilaterally but will find

reciprocal liberalization that does not go all the way to free trade desirable.

2. Reciprocal liberalization is valuable to a government because negotiated foreign

liberalization allows that government credibly to claim credit, in an environment

where many things change at once, for the good fortune experienced by ex-


3. If reasonably close substitutes for each country’s products exist elsewhere,

reciprocal trade negotiations of any sort, that do not intend to go all the way to

free trade, require that negotiating government’s subsequent behavior toward

third countries be constrained in some way.

23 Prima facie it would seem that a small country would have an incentive to join a trade agreement,
with MFN provisions, which includes large countries since it will enjoy the benefits of a reduction in its

tariffs by a large country that raises the world price of exports of the small country. Ethier’s argument does

not appear to dispute this. It suggests that if constraining terms-of-trade externalities is the sole reason for
a trade agreement, small countries, whose trade policies create no such externalities by definition, and who
might wish to use such policies for domestic objectives, would have no reason to join such agreements.


4. MFN status is a useful way, and the simplest way, to constrain negotiating

governments’ subsequent behavior toward third countries. And, if used enough,

it works, not because of the direct assurance it offers individual countries, but

because it functions as an externality.

5. Multilateralism can internalize this crucial externality. Thus it is appropriate that

Article I of the GATT establishes nondiscrimination as a basic principle, instead

of defining it as a subject for negotiation.

6. If trade agreements do feature MFN, negotiations must, sooner or later, be

multilateral in any event. (Ethier, 2004: 317–318)

Horn and Mavroidis (2001) are right in their assessment that:

The models of Bagwell and Staiger, and of Ethier, are rather different and
emphasize rather different aspects of MFN. The former shows how the sole role
of trade negotiations is to address international externalities that go through
terms of trade, while Ethier only focuses on political externalities that do not
go through terms of trade. Interestingly, both approaches suggest a positive role
for MFN, in both cases when complemented with another salient feature of
the trading system. In the case of Bagwell’s and Staiger’s analysis, MFN is com-
plemented by reciprocity: MFN and reciprocity jointly imply that there need
not be any negative externalities from bilateral renegotiations. Ethier, on the
other hand, puts no restrictions on negotiations other than MFN, but instead
argues that multilateralism is what prevents MFN from being eroded through
concession diversion.

Both analyses, in essence, are about characterization of equilibrium: to the

best of my reading, there are no real disequilibrium dynamics in either.24 However,

real negotiations in WTO have dynamics of their own, with suggestions of brink-

manship as a strategy.25 The processes of first agreeing on a framework for nego-

tiations, then on their modalities, and eventually to actual negotiations, are

not easily modeled and, again to the best of my knowledge, have not been. It is

not easy to discern any ‘convergence’ to an equilibrium property in these pro-

cesses.26 Although I have dismissed associating notions of fairness with MFN and

reciprocity, perhaps it is the perception of fairness of both in some vague sense

and the belief that insisting on both will ensure that whatever be the end outcome

24 Many of the game theoretic analyses of GATT, like most such analyses, implicitly or explicitly

assume ‘common knowledge’, an assumption that is far from reality.

25 In the GATT ministerial meeting that launched the Uruguay Round at Punta del Este, Uruguay,

at 6 p.m. the day before the meeting was scheduled to end, no agreement had been reached. The US
delegation announced with great fanfare that they had their bags packed and would depart for the US the

next morning whether or not there was an agreement. This threat seemed to have worked – agreement on

all issues was reached by 4:30 a.m. the next morning (Srinivasan, 1998: 31).
26 Preferential agreements are said to be of help in persuading trade negotiators that their failure to

conclude a multilateral agreement will unleash regional agreements. For example, it is claimed that the

North American Free Trade Agreement helped in concluding the Uruguay Round Agreement and the

plethora of agreements now in place would help an agreement in the Doha Round. Although politicians
seem to hold this view, there is no convincing evidence for it..

Nondiscrimination in GATT/WTO 89

of these unbelievably complex and inexplicable processes, it will turn out to be fair

for all participants.

5. MFN and NT once again

The following simple model is meant to illustrate a fundamental problem with

the exceptions to nondiscrimination in Articles III and XX(b). It goes beyond the

relatively easier one to solve of defining more precisely the meaning of ‘ like’ or

‘similar’ products. There are two goods, x and y. Good x has one domestic and

two foreign, sources of supply. Good y has just two, one domestic and one foreign,

sources of supply. Let the social welfare function of the economy be addititively

separable in utility of private consumers and social disutility, possibly from

environment pollution it creates, from private consumption of good x. More

specifically, social welfare W is given by:

W=U(XD+X1F+X2F,Y)xV(XD+h1X1F+h2X2F) (1)

where XD is consumption of domestically produced x and X1F and X2F are

consumption from the first and second foreign sources of good x, respectively, and

Y is the total consumption of good y.

It is to be noted that in private welfare and in social disutility the three sources

of supply of good 1 are perfect substitutes, except that the marginal rate of sub-

stitution (MRS) in private utility between any two sources of supply is unity,

while in social disutility, the MRS between domestic and the first (second) foreign

sources of supply is h1(h2).

The production side of the economy is described by the transformation


F(Qx,Qy)=0 (2)

whereQx(Qy) is the domestic output of good x(y). Assume the economy is a small

open economy that faces an international relative price of unity for all goods in

terms of the domestic good x and trade is balanced so that:

XD+X1F+X2F+Y=Qx+Qy (3)

By assumption consumers maximize private welfare U and ignore the social

disutility xV. Then it is straightforward to note that a laissez faire free trade
competitive equilibrium (LFTCE) will be characterized by:

=1 (4)

where U1 is the private marginal utility of total consumption of good x from

all sources and U2 is the private marginal utility of consumption of good 2.

Equation (4) is the standard characterization of LFTCE by the equality of the


MRS in consumption (i.e., U1/U2) and the marginal rate of transformation (MRT)

in production (i.e., F1/F2), and equality of their common value to the world

relative price (i.e., 1). If good x is exported in equilibrium, clearly X1F and X2F
would be zero, and, if it is imported, the consumer would be indifferent

between dividing total imports in any fashion between the two sources of supply.

Clearly, with free trade and no domestic taxes, both MFN and NT hold.

Consider now a social optimum. Now the social MRS between consumption of

good x and good ywill depend on what sources of supply are being used in positive

amounts. In the optimum, productions levels Qx
o and Qy

o will be the same as in

the laissez-faire equilibrium. However, there are several possible consumption

optima. Without loss of generality, assume h1<h2.

Case I : In this optimum, good x is exported. At this optimum:




=1 (5)

wherexV1 is the social marginal disutility from consumption of good x.

x (6)



D (7)

For no imports of good x to take place, it is clear that h1 has to exceed 1.

Since consumers who ignore social disutility would be equating U1/U2 to

the relative price of good x they face, implementation of this optimum as a com-

petitive equilibrium will involve producers’ facing the world relative price of

unity for good x, and consumers facing 1+(V1/U2). Thus, free trade plus a tax
of V1/U2 on private consumption of good x will sustain this equilibrium. Once

again, MFN holds and NT holds trivially, since only the domestically produced

good x is consumed. This being the case, the consumption tax applies only on

the domestic supply of good x.

Case II : In this optimum, part of the consumption of good x is from imports.

It is straightforward to see that only the first foreign source of supply of imports

will be used in this optimum, given h1<h2. The relevant optimality conditions are:





=1 (8)

o=Qoy (9)

For sustaining this optimum as a competitive equilibrium the relative price in terms

of good 2 of domestic good x facing producers has to be unity, while that facing

Nondiscrimination in GATT/WTO 91

consumers has to be 1+(h1V1/U2), implying that consumption of good x has to be
taxed to raise its consumer price by h1V1/U2 over the producer price of unity.

However, this is not enough to sustain the optimum, since at these prices there

is no way to prevent consumers from choosing to split the imported component of

their total consumption of good x in any way they like between the two foreign

sources of supply. However, the social optimum requires them to use only the

first source of supply. To ensure that this happens, the second foreign source

of supply has to bear a marginally higher tax (or to receive a lower subsidy).27

Such a tax or subsidy scheme will prima facie violate NT, although MFN treat-

ment holds. For this violation to be within the permitted exceptions under Article

III or Article XX(b), it has to be shown that the additional consumption tax

on the supply of good x from the second foreign source, which is levied only

to exclude it from the domestic market, is not being ‘applied so as to afford

protection to domestic production’ of good x to a minimum consistent with

maximization of social welfare. The effect of the discriminatory tax is certainly

the exclusion of imports from the second source. Hence, without going into the

‘aim’ of the discrimination, there is no way to establish that discrimination is

not protectionist.

The example was deliberately chosen for two reasons. First, it sidesteps the

issue of ‘ like’ product: from the consumer perspective, good x from any of

the three sources is identical. From a social perspective, although they are not

identical, they are nonetheless perfect substitutes. Second, the effect of a discrimi-

natory consumption tax on one of the foreign sources of supply, by excluding

it from the market, prevents it from competing with its domestic and the other

foreign source of supply. The example clearly shows that the problem with the

exception to nondiscrimination in Articles III and Article XX(b) is fundamental

and not just lack of clarity about the content of phrases such as ‘ like products’,

‘similar products ’, and others. The fundamental problem is that there is no getting

around an inherently difficult ‘aim’ test.

6. Conclusions

Several conclusions emerge from the discussion in the previous sections. First,

although the principle of nondiscrimination as enunciated in Articles I and III of

GATT is considered by most analysts as its very foundation, the exceptions

to nondiscrimination in other articles of GATT considerably qualified its force

ab initio.

27 Note that an alternative scheme of setting an import quota of the amountX1F
o and auctioning it will

run into the same problem. The quota holder has no incentive to import all of his quota from the first

source. However, two quotas, an amount of X1F
o from the first source and nothing from the second will

work. But this would be a violation of MFN in the application of quotas.


Second, the so-called ‘free trade’ areas and customs unions, for which

Article XXIV allows (an exception to the MFN principle) under certain stringent

conditions, and other preferential trade agreements (PTAs) have proliferated

and continue to proliferate, so that all the 147 members of the WTO are either

members also of a PTA or negotiating to become one. During the nearly six

decades of the history of the GATT/WTO, only six such agreements were pro-

nounced to be compatible with Article XXIV by a GATT working party, of which

only two still exist. Interestingly, the European Union is yet to be pronounced to

be compatible. Whether or not the PTAs are largely beneficial to their members,

without inflicting serious damage to non-members, whether they promote ‘deeper

integration’ of the economies, in some desirable sense, of members than non-

discriminatory multilateral agreements, such as WTO, promote, or whether they

are simply alternative means for powerful members of the WTO for extracting

concessions from the less powerful on issues such as labor standards, investment,

etc., on which there are no multilateral agreements, are open questions on

which there is an enormous and growing literature. Regardless of the merits of

PTAs, the fact that their proliferation has undermined, almost completely, the

nondiscrimination principle of WTO is beyond doubt.

Third, the proliferation of PTAs, along with continuation of tariff quotas

and Special and Differential Treatment of Developing Countries, and weaknesses

regarding MFN and NT in GATS, have rendered nondiscrimination as almost

obsolete in the WTO.

Fourth, there is no strong historical foundation for the modern concept of un-

conditional MFN. In game-theoretic approaches to trade agreements (Coneybeare,

1987; Caplin and Krishna, 1988), MFN norms could be counterproductive.

As Caplin and Krishna (1988: 281–282) correctly point out :

If we view the bargaining process as yielding efficient outcomes, as for example
with the Nash bargaining solution, then MFN simply limits the tools available
to different countries, shifting in the utility possibility frontier. Hence the most
positive aspects of MFN can only be illustrated when the bargaining process
absent-MFN yields inefficient outcomes.

Problems of incentives created by MFN for free riding and delaying partici-

pation in negotiations by large traders and other negative aspects of MFN are

extensively discussed in the literature, along with its positive efficiency promoting

dimension. Although the work of Bagwell and Staiger (2002) and Ethier (2004)

has restored an essential role (different in the two contributions) for MFN in

achieving and sustaining a trade agreement, the fact that, as noted above, dis-

crimination rather than MFN-based nondiscrimination is the reality in global

trade suggests that the role that these authors attribute to MFN may no longer be

that relevant.

Fifth, tariff and other barriers at the border have largely been reduced to insig-

nificant levels thanks to the eight rounds of multilateral trade negotiation under

Nondiscrimination in GATT/WTO 93

the auspices of GATT. Inside the border, regulatory measures have become

far more salient with the conclusion of GATS and also with the increasingly

successful attempts to introduce non-trade matters under the rubric of trade-

related this or that. It is an open question whether nondiscrimination per se is an

important issue in thinking about multilateral disciplines in these new areas.

I am sure, had Bob Hudec been with us, he would have authoritatively and

illuminatingly analyzed the emerging issues. My reading of his work is that he

was not wedded to any dogmatic adherence to any principle, be it of non-

discrimination or of free trade and was skeptical of all forms of dogmatism. We all

miss him, his intellect, and his charm.


Bagwell, Kyle and Robert W. Staiger (2002), The Economics of the World Trading System, Cambridge,

MA: MIT Press.

Bhagwati, Jagdish and Robert Hudec (1996), Fair Trade and Harmonization, Vols I and II, Cambridge,

MA: MIT Press.

Bronckers, Marco and Natalie McNelius (2000), ‘Rethinking the ‘Like Product’ Definition in GATT

1994: Anti-Dumping and Environmental Protection’, in Cottier and Mavroidis 2000.

Caplin, Andrew and Kala Krishna (1988), ‘Tariffs and the Most-Favored-Nation Clause: A Game-

Theoretic Approach’, Seoul Journal of Economics, 1 : 267–289.

Coneybeare, John A. C. (1987), Trade Wars: The Theory and Practice of International Commercial

Rivalry, New York: Columbia University Press.

Cottier, Thomas and Petros C. Mavroidis (2000), Regulatory Barriers and the Principle of Non-

Discrimination in World Trade Law, Ann Arbor: University of Michigan Press.

Ethier, Wilfred J. (2002), ‘Unilateralism in a Multilateral World’, The Economic Journal, 112(179):


Ethier, Wilfred J. (2004), ‘Political Externalities, Nondiscrimination, and a Multilateral World’, Review

of International Economics, 12(3): 303–320.

GATT (1994), The Results of the Uruguay Round of Multilateral Trade Negotiations: The Legal Texts,

Geneva: GATT Secretariat.

Grandmont, Jean-Michel and Daniel McFadden (1972), ‘A Technical Note on Classical Gains from

Trade’, Journal of International Economics, 2(2): 109–125.

Hoekman, Bernard (2004), ‘Overcoming Discrimination Against Developing Countries : Access, Rules

and Differential Treatment’, Paper presented at the conference ‘Role of the WTO System in the

World Economy’, Cornell Law School and Cordell Hull Institute, 9–10 July 2004, Paris.

Hoekman, Bernard, Constantine Michalopoulos, and L. Alan Winters (2004), ‘Special and Differential

Treatment of Developing Countries in the WTO: Moving Forward After Cancún’, World

Economy, 27(4): 481–506.

Horn, Henrik and Petros Mavroidis (2001), ‘Economic and Legal Aspects of the Most-Favored-Nation

Clause’, European Journal of Political Economy, 17 : 233–279.

Horn, Henrik and Petros Mavroidis (2004), ‘Still Hazy After All These Years: The Interpretation of

National Treatment in the GATT/WTO Case Law on Tax Discrimination’, European Journal

of International Law, 15(1): 39–69.

Hudec, Robert (1988), ‘Tiger, Tiger in the House: A Critical Appraisal of the Case Against Discriminatory

Trade Measures’, The New GATT Round of Multilateral Trade Negotiations: Legal and

Economic Problems, Boston: Kluwer.

Hudec, Robert (1999a), Essays on the Nature of International Trade Law, London: Cameron May.


Hudec, Robert (1999b), ‘The New WTO Dispute Settlement Procedure: An Overview of the First Three

Years’, Minnesota Journal of Global Trade, 8(1): 34–56.

Hudec, Robert (2000), ‘ ‘‘Like Product’’ : The Differences in Meaning in GATT Articles I and III ’, in

Cottier and Mavroidis 2000.

Keck, Alexander and Patrick Low (2004), ‘Special and Differential Treatment in the WTO: Why, When

and How?’, Staff Working Paper ERSD-2004–03. World Trade Organization, Washington, DC.

League of Nations (1932), ‘Procedure for the Friendly Settlement of Economic Disputes between States’

(Official No. C.57.M.32.1932.Il.B), at 4.

Mattoo, Aaditya (2000), ‘MFN and the GATS’, in Cottier and Mavroidis 2000.

Neven, Damien J. (2000), ‘The Economic Evaluation of Protection Under Art III : Some Principles

and Suggestions for the Definition of a Trade Relevant Market’, in Cottier and Mavroidis 2000.

Srinivasan, T. N. (1998), Developing Countries and the Multilateral Trading System: From the GATT

to the Uruguay Round and the Future, Boulder, CO: Westview.

US Department of State (1948), ‘Suggested Charter for an International Trade Organization of the United

Nations’, Article 76; reprinted in US Department of State Publication 2598, Commercial Policy

Series 93, United States Department of State, Washington, DC.

WTO (2004), ‘Annual Report 2004’, http://www.wto.org/english/res_e/reser_e/annual_report_e.htm,

Accessed October 12, 2004.

WTO (1995), Regionalism and the World Trading System, Geneva: World Trade Organization.

Nondiscrimination in GATT/WTO 95