Aaditya Mattoo *

January 1999

The GATS and its MFN obligation came into effect before WTO Members were willing to completely eliminate discriminatory measures in services trade. The Agreement therefore had to strike a difficult balance between creating meaningful multilateral disciplines and accommodating discriminatory trade practices. The challenge to multilateral disciplines posed by the explicit departures from the MFN obligation, such as the exceptions for regional integration agreements and the MFN exemptions listed by Members, are widely recognized. However, the difficulties arising from less visible, implicit discrimination have not been adequately appreciated. It is this second class of problems that this paper seeks to highlight, both from an economic and legal perspective.

The next three sections constitute the descriptive part of the paper. Section 2 contains a description of the basic MFN obligation in the GATS drawing upon recent jurisprudence. Section 3 briefly describes other GATS provisions which reiterate or extend the basic obligation, notably Articles XVI and XVII on market access and national treatment, and Article VIII dealing with monopolies and exclusive service suppliers, as well as provisions which allow for departures from the basic obligation, notably Article V dealing with regional integration and Article XIV with general exceptions. Article VII dealing with recognition receives special attention because the interpretation of this provision may well determine how GATS resolves the potential conflict between selective and multilateral liberalization. Then Section 4 describes the pattern of the MFN-exemptions listed by Members and assesses their potential impact.

Section 5 contains an analysis of non-discrimination in services trade, from an economic and legal perspective. The implications of discrimination through domestic regulations, arguably the central issue in services trade, are analysed in Section 5.1. It is argued that making distinctions between services and service suppliers in the pursuit of certain domestic policy objectives, such as to ensure the quality of professional services, financial stability, and competitive market conditions, is economically sensible. It would, therefore, be desirable to allow Members the legal freedom to pursue such objectives, but to discipline the exercise of such freedom by ensuring that the choice and level of instruments is not more burdensome than necessary - with economic efficiency considerations playing a role in this assessment. Section 5.2 addresses the other major form of protection in services, quantitative restrictions, and suggests that there may be need to clarify the rules governing the non-discriminatory allocation of quotas.

The next sub-sections examine three of the most important sources of tension between the MFN obligation and Members' desire to discriminate. Section 5.3 addresses the issue of reciprocity, the basis for several MFN exemptions and possibly the main reason for the failure of the maritime negotiations. Distinctions between services and service suppliers based on the openness of the home market raise complex economic questions, related, on the one hand, to avoiding adverse terms of trade movements and free-riding in bargaining, and on the other to preventing costly trade diversion and maintaining the security of concessions. This economic ambiguity is reflected in a compromise in the GATS: distinctions conditional on home-market openness are illegal, unless covered by an MFN exemption or the exception for regional integration agreements.

Section 5.4 discusses the scheduling "innovation" which played a crucial role in the financial services negotiations: grandfathering the superior conditions of operation for firms which were already established. It is argued that distinctions between service suppliers based on criteria such as time of entry into the market can only have a protectionist effect and are not economically desirable. Furthermore, there may be a conflict between the basic MFN obligation and the schedules of specific commitments containing grandfather provisions.

Finally, Section 5.5 addresses an issue which was particularly important in the context of the telecommunications negotiations: whether the MFN obligation in the GATS leaves Members adequate scope to take action against foreign anti-competitive practices. It is argued that if competition policy itself is based on non-discriminatory principles, then it would be acceptable for it to impact differentially on particular services or service suppliers in specific instances provided they themselves manifested the characteristics which aroused concern. However, any discrimination in treatment, based not on competition policy-related attributes of the service or the service supplier, but on unrelated attributes such as the fact of protection in the home market, would seem to violate the MFN obligation.

Article II of the GATS constitutes a general obligation which is, in principle, applicable across the board by all Members to all services sectors. Article II:1 of GATS states:

"With respect to any measure covered by this Agreement, each Member shall accord immediately and unconditionally to services and service suppliers of any other Member treatment no less favourable than that it accords to like services and service suppliers of any other country".

While the meaning of the basic MFN obligation is well understood, there remains scope for clarifying certain aspects in the services context - particularly in the light of the recent rulings of a WTO Panel and the Appellate body on the dispute concerning the European Communities' regime for the importation, sale and distribution of bananas (henceforth, Bananas Case).¹ Consider each of the following terms:

any measure covered by this Agreement: The scope of the GATS is defined in Article I:1 of the Agreement to encompass "measures by Members affecting trade in services". A closer look at the elements of this definition reveals the wide domain of the Agreement. First of all, the Agreement takes a view of "trade in services" that includes not only cross border supply, but also the supply of a service through consumption abroad, commercial presence and presence of natural persons. Hence, while the MFN obligation under GATT 1994 is concerned with measures affecting products per se, the domain of this obligation in the GATS includes also measures affecting service suppliers. Secondly, "measures by Members" are defined broadly to include measures taken by central, regional or local governments, as well as non-governmental bodies in the exercise of powers delegated by the government - even though the GATS only requires Members to take "such reasonable measures as may be available to it" to ensure compliance by sub-central entities.² Finally, the Bananas Case confirmed a broad interpretation of the term "affecting", i.e. measures concerned need not affect trade in services as such but could also be measures taken in other areas with repercussions on services - such as measures in respect of the purchase of goods.³

treatment no less favourable: The key issue here is whether this should be interpreted to mean only de jure, or formal, discrimination or whether it also pertains to de facto discrimination. The Appellate Body in the Bananas Dispute upheld the Panel's conclusion that the provision prohibits both forms of discrimination, but differed on the basis for this conclusion.

The Panel argued that the basic requirement to accord "treatment no less favourable" was the same in GATS Article II:1 and GATS Article XVII:1, the national treatment provision. The standard of "no less favourable treatment" in paragraph 1 of Article XVII was meant to provide for no less favourable conditions of competition regardless of whether this was achieved through the application of formally identical or formally different measures. Paragraphs 2 and 3 of Article XVII served the purpose of codifying this interpretation, established in previous jurisprudence pertaining to the GATT national treatment provision, and did not impose new obligations on Members additional to those contained in paragraph 1. The absence of similar elaboration in Article II was not a justification for giving a different ordinary meaning to the words "treatment no less favourable", which were identical in both Articles II:1 and XVII:1. It could be legitimately inferred, therefore, that Article II, also proscribed both forms of discrimination - notwithstanding the absence of clarifying elaboration.

The Appellate Body found the Panel's reasoning on this issue to be "less than fully satisfactory".⁴ It argued that the MFN obligation in GATS Article II should be interpreted, not in the light of the national treatment obligations in GATS Article XVII or GATT Article III, but in the light of the MFN obligation in GATT Article I. Article I of GATT 1994 had also been applied, in past practice, to measures involving de facto discrimination.⁵ Significantly, the Appellate Body agreed with the Panel that if Article II were to be interpreted to require only formally identical treatment, "it would not be difficult - and, indeed, it would be a good deal easier in the case of trade in services, than in the case of trade in goods - to devise discriminatory measures aimed at circumventing the basic purpose of that Article" which was to ensure equality of competitive opportunities. For these reasons, it concluded that "treatment no less favourable" in Article II:1 of the GATS should be interpreted to include de facto, as well as de jure, discrimination. The Appellate Body went further than the Panel in clearly stating that its conclusion was not limited to the case under consideration.

like services and service suppliers: In the Bananas Case, the question of how likeness of services and service suppliers should be established needed to be addressed to only a limited extent. First, and relatively straightforward, the Panel found that wholesale services of bananas are like, irrespective of the origin of the bananas supplied. Secondly, in the Panel's view, "to the extent that entities provide these like services, they are like service suppliers".⁶

It is relatively straightforward to apply the MFN provision to measures taken by governments that are overtly discriminatory, i.e, those that discriminate on the basis of the provenance of the good. The problem arises in relation to what might be termed as apparently origin-neutral (AON) measures; namely, those where the basis for applying a measure is not the provenance of the service or service supplier but some other characteristic or set of characteristics, such as the place of training. In such cases, a determination under Article II hinges on determining whether the imported products are "like" each other.⁷ The wider the definition of "likeness," the greater will be the set of measures that are inconsistent with Article II. If a doctor is a doctor, a regulation which imposed greater qualification requirements on a doctor trained in Country A than on a doctor trained in Country B would violate Article II. On the other hand, the narrower the definition of "likeness" the more measures will conform with Article II. If a doctor trained in one country is not "like" a doctor trained in another country, then the treatment of the doctor trained in country A would have to be compared with the treatment of a doctor trained in a country with similar training standards, and the additional qualification requirement could then be found consistent with Article II. Similar questions also arise in establishing likeness of service suppliers. These are discussed in Section 5.1 below.

Certain sectoral sensitivities that emerged in the Uruguay Round raised the spectre of wholesale sectoral exclusions from GATS as a means of avoiding the MFN rule. In order to prevent this, it was agreed to permit limited exemptions to MFN under GATS. Such exemptions, however, had to be taken at the time the negotiations were concluded. Any exception to this general obligation would have to be provided for explicitly in accordance with the terms of the GATS. Article II:2 states:

"A Member may maintain a measure inconsistent with paragraph 1 provided that such a measure is listed in, and meets the conditions of, the Annex on Article II Exemptions".

The annex makes it clear that no new exemptions can be granted, at least by this route: any future requests to give non-MFN treatment can only be met through the WTO waiver procedures.⁸ Some listed exemptions are subject to a stated time limit. For those that are not, the annex provides that in principle they should not last longer than ten years (that is, not beyond 2004), and that in any case they are subject to negotiation in future trade-liberalizing rounds. Meanwhile, all remaining exemptions will be reviewed in the Council for Trade in Services before the end of 1999 to see whether they are still needed.⁹

A limited exception to the rule that no new exemptions could be listed after the entry into force of the WTO applied to financial services, maritime transport and basic telecommunications: GATS annexes providing for continuing negotiations on these subjects, allowed exemptions to be listed only at the end of the negotiations. Now that negotiations in basic telecommunications and financial services have been concluded, it is only in maritime transport services that it is still possible to take an MFN exemption.

This Section discusses other GATS provisions which either extend/reiterate the scope of the basic MFN obligation or provide for departures from it.

While Article II applies across the board, Article XVI on market access and Article XVII on national treatment only apply to sectors included by a Member in its schedule of specific commitments. The liberalizing content of the GATS depends to a large extent on the nature of these specific commitments. Article XVI prohibits a range of measures (mostly quantitative restrictions) unless explicitly listed in a Member's schedule, and Article XVII prohibits measures which discriminate against foreign services or service suppliers, again unless listed in a Member's schedule.

The texts of Articles II, XVI and XVII do not create an explicit hierarchy of obligations. Thus, it is not completely clear from the text itself whether a Member taking an MFN exemption would also be free to discriminate with respect to the commitments made under Articles XVI and XVII. However, the "Scheduling of Initial Commitments in Trade in Services: Explanatory Note" states that "where an MFN exemption has been granted for a measure, a Member is free to deviate from its Article II obligations, but not from its Article XVI and XVII commitments. Therefore, in such cases, a Member may accord treatment more favourable than the minimum standard to some Members, as long as all Members receive at least the that minimum standard of market access and national treatment appearing in its schedule."¹⁰

It follows that an MFN exemption would give a Member who had made no commitments in a sector considerable freedom to discriminate. However, an MFN exemption which coexists with fully liberal market access and national treatment commitments would have little meaning - unless a Member wished to provide more favourable treatment to certain foreign suppliers than it does to its own.

Apart from services specified in individual MFN exemption lists, the only permitted departure from most-favoured-nation treatment under the GATS covers preferential treatment among countries that are members of regional trading arrangements. The GATS rules on "Economic Integration", in Article V, are modelled on those in Article XXIV of the GATT. Article V:1 permits any WTO member to enter into an agreement to further liberalize trade in services with the other countries that are parties to the agreement, provided the agreement has "substantial sectoral coverage", eliminates measures that discriminate against service suppliers of other countries in the group, and prohibits new or more discriminatory measures. Recognizing that action to open up services markets may well form part of a wider process of economic integration, Article V:2 allows the liberalization achieved to be judged in this light.

The drafting of these rules improves on their GATT equivalents in some respects: for instance, "substantial sectoral coverage" is more carefully defined, since it disqualifies agreements that exclude any of the four modes of supply. Some other provisions are carried over from Article XXIV with little change. An approved agreement must be designed to help trade among its members, and must not result in an increase in the overall barriers faced by non-members in trading with the group within the respective sectors or sub-sectors (Article V:4). If the establishment of the agreement, or its subsequent enlargement, leads to the withdrawal of commitments made to non-members, there must be negotiations to provide appropriate compensation (Article V:5). No compensation, on the other hand, is due from non-members for any trade benefits they may gain from the agreement (Article V:8).

Article V:7 creates a transparency obligation, requiring Members who are party to regional agreements to notify promptly any such agreement and any enlargement or significant modification to it. The agreements which have been notified so far include the agreements establishing NAFTA, the European Communities and their Member States, as well as their agreements with the Slovak Republic, Hungary, Poland, the Czech Republic, Romania, Norway, Iceland, Liechtenstein and Bulgaria, and agreements between Canada and Chile and between Australia and New Zealand.

A related exception from the MFN rule, for the movement of natural persons, is permitted by Article V bis of the GATS. This allows countries to take part in agreements which establish full integration of labour markets. The only such agreement notified so far is the one involving Denmark, Finland, Iceland, Norway and Sweden.

Article VII of the GATS, dealing with recognition, attempts to strike a difficult balance. On the one hand, it is permissive. Thus, Article VII:1 states:

For the purposes of the fulfilment, in whole or in part, of its standards or criteria for the authorization, licensing or certification of services suppliers, and subject to the requirements of paragraph 3, a member may recognize the education or experience obtained, requirements met, or licenses or certifications granted in a particular country. Such recognition, which may be achieved through harmonization or otherwise, may be based upon an agreement or arrangement with the country concerned or may be accorded autonomously.

This provision would seem to allow a Member at any point of time to recognize standards of one or more Members and not of others, without violating its GATS obligations - even though services and service suppliers of the former group have easier access than those of the latter group. The remaining paragraphs of Article VII seek to ensure that this freedom is not abused. Article VII:2 requires a Member who enters into a mutual recognition agreement (MRA) to afford adequate opportunity to other interested Members to negotiate their accession to such an agreement or to negotiate comparable ones. If a Member grants recognition autonomously, then it is obliged to give any other Member adequate opportunity to demonstrate that education, experience, licenses, or certifications obtained in that other Member's territory should be recognized. In this respect, Article VII mandates an openness vis-à-vis third countries in a way that Article V, dealing with economic integration agreements, does not.

Article VII:3 stipulates that a Member must not grant recognition in a manner which would constitute a means of discrimination between countries in the application of its standards or criteria for the authorization, licensing or certification of services suppliers, or a disguised restriction on trade in services. This raises the question of how it could be established whether selective recognition is discriminatory. This question we discuss more fully below. It is worth noting that Article VII:5 states that "wherever appropriate, recognition should be based on multilaterally agreed criteria" and requires Members to work towards the establishment and adoption of such criteria. The issue, of course, is how much discretion the phrase "wherever appropriate" gives Members in deciding whether to follow their own rather than internationally agreed criteria.

What is the empirical significance of MRAs? It is possible to provide a preliminary answer thanks to the transparency requirement created by Article VII:4: Members must inform the Council for Trade in Services about existing MRAs and of the opening of negotiations on any future ones. So far, 21 notifications have been received under Article VII:4, of which 10 are from Latin American countries, 4 from the United States, 3 from Switzerland, and 1 each from the European Commission, Australia, Norway and Macau. Not surprisingly, all but one of these pertain to the recognition of educational degrees and professional qualifications obtained abroad.¹¹ The only other MRA is in the domain of financial services: owing to reciprocal recognition of the proof of solvency between the EU and Switzerland, foreign direct (other than life) insurance companies having their principal place of business in the territory of one of the contracting parties are not obliged to localise funds to a significant extent and are not obliged to furnish security.

Interestingly, mutual recognition of qualifications is also mentioned as an element of 11 regional integration agreements, notified under GATS Article V:7(a). These agreements include the one establishing the European Union, agreements between the European Union and neighbouring countries, and the Closer Economic Relations Treaty between Australia and New Zealand. This raises the question of whether MRAs concluded in the context of a regional integration agreements are still subject to the disciplines in Article VII:2 and 3. One view may be that Article V provides an exception to the fundamental non-discrimination obligation in Article II and therefore an exemption also to similar obligations contained in other GATS provisions, including Article VII. Alternatively, it could be argued that all MRAs, regardless of whether they are concluded by parties to a regional integration agreement or other Members, are covered by Article VII and its disciplines cannot be circumvented by appealing to Article V.

Article VIII requires each Member to ensure that any monopoly supplier of a service does not "in the supply of the monopoly service in the relevant market" act in a manner inconsistent with the MFN obligation and the Member's specific commitments. A monopoly supplier of a service is defined in the GATS as any person, public or private, which in the relevant market of the territory of a Member is authorized or established formally or in effect by that Member as the sole supplier of that service (Article XXVIII(h)).

Article XIII of the GATS exempts government purchase of services for its own use from the MFN obligation (as well as from the market access and national treatment rules). However, the same provision mandates negotiations on government procurement in services which may lead to commitments to open up some government purchases to foreign service suppliers.

The GATS provisions on general and security exceptions resemble their GATT equivalents. The general exceptions are, as in the GATT, preceded by a headnote that makes the right of a member to adopt or enforce measures necessary for the purposes listed subject to the condition that they not be applied as a means of "arbitrary or unjustifiable discrimination between countries where like conditions prevail, or as a disguised restriction on trade in services". The list of purposes that follows includes the protection of public morals and maintenance of public order;¹² protection of human, animal or plant life or health; securing compliance with laws or regulations which are not inconsistent with the provisions of GATS including those relating to the prevention of deceptive and fraudulent practices, the protection of individual privacy in the handling of personal data, and safety; ensuring equitable and effective taxation and the avoidance of double taxation. A footnote spells out a number of ways in which a country's taxation practices may treat foreigners differently from its own nationals. The security exceptions are virtually identical with those in GATT 1994.¹³

The Annex on Air Transport specifically excludes the complex network of bilateral agreements on air traffic rights from GATS rules.¹⁴ In consequence, the GATS applies at present only to aircraft repair and maintenance services, the selling and marketing of air transport services (a function defined as not including the pricing or conditions of transport services), and computer reservation systems.¹⁵

In the maritime transport sector, the failure to conclude negotiations successfully led to the June 1996 Decision to suspend the application of the MFN obligation to the sector until the end of the next round of comprehensive services negotiations. The suspension was prompted by the difficulty in eliminating MFN-inconsistent measures in the maritime sector. Even though Members could have listed exemptions from the MFN obligations, the dominant view was that the continued suspension of the MFN obligation would avert the need for many countries to take MFN exemptions which may be more difficult to negotiate away once explicitly listed.

MFN exemptions have been listed by some 70 Members (Table 1). There are around 380 measures involved, with many Members listing several measures in the same sector.¹⁶ Table 2 attempts to distinguish between measures according to the conditions specified by Members as creating the need for the exemption. The first category includes measures the reason for which is the existence of sector-specific preferential regional agreements; the second category consists of measures taken under other bilateral or plurilateral agreements - usually designed to preserve certain historical preferences; the third category includes measures based on a unilaterally imposed reciprocity condition and not an agreement with trading partners. These distinctions are sometimes fuzzy and Table 2 should be used to infer broad patterns rather than precise details.

Nearly two-thirds of the exempted measures are to be found in communication services and in transport services. The largest number of MFN-inconsistent measures (69) were listed in the audiovisual sector. More than half of them mention promotion of common (regional) culture as a motive for limiting access to joint programmes to finance and diffuse audiovisual works, while most of the others have been claimed for bilateral agreements on co-production of films and television programmes. In maritime transport, many Members chose to maintain MFN exemptions despite the suspension of the obligation for the sector. Of the 55 measures listed, nearly half are exemptions by developing countries for measures implementing the provisions of the United Nations Convention on a Code of Conduct for Liner Conferences. These provisions, in principle, divide 80 per cent of the liner trade on a traffic route between the shipping companies of the two states at each end, leaving only 20 per cent for shipping companies of other nationalities, but full implementation of this rule is apparently rare. Many of the rest provide cover for existing and future measures for regional preferential treatment. In road, rail or inland waterways transport, governments have sought exemptions for preferential access granted to cross-border access to hauliers of passenger and freight transport of neighbouring countries or regional partners, sometimes through bilateral agreements. More distant countries are probably not greatly affected by such preferences. It is worth noting that these sectors, which account for a substantial share of MFN exemptions, are also those in which Uruguay Round commitments were particularly limited.

Reciprocity conditions - which specify that a Member is willing to guarantee access to its market only to those Members who provide it with access to their markets - are particularly significant in air transport services and financial services. The exemptions listed for air transport services pertain to the services falling within the scope of the GATS, i.e repair and maintenance, selling and marketing of air transport services, and computer reservation system (CRS) services. In financial services, at the end of the first round of extended negotiations in April 1995, many countries followed the lead of the United States in listing MFN exemptions for the licensing of foreign service suppliers on the basis of reciprocity.¹⁷ While several Members removed such exemptions at the end of the negotiations in December 1997, a number were maintained by countries, including Hungary, Pakistan, Peru, Philippines and Venezuela - with the stated objective of obtaining equal access opportunities for domestic suppliers in foreign markets. It is notable that many of these countries made significant commitments in financial services, reducing significantly the discriminatory potential of the MFN exemption. The same is true of the United States, which narrowed the scope of its MFN exemption considerably, retaining the right to discriminate only against countries in which United States financial institutions were forced to disinvest on the basis of their nationality. In maritime transport, several Members have sought exemptions for MFN-inconsistent measures such as retaliatory action against trading partners who are perceived to restrict access to their markets. The most significant of such exemptions, for the United States retaliatory legislation in the maritime sector, was listed and then withdrawn when the application of the MFN obligation to the maritime sector was suspended.¹⁸ Unlike in the case of financial services, only limited commitments have been made in maritime - primarily, as in the case of the United States, to retain the right to discriminate with respect to a broad class of measures.

In the case of basic telecommunications, relatively few MFN exemptions have been inscribed. Among them is the MFN exemption of the United States on one-way satellite transmission of DTH and DBS television services and of digital audio services, and a similar exemption by Brazil. A particular application of the MFN requirement in the basic telecommunications sector, however, concerns accounting rates - particularly through Article VIII dealing with monopolies and exclusive service suppliers.¹⁹ Although there was general appreciation among most negotiators that competition resulting from liberalization would lead over time to the destruction of the accounting rate system, it was nevertheless decided to secure a shared understanding that Members would not challenge one another’s accounting rate regimes under the WTO’s dispute settlement provisions.²⁰ Despite the understanding, several countries, including Bangladesh, India, Pakistan, Sri Lanka and Turkey, listed exemptions to the MFN rule for their accounting rate systems.

Several MFN exemptions were claimed in professional services for bilateral agreements on the mutual recognition of qualifications - presumably because they did not wish such agreements to be subject to the disciplines of Article VII on recognition. But the number of countries claiming MFN exemptions for professional services is small in relation to the large number of WTO members which made specific commitments in this sector.

Measures for which MFN exemptions have been claimed affecting all sectors mainly concern bilateral agreements for the entry of workers or the protection of investments. Some Members have, however, claimed wide-ranging exemptions for preferential regional agreements - presumably because of uncertainty regarding whether they would fulfil the conditions of Article V dealing with regional integration agreements.

On the whole, MFN exemptions would seem to matter most in sectors like audiovisual services and maritime transport where few specific commitments have been made and discriminatory practices seem to be empirically important. In other cases, where the exemptions coexist with specific commitments (as in financial services) or legitimise preferences which do not greatly affect the pattern of trade (as in cross-border supply of land transport services), there is probably less cause for concern. Finally, the exemptions which cover retaliatory legislation need not lead to actual discrimination in trade policy though the credible threat itself may have real effects.

This Section begins with an analysis of the economic and legal implications of discrimination in services through domestic regulations and quantitative restrictions. It then examines, in turn, three instances of potential conflict between the MFN obligation and Members' objectives: reciprocity-based trade policy, grandfathering of the rights of existing suppliers, and the pursuit of competition policy.

In order to examine the implications of discrimination in services, we must first consider the instruments of protection that are usually employed. In general, governments rely more on domestic regulations, such as technical standards, licensing and qualification requirements, and quantitative restrictions than on discriminatory taxes. The consequences of discrimination through taxation (or duties) are well understood. Does discrimination through alternative instruments raises new analytical issues from the legal and economic point of view?

To begin with, consider domestic regulations pertaining to technical standards and qualification requirements. The requirement to meet a certain standard may involve a change in variable costs, fixed costs, or both. The requirement for financial institutions to maintain local reserves is an example of a standard that may increase variable costs of operation. The requirement for foreign professionals to requalify increases the fixed costs of entry for each individual, but if we define units so that each professional is assumed to provide a single unit of service, then even qualification requirements can be seen as affecting variable costs. The problem is that it would not usually be correct to treat all the additional costs imposed on foreign services or service suppliers as a form of protection. It is necessary to distinguish between the regulatory burden imposed on the foreign supplier which is necessary to ensure the desired quality of the service and that which is excessive.

This issue can be illustrated by considering the qualification requirements imposed on foreign professionals. Even though professionals obtain composite degrees and qualifications in each country, it should usually be possible to distinguish between a basic "universal" part of the training, consisting of u units, which is identical between countries, and country specific training (say in local law) equal to v_i units. In some professions, like medicine and engineering, the universal component is likely to be high, whereas in other professions, like law and accountancy, the country-specific component is likely to be high. Outside of professional services, for instance in construction, financial and transport services, it is reasonable to presume that there is a high universal component to standards though there is usually also a variable country specific elements.

Let us also assume that the cost per unit of training in country i is constant at c_i. The variations in c_i are meant to capture inherent advantages that certain countries have in certain areas. It would seem obvious that if a foreign lawyer wished to practise in country i, he would necessarily have to incur an entry cost equal to c_iv_i. But it is possible that the first country also refuses to acknowledge the equivalence of the universal part acquired in the home country, and insists on full requalification, implying entry costs c_i(u + v_i). In this case, c_iu would be a measure of the excessive regulatory burden. More subtle forms of protection could involve understatement of the universal element u, and exaggeration of the country specific element v_i.

Recognition agreements may exempt certain suppliers from incurring whole or part of these costs.²¹ In situations where the impact of regulation is on variable costs, as is assumed here, the analysis of discriminatory regulation can proceed in a manner analogous to tariffs.²² Assume that there are three countries: country X is an importing country, facing an upward sloping domestic supply of professionals (say because of increasing opportunity costs) and countries Y and Z are potential exporters. Let us say that c_x > c_y > c_z, and that in the absence of any trade the prevailing price in country X is p*. Assume now that country X recognizes the equivalence of the universal component of training obtained in country Y, but not in country Z. There are several possibilities, but we consider only two:

No trade prior to recognition: Prior to recognition, professionals from both countries Y and Z were required to obtain country-specific training in country 1 but neither found it worthwhile to do so, i.e. c_y(u + v_y) + c_x(u + v_x) > p* and c_z(u + v_z) + c_x(u + v_x) > p*.²³ But c_y(u + v_y) + c_x v_x < p*, i.e. when the universal component of training in country Y is recognized as equivalent to that in country X, then professionals from Y find it worthwhile to export to country X, and the price in X will fall. Hence, if all foreign professionals had been completely deterred from practising in country X by the absence of recognition, then any recognition agreement is necessarily trade creating.

Country Z exports to country X prior to any recognition: In this case, c_z(u + v_z) + c_x(u + v_x) < p* < c_y(u + v_y) + c_x(u + v_x), i.e. when all foreign professionals were required to requalify, only those from the third country were willing to supply country 1. But c_y(u + v_y) + c_x v_x < c_z(u + v_z) + c_x(u + v_x), i.e. once the second country professionals are exempted from the basic qualification requirement, they gain a competitive edge over the third country. That is, if lawyers from the third country were already present in the first country, then the recognition of second country professionals would put them at a competitive disadvantage and could lead to trade diversion.

This situation is depicted in Figure 1 (the following discussion resembles closely the discussion in Pelkmans and Winters, 1988). The pre-recognition situation involves domestic output q₂, consumption q₃, and imports from Z, q₃ - q₂. After country X recognises qualifications in country Y, domestic output declines to q₁, consumption increases to q₄, and imports from Y, q₄ - q₁, displace imports from Z. So we are witnessing both trade creation and trade diversion. The welfare effects are straightforward. Consumer surplus rises by A + B + C + D as consumption expands from q₃ to q₄. The area A is a gain at the expense of domestic suppliers, whose surplus falls with their output. Area D is the gain from the better allocation of consumption expenditures; area B the gain from the resources released as inefficient domestic supply contracts; and area C the gain arising from the elimination of wasteful requalification.

The area C + E is of crucial significance, and can be interpreted in several different ways. It helps to recall the analysis of preferential arrangements when tariffs are the instruments of protection. In that case, area C + E would be the loss in government revenue because preferential imports displace high tariff imports. While C is gained by consumers, E is completely lost because supply comes from the more expensive source, and is the loss due to trade diversion. The net gain to the country is only B + D - E, and could be positive or negative. In the example here, C + E were the costs of requalification for country Z professionals when they supplied country X. If these costs were completely dissipated, then they do not enter the welfare calculus of country X. That is, there is no cost of trade diversion for the importing country and the net gain to the importing country from the recognition agreement is B + C + D. If, however, part of these requalification costs (say a fraction α) were appropriated by country X, perhaps as the producer surplus of its training industry or as some form of regulatory rent, then they would be foregone with trade diversion and would need to be taken into account: the net gain to the importing country from the agreement would be B + C + D - α(C + E).

When tariffs are the instruments of protection, the costs of trade diversion for the importing country may be an important deterrent to preferential liberalization agreements. Despite the increase in consumers' surplus from any liberalization, governments may nevertheless be averse to such agreements because the displacement of high-tariff imports from third countries by low or no-tariff imports from preferential sources implies lost revenue. The same reasoning also applies to other regulations which imply a transfer from foreign suppliers to domestic interest groups. However, the situation is different when the protectionist instrument is a regulatory barrier which imposes a cost on the exporter without yielding a corresponding revenue for the importing government or other interest group. There is then no cost to granting preferential access because there is no revenue to lose. Therefore, preferential liberalization is necessarily welfare enhancing for the importing country.

In this simple model, trade diversion reduces the third country's sales to the first country. The simplest way of depicting the negative welfare effects on the third country is by assuming that its supply curve is upward sloping, so that lower sales imply lower prices and loss of producers' surplus. Interestingly, it is possible to show that recognition agreements may well increase global welfare even though suppliers who are left out of such agreements lose. In effect, a recognition agreement is like a positive cost shock to a certain class of suppliers leading to a reduction in price in the importing country. The gain to consumers from any decline in price is necessarily greater than the loss to a subset of suppliers. This makes intuitive sense: eliminating wasteful duplication of training should enhance global welfare. Though, of course, a non-preferential recognition agreement would enhance welfare even more because the service would be supplied by the most efficient locations.

In this Section, it is argued that there is a need to depart from the GATT precedent in examining consistency of measures with the MFN obligation.²⁴ First, the conventional criteria used to judge whether two products are like are not particularly useful in the services context and need to be replaced by a consideration of the end-uses of services. Secondly, the sequential procedure of first determining whether two products are like and then if one has been treated less favourably leads to a legal cul-de-sac, and needs to be replaced by a simultaneous consideration of the degree of unlikeness and differences in treatment. Thirdly, there is a need to read in the context of Article II a necessity test of the kind previously associated with the GATT exceptions provision. Finally, the necessity test itself needs to be refined to take into account the economic efficiency of instruments.

Consider, first, how the GATT precedents for defining "likeness" a priori would fare in the GATS context. In the GATT, likeness has been determined on a case-by-case basis in terms of criteria such as "the product's end-uses in a given market, consumers' tastes and habits, which change from country to country; the product's properties, nature and quality, and the product's tariff classification".²⁵ In the context of Article I, considerable importance has been given to a country's tariff classification. Even though the Harmonized System has brought about a large degree of harmonization in the field of customs classification of goods, this system "did not entail any obligation as to the ultimate detail in the respective tariff classifications" and "a tariff classification going beyond the Harmonized System's structure is a legitimate means of adapting the tariff scheme to each contracting party's trade policy interests, comprising both its protection needs and its requirements for the purposes of tariff and trade negotiations..."²⁶ However, even though Members have significant discretion in the choice of classifications, the MFN obligation required "that the same tariff treatment be applied to 'like products'" irrespective of the classification adopted.²⁷

Corresponding to the tariff classification in the case of goods, commitments in services have generally followed a classification scheme based on the United Nations Central Product Classification, which is in the process of being revised.²⁸ The status of this classification, and the closeness with which it has been followed in scheduling commitments, do not compare favourably with the well-developed Harmonized System classification. However, in so far as the services classification was developed and used for negotiating commitments on both market access and national treatment, it could be argued that it represents mutually accepted product distinctions for regulatory purposes. Thus being classed in separate sub-sectors in the services classification may be regarded as a sufficient condition for "unlikeness". However, given the high level of aggregation of the services classification, being classed in even the finest sub-division can only be considered as at most creating a presumption that the services are like.²⁹

In the goods context, the elevation of certain physical characteristics of the product to decisive status, regardless of consumer perception or behaviour in the market-place, does not seen entirely persuasive nor is it capable of consistent application. In the case of services, physical characteristics are, perhaps fortunately, irrelevant. There may also be a need to consciously deviate from the narrow interpretation of the MFN obligation in the goods context based on the fact that the provision mentions "like products" but not the wider category of "directly competitive or substitutable products" as does the national treatment obligation.³⁰ It is the notion of "end uses" which offers the most appropriate basis for comparing services. This notion is related to the economically meaningful concepts of directly competitive or substitutable, and follows the logic of the market place. After all, what matters is whether consumers treat the services in question as substitutes.

But irrespective of the criteria chosen to arrive at an a priori notion of like services, it would be difficult to deny regulators the right to make further distinctions between services. Furthermore, the scope for making such distinctions is arguably greater in the case of services than in the case of goods. It may be useful to begin with an example which illustrates why these issues might be particularly difficult to tackle in the GATS context. Country X, imposes the requirement that only foreign doctors trained in country Y will be allowed to practice in X. It argues that the measure is not inconsistent with MFN because all foreigners have access to medical colleges in Y, and even citizens from Y trained elsewhere are not allowed to practice in X. Even though the measure is not de jure discriminatory, surely it is de facto discriminatory since the incidence of the measure is much greater on foreign doctors from countries other than Y? But X argues that a determination of de facto discrimination must be consequent upon a determination of likeness, and in its view the service provided by doctors trained in countries other than Y is unlike a service provided by a doctors trained in Y. Say country Z were to complain about the measure imposed by X. How would GATS disciplines deal with this problem?

The burden of proving likeness would fall on the complainant. County Z would argue that the services of doctors trained in Y and Z are indeed like products for the purpose of Article II. However, country X could argue could that training in Y is more suited to the needs of X (in terms of the earlier example, v_x is similar to v_y but not v_z) or that the examination system in Y is more reliable (the u obtained in country Y can be trusted but not that in Z). Since standards and qualification requirements are unlikely to be exactly the same in any two countries, the question is how much difference justifies a pronouncement of unlikeness? A WTO panel would find it difficult to appropriate the right to make this judgement from the medical authorities of country X: if it did, and judged the services to be "like", it would be forced to recommend that X should provide access to doctors trained in Z on the same term as those trained in Y. If, on the other hand, it accepted that the services were "unlike" then no matter how great the difference in treatment, the disciplines of Article II would simply not apply.

In these circumstances, the most reasonable argument that a panel could advance would be that country X should subject doctors trained in countries other than Y to tests of competence (establish the authenticity of u) or to training only in country-specific areas (vx) rather than insist on their undergoing full retraining in Y. In other words, full requalification is not necessary to meet the objective of ensuring high quality medical services since the same objective could be met by the less trade-restrictive means. But here is the key question: can we read Article II to contain a necessity test which would enable a panel to pronounce judgement on the stringency of a measure? Of course Article XIV, dealing with general exceptions contains such a test, but to move to Article XIV would require the panel to establish first that the measure is inconsistent with Article II. It is only then that a country would need to invoke the general exceptions provision. Thus, it would be difficult for a panel to circumvent the need to pronounce judgement on "likeness."

There is, in principle, a more direct route under the GATS to address the problem of standards which are unduly trade-restrictive. This is through Article VI which deals with domestic regulation. Today Article VI primarily provides a mandate (in Article VI:4) to develop disciplines under the GATS to ensure "that measures relating to qualification requirements and procedures, technical standards and licensing requirements do not constitute unnecessary barriers to trade in services." Pending the entry into force of disciplines developed pursuant to Article VI:4, the only current disciplines are contained in Article VI:5, which requires that

"... the Member should not apply licensing and qualification requirements and technical standards that nullify or impair any existing sectoral commitments in a manner which:

1. does not comply with the criteria outlined in subparagraphs 4(a) [objectivity and transparency], 4(b) [not more burdensome than necessary to ensure the quality of the service], and 4(c) [in the case of licensing procedures, of not being in themselves a restriction on the supply of the service]; and
2. could not reasonably have been expected of that Member at the time the specific commitments in those sectors were made." (emphasis added)

The requirements in (i) could have constituted a powerful discipline, but element (ii) renders them toothless. In the extreme, it could be read as "grandfathering" all existing restrictive requirements. Thus country X could argue that training in medical schools in Y had always been a requirement to practice in X and it could, therefore, "reasonably have been expected" when the specific commitments were made. This provision seriously limits the scope for translating the commitments under GATS into non-discriminatory market access.

In this respect, it is worth comparing Article VI of the GATS with the newly strengthened Agreement on Technical Barriers to Trade (TBT). The key provision is Article 2. Article 2.2 states that:

"Members shall ensure that technical regulations are not prepared, adopted or applied with a view to or with the effect of creating unnecessary obstacles to international trade. For this purpose, technical regulations shall not be more trade restrictive than necessary to fulfil a legitimate objective, taking into account the risks non-fulfilment would create ..." (emphasis added)

Thus, not only does the TBT deter discriminatory measures, as does GATT Article III, it goes further in imposing additional disciplines on non-discriminatory measures. It makes possible a critical assessment of non-discriminatory standards (i.e. the choice of v₁ itself) from the point of view of their impact on trade in a way that is not currently possible under GATS.³¹

It could be argued that Article VI will eventually contain a second line of defence against origin neutral measures which escape Article II. But there are two problems. First, even though the domain of Article VI is all domestic regulations, significant disciplines are to be developed under the Article VI:4 mandate only for qualification requirements and procedures, technical standards and licensing requirements. Secondly, there is an intertemporal problem. It is reasonable to assume that at some point in the future Article VI of GATS will be as strong as the TBT, and note furthermore, that its domain is wider than that of the latter. If this were to happen, then the escape of certain measures from Article II would indeed be less cause for concern. However, today, with Article VI not creating a strong discipline, reading Article II narrowly would mean an indefinite wait for meaningful disciplines.

One solution to the problem would be to take a stringent view of likeness in the context of Article II and then Members found to be acting inconsistently could invoke Article XIV, the exceptions provision in defence. However, the list of regulatory objectives in Article XIV as it currently stands is far too narrow, and cannot be taken as the universe of objectives that can be legitimately pursued by WTO Members.³² Furthermore, the prospect that these objectives will be broadened in the near future through political negotiation does not appear very promising. The central problem is that on the one hand, it would be difficult to prevent Members from invoking objectives other than those listed in Article XIV (dealing with general exceptions) as a basis for taking measures which may have a discriminatory effect; on the other hand, allowing them to do so may make it possible to circumvent the disciplines of Article XIV and open the door to all manner of discrimination being justified as incidental to the pursuit of supposedly non-discriminatory aims.

There is a solution to this problem: to read disciplines in GATS Article II of the type previously associated only with GATT Article XX and GATS Article XIV, both of which deal with general exceptions. This implies, on the one hand, accepting the right of regulators to pursue a legitimate objective, but on the other hand, ensuring that the objective is not pursued in a manner which discriminates between foreigners. In effect, the question of whether two services or service suppliers are treated differently must not be separated from how they are treated differently.

On this basis, a two stage test can be suggested:

1. Stipulate an a priori definition of like services based on similarity of end-uses, and a clear relationship of substitutability and direct competition based on market conditions.

The criterion of end-uses serves to demarcate the class of services or service suppliers within which a particular measure may give rise to protectionist effect. For example, a higher regulatory burden on banks than on insurance companies would clearly not arouse concern in the same way that a higher burden on banks regulated in one country rather than another would. But, even within the class of similar end-use, a criterion is needed to distinguish between situations in which discriminatory effect is an incidental consequence of a domestic measure and those in which it is not.

2. If a Member takes measures which distinguish between what could be regarded as a priori like services or service suppliers, then that Member must assume the burden of proving that any resultant unfavourable treatment of foreigners is necessary. In other words, that the Member could not have achieved the stated objective through any other reasonably available measure which did not disadvantage foreign services or foreign suppliers, or did not disadvantage them as much.

There is no precedent of GATT panels applying a necessity test in the context of Article I (MFN). There is a strong case that the GATT precedent of uncritical acceptance of the choice of instruments in the context of Article I needs to be disregarded. In the GATS context, since it will be difficult to resist the invocation of objectives in the context of II, it is imperative to apply a more stringent test. If such a test cannot already be read in Article II, then it would seem desirable that its explicit inclusion be considered in future negotiations.

Moreover, the necessity test, as interpreted by previous panels in the context of GATT Article XX, may itself may be in need of refinement.³³ In their exclusive concern with GATT-consistency, these interpretations do not explicitly address the issue of efficiency of measures. Economic principles can provide meaningful rules for the choice of instruments. Ideally, state intervention is meant to remedy distortions. The economic theory of optimal policy intervention provides an hierarchy of instruments to deal with the problem of domestic distortions. The "first-best" instrument is the one which attacks the divergence between the private and the social cost at the source. There is a case for interpreting the necessity test to create a presumption in favour of the choice of the economically optimal policy to achieve a legitimate objective. A Member choosing to pursue an objective by a measure other than the economically optimal policy would need to justify this choice.³⁴

Thus, in ruling on the foreign doctor dispute, a Panel could argue: considerations of end-use and substitutability create a presumption that all foreign doctors provide like services. Country X, in the light of its objective of ensuring high quality medical services, does not consider the services provided by doctors trained in Y and those by doctors trained in Z like services. The objective is legitimate. But if the disciplines of Article II are not to be circumvented, then the judgement of unlikeness itself and the manner in which supposedly unlike services are treated must not be such as to accord less favourable treatment to some foreigners as compared with others. This implies that the instrument chosen, and the manner of its application, should not unnecessarily discriminate between foreign doctors. There are, in principle, a range of instruments which could achieve the objective of ensuring adequate quality of medical services. The optimal instrument would be one which achieved the objective of remedying the problem of asymmetric information about foreign suppliers' abilities at least cost: say through a test of competence. Even if Country X's distinction is accepted, the instrument chosen, full training in Y, modifies conditions of competition excessively even in the light of the objective, which could be attained by means less unfavourable to doctors from Z, such as tests of competence with regard to universal skills and additional training only in country-specific skills.³⁵

The approach suggested here could be applied not only to establishing the likeness of services but also to the question of who can be regarded as like suppliers - a subject on which GATS offers little guidance. The logic of the MFN obligation would suggest beginning with the presumption that suppliers of like services are like suppliers. Any distinction made between suppliers presumed to be like would need to be justified. But there are indications that Members may wish to take a more flexible approach to this issue. First of all, by scheduling limitations under GATS Article XVI:2(e), a Member specifies the types of legal entity through which a supplier may supply a service. Thus a Member may allow the supply of a service through a subsidiary but not a branch. Secondly, the treatment of taxes under the GATS seems to suggest a willingness to accept that suppliers of like services located in different tax jurisdictions can be legitimately treated differently. Nevertheless, Section 5.5 illustrates how the suggested approach could be applied to the like supplier issue.

The previous reasoning can also be applied to resolve the difficulty of applying Article VII. An MRA essentially helps beneficiaries to circumvent the equivalence establishment process. The key question concerns the treatment of those with whom an MRA has not been concluded or to whom recognition has not been granted. Article VII:2 requires that an opportunity be afforded to other Members to conclude MRAs or to demonstrate equivalence, but says little about the rights of individual suppliers. The more powerful discipline is contained in Article VII:3 which prohibits the use of recognition as a means of discrimination. While recognition, unilateral or through an MRA, amounts to an acceptance of likeness vis-à-vis certain suppliers, it also defines a standard of treatment vis-à-vis other suppliers. In terms of the example, if X recognizes the universal element u obtained in Y, then an individual supplier from Z should have the possibility of demonstrating equivalent training obtained in Z. If X has legitimate doubts about the authenticity of the training, then the incremental burden imposed on such a supplier should be no greater than that necessary to achieve the regulatory objective. This would amount to a presumption in favour of a test of competence. Ideally, local qualification or training should only be used as a means of remedying differences in the country-specific element, v_i.

Such an interpretation of Article VII:3 would ensure that the liberalizing effect of MRAs can be generalized. At present, given the weakness of Article VI, if a Member has not made a commitment to grant national treatment, then the Member is free to impose uniformly burdensome regulations on all foreigners. However, if a Member concludes an MRA or grants recognition to one foreign country, then this provides all others with a potentially valuable foothold by defining a standard of treatment. The problem is that if it is rationally expected that extending recognition to one Member would eventually require extending it to many, then even the recognition of one may be deterred. This would be undesirable since even limited preferential liberalization through MRAs may be welfare enhancing in a second-best world, as the previous sub-section has demonstrated. The tendency of Members to notify MRAs under Article V rather than Article VII may reflect an attempt to share these gains on a limited reciprocal basis by avoiding the obligation to extend recognition more widely.

While quotas are an aberration in goods trade and have been more or less outlawed, they are a pervasive form of protection in services and are only inconsistent with the GATS in sectors where a Member has undertaken not to use them. Furthermore, quotas under the GATS restrict not only the quantity or value of services output, but, also the number of (foreign) service suppliers.

There is a huge economic literature examining almost every aspect of quotas so this section makes only a only a few brief remarks. The earlier analysis of discriminatory regulation is also relevant here. It could be argued that in services, since intermediation is difficult, rents are usually appropriated by exporters rather than domestic importers. As in the case of certain restrictive regulations where the protective instrument does not generate domestic revenue (or rents), there is no cost to granting preferential access because there is no revenue (or rents) to lose. Therefore, preferential liberalization is necessarily welfare enhancing for the importing country. However, we can conceive of certain situations, such as when there are quotas on workers from countries with low labour costs, where the rents accrue to the domestic employer/supplier. Such an employer/supplier would oppose a preferential allocation of the quota to countries with higher labour costs. It may also be the case that the importing country is not indifferent to the allocation of rents between foreign suppliers - as was the case in the European Union's régime for the distribution of bananas. In such cases, there may be welfare costs to preferential allocation of quotas as well as political economic opposition to such arrangements.

One central legal tension in the GATS, which has received surprisingly little attention, is between quotas on market access, particularly those specifically listed under Article XVI, and Article II concerning non-discrimination. Quantitative restrictions include those on the number of service suppliers, volume of output, value of output, and number of service suppliers. The problem concerns how these quotas are to be allocated in a manner consistent with the non-discrimination obligation. In the past, this was not a major issue because commitments reflected the status quo and the quotas, particularly with regard to service suppliers, were descriptions of the existing market structure. Thus when Bangladesh committed to "four licenses issued" in cellular telephony, the ambiguity in the choice of tense was not an accident: the licenses in question had already been issued. But in the future, as genuine liberalizing commitments are made, the non-discriminatory allocation of quotas is bound to be an important issue.

Does the goods precedent offer any guidance? GATT Article XIII, on the "non-discriminatory administration of quantitative restrictions", states in paragraph 2 that: "In applying import restrictions to any product, contracting parties shall aim at a distribution of trade in such a product approaching as closely as possible the shares which the various contracting parties might be expected to obtain in the absence of such restrictions ..." It also stipulates that the country implementing a quota should ideally reach an agreement regarding its allocation with all countries which have a substantial interest in supplying the product. If this is not "reasonably practicable", then allocation should be based on proportions supplied by the countries during a previous representative period, "due account being taken of any special factors which may have affected or may be affecting trade in the product."

In the services context, the requirement to replicate historical shares may have no relevance if there was no previous foreign presence, or even perpetuate historical discrimination if previous quotas were allocated to favoured suppliers.³⁶ Furthermore, neither of the instinctive candidates for a non-discriminatory rule, first-come, first-served and a system of auctions to the highest bidder, would necessarily lead to distributions "which ... might be expected to obtain in the absence of such restrictions." It is obvious that first-come, first-served favours the proximate. Auctions would give the relatively efficient producers larger shares than they would have obtained in the absence of quotas (when quotas are set at below unrestricted trade levels).³⁷ It would seem, therefore, that the rules for ensuring non-discriminatory allocation of quotas under GATS would need to look beyond the GATT-precedent. It is possible that a less elaborate variant of the disciplines in the Agreement on Government Procurement, designed to ensure competitive tendering on non-discriminatory basis, will need to be considered.

Once we introduce service suppliers and time into the picture, the picture gets more complicated. Consider the example of the several governments who have made market access commitments to issue an exclusive license to a telecommunications services provider and guaranteed no new entry for a fixed period, say ten years. This licence may have been auctioned off in a non-discriminatory manner to a supplier at a price which reflects the future discounted monopoly profits over ten years. Now suppose a year later, the government decides to liberalize further and issues one more license.³⁸ The government is not prevented from doing so by Article XVI because limitations are minimum guarantees and not maximum limits. Say the new license is also issued through a non-discriminatory auction. The additional license is likely to be sold at a price lower than the previous price, because the expected profits of the new entrants are lower than those of the first entrant who believed that it would be a monopoly - first, because the new situation is a duopoly and the new entrant may build into its bid the expectation of more competition.³⁹ Quite apart from the first entrant's rights under the terms of its contract with the government, could it be claimed that the government's actions violated Article II because it had been charged a higher entry price than the new entrant? Would the fact that the government obeyed a non-discriminatory rule at each point of time be a sufficient defence or is it obliged to extend the same treatment across time after the GATS came into effect?

As described in Section 3, a number of Members have sought MFN exemptions on the grounds of reciprocity. Furthermore, initially in sectors like financial services and telecommunications and until now in maritime transport, MFN-based outcomes could not be attained because certain Members wished to retain the right to grant market access conditional on the treatment of their suppliers in other markets. It is important to distinguish between reciprocity as an aspect of the bargaining process and reciprocity as an aspect of the trade regime. In the former case, concessions are negotiated on a reciprocal basis and the results are multilateralized, so discrimination is not practised in actual trade. In the latter case, reciprocity is the basis for the implementation of trade policy, so discrimination is a feature of actual trade policy.

Recently, Bagwell and Staiger (1997) have provided a rationalization of reciprocity-based negotiations which is independent of the national policy formulation process, i.e. of whether government policy is aimed at maximizing national welfare or promoting the interests of particular groups. They begin with the recognition that the spillover effects of policy in one country on other countries necessarily take place through changes in the terms-of-trade. They find, first in a two-country game, that a reciprocity rule in negotiations - in the sense, that I will reduce my protection if you will - can take the countries from an inefficient Nash equilibrium to the efficient policy locus. This result follows from the fact that reciprocity insulates each country from the terms-of-trade loss consequent upon unilateral liberalization and hence eliminates the negative externality which leads to the inefficient Nash equilibrium. Furthermore, if we include a rule for reciprocity-based renegotiation of concessions - in the sense, that if I increase my protection, so can you - then, the only renegotiation-proof equilibrium is, in fact, the Pareto optimum. If governments were welfare maximizing, then the Pareto optimum would be free trade. If they were not, then the Pareto optimum implies some positive level of protection.

In their model, given that reciprocity is the basis for negotiations, in a multi-country world, the MFN principle implements the efficient outcome. The intuition is that with discriminatory tariffs, countries have an incentive to import relatively more of a given volume of imports from countries on whom higher tariffs are imposed since this increases tariff revenue and hence welfare. The volume of imports from a particular source depends, of course, on the local price in the source country. Thus, under discriminatory tariffs both world and foreign local price externalities arise, whereas under the principle of non-discrimination, the only externality between government is the world-price externality. In the latter case, the principle of reciprocity works well to neutralize externalities and deliver efficient outcomes. However, in the former case, the principle of reciprocity neutralises world price externalities but not local price externalities, and, therefore, no longer serves as a means to implement efficient trade-policy outcomes.

Bagwell and Staiger thus provide a useful rationalization of the non-discrimination principle as complementing reciprocity-based negotiations to deliver efficient outcomes. What they do not do is explore fully the incentive properties of the MFN rule when there is a possibility of strategic behaviour in bargaining. For instance, how is the free-rider problem overcome when the results of reciprocity are immediately multilateralized? When do large countries have an incentive to play by such rules rather than do either bilateral deals or indulge in aggressive unilateralism? As noted above, these questions assume particular relevance in the context of services negotiations

The MFN principle is not simply a rule that constrains trade discrimination, but also a rule that influences the multilateral bargaining process (Schwartz and Sykes, 1996).⁴⁰ Acceptance of the MFN principle implies that ex ante the participants in the negotiations know that any concession negotiated with a single trading partner must be unconditionally extended to all other trading partners, regardless of their ability to extract a quid pro quo. At the same time, the benefit from any concession extended by others may be obtained without having to give a quid pro quo. Ex post, the participants know that the value of any concession that they obtain under MFN will not be impaired by subsequent more generous concessions to other countries - except, of course, in the context of a regional integration agreement.

The problem is that Members may try and free ride on each other during bargaining - as was alleged in the context of the financial services and telecommunications negotiations. Each of the beneficiaries of a concession from a trading partner may be tempted to understate their willingness to pay for it, hoping that offers of reciprocal concessions from other Members will be sufficient to induce the concession. If each Member behaves in this way, the result could be that mutually beneficial deals will not be struck. This was reflected in the unwillingness in some of the services negotiations of some Members to make binding commitments unless a certain "critical mass" of Members was willing to make significant liberalization commitments. But it is not clear how well-founded these fears of free-riding were. Table 3 presents estimates of how concentrated trade in services is on the basis of balance-of-payments statistics. These figures do not include trade through commercial presence, but such trade is probably even more concentrated than cross-border trade. It is evident that in most sectors, the top 10 exporters and importers account for a large part of total exports and imports, respectively, even taking into account the limitations in reporting. A more detailed analysis reveals that the major exporting and major importing countries tend to be the same. This degree of concentration does not provide strong support for the free-riding view.⁴¹

Nevertheless, in so far as free-riding is a real problem, two possible solutions exist: for beneficiaries to form a caucus to work out a collective offer for a desired concession - which may describe the European Union's efforts to form a package of offers in financial services and maritime negotiations. The second is to abandon product-by-product negotiations and agree instead that every Member will reduce barriers according to a mutually agreed formula - which may be difficult in services given the nature of the instruments of protections (see Sapir, 1998).

However, because there is no complete solution to the free-rider problem, it is likely that the MFN principle prevents the parties in large multilateral trade negotiations from exhausting all joint political gains and makes the first-best political optimum unattainable. In the absence of such a principle, a country that secured a discriminatory concession would realise the full benefit of it and seek no other concession from an alternative source to induce it. Ceteris paribus, the desirability of discriminatory arrangements would be greater, the more serious the free-rider problem for non-discriminatory arrangements, and the less the amount of trade diversion that results from discrimination. Under these circumstances, preferential trading arrangements may make some political sense as a second best alternative, even though such arrangements tend to dissipate political surplus through trade diversion, which affects producers and national treasuries; they also tend to undermine the security of trade concessions.

As Schwartz and Sykes argue, it is in averting this last cost that the MFN principle also brings benefits for the bargaining process. A trading regime without an MFN obligation creates an opportunity for a subset of Members to threaten to create a discriminatory arrangement to induce other nations to make concessions. The possibility of such threats may create an unstable trading regime. Furthermore, the MFN obligation protects the value of concessions against future erosion through discrimination. My willingness to pay for a concession may be undermined if I fear that you might subsequently grant preferred access to someone else. Faced with this possibility I would offer less for the concession in the first place and fewer mutually beneficial deals would be struck. A blanket MFN provision is of course not the only solution to the problem. Each party to a trade agreement could secure in conjunction with the concession additional promises limiting the future concessions that a trading partner could subsequently make in the same domain to other trading partners.⁴² However, the transaction and information costs of proceeding in this fashion would be high. Alternatively, members could confine all concessions to well defined multilateral negotiating rounds - as is generally the case, with the exception of regional initiatives. No nation would commit to anything until it could review the entire schedule of proposed discriminatory concessions by each of its trading partners. Both these approaches would compare unfavourably with the simple MFN rule in terms of ease of administration and the maximization of joint surplus.

In the services context, the possibility of seeking MFN exemptions meant that ex ante bargaining often took place without the guarantee of the MFN principle. The United States and others held out against the acceptance of the principle until they were satisfied with the liberalization commitments obtained from others. The others in turn were unwilling to permanently bind these commitments unless the United States and others accepted adherence to the principle, i.e. guaranteed them a certain security of concessions. To an extent therefore the ex ante costs of the MFN principle were avoided while the ex post benefits were reaped - at least in areas where full MFN-based liberalization was achieved.

Is there any restriction on the type of measures for which an MFN exemption can be obtained? There is little doubt that a Member can list measures which grant discriminatory access in a particular sector. But can a Member, by taking an MFN exemption, maintain the right to take retaliatory measures on a discriminatory basis against, what it unilaterally judges to be, foreign restrictive trade practices? Examples of MFN exemptions of which reserve the right to do so are the European Union's exemption in audiovisual services states and the United States legislation in maritime transport for which an exemption was not listed because the MFN obligation was suspended for the sector.⁴³

It would need to be examined whether an interpretation of Article II which allows this would be inconsistent with the Understanding on Rules and Procedures Governing the Settlement of Disputes (DSU), in particular, Article 23 which deals with strengthening of the multilateral system. This Article states that "when Members seek the redress of a violation of obligations or other nullification or impairment of benefits under the covered agreements or an impediment to the attainment of any objective of the covered agreements, they shall have recourse to, and abide by, the rules and procedures of this Understanding." To abide by the rules and procedures of the DSU would seem to preclude a unilateral determination of a violation.

Audiovisual services and maritime transport services are a part of GATS and thus covered by the DSU. Could it be argued that Article 23 of the DSU does not prevent a Member from taking unilateral action to obtain benefits beyond those accruing under the covered agreements? That is, could a Member unilaterally attempt to induce greater liberalization in another Member's market than is granted in its schedule of commitments? An argument could be made that this would imply seeking the redressal, without abiding by the rules and procedures of the DSU, of "an impediment to the attainment of any objective" of the GATS. The preamble to the GATS states that Members desire "the early achievement of progressively higher levels of liberalization of trade in services through successive rounds of multilateral negotiations." Thus, inducing further liberalization in all services sectors is an aim which has, in a sense, been appropriated by the GATS, and so Members are no longer free to pursue this aim by actions which are inconsistent with the DSU. This interpretation, however, needs closer examination.

One of the central problems in the financial services negotiations was solved by a scheduling innovation. The conflict arose because certain countries were unwilling to make commitments which reflected the status quo with respect to commercial presence. Thus, they were either inclined to bind foreign ownership levels below those which currently prevailed, or insist on legal forms (local incorporation) other than those currently in the market (branches), or both. In some cases, the problem arose because domestic law had changed since the foreign firms first established commercial presence, e.g. in Malaysia, where the indigenisation policy was being implemented after the establishment of many foreign firms.

To see the nature of the problem, consider the following example. Say country A had allowed a firm from country B to establish a fully owned subsidiary in 1990. If it made a specific commitment under GATS to allow commercial presence through fully-owned foreign subsidiaries, then it would be obliged to allow firms from all Members to establish under similar conditions. The implied level of openness was apparently unacceptable for some Members in situations similar to country A. If, however, market access was bound only for, say, minority owned foreign firms, then the existing firm from country B would have no guarantee that it would not one day be asked to disinvest. Such an uncertain situation evoked significant concern in existing investors from countries in type B situations. The solution was to drive a wedge between the conditions facing firms which were already present and those which would enter when the commitments came into effect. In effect, the situation of existing firms was "grandfathered".

The three types of grandfather provisions, foreign equity-related, legal form-related and general, which are to be found in the financial services schedules are shown in Table 4. It is evident that grandfathering was primarily an Asian phenomenon, prompted either by the introduction of more restrictive regimes pertaining to foreign equity and legal form than had prevailed when the foreign firms first entered, or simply an unwillingness to guarantee what had previously been allowed ad hoc.

The grandfather provisions reflect the relative emphasis in these negotiations on guaranteeing the rights of incumbents. They provide the benefits of security to investors who are already present in the market rather than to new investors and thus do little to make markets more contestable. Furthermore, they may even place new entrants at a competitive disadvantage where differences in ownership and legal form affect firm performance. For instance, larger equity shares may make it easier to exercise effective control over the operations of a firm and ensure efficient production, and the marginal cost of providing a service through a branch may be lower than through a subsidiary. Thus, Members who incorporated such provisions into their schedules may have succeeded in using the GATS to discourage rather than encourage competition.

The legal question is of course the consistency of these provisions with Article II. This question can be posed in two forms: first, would it be consistent with MFN to allow new (post-GATS) foreign firms to operate under less favourable conditions than old (pre-GATS) foreign firms? Secondly, would it be consistent with MFN to provide a binding of such differences in treatment through the GATS schedules? The instinctive answer to both questions is probably negative. After all, the whole point of the GATS was to level the playing field and eliminate all forms of discriminatory treatment, historical or otherwise. However, regarding the first question, it could be argued that the GATS only applies to measures affecting trade in services after it came into affect. The conditions of operation of the old firms are governed by licenses issued before the Agreement came into effect and so are not covered by Article II of GATS. The question arises, of course, whether the continued operation of firms under such conditions does not in fact require continued measures to which the GATS applies. But regardless of how one treats the first question, a positive answer to the second question is difficult. For even if we accept that identical conditions of competition cannot be ensured because the inequity is caused by actions predating the GATS, to guarantee the perpetuation of the inequity under the GATS by scheduling different levels of commitments for old or new firms must surely raise questions of MFN-consistency. Or is the notion of likeness to be contingent on the time of a firm's entry into the market?

Even though it is evident that the GATS itself contains only limited obligations on Members to curtail anti-competitive practices, the question arises whether it leaves Members adequate scope to take action against the anti-competitive practices of the service suppliers of other Members.⁴⁴ An interesting issue arose in the context of the basic telecommunications negotiations pertaining to international services which illustrates this issue. Some Members were reluctant to grant unconditional MFN access to their liberalized markets to operators originating from non-liberalized markets. Apart from the desire to retain negotiating leverage, there were two main concerns. The first was the fear of anti-competitive cross-subsidization and the second was the problem of "one way monopoly by-pass." The first concern is relatively straightforward, and not unlike certain concerns about predatory pricing that have been raised about the dumping of goods. It was feared that the competitive structure of the liberalized market could be undermined if an operator from a closed home-market were to subsidize its activity in the liberalized market, using revenues generated in the closed home market, including revenue from excessively high accounting rates.⁴⁵ The second concern was that the operator from the closed market would be able to by-pass the accounting rate system by establishing commercial presence and its own facilities in the liberalized market and using this to terminate its own calls. Operators of the liberalized market would, meanwhile, still have to depend on the closed-market operator to terminate their calls. The imbalance in payments between the two would worsen, since the liberalized market operators would no longer be able to offset some of their payments for outgoing calls by receipts for incoming calls.

Several solutions were considered to the problem of achieving MFN-based liberalization while allowing Members the freedom to deal with possible anti-competitive practices by exclusive operators from closed markets. One was taking measures ex ante, consisting in limiting market access for operators capable of engaging in anti-competitive practices because their home-markets were closed. Second was taking measures ex post, after market access is granted, if there was evidence of actual anti-competitive practices. And finally, there was the possibility of commitments by non-liberalizing Members to restrain exclusive operators from engaging in anti-competitive practices in international services. The limited time available did not permit a full exploration of the alternatives, and eventually it was the second option which seems to have been adopted by default.

The ability to exercise national competition policy depends again on the interpretation of the notion of "likeness" in the MFN obligation. Is a long-distance call provided at a low price by the subsidiary of a protected monopolist like a long-distance call provided by any other supplier? And is the subsidiary of a protected monopolist like any other supplier? If the reply to these questions is in the affirmative, then the MFN obligation would preclude any discrimination between these services and service suppliers, including on the basis of competition policy considerations. Alternatively, it could be argued that if competition policy itself is based on non-discriminatory principles, then it would be acceptable for it to impact differentially on particular services or service suppliers in specific instances provided they themselves manifested the characteristics which aroused concern. For instance, if competition policy had general restrictions on cross-subsidization or on the expansion of dominant suppliers, then specific actions which happened to be directed against foreign services or service suppliers which had these attributes would not constitute infringements of MFN. However, it would be required, in line with the argument in Section 5.1, that the actions should not be more burdensome than necessary to achieve the relevant competition policy objective.

However, any discrimination in treatment, based not on competition policy-related attributes of the service or the service supplier, but on unrelated attributes such as the fact of protection in the home market, would seem to violate the MFN obligation. There would seem to be an acknowledgement of this distinction in the case of maritime transport. This time the concern was that anti-competitive practices in the provision of port services in certain countries hampered access of foreign maritime transport suppliers to their markets. How far these practices are facilitated by government action (or inaction) is subject to dispute. In this case, the MFN obligation was not a constraint on retaliatory action since the application of the obligation to the sector has been suspended. Indeed, one of the reasons the United States presented for seeking an MFN exemption when the application of the obligation was a possibility, was precisely to preserve its right to retaliate against foreign restrictive practices.⁴⁶

Altinger, L. and Enders, A. (1996), The Scope and Depth of GATS Commitments, World Economy, 19, 65-94.

Bagwell, K. and Staiger, R. (1997), GATT-Think, mimeo.

Farber, D.A. and Hudec, R.E. (1996), GATT Legal Restraints on Domestic Environmental Regulations, in Bhagwati, J. and Hudec, R.E. (eds.), Fair Trade and Harmonization: Prerequisites for Free Trade? Volume 2, Legal Analysis, MIT Press, Cambridge, Massachusetts, and London, England.

Hoekman, B. (1996), Tentative First Steps: An Assessment of the Uruguay Round Agreement on Services, in Martin, W. and Winters, L.A. (eds.), The Uruguay Round and the Developing Countries, Cambridge University Press, Cambridge.

Mattoo, A. (1997), National Treatment in the GATS - Corner Stone or Pandora's Box?, Journal of World Trade, 31, 107-135.

Mattoo, A. and Subramanian, A. (1998), Regulatory Autonomy and Multilateral Disciplines, Journal of International Economic Law, 1, 303-322.

Pelkmans, J. and Winters, A. (1988), Europe's Domestic Market, Chatham House Papers 43, Royal Institute of International Affairs, Routledge, London.

Roessler, F. (1996), Diverging Domestic Policies and Multilateral Trade Integration, in Bhagwati, J. and Hudec, R.E. (eds.), op. cit.

Sapir, A. (1998), GATS 1994-2000, mimeo.

Schartz, W.F. and Sykes, A.O. (1996), Toward a Positive Theory of the Most Favoured Nation Obligation and Its Exceptions in the WTO/GATT System, International Review of Law and Economics, 16, 27-51.

Source: Modified version of Table 9 in Altinger and Enders (1996)