Patrick Low and Aaditya Mattoo *

The General Agreement on Trade in Services (GATS) was a bold and long overdue innovation in international rule-making. While many writers and observers of trade policy readily acknowledge this, they are also inclined to note that the GATS is in need of improvement in several important areas. This includes both the rules and continuing betterment of the somewhat modest degree of liberalization achieved or committed to so far under the agreement (see, for example, Hoekman (1995), Feketekuty (1998), Sauvé (1994), Snape (1998)). The purpose of this paper is to identify some aspects of the GATS structure and approach to liberalization that seem particularly deserving of attention. If the rules do not impose unambiguous disciplines upon the design and conduct of policy at the national level, they will not be enforceable internationally. Nor will they foster a strong commitment to continuing trade liberalization.

The paper is divided into six sections, and is organized around four main themes. The next section (Section II) takes up the first of these themes, dealing with ways in which the clarity of the agreement and schedules commitments could be improved. The second theme (Section III) considers how the GATS might be better used as a vehicle for liberalization, and in particular, for creating effective conditions of competition in services markets. The third theme (Section IV) analyses a number of ways in which the disciplines of the GATS could be deepened in order to provide an improved legal framework and better basis for extending trade liberalization. Finally, Section V briefly examines the scope for carrying out negotiations in ways that might promote a more vigorous negotiating dynamic. Section VI concludes.

Many practitioners and observers have pointed to aspects of the design of schedules and scheduling techniques that introduce an unwelcome element of opacity and interpretative ambiguity into the GATS, making the agreement less effective as a system of rules and vehicle for further liberalization. In this section, we specify some of the more important areas where modifications - sometimes of a relatively straightforward nature - could make a significant contribution to the effectiveness of GATS.

GATS schedules of specific commitments consist of market access undertakings potentially subject to six limitations (Article XVI)¹ and national treatment undertakings which may be conditioned by any kind of specified discriminatory measure (Article XVII). This bifurcation between market access limitations and discriminatory measures potentially raises some confusion about the true nature of Members' scheduled commitments.² The problem is confounded by a scheduling convention set out in Article XX:2. This provision is intended to deal with situations where discriminatory market access limitations are scheduled, or in other words where restrictive measures fall within the scope of both Articles XVI and XVII. In such cases, Article XX:2 states that the relevant measures should be inscribed in the market access (Article XVI) column of the schedule and would be understood to provide a condition or qualification to Article XVII as well. Thus, the market access column contains measures which are inconsistent with Article XVI only (non-discriminatory market access limitations) and with both Article XVI and XVII, but there is frequently no indication as to whether the measures concerned are non-discriminatory or discriminatory.

Since the precise overlap between Article XVI and Article XVII is not identified, the scope of the national treatment obligation remains ill-defined. This is already a problem. But in addition, suppose a Member only undertakes to provide national treatment, and not full market access. In this case, there is no way of knowing (in the absence of a proper definition of national treatment in relation to market access) whether any unscheduled improvements in market access would have to respect full national treatment. The problem of identifying the scope of national treatment is most acute in mode 3, bearing in mind that securing market access under this mode is in practice a two-stage process - one set of measures will define the terms of entry for a foreign investor and another will establish the conditions for post-entry activity.

At least three possibilities suggest themselves in relation to the scope of national treatment in these circumstances (Mattoo, 1997). First, national treatment may be deemed to apply to all present and future market access commitments with respect both to entry and post-entry operations (full national treatment). This interpretation is in line with the text of Article XVII which states that this provision applies to "all measures affecting the supply of services." Second, national treatment may apply only to market access commitments entered into at the time the national treatment commitment itself was made, but not to any subsequent entry beyond the scheduled commitment, while at the same time, national treatment could still apply to all post-entry operations of present and future foreign entrants (post-entry national treatment). Third, national treatment may only apply with respect to initial market access commitments and only to the post-entry operations of these entities.

The differences among these three options are stark in terms of the value of a national treatment commitment, and the fact is that there are many entries in Members' schedules where it would be impossible to determine which of these definitions of national treatment is to apply. What would be the value, for example, of a schedule entry in mode 3 recording no commitment under market access ("unbound"), or specifying the requirement of an economic needs test, and a full commitment under national treatment ("none", meaning no limitations)? Under one interpretation, national treatment would apply to all measures covered by Article XVII. An alternative interpretation is that national treatment would only provide a guarantee of non-discrimination in relation to measures covered by Article XVII other than those covered by Article XVI. Clearly, this is an issue in need of attention.

The GATS defines trade in services in terms of the four modes of supply - cross-border supply (mode 1), consumption abroad (mode 2), commercial presence (mode 3), and the movement of natural persons (mode 4). These modes are also used for scheduling purposes. Two issues that have arisen in regard to the use of modes for scheduling purposes are considered here. First, there is the question of overlap between modes, specifically modes 1 and 2. Second, the definition of "likeness" of foreign and national services and service suppliers across modes of delivery raises a number of interpretative difficulties with regard to the rights acquired through specific commitments at the modal level.

The problem of modal overlap attracted particular attention in the context of the negotiations on trade in financial services. In essence, the question at issue is whether a cross-border financial service transaction should be classified as a mode 1 or a mode 2 transaction. At the margin, this can be virtually impossible to determine under existing definitions. If the transaction is deemed to have originated with a supplier in one jurisdiction selling a service to a consumer in another, then from the point of view of the jurisdiction in which the consumer is located, this would be classified as cross-border delivery, or a mode 1 transaction. If, on the other hand, the consumer initiates the transaction or solicits the service, it could be classified as consumption abroad. This potential confusion between mode 1 and mode 2 transactions obviously becomes important if the commitments scheduled by a Member are not identical in both modes.

A number of solutions to this problem have been mooted. One is to work towards ensuring that mode 1 and mode 2 commitments are indeed identical. The choice of modal definition could then turn on other matters - in particular, the question of whose regulatory system, or which territory, would have legal jurisdiction in respect of a transaction. The issue of jurisdiction has not been addressed in the GATS context, and it is going to become increasingly important with the growth of electronic commerce. Part of the solution will presumably be to clarify the relationship between mode 1 and mode 2. Other suggestions for dealing with the modal definition issue include the amalgamation of both modes into a single one, the redefinition of mode 2 to require the physical movement of a consumer, and the requirement that every sectoral activity involving an inter-jurisdictional transaction is pre-defined in terms of the mode to which it belongs. Space limitations preclude a more systematic analysis of this issue, but in the absence of clarification, doubt will remain as to the value of certain commitments.

The second problem referred to above concerns the possibility that a commitment on a particular service in one mode can be undermined by the absence of a commitment in another mode, or by an interpretation of the relationship among modes that treats a given service as an "unlike" product by virtue of the fact that it is delivered via one mode rather than another (Mattoo, 1997). If, for example, a Member has accorded unrestricted access to the foreign supply of a service under mode 1 in respect of both market access and national treatment, and then offers a subsidy to national producers of the same product in the domestic market - a measure seemingly consistent with the GATS in the absence of a national treatment obligation under mode 3 - then clearly the subsidy will alter the conditions of competition and undermine the value of the mode 1 commitment. Similarly, even if the subsidy was granted to both foreign and national producers operating in the national economy, or in other words, even if the Member concerned had a national treatment commitment under mode 3 as well as mode 1, the grant of a subsidy to domestically-based producers would undermine the value of the mode 1 commitment. Conversely, a tax on mode 3 production would undermine the rights of suppliers under this mode in relation to suppliers under mode 1.

There is nothing in the national treatment provisions of the GATS (Article XVII) which suggests that the mode of supply is a determining factor in defining the "likeness" of a service - alternative modes of delivery may be used to supply "like" services. If this interpretation were adopted, then something would need to be done about the effects of an intervention under one mode on the value of a commitment under another. At present, probably the only recourse would be a non-violation complaint under the WTO's Dispute Settlement Understanding.³ On the other hand, the manner in which trade in services is defined through the modes, and the schedules designed, suggest that commitments are indeed mode-specific. This is an issue in need of attention.

The description of a service under GATS may not be sufficiently developed or explicit for it to be clear whether a commitment is intended to be technology-neutral. Technological neutrality refers to the idea that a commitment covers all means by which the service in question might be delivered within a mode of supply. It became apparent that WTO Members were aware of these kinds of difficulties in the negotiations on basic telecommunications. A Chairman's understanding was developed during those negotiations in order to clarify the coverage of scheduled commitments. The understanding established a presumption that unless indicated to the contrary, the description of a basic telecommunication service in a Member's schedule encompassed the full spectrum of ways in which a service could be supplied. A commitment on voice telephony, for example, would cover radio-based as well as wire-based technologies unless otherwise indicated. Similarly, in discussions in the WTO on electronic commerce and in the Committee on Specific Commitments, Members seem to have agreed that a commitment on a service should be invariant with respect to the means by which the service is delivered.

In considering whether the means of conveying a service should be regarded as a distinguishing feature from a legal perspective, the concept of like product is crucial. It can be argued that products should be deemed alike regardless of the means by which they are conveyed from the supplier to the consumer. Suppose, for example, that a Member claimed that legal services could be supplied cross-border through mail delivery, but not through electronic delivery. In order to sustain the argument that such a regime is non-discriminatory, it would be necessary to assert that identical products delivered by different means of conveyance were not like products in a legal sense. In the sphere of goods, a comparable case would be one in which garments transported by road would be subject to one regime and those transported by air would be subject to another. In order to justify this differentiated regime against a charge of MFN-inconsistency, garments entering by road and identical garments entering by air would have to be deemed unlike products. It is highly unlikely that a WTO panel would see matters in this light.

A further consideration is whether the market access restrictions permitted under Article XVI:2 of GATS, which are all expressed in quantitative terms, could encompass restrictions on the means of delivery of a service. It would appear that such restrictions are not covered. Since the only restrictions that may be scheduled under Article XVI:2 are those that are listed, this implies that limitations on the means by which a service is delivered can neither be scheduled nor directly disallowed. In the absence of an explicit understanding by governments in favour of neutrality with respect to alternative means of delivering a service, along the lines of the Chairman's understanding developed in the basic telecommunications negotiations, such limitations could presumably be challenged through the two provisions in GATS dealing with non-discrimination - the Article II MFN requirement and the Article XVII provisions on national treatment. These provisions would apply, for example, where some suppliers were permitted to use electronic means and others were not, and in the case of national treatment, if a service had been scheduled without national treatment limitations. Clarity and legal certainty would be improved if these different aspects of technological neutrality were clearly addressed.

The structure of GATS requires that a tenable distinction be made between measures intended as limitations on access to the domestic market by foreign-produced services and service suppliers, and measures adopted in pursuit of public policy objectives. This is the distinction between measures falling under Articles XVI and XVII on the one hand, and Article VI on the other. The approach has been to maintain that if a regulatory intervention embodies a market access restriction, the measure should be inscribed under Article XVI. If there is a discriminatory element, then the measure should be scheduled under Article XVII. In all other cases, the disciplines of Article VI apply. Regulatory interventions cannot, of course, be characterized as either restricting trade or not restricting trade, which is why there exists the requirement that regulatory interventions must be the least-trade restrictive possible.

This does raise the question of how clear in practice the distinction will be between Article XVI and Article VI. Whether a public policy measure falls under one or other of these provisions may not always be clear at the margin and will require legal interpretation. Similarly, a discriminatory measure that is "excessive" in the sense of going beyond what is necessary to achieve a public policy objective would not be dealt with as an unnecessary barrier to trade under Article VI, but as a measure in need of liberalization under Article XVII. In the WPPS work on accountancy, it became apparent that some Members felt that embracing certain regulatory disciplines could imply de facto acceptance of market-access commitments. A range of measures considered to fall within the ambit of Article XVI and/or Article XVII were consequently excluded from consideration. Furthermore, the Disciplines on Domestic Regulation in the Accountancy Sector drawn up by the WPPS apply only in cases where Members have entered into specific commitments. While it might be argued that the application of regulatory disciplines independently of the existence of specific commitments would be desirable because it would enhance the conditions of competition in the market, as a practical matter the value of regulatory disciplines in the absence of trade liberalization commitments would be very limited. The legal device of maintaining Articles XVI and XVII entirely separate from Article VI facilitates interpretation, but this does not alter the fact that in the absence of specific commitments, limitations on market access and national treatment can render Article VI disciplines ineffective. This conclusion suggests the need for a parallel approach to regulation and liberalization.

In many schedules, Members have provided information that does not, on the face of it, concern the substance of the legal commitment being undertaken. The schedules are intended to specify in precise terms the nature of commitments on market access (Article XVI), national treatment (Article XVII), and additional commitments (Article XVIII). If descriptive material with no bearing on the legal commitments is included in the schedules, this may lead to legal uncertainty. Members may have chosen to include explanations and additional information because they felt it would clarify the nature of the policy regime to which they were making a commitment. To the extent that this is the case, it indicates a need for better mechanisms to provide such transparency. The schedules are not the place to do it.

At the same time, there are many examples in Members' schedules where the precise nature of legal obligations is not spelled out. Perhaps the most egregious of these are the references to economic needs tests, permitted under Article XVI:2. When the nature of these needs tests is unspecified, uncertainty and scope for arbitrariness follow. Even if such measures were spelled out fully, they would raise problems regarding the security of market-access commitments if they were in any way made contingent upon unforeseeable circumstances. And if this is not their intent, then it is unclear why the concept is necessary in the first instance.⁴ No Member inscribed an economic needs test in its schedule of commitments in the post-Uruguay Round negotiations on basic telecommunications and financial services, perhaps indicating a willingness on the part of Members to dispense with this instrument of trade restriction.

A more general point about clarity in schedules is that the product nomenclature used in GATS has severe shortcomings. First, many Members have used their own definitions of certain sectors, subsectors or particular services activities, sometimes with no reference to any established nomenclature. Second, the nomenclature developed by the Group of Negotiations on Services during the Uruguay Round (document MTN.GNS/W/120), which was relied upon by many Members to a significant degree, is too aggregated to provide adequate standardized definitions in many sectors. Work is underway under UN auspices to improve nomenclature. The development of an adequate system of nomenclature is an essential part of improving the clarity of scheduled commitments.

This section takes up three questions that help to determine the quality of liberalization commitments. The first concerns the relationship between foreign equity participation and the conditions of competition in the market. The second relates to the use of GATS schedules as a mechanism for pre-commitments to future liberalization, and the third relates to the gap between commitments and actual policies (ceiling bindings).

One aspect of the existing commitments which evokes concern is the emphasis on change of ownership rather than on the introduction of entry. For instance, if we look at the commitments on financial services, the pattern differs across regions: in Latin America, many Members have retained discretion on whether to allow entry, but few have imposed limits on foreign equity participation;⁵ in Asia, the two types of limitations are frequently encountered together. Apart from economic considerations, these differences in policy reflect differences in political attitudes to foreign direct investment, and varying degrees of concern about the prospect of foreign ownership and control in financial services.

A multilateral commitment by a government to allow entry influences the degree to which markets are contestable. Regardless of the existing market structure, established suppliers in the market are likely to behave more competitively if there are no legal barriers to entry. Increased competition brings benefits both through promoting allocative efficiency, i.e. pricing close to costs, and internal efficiency, producing at least cost. Conversely, privately efficient profit-seeking behind protective barriers cannot be expected to lead to socially efficient results. Restrictions on entry benefit producers at the expense of consumers. The earnings of producers are then greater than the social productivity of the inputs because there is a component which is a transfer from consumers. It is therefore desirable for the scope of competitive forces to be enhanced by the effective removal of barriers to entry.

In light of the emphasis in the GATS negotiations upon increasing permitted (or maintaining existing) levels of foreign equity participation, it is interesting to consider the implications of a situation in which foreign participation has been permitted, without an increase in the degree of competition allowed to occur in the market. In other words, what are the welfare consequences of foreign ownership without adequate competition? Foreign investment clearly brings benefits even in situations where it does not lead to enhanced competition. First, allowing foreign equity participation may relax a capital constraint which could otherwise result in socially suboptimal levels of investment in the sector. Furthermore, the benefits of increased investment in helping to recapitalize troubled financial institutions in many developing countries cannot be underestimated. In fact, one reason why countries may have chosen this particular combination of policies, i.e. to restrict new entry while allowing foreign equity participation, is probably because they would like new foreign capital to help strengthen weak domestic financial institutions rather than to come in the form of highly competitive new banks and insurance companies which might drive their domestic rivals out of business. Second, foreign equity participation may serve as a vehicle for transferring technology and know-how. The benefits come not only in the form of technological innovations, such as new methods of electronic banking, but also in terms of improved management and credit assessment techniques, as well as higher standards of transparency and self-regulation.

Against these benefits, there may well be costs associated with foreign direct investment when competition is restricted. If foreign investment comes simply because the returns to investment are artificially raised by restrictions on competition, then the cost to the host country may exceed the benefits, because the returns to the investor will be greater than the true social productivity of the investment. The argument may be presented in an alternative form. Aggregate national welfare in a particular sector can be seen as the sum of consumers' surplus and national producers' profits (plus government revenue). In competitive markets, welfare is greatest because marginal social benefit is equated to marginal social cost. In imperfectly competitive markets, welfare is reduced because output is restricted to a level where marginal social benefit exceeds marginal social cost. Producers gain at the expense of consumers. Now if foreign participation enhances competition, welfare may increase, but if foreign participation takes place with limited change in competition, then there is a further reduction in national welfare because of the transfer of rents from national producers to foreign producers.⁶

In this context, it is important to consider a scheduling innovation which helped solve one of the central problems in the financial services negotiations. 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. In other cases domestic law became less restrictive than the binding, e.g. in the Philippines, where the law enacted in 1994 stipulates maximum foreign equity of 60% in banking but new entry is bound at 51%.

The three types of grandfathering provision, foreign equity-related, legal form-related and general, are to be found in the financial services schedules. It is evident that grandfathering was primarily an Asian phenomenon, prompted presumably by the introduction of more restrictive regimes pertaining to foreign equity and legal form than had prevailed when the foreign firms first entered. The grandfathering 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. Furthermore, they may even place new entrants at a competitive disadvantage where differences in ownership and legal form affect firm performance.

It would seem desirable in future negotiations to put a greater emphasis on the introduction of competition rather than maintaining or creating foreign ownership. One way in which this could be done is by creating a stronger presumption against restrictions on the number of suppliers even when they are accompanied by a relaxation of restrictions on the number of foreign suppliers. But it is not clear whether any rule could prevent countries from negotiating patterns of liberalization which are mutually acceptable but not economically optimal.

What is the potential role of the WTO as a vehicle for promoting future liberalization? One of the reasons why governments are unwilling to liberalize immediately can be seen as a variant of the traditional "infant industry" or "infant regulation" arguments. The first is based on considerations of potential comparative advantage, whereby currently disadvantaged national suppliers, if provided with protected markets, are expected to learn-by-doing and eventually become internationally competitive. The failure of these policies in the past, and the innumerable examples of perpetual infancy, may well be attributed to the inability of a government to threaten credibly to liberalize at some future date - either because it has a stake in the national firm's continued operation, or because it is vulnerable to pressure from interest groups which benefit from protection.⁷

The GATS offers a valuable mechanism to overcome the difficulty of making credible commitments to liberalize. Commitments to provide market access and national treatment at a future date are binding under WTO law. Failure to honour these commitments would create an obligation to compensate those who are deprived of benefits. This need to compensate does in fact make the commitment more credible than a mere announcement of liberalizing intent in the national context.

Several governments have already taken advantage of this mechanism to strike a balance between, on the one hand, their reluctance immediately to unleash competition on protected national suppliers, and, on the other hand, their desire not to be held hostage to these suppliers in perpetuity. For instance in basic telecommunications, several governments made commitments to future liberalization, including a number of Asian and Latin American countries (see Low and Mattoo, 1998). In financial services there was less evidence of such commitments. This could be seen as a consequence of the uncertain economic climate in which the negotiations were concluded, but it could also be argued that precommitment to future liberalization would have contributed to creating stability. In future negotiations, it would seem desirable to encourage such precommitments where there are difficulties in immediate liberalization. In fact, it may be required of countries to demonstrate why it is not possible to make such commitments, at least on the basis of a pessimistic, delayed liberalization scenario - unless it is the intention to provide protection in perpetuity.

Many Members, especially developing countries, have undertaken specific commitments that define minimum access guarantees which are less than the status quo in terms of access permitted to foreign services and/or foreign service suppliers. Commitments of this kind are to be contrasted with those that define actual levels of access, and those that promise improved access levels in the future. In both financial services and basic telecommunications, there are examples of each of these levels of commitment (Kono et al., 1997; Low and Mattoo, 1998; Mattoo, 1998). In the sphere of trade in goods, Francois and Martin (1994) have used a model constructed on a probability distribution of policy outcomes to argue that ceiling bindings are likely to give rise to more liberalization than situations in which there are no bindings at all.

While ceiling bindings may have positive value, however, they clearly offer less guaranteed access or legal security than bindings that reflect actual or future additional levels of binding. There is an argument, therefore, for seeking ways of bringing bindings closer to the policy status quo. In practical terms, of course, the gap can widen again as markets become more open, unless an undertaking exists automatically to ratchet up scheduled commitments in the face of liberalizing policy changes. In discussions on the possible development of a safeguard mechanism in GATS, it has been suggested that the possibility of taking safeguard action might be linked to commitment levels that reflect the status quo, or entail a pre-commitment to future liberalization.

This section looks very briefly at what is arguably the single most important area in which GATS disciplines are in need of deepening - that of domestic regulation. However, we only seek to lay out an analytical framework that seems to us useful in addressing the issues involved. A considerable amount of further analysis is needed of these issues, particularly in light of the importance of adequate disciplines on domestic regulation as an accompaniment of liberalization.

Trade in services, far more than trade in goods, is affected by a variety of domestic regulations. A central task in the coming GATS negotiations will be to develop disciplines which ensure that such regulations support rather than impede trade liberalization. One basic discipline, the national treatment obligation, requires that regulations do not discriminate in any way against foreigners. However, trade can be inhibited even by regulations which do not discriminate, like certain standards and licensing requirements, and by the absence of pro-competitive regulations. While important initiatives have recently been taken to remedy these problems in the areas of accountancy and telecommunications, the general disciplines on domestic regulations in the General Agreement on Trade in Services remain weak. As we approach the next round of services negotiations, the question arises as to whether it is best to rely on further sectoral initiatives, or whether it is possible to adopt a more general approach.

The diversity of services sectors, and the difficulty in making certain policy-relevant generalisations, has tended to favour a sector-specific approach. However, it can be argued that even though services sectors differ greatly, the underlying economic and social reasons for regulatory intervention do not. And focusing on these reasons provides the basis for the creation of meaningful horizontal disciplines. Before elaborating on the suggested way forward, it is worth noting that such a generic approach is to be preferred to a sectoral approach for at least three reasons: it economises on negotiating effort, leads to the creation of disciplines for all services sectors rather than only the politically important ones, and reduces the likelihood of negotiations being captured by sectoral interest groups.

The basic argument can be summarised as follows (Gamberale and Mattoo, 1999). The economic case for regulation in all services sectors arises essentially from market failure attributable to three kinds of problem - natural monopoly or oligopoly, externalities, and asymmetric information. The social case for regulation, on the other hand, is based primarily on considerations of equity. Market failure due to natural monopoly or oligopoly may create trade problems because incumbents can impede access to markets in the absence of appropriate regulation. Because of its direct impact on trade, this is the only form of market failure which needs to be addressed directly by multilateral disciplines. The relevant GATS provision, Article VIII dealing with monopolies, is limited in scope. As a consequence, in the context of the telecom negotiations, the reference paper with its competition principles was developed in order to ensure that monopolistic suppliers would not undermine market access commitments. These principles should be generalised to a variety of other network services, including transport (terminals and infrastructure), environmental services (sewage) and energy services (distribution networks).

In all other cases of market failure, multilateral disciplines do not need to address the problem per se, but rather to ensure that domestic measures to deal with the problem do not serve unduly to restrict trade. The same is generally true for measures designed to achieve social objectives. Such trade-restrictive effects can arise from a variety of technical standards, prudential regulations, and qualification requirements in professional, financial and numerous other services, as well as from the granting of monopoly rights to complement universal service obligations in services like transport and telecommunications. The trade-inhibiting effect of this entire class of regulations is best disciplined by complementing the national treatment obligation with a generalization of the so-called "necessity" test. This test essentially leaves governments free to deal with economic and social problems provided that any measures taken are not more trade restrictive than necessary to achieve the relevant objective. Such a test already appears in the GATS, weakly in Article VI on domestic regulations and more effectively in Article XIV on general exceptions, and has been further developed in the accountancy negotiations. One of the shortcomings of the necessity test is that it has often been less than clear what criteria are to be used in determining whether a measure can be judged "necessary" in relation to alternative interventions (Mattoo and Subramanian, 1995). Criteria built explicitly on the notion of economic efficiency could serve to reduce uncertainty on this point, offering a sound and consistent methodology for choosing interventions to remedy market failure or to pursue non-economic objectives.

Part IV of the GATS carries the title Progressive Liberalization and the associated text talks of achieving "a progressively higher level of liberalization" through successive rounds of negotiation. As noted at the outset, many commentators have observed that the level of liberalization achieved or committed to is quite modest so far. The GATS is certainly designed to accommodate gradualism. Is there, nevertheless, potential for concerted or more coordinated approaches to liberalization, such that Members end up making more far-reaching commitments, both generally and at the sectoral level?

In the sphere of trade in goods, governments have sometimes agreed to a formula on the basis of which they cut tariffs across-the-board by a uniform amount. In the Tokyo Round (1973-79), a weighted formula was devised so that higher tariffs would be cut more deeply than lower ones. This approach had the effect of moving the liberalization process ahead on multiple fronts, although exceptions to the formula applied in certain sectors. Such an approach is easier to conceive where there is a single policy instrument - the tariff - to consider. In the sphere of services, the situation can be more complex. Many different instruments determine the prevailing degree of liberalization, and the lack of adequate nomenclature may complicate efforts to ensure uniformity of commitments among Members. But already in services, there are examples of concerted liberalization packages through which a number of Members have agreed to a uniform set of commitments. These are approaches that could be built upon.

Three different "models" of a concerted approach to liberalization suggest themselves from the experience accumulated so far in the GATS context. First, model schedules were developed in the maritime and basic telecommunications negotiations. The basic purpose of the model schedules was to identify a set of subsectors and commitments that would be assumed by all parties to the negotiations. The premise was that agreement on standardized commitments would secure a higher level of commitment overall than if Members devised their liberalization offers independently. A package would not only guarantee an acceptable degree of reciprocity, but also define areas for exclusion, and areas where differing degrees of liberalization were feasible. In maritime services, for example, there was agreement to exclude cabotage altogether from the picture. And the separation of bulk and liner shipping services permitted Members to offer more in the former area than they would have been willing to do without this separation. Similarly, it was essential in basic telecommunications to separate out international, domestic long-distance and local loop telephony services. In the event, the maritime negotiations did not come to closure, and nor did the draft model schedule in basic telecommunications prosper greatly among Members, but the approach clearly holds significant promise in future negotiations.

Second, in financial services, a formula approach was developed, although only a restricted number of (mostly developed) countries adopted the formula. The Understanding on Commitments in Financial Services specified the content of market-access commitments, and in addition contained provisions dealing with procurement, the treatment of new financial services, and a standstill commitment (Kono and Low, 1997). Two points are particularly significant about this initiative. First, by addressing such questions as a standstill, procurement rules, and the treatment of financial services, the Understanding took Members further than they would otherwise have been able to go in terms of the GATS framework of rules and the normal approach to liberalizations commitments. Second, the Members that accepted this approach were willing to do so on an MFN basis, thereby ensuring a higher level of openness without compromising the basic non-discriminatory structure of a multilateral agreement.

Third, the reference paper in basic telecommunications was a set of regulatory commitments entered in the additional commitments column (Article XVIII) of the schedules of those Members that accepted the disciplines. This instrument attracted many more signatories (over 50) than the Understanding in financial services, and was regarded by many as a sine qua non for rendering meaningful the market access and national treatment commitments contained in Members' schedules. Without effective regulation, it was feared that monopolistic market power in the telecommunications sector could be used to neutralize the competitive and market-opening effects of market access and national treatment commitments. The reference paper was about more than market access, in the sense that it entailed regulation, but it is a good example of a cooperative effort around a common set of commitments that led to more effective market opening than would have otherwise been possible. A similar effort was in the making in the maritime negotiations, where additional commitments were drawn up to safeguard access to and use of port services.

These experiences demonstrate different ways in which formulae approaches might be crafted. We would suggest an appropriate generalization of the above three models along the following lines. The model schedule approach is most suitable to situations where there may be ambiguity about sectoral definition and/or a degree of consensus between Members on the areas where liberalization is feasible. One area in which such an approach may be fruitful if with respect to the presence of natural persons: clear definitions could be agreed for the categories of skilled workers (currently listed as managers, executives and specialists) and a strong presumption could be created in favour of a certain threshold level of liberalization built into the model schedule. Then the burden would be on a Member to justify its refusal to concede the threshold level rather than for other Members to extract the minimum concessions through painful negotiations. The approach may also be useful in other sectors where segmentation facilitates the establishment of clear focal points on the acceptable level of liberalization. It would seem appropriate for Members with special interests in particular sectors to prepare such model schedules, and to secure their general acceptability by striking the necessary balance between their own ambitions and the feasibility of widespread liberalization.

The Financial Services Understanding approach is most suitable for areas where only a sub-set of Members are willing to develop deeper disciplines, but are, nevertheless, willing to extend the benefits on an MFN basis to all other Members. Whereas the model schedule reflects a situation in which all Members simultaneously accept certain commitments, the Understanding is more on the lines of a leader-follower model - though, given the nature of the negotiating process, the leaders would understandably be reluctant to reveal their identity too early. This approach (or outcome) would obviously be feasible where participating Members are satisfied by the acceptance of disciplines by a critical mass of Members rather than all of them. The approach has the advantages of the Tokyo Round Codes without their disadvantages: it would facilitate effective and quick negotiations among like-minded Members, and create a positive demonstration effect on initially non-participating Members, without creating a discriminatory arrangement. Thus, the acceptance of disciplines in more sectors on government procurement, the treatment of new services, and a standstill on new restrictive measures are among the possible candidates for such an approach.

Finally, as has been argued in Section IV, it would make sense to generalise the basic telecommunication reference paper model to other network-based industries, where pro-competitive regulations are called for at the sectoral level in order to underwrite effective market access.

Much discussion has taken place of about the difference between a positive and negative list approach to identifying specific commitments. A positive list approach is one where parties to an agreement specify which sectors are covered. A negative list approach, by contrast, requires that parties specify the sectors that are not covered by commitments. The GATS uses a positive list approach to identify sectoral coverage and then a negative list approach to indicate limitations to market access and national treatment commitments in respect of sectors listed in schedules. NAFTA, by contrast, relies on a negative list approach. Sauvé (1998) has argued strongly in favour of a negative list approach, which has also been advocated by Snape (1998).

The issue can be considered at two levels. One is somewhat trivial, in that it concerns little more than the choice between saying what is to be done and what is not to be done. In these terms, the case for a negative list turns on three arguments (Low, 1997). First, a negative list approach will foster greater transparency, as it will be immediately obvious which sectors or activities are excluded from coverage. On the other hand, exclusions can be identified by deduction under a positive list approach, and arguments about the need for transparency might be better addressed more directly through appropriate transparency provisions. Second, a negative list approach may generate a greater pro-liberalization dynamic, as governments might be embarrassed by long lists of exceptions. It is not obvious, however, why governments would be more embarrassed by long negative lists than by short positive ones. Finally, a negative list approach would imply that any new services developed as a result of innovation or technological advancement, or for any other reason, would automatically be subject to established disciplines. This is potentially a strong liberalizing argument for a negative list approach, but may also be the one that makes governments cautious about adopting this approach.

A more fundamental interpretation of a negative list approach is that, unlike a positive list, it would entail an across-the-board binding with respect to all services activities. The negative listing would then apply only with respect to specifically identifiable non-conforming measures in particular sectors. The liberalizing consequences of such an approach would obviously be more far-reaching than what exists today under GATS. But as soon as a negative listing permits the exclusion of sectors, the difference between the two approaches becomes less interesting (as suggested above). We would argue that Members are simply not ready to make commitments in all services sectors, and that even if they did, they would be tempted to specify heavy-handed restricting measures in their negative lists that would take the substance out of commitments in sectors that they regarded as sensitive. Instead of arguing for what would amount to a significant structural change in GATS, therefore, we conclude that the preferable approach for the time being would be to emphasize reductions in limitations currently inscribed in schedules - especially with respect to the "unbound" entries - and at the same time to press for widening the scope of sectoral coverage in existing schedules.

A basic conclusion of our paper is that it is possible to make significant improvements in the GATS, and to make it a much more effective instrument of liberalization, without fundamental structural changes, which are, in any case, of doubtful political feasibility. This paper has identified four broad themes: improving the clarity of the Agreement, using the existing structure to generate more effective liberalization, deepening the disciplines on domestic regulations, and improving the negotiating dynamic. To an extent, our proposals build on the existing commitment to successive rounds of negotiations aimed at "achieving a progressively higher level of liberalization" (Article XIX:1 of GATS), and on the existing negotiating mandates to develop new rules. However, we believe that there is also need for an explicit mandate for work on improving the clarity of the Agreement. Some of the ambiguities in the GATS are quite fundamental, turning on questions of interpretation that make a significant difference to the nature of Members' obligations and the extent of market access and rule-based certainty delivered by the Agreement. The main suggestion made in the paper are summarized below.

• Clarify the relationship between market access and national treatment in order to specify precisely the scope of existing and future national treatment commitments. At present, credible alternative interpretations carry starkly different implications.
• Clarify the relationship among modes of supply with respect to commitments on a given services activity. Is a service delivered under different modes considered a "like" service regardless of the modal distinctions made in schedules? If not, as the structure of the schedules suggests, then the rights of Members should be clarified with respect to policy interventions in one mode that may undermine the value of a commitment in another.
• Confirm that the principle of technological neutrality applies within modes. In other words, within a mode of supply, a service is to be regarded as "like" independently of the means by which it is delivered. It should also be confirmed that no scope exists under Article XVI for scheduling restrictions based on the manner of conveyance of a service within the same mode. This would confirm, for instance, that existing commitments cover electronic delivery of services.
• In view of the clear separation that the GATS seeks to establish between market access and national treatment on the one hand, and domestic regulation on the other, it is clear that meaningful progress in strengthening the domestic regulation provisions of GATS Article VI only makes sense where specific market-access commitments have been undertaken. Otherwise, the value of good regulation can simple be nullified by restrictions on access to the market. For this reason, work on domestic regulation and improving the quality of liberalization commitments should proceed in tandem.
• Eliminate extraneous descriptive material from the schedules of commitments, to prevent interpretative confusion regarding the true scope of bound market access and national treatment commitments. Only commitments should be inscribed in schedules and Members should consider whether it would be useful to supplement existing transparency provisions with an additional mechanism of a non-binding nature for information that will contribute to a better understanding of commitments.
• Accord priority to the development of a detailed nomenclature which should be generally applicable. The absence of an adequate nomenclature works against transparency and the enforceability of commitments.

• Attention should be focused on the manner in which market access liberalization is defined. In particular, emphasis should be focused upon ensuring foreign market entry to enhance contestability and improve the conditions of competition in the market. Merely increasing foreign equity participation in existing enterprises is less likely to achieve these beneficial effects and could lead to losses in national welfare in imperfectly competitive markets.
• Members should use services negotiations more to make phased-in commitments to liberalization, as a number of countries did in the basic telecommunications negotiations. Pre-commitment is a valuable instrument for planning future market-opening in a credible manner and guaranteeing adequate time to ensure that the necessary conditions are in place when additional competition is introduced into the market.

• Meaningful liberalization requires that the provisions on domestic regulations are strengthened. The basic approach should be horizontal in nature, to take advantage of economies of scale in rule-making and lessen the risk of regulatory capture. Sector-specific regulatory provisions may sometimes be necessary to supplement the horizontal approach. Horizontally-based regulation implies generalization of existing initiatives, depending on the source of market failure that the regulatory intervention is designed to address. Where a problem arises from monopolistic or oligopolistic control over essential facilities, the approach should be to develop regulatory principles along the lines of those negotiated in the basic telecommunications sector. Where other market failures are present, the "necessity" test should be applied, on the basis of criteria built explicitly on economic efficiency criteria

• Different formulae have been used in sectoral liberalization negotiations so far, and these approaches should be built upon. A model schedule can be helpful where ambiguity over sectoral definitions coexists with potential consensus on the degree of liberalization that is achievable. Providing collective focal points (that function like a negative list approach) is likely to lead to more liberalization. Understandings such as those developed in the financial services sector make sense where some Members are willing to undertake deeper commitments or accept stricter rules, but need to develop a common understanding as to the content of such liberalization or rules. Such initiatives will unambiguously raise the level of liberalization, provided they are built around the MFN principle. The idea of developing standardized sets of additional commitments, such as the pro-competitive regulations in the basic telecommunications reference paper, has significant appeal and may find a place in negotiations in other network-based industries, as discussed above.

* The authors work, respectively, for the World Bank and the WTO Secretariat. The views expressed in this paper are their own and should not be attributed to the institutions with which they are associated.

1 The six permitted restrictions encompass limitations of the number of service suppliers, the value of transactions or assets, the number of operations or total quantity of output, the number of natural persons that may be employed, the nature of legal entities permitted to supply services, and the extent of participation of foreign equity in an enterprise.

2 For a full analysis of these issues, see Mattoo (1997).

3 For a non-violation complaint to prosper, it would need to be shown that the source of the difficulty could not have reasonably been anticipated at the time the commitment was entered into. This ay prove a high standard t meet in the situations described above.

4 The question whether market access should be made contingent on unpredictable circumstances is under consideration in the ongoing negotiations on a possible emergency safeguard mechanism under the GATS. If such a mechanism were to be agreed, it would surely contain provisions against its careless or excessive use.

5 It is possible that the discretionary licensing in some Latin American countries could pertain to both entry and equity participation.

6 To some extent rent appropriation can, of course, be prevented by profit taxation or by holding competitive auctions of licenses or equity. The rents would then accrue either to the government or to existing national shareholders. But the static and dynamic inefficiencies consequent upon lack of competition would still exist. Creating discriminatory profit tax regimes would have negative incentive effects on new foreign investment, but such regimes are ruled out, of course, where commitments are undertaken to provide national treatment. Furthermore, while equity auctions may prevent net profit transfers abroad through new acquisitions, and license auctions achieve the same vis-a-vis new entrants, neither addresses appropriation by existing foreign share owners. In this context, the grandfathering commitments assume particular significance.

7 National firms often behave as if they prefer to operate as high cost, poor quality producers in protected markets than as low cost/high quality producers facing international competition. This may be because of the profitability of protection, or the greater utility that managers and workers derive from operating in sheltered environments. In any case, when the Government cannot credibly threaten to liberalize, then national firms may have an incentive to precommit to high costs or poor quality, in an environment of slow learning and under investment in research and development. Such behaviour by the firm, either for strategic reasons or on account of inertia, forces Governments to prolong socially costly protection. See Staiger and Tabellini (1987) for a variant of this argument.