2. Commonalities and differences between PTAs and PAs

Deadlock in the Doha Development Agenda (DDA) has been accompanied by steadily intensifying efforts to negotiate new PTAs by WTO Members. The pursuit of mega-regional agreements by the US and other major players,Footnote ² and the launch of negotiations on a trade in services agreement (TiSA) by a group of WTO Members are important illustrations.Footnote ³ Many PTAs that have been negotiated since the WTO entered into force go beyond the WTO in terms of coverage of policy areas (Horn et al., Reference Horn, Mavroidis and Sapir2010). The shift towards PTAs reflects both the heterogeneity of the WTO membership and a desire by high-income nations to address spillovers created by policies that are not on the table in the DDA. There are important potential down-side effects of the proliferation of trade initiatives outside the WTO. Most obviously, it may lead to the balkanization of trade, especially insofar as firms are induced to invest in and source from within a bloc. Greater recourse to PAs under auspices of the WTO can help attenuate or avoid such outcomes and support more multilateral cooperation (multilateralization) over time.

PAs and PTAs are similar in that both can be, and mostly are, applied on a non-MFN basis.Footnote ⁴ Thus, both are mechanisms through which signatories can prevent free riding by non-participants. Another common element is that both instruments involve binding commitments that are enforceable. Neither Art. XXIV GATT nor Art. V GATS nor the various WTO provisions regulating PAs discuss this point explicitly, but if PTAs or PAs are not legally enforceable, there is little point in negotiating them. Yet another commonality is that neither is a very effective instrument to address the major market access issues that have been a key source of the deadlock in the Doha round – i.e., the fact that at the end of the day large trading powers want more liberalization of agricultural trade policies and more non-agricultural market access from each other than they are willing to give. A situation where large players all want to see more market access concessions tabled can only partially be addressed by shifting the focus to PTAs, and cannot be addressed at all through PAs. As argued by numerous commentators, what is needed is to agree on a negotiating set that has enough to interest the major powers (see, e.g., Evenett, 2014; Hoekman, Reference Hoekman2014b, Wolfe, Reference Wolfe2015; Decreux and Fontagné, Reference Decreux and Fontagne2015).

PAs differ from PTAs in important respects. PTAs often will have as an objective the integration of the markets of the participating countries on an explicitly discriminatory basis – something that is recognized and accepted by all WTO Members as a legitimate endeavour. PTA membership will be limited to a set of countries that satisfy whatever criteria are established by the bloc. PTAs tend to be closed clubs – most PTAs do not include an accession clause. Those PTAs that do allow for accession often restrict this to countries in a specific geographic region. This helps explain the proliferation of PTAs – a new agreement tends to be negotiated between members of any given PTA and non-members. PAs in contrast are ‘open’ – in principle, no WTO member can be excluded from a PA once it has been negotiated and accepted as an Annex 4 agreement by the WTO membership.Footnote ⁵ Insofar as WTO members can satisfy whatever conditions apply for membership (i.e., conform to the disciplines that constitute the substantive provisions of the PA), they cannot be excluded. Indeed, a rationale for negotiating a PA in the first place is to encourage wider membership over time – that has certainly been the goal of the signatories of the GPA.

PAs also differ from PTAs in that they will be narrower in scope. A PTA will usually cover many policy areas, ranging from trade in goods and services to investment, IPRs, and development assistance, and other forms of ‘soft law’ cooperation. The condition imposed on WTO Members that PTAs cover substantially all trade (have substantial sectoral coverage in the case of services) is intended to ensure that a PTA is not used as a mechanism to engage in selective discrimination for just a few products or a specific sector. A PA in contrast may deal with just one issue. The chosen issue may have many dimensions and cover many types of activities, as is the case with procurement, but it need not. This could in theory lead to PAs being used to very selectively circumvent the WTO MFN rule – which is exactly what the WTO requirement that PTAs cover substantially all trade is intended to preclude. However, WTO members cannot use a PA to selectively violate existing WTO commitments, because PAs that do so will not be approved. Thus, PAs that limit benefits to signatories can only address matters that are not subject to existing WTO disciplines. If they deal with matters where the MFN rule applies, benefits must be extended to all WTO members, i.e. they must be critical mass agreements, such as the ITA.

Art. X.9 of the Agreement Establishing the WTO stipulates that the Ministerial Conference of the WTO may decide to add an agreement to the existing set of PAs ‘exclusively by consensus’. In contrast, with the advent of the Transparency Mechanism in 2006, there is no longer any effort by WTO Members to approve new PTAs (Mavroidis, Reference Mavroidis2011). Consistency with the WTO is instead left to be determined through dispute settlement: if a WTO member believes a PTA is inconsistent with the WTO, it can bring a claim to that effect and ask for a Panel to rule on the matter, a rare occurrence as practice to date reveals that the incentives to litigate are missing. The fact that there are no provisions or criteria on what is (should be) permitted in terms of sectors or their content/coverage implies that there is great flexibility in principle for those aspiring to establish a PA, but utilization of this flexibility is constrained by the need to obtain approval by all WTO Members to move forward, even if many or most do not intend to join.

PAs may deal with issues that are already subject to WTO disciplines or cover matters that are not included in the WTO. The two extant PAs offer an example of each possibility. The Civil Aircraft Agreement deepened disciplines on trade policy instruments affecting trade in aircraft – removing a set of tariffs on a MFN basis (as was required given that tariffs are covered by the WTO) and laying out more specific disciplines on the use of subsidies than prevailed at the time this PA was negotiated. The GPA addresses a policy that is explicitly excluded from the reach of Art. III GATT (national treatment) and the GATS (Art. XIII.1). The GPA does not undercut existing WTO commitments, as there are none. This may not be the case for PAs dealing with subjects that are already covered by the WTO but that go further. An example would be a PA on services trade. In such instances, a PA may be detrimental to non-participants if provisions are applied on a discriminatory basis, but as long as what is agreed does not violate existing specific commitments, the question non-participants should ask is what the outside option is for those that are proposing a PA and whether the outside option will be better or worse for them.

In the hypothetical case of a proposed services PA, a PTA is a feasible outside option. If, however, countries want to cooperate in a regulatory area such as competition or investment policy, a PTA is not a feasible alternative, but countries might pursue cooperation outside the WTO. In assessing the implications of a potential or proposed PA, it is therefore necessary to consider what the alternative options are for those proposing a PA. The outside options may in practice be worse for non-members than accepting a PA. This is because the PA route has benefits that outside options such as PTAs do not have – including greater transparency, access to and use of the DSU, and the possibility of eventual accession if and when countries decide to join at a future date.

Transparency of PAs is ensured through the process of notification to the General Council and the need for the Council to approve any PA that is brought forward. If approved, a PA will result in the establishment of the types of WTO bodies that assist Members in the implementation of agreements, such as a Committee, with regular (annual) reporting on activities to the Council, and documentation that is open to all WTO Members. Moreover, disputes under a PA will (must) be submitted to WTO Panels (and eventually the Appellate Body). This ensures that PA-related case law will develop harmoniously with case law pertaining to the multilateral WTO agreements. Thus, PAs ensure a much closer ‘connection’ with day-to-day WTO activities and processes, and greater coherence when it comes to case law/dispute settlement.

3. Arguments against PAs

The fact that PAs can be adopted into the WTO exclusively by consensus (after they have been negotiated, i.e., once their content has been established) and that participation is voluntary (whatever disciplines are negotiated only apply to signatories) would appear to offer substantial assurance to WTO Members that they have nothing to fear from efforts by some countries to negotiate PAs. Why are they such a sensitive issue then?

Objections raised against PAs include worries that they will be used for controversial issues, such as trade and labor standards; that they will erode the MFN principle; that the rules that are negotiated will be precedent setting; that asymmetric negotiating capacity will bias disciplines towards what powerful countries want; and that PAs will reduce the prospects for issue linkages needed to conclude deals on policies that are supported by strong vested interests.

The salience of many of these concerns depends on the substance and coverage of a PA. If the PA is WTO-X (that is, deals with a new issue),Footnote ⁶ it may be precedent setting but there is no issue of fragmentation or undercutting MFN as this currently does not apply. Countries that stay out may object to the PA precisely because it is precedent setting, and they fear the disciplines eventually being ‘imposed’ on them, as was done in the Uruguay Round as a result of the creation of the WTO and the take it or leave it choice that this created (see, e.g., Hoekman and Kostecki, Reference Hoekman and Kostecki2009). However, a repeat of this scenario seems very unlikely. If countries do not believe the rules of the game of a PA are in their interest, they can stay out. Such a sanguine assessment also applies in the case of so-called WTO+PAs – i.e., agreements that deepen existing rules. WTO+PAs that undercut existing WTO disciplines can, and presumably will, be rejected by the membership, and can also be contested through the DSU. WTO+PAs will be subject to the MFN rule, i.e., in principle the PAs will have to satisfy the critical mass constraint. In practice, such agreements may be discriminatory in effect, but this will be de facto and not de jure in that benefits may be conditional on satisfying certain technical preconditions – e.g., attainment of a certain minimum regulatory performance.Footnote ⁷

That said, PAs will define the rules of the game in a specific area for signatories. Wolfe (Reference Wolfe2009) notes that any PA will invariably include Organization of Economic Cooperation and Development (OECD) member countries that may already have achieved much of whatever level of cooperation-cum-discipline that is agreed for an issue, and that many non-OECD countries are not going to have the capacity to participate in negotiations that will set a precedent. PAs are likely to reflect the interests and current practices of the initial signatories, which may not be appropriate for all countries. Capacity constraints and resulting non-participation by developing countries in a negotiation makes it less likely that an agreement will address issues that are of primary concern to low-income economies. Clubs will define the rules of the game in an area that will be difficult to change subsequently if and when initial non-signatories decide to participate. Experience illustrates that it is very difficult to amend (re-negotiate) disciplines, so that a plurilateral approach may well become analogous to the Acquis Communautaire for prospective members of the European Union – i.e., non-negotiable.

Thus, even if countries opt out, in practice the advantage may be in favor of OECD countries and major emerging markets that have the capacity to engage effectively on the substance of proposed rules. However, the presumption that PAs will define the rules of the game forever is not necessarily true. If large countries initially decide to stay out but subsequently become willing to join but have serious concerns regarding specific provisions, they can seek to renegotiate these. Small countries will be ‘price takers’ but that is the case in any event – that is not an issue that is specific to PAs. The same considerations apply regarding fears that PAs will be dominated by large OECD countries. This need not be the case. PAs also offer a mechanism that a broad set of WTO Members could use to move forward in an area where one of the large WTO members is not willing or able to participate. A PA that centers on operationalizing 100% duty-free, quota-free access for LDCs is a possible example. This is something that is currently not feasible for the US to agree to, but that has already been implemented by many other countries. Greater cooperation through a PA on issues like simplification, cumulation, or harmonization of rules of origin among countries offering duty-free, quota-free access could enhance the benefits for LDCs.

More intensive use of PAs may result in a long-term bifurcation in the WTO Membership, splitting ‘insiders’ from ‘outsiders’. This was the pattern that emerged in the GATT years, with very few countries subsequently joining the Tokyo Round codes after their initial negotiation. If so, greater use of the PA approach would move the WTO towards a two-track regime with subsets of countries (clubs) playing in their own sandbox. Many developing countries have argued that this is contrary to the basic character of the WTO and conflicts with the consensus-based approach that has historically been the norm (Hoekman and Kostecki, Reference Hoekman and Kostecki2009).Footnote ⁸ However, this argument applies even more to PTAs, which may be designed with the strategic objective of excluding specific countries. If PA club members end up agreeing to disciplines that are unacceptable to countries that are not intending to join, the question is whether a PA would be worse from a global welfare/systemic perspective than if countries concluded a PTA or embedded their proposed cooperation in existing PTAs. A world in which there are many PTAs that deal differently with a specific subject area could well be worse for global welfare (efficiency) than one in which the issue is addressed through a PA.Footnote ⁹

PAs will impose additional costs on the rest of the WTO Membership by utilizing the WTO ‘infrastructure’ – including operation of a Committee, making use of the WTO facilities, potential invocation of the DSU, calling on the Secretariat for support, etc. The fact that the operation of a PA is centered in the WTO as opposed to occurring outside it is a positive feature, but this does come with additional direct costs, as well as potential opportunity costs given limited Secretariat resources. There is a straightforward solution to this problem: signatories can be required to provide additional contributions to the WTO in order to cover the costs of implementing and administering PAs. They would need to incur these costs in any event if the PTA route was chosen instead or if cooperation is pursued outside the WTO.

Moving down the PA track may imply that countries give up negotiating chips that could be used to obtain concessions in other areas in a multilateral negotiation. The fundamental premise underlying the Single Undertaking is that it permits issue linkage: country A can get something it wants by giving up something that country B wants and the trade may involve subjects that have nothing to do with each other. If PAs are negotiated for specific issues, the scope for such linkage may decline. Much depends here on the subject matter of a potential PA and on contracting costs.Footnote ¹⁰ If it does not offer much in the way of negotiating leverage for the countries that are involved – i.e., nobody is inclined to ‘pay’ much if anything for a deal on the subject covered by a proposed PA – the ‘linkage downside’ will be small. The absence of linkage potential might, under some circumstances, act as an incentive to join the PA in the first place by lowering the opportunity cost of participation.Footnote ¹¹

Summing up, as argued at greater length in Hoekman and Mavroidis (Reference Hoekman and Mavroidis2013), there are good reasons for privileging PAs over PTAs.

• • PAs are Pareto-sanctioned because their content is approved by the WTO Membership. PTAs are reviewed by the WTO, but there is no sanctioning of their content; the process is limited to supply of information (Mavroidis Reference Mavroidis2013).

• • Information provision stops at the moment a PTA has been reviewed. There is no obligation to continue to supply information to the WTO after that. In contrast, PAs involve regular reporting on activities to the WTO Membership as a whole.

• • The plethora of PTAs results in significant dispersion in rules and approaches and thus transactions costs and trade diversion – Bhagwati's famous spaghetti bowl analogy. Multiple PTAs deal with the same subject-matter, so that rules of the game for firms from a country often differ for the same issue depending on the PTA that applies for a given trade flow. In the case of a PA, there will only be one regime regulating a given subject matter.

• • PTAs are mostly closed to accession by new members. PAs in contrast are required to be open.

• • PTAs often have their own dispute settlement procedures. Dispute settlement is the most frequent form of ‘contract completion’, and as a result the WTO misses out on important information. In contrast, PAs will use WTO dispute settlement mechanisms.

4. Supporting greater use of plurilateral agreements in the WTO

For the reasons discussed above and in light of the heterogeneity of the WTO Membership, we take the view that it would be beneficial to allow for more ‘variable geometry’ to be pursued under the umbrella of the WTO. This is preferable to a situation where countries are pushed into ever-greater reliance on PTAs − which effectively escape multilateral disciplines − or are induced to engage in cooperation outside the WTO – as was the case in the negotiations on an Anti-Counterfeiting Trade Agreement (ACTA).

There are two important factors that impede the feasibility of pursuing plurilaterals under WTO auspices. The first of these was mentioned in the Introduction: there is no straightforward way for WTO Members to pursue WTO+ critical mass agreements (CMAs) that involve deepening of disciplines on policies that are already subject to WTO rules and that they are willing to apply to all WTO Members on a MFN basis. The second is that incorporation of a PA into the WTO requires unanimity (‘exclusively by consensus’), which in practice is a major disincentive for countries to pursue this type of cooperation.

Which path to take – CMA or PA – will depend on whether governments interested in cooperating in a given area are worried about free riding and whether negotiated disciplines can be applied on an MFN basis. For some issues, there may be no free rider concern. A recent example is the Bali Trade Facilitation Agreement (TFA), which involves measures that every WTO member will benefit from implementing. Moreover, implementation will be on a MFN basis by construction: changes in Customs procedures will apply to all trade, not just to signatories of the TFA. At the time of writing, adoption of the TFA Protocol has been blocked by India. Assuming this continues to be the case and that the majority of the WTO membership wants to implement the TFA, the question is how it might be pursued on a plurilateral basis inside the WTO. The PA track is not available, as presumably India would block it – after all, why accept the TFA as a PA if you do not accept the Protocol? There is no obvious way of making the TFA a CMA either. WTO Members could in theory try to use Art. XXVIII GATT (Modification of Schedules) to incorporate the substantive elements of the TFA into their GATT schedules. However, this approach will not allow inclusion of the special and differential treatment elements of the TFA, which are a core part of the agreement (see Hoekman, Reference Hoekman2014a). This suggests that for the CMA path to be used to deepen disciplines in areas that are already covered by the WTO – as is the case for the TFA – the modalities of using GATT Art. XXVIII need to be revisited. There is currently a disconnect in this regard between the GATS and the other WTO agreements − as Art. XVIII GATS makes explicit, provision for WTO Members to make ‘additional commitments’ that are applied on a non-discriminatory basis.Footnote ¹²

With respect to the consensus requirement for new Annex 4 PAs, some are of the view that no change in the consensus rule is needed and that non-members should be comfortable with the terms of any PA that is tabled (Lawrence, Reference Lawrence2006). Others argue that greater use of PAs is conditional on a relaxation of the consensus constraint (Tijmes-Lhl, Reference Tijmes-Lhl2009). Discussions in the TISA context and the July 2014 veto by India on adoption of the protocol needed to incorporate the Bali Trade Facilitation Agreement into the WTO suggest that consensus is likely to be a binding constraint on greater use of PAs. The incentives to seek side-payments and attempt to link approval to issues that are not related to the subject matter of the PA are likely to make it too difficult to add new PAs to the WTO.

Maintaining the consensus rule is arguably too strong a constraint. Consensus is not needed to ensure that PAs that raise legitimate concerns for non-participants can be blocked. Relaxing the consensus requirement – for example through agreement that ‘substantial coverage’ of world trade or production is sufficient (Hufbauer and Schott, Reference Hufbauer and Schott2012)Footnote ¹³ or acceptance that a two-thirds majority suffices – would still ensure that proposed PAs can be rejected while removing the ability of just a few countries to block a PA that the majority of the WTO Membership finds acceptable. There is however very strong opposition to moving away from consensus, not just from developing countries but also from countries like the US which fear being outvoted. In practice, it may not be necessary to abandon consensus if the WTO can agree in principle that more should be done to allow plurilateral cooperation to occur under WTO auspices, and Members work together to define a code of conduct as regards new PAs.

We agree with Lawrence (Reference Lawrence2006), WEF (2010), Draper and Dube (Reference Draper and Dube2013), and others that a necessary condition for greater use of PAs is to address the concerns that have been expressed by WTO Members. What is needed are clear(er) ex ante rules that ensure that PAs are not vehicles for some countries to escape their general or specific WTO obligations, and that the interests of small/poor countries are protected. One way of doing this is to launch an effort to agree on a set of principles for new PAs to be negotiated under the umbrella of the WTO. Such principles could include the following:

1. (i) membership of PAs is voluntary and will always remain so;

2. (ii) the subject of any PA must be clearly trade-related in the sense of involving a policy that has a direct impact on trade costs and incentives;

3. (iii) the issue under negotiation should enjoy substantial support from the WTO's membership;

4. (iv) any WTO Member may seek to join an existing PA on the same conditions that applied to original signatories;

5. (v) WTO members opposing a PA commit to explain their substantive reasons for their opposition, and to refrain from ‘tactical’ opposition to extract promises (side-payments) in other areas;

6. (vi) only parties to a PA can participate in WTO dispute settlement, and retaliation in the case of non-implementation of dispute settlement recommendations is limited to the policies covered by the PA (as is the case now under the GPA);

7. (vii) all PAs must include mechanisms to provide assistance to countries that are not in a position to implement the provisions of the agreements but want to do so;Footnote ¹⁴ and

8. (viii) all PAs include transparency mechanisms to ensure that non-participants have full information on what is being and what has been achieved.

These are just examples of provisions that could help address concerns that have been expressed regarding PAs – no doubt others will have additional suggestions. What is needed is to launch a process to define a set of such criteria.

Lawrence (Reference Lawrence2006) has suggested that one criterion for PAs should be that any new PA be open to all WTO Members at the negotiation stage – i.e., participation in the development of rules should not be limited to likely signatories. This is very important, as are commitments to make discussions transparent, as this will facilitate decision making by governments who initially might not join the negotiation. An ‘open to all’ norm may be problematical, however, in that it assumes that participation will be in good faith. But what happens if countries stonewall and seek to block progress on an issue agreed by a majority of participants'? WTO Members that have no intention to participate might behave ‘strategically’ and try to raise the costs for those that are eager to establish new rules. Some countries engaged in the TISA talks have indicated that they are not seeking to pull in countries that showed very limited interest in making substantial liberalization commitments in the course of the Doha negotiations. This suggests an ‘open access’ criterion for the initial negotiation phase of a PA should be tempered by recognition and acceptance that at some point a PA negotiation must be limited to those countries that are serious about making commitments on a specific matter.

As already discussed, open access, in the sense that once negotiated any PA allows for accession by any WTO Member, is not explicitly required in Art. X.9 WTO. Instead, accession provisions are defined in the individual PAs. It would be desirable to agree explicitly that ‘open access’ ex post be a precondition for approval of any PA. Other criteria might be considered as well, e.g., prohibiting incumbents from ratcheting up the entry price for latecomers and making this enforceable through binding arbitration if contested.

In practice, not all countries will be able to engage on an equal footing in the negotiation of a PA. There are major differences in capacities to engage on regulatory matters and the ability to participate in a fully informed way. Some governments may find it difficult to determine the ‘return’ to applying a proposed rule (e.g., the direct administrative costs or the size – and perhaps even the sign – of the net economic impact of implementing a proposed set of disciplines). LDCs are likely to be among the least able to engage in PA talks that focus on regulatory issues or matters that are not covered by the WTO. Whatever the subject of a PA, consideration could be given to extending whatever is negotiated amongst a club of WTO Members to all LDCs on a non-reciprocal basis. Of course, the value of such action will depend on the capacity of the LDCs to benefit from (make use of) whatever is agreed among the PA members. In practice, LDCs may not have the capacity to benefit, especially if a precondition is satisfying specific minimum standards.

This suggests that any PA should include an aid-for-trade component – mechanisms to assist the LDCs improve their standards, regulation, etc. to the level that is required to benefit from the PA. Such mechanisms will need to be tailored to address whatever the associated capacity-building needs are. One possibility would be to develop PA-specific ‘platforms’ that help LDCs, as well as other developing countries with an interest in acceding to the PA, to undertake diagnostic analysis, identify action plans, and implement needed reforms with funding and assistance from high-income PA signatories.Footnote ¹⁵ Including an operational aid for trade dimension in PAs could enhance the relevance of PAs for LDCs and other low-income countries and give them a development dimension.Footnote ¹⁶