ORIGINAL PAPER

Explaining rising regionalism and failing multilateralism:consensus decision-making and expandingWTO membership

Euan MacMillan

# Springer-Verlag Berlin Heidelberg 2014

Abstract The beleaguered progress of the Doha Development Agenda of the WorldTrade Organisation (WTO) presents something of a puzzle for economic theory: ifmultilateralism is an effective forum for liberalisation (as it has been in the past), thenwhy have the current round of talks faltered amid the proliferation of preferential tradeagreements (PTAs)? This paper builds a ‘hub and spoke’ version of the Maggi (AmEcon Rev 89(1):190–213, 1999) model of trade negotiations to shows that the combi-nation of the WTO single-undertaking and consensus decision-making principles withan expanded and more diverse membership can render multilateralism less desirable forhub countries than bilateralism. It is argued that these principles give spoke countries defacto veto power meaning that their threat point duringWTO negotiations is a reversionto PTA negotiations between all parties. Accordingly, spoke countries with relativelyless to gain from the WTO can use their veto power to extract gains from those thatwould benefit substantially. If an expanding WTO membership has increased thenumber of such countries, then the benefits of multilateralism versus regionalism fromthe perspective of hub counties may have been diminished to such an extent that theyare no longer willing to wait for the conclusion of the Doha round before engaging inPTA negotiations.

Keywords Trade negotiations .WTO . International institutions .Multilateralism .

Regionalism

Int Econ Econ PolicyDOI 10.1007/s10368-013-0263-6

Acknowledgments This paper stems from PhD work undertaken at the University of Strathclyde under thesupervision of Professor Ian Wooton. Economic and Social Research Council funding is gratefullyacknowledged. Earlier drafts benefitted from comments from Professors Ron Davies, James Markusen, andAlan Woodland, Dr. Mark Melatos, and an anonymous reviewer. Responsibility for remaining errors liessolely with the author. The views reflected herein are those of the author and do not necessarily reflect those ofthe Department for International Development.

E. MacMillan (*)Department for International Development, UK Government, London, UKe-mail: euan.macmillan@gmail.com

1 Introduction

The beleaguered progress of the Doha Development Agenda (DDA) of the WorldTrade Organisation (WTO) presents something of a puzzle for economic theory: ifmultilateral trade negotiations are an effective forum for liberalisation (as has been thecase historically), then why have the current round of talks struggled amid theproliferation of preferential trade agreements (PTAs)? Numerous authors have positedthat this recent phenomenon may in part be precipitated by the combination of thespecific decision-making process of the WTO and the fact that the current membershipof the WTO is larger and more diverse than it was during previous rounds. 1 Twofeatures of WTO jurisprudence underpin this diagnosis: the requirement that WTOdecision-making be formed by consensus; and the single-undertaking principle. Thecombination of these two elements means that all aspects of a WTO negotiating roundmust be acceptable to all WTO members before the round can be concluded. Accord-ingly, it is argued that reaching unanimous decisions amongst an expanded and morediverse membership will inevitably be frustrated by coordination failure. However, it isthe contention of this paper that the combination of the single-undertaking and con-sensus decision-making principles with an expanded and more diverse membership canlead to more than just coordination failure; it can render multilateral negotiations lessdesirable for some parties than bilateral negotiations.

This contention stems from an extension of the Maggi (1999) optimal tariff drivenprisoners’ dilemma model of trade negotiations to a ‘hub and spoke’ setting in whichone large hub country must negotiate with multiple smaller spoke countries. In theMaggi (1999) model, which assumes symmetry, all parties prefer multilateral (i.e.WTO) bargaining to bilateral (i.e. PTA) bargaining. The difference being that multilat-eral bargaining, as facilitated by the single-undertaking and consensus decision-makingprinciples, maximises a single global Nash bargaining product (NBP), whereas bilateralbargaining maximises relationship-specific NBPs. In a hub and spoke setting, prefer-ences are opposed; the large country prefers bilateral bargaining, the smaller countriesprefer multilateral bargaining. Under these circumstances, multilateralism can onlypertain where it confers exogenous benefits, such as: access to common resources, aneffective dispute settlement mechanism, etc. This paper constructs a 2-period game inwhich the hub and spoke countries first decide whether to form PTAs or to endure theprisoners’ dilemma outcome for an additional period in order to engage in time-consuming, but beneficial, WTO negotiations. Successful conclusion of WTO negoti-ations confers global gains which must be distributed to all WTO members given themultilateral NBP. Two countervailing forces affect the hub country’s 1st perioddecision: the cost of waiting for WTO negotiations to conclude; and their share ofthe benefits accruing from WTO negotiations. It is shown that the second effect isdiminishing in the number of spoke countries that are WTO members. Hence, over-accession can occur, whereby the hub country does not gain enough frommultilateralismto make it worth the wait.

The driving factor underpinning this result is that the threat point of each countryduring multilateral bargaining is a reversion to bilateral bargaining between all parties.

1 For examples see: Jackson (1998), Schott and Watal (2000), Steinberg (2002), Sutherland et al. (2004) andEhlermann and Ehring (2005).

E. MacMillan

Accordingly, multilateral bargaining affords great power to countries that have less togain from multilateralism relative to bilateralism. Hence, it is possible that risingregionalism and failing multilateralism could be explained thus: the recent expansionof WTO membership to include numerous small countries, in combination with thenature of multilateral bargaining facilitated by the single-undertaking and consensusdecision-making principles, has eroded the benefit of the WTO relative to PTAs fromthe perspective of larger countries.2 This analysis adds credence to suggestions, madeby Sutherland et al. (2004) and Lawrence (2006) amongst others, that some form of‘variable geometry’ be considered within the WTO framework in which the single-undertaking and consensus decision-making principles are relaxed to allow a mixture ofmultilateral, plurilateral and bilateral negotiations to take place under the umbrella ofthe WTO.

The structure of this paper is as follows. Section 2 provides more detail on thedifference between multilateral and bilateral negotiations in the current context.Section 3 outlines the basic model. In section 4 multilateral and bilateralnegotiations are formally modelled. Section 5 utilises numerical simulations to assessthe welfare differences between the two negotiating forums when WTO membershipexpands. Section 6 provides a discussion of the implications of these results.

2 The difference between WTO and PTA negotiations

The dominant theoretical approach to the logic underpinning trade agreements,epitomised in the work of Bagwell and Staiger (1999, 2002), posits that trade agree-ments facilitate avenues of escape from terms-of-trade driven prisoners’ dilemmas thatarise when countries with market power engage in optimal tariff setting. Thus, tradeagreements represent the outcome of bargaining games between countries over self-enforcing tariff combinations. However, this approach does not in itself explain whycountries would choose to engage in multilateral trade liberalisation let alone why aninstitution such as the WTO is required to mediate the process. Accordingly, in order tohighlight the distinction between negotiating in the WTO and negotiating a web ofbilateral PTAs, it is necessary to provide explanations of these two issues.

2.1 Multilateral versus bilateral bargaining

The issue of the benefit of negotiating multilaterally rather than bilaterally has beenaddressed by Maggi (1999). Maggi developed a 3 country model, as illustrated in Fig. 1below, in which multilateral bargaining mitigates power imbalances between negotiatingparties and thereby leads to global welfare improvements.

In Fig. 1 preferences and endowments are such that each country is a net exporter tothe country on its right and a net importer from the country on its left. Given thissituation, each bilateral pairing contains an imbalance of power. For example, country a

2 Note that this argument abstracts from the Most Favoured Nation (MFN) principle of the WTO. Whilst, theMFN principle undoubtedly plays a pivotal role in determining the relative merits of multilateralism, this topichas been extensively tackled elsewhere. For examples see, Bagwell and Staiger (2002 and 2004), Ludema andMayda (2009), Saggi (2009) and Saggi and Sengul (2009). Horn and Mavroidis (2001) provide a usefulsummary of the economics of the MFN clause.

Explaining rising regionalism and failing multilateralism

can impose import taxes on country c without risk of commensurate retaliation and thusstands to lose less from a trade war. However, country a faces the opposite situationwhen it engages with country b.

Bilateral Nash bargaining would maximise three separate NBPs of the form:

NBPij ¼ uij − uNij� �

uji − uNji� �

ð1Þ

where: uij denotes the utility of country i with respect to country j and uijN denotes the

Nash equilibrium utility of country i with respect to country j.Conversely, (assuming symmetry) multilateral bargaining maximises a single NBP

given as:

NBPM ¼ uiM − uNiM� �3 ð2Þ

where: uiM is the total utility of country i from trade with both other countries and uiMN is

the total utility of country i at the Nash equilibrium.3 Note that the outside option foreach country under multilateral bargaining is a reversion to Nash equilibrium for allparties.

In order to highlight the difference between these two approaches it is useful toconsider the case of extreme power imbalances where, as first argued by Johnson(1953–54), one country in each pair will prefer the Nash equilibrium outcome to that ofbilateral reciprocal free trade. In this situation bilateral bargaining would not lead to freetrade. However, given that multilateral bargaining maximises global welfare, and

3 It is important to note that this approach to multilateral bargaining implicitly assumes that the WTOnegotiations are not constrained by the reciprocity rule as it is interpreted by Bagwell and Staiger (1999,2002), i.e. WTO negotiations do not necessarily maintain world prices. However, whether the reciprocity ruleas described by Bagwell and Staiger pertains in reality or not is currently an open question (WTO 2007).

b c

a

Fig. 1 Multilateral bargaining according to Maggi (1999, p. 196)

E. MacMillan

thereby balances power across all three countries, it would lead to global free trade.Thus bilateral bargaining is globally inefficient compared to multilateral bargaining.

By ensuring that all issues are treated as part of a single package that membercountries either unanimously accept as a whole or do not accept at all, the single-undertaking and consensus decision-making principles of the WTO can be thought ofas facilitating multilateral bargaining as described by Maggi (1999). However, it isimportant to recognise that the global welfare gains afforded by multilateralism in thiscontext only arise when there is a ‘balance of imbalances’, i.e. when there are localimbalances in power within a globally symmetric system. When one country is morepowerful than the rest, multilateral bargaining would benefit the weaker parties at theexpense of the strong party. Hence, the fact that hegemonic countries like the USA havehistorically chosen to fully participate in the WTO system suggests that it yieldsbenefits over and above those espoused by Maggi (1999).

2.2 The benefit of negotiating via an institution

An extensive literature exists regarding the myriad ways in which a central institutioncan mitigate the transaction costs associated with negotiating international trade agree-ments. For example, as noted by authors such as Koremenos et al. (2001) and Abbotand Snidal (1998), a central institution can provide common support services such astranslators, the provision of background data, and technical assistance, etc. An institu-tion can also lessen the chance of coordination failure amongst negotiating parties byorganising and overseeing negotiations according to some pre-established rules (Fearon1998). Moreover, as argued by Maggi (1999) among others, ceding control to a thirdparty mediator can more efficiently settle trade disputes in the presence of imperfectinformation. Indeed the Dispute Settlement Mechanism (DSM) of the WTO,which serves this purpose, is seen as one of the central pillars of the WTO system(WTO 2007).4

Therefore, it may be the case that for large countries, the benefits from negotiatingvia a central institution outweigh the cost of multilateral bargaining, as opposed tobilateral bargaining, with respect to less powerful countries. The fact that the originalcontracting parties to the GATT welcomed the accession of smaller countries through-out its near 50 year history suggests that this is the case. However, the recentproliferation of PTAs coupled with deadlocked negotiations in the WTO raises thequestion of whether the stream of developing country accessions that has taken place inrecent years has reversed this outcome. It is to this question that the analysis now turns.

3 Model setup

3.1 Basic setup

The model is an adapted version of the Kennan and Riezman (1990) n×m endowmentmodel with two periods. The current model consists of N+1 countries, indexed by an i

4 See Jackson (1998), Hoekman and Kostecki (2001) and Bagwell and Staiger (2002) for further discussion ofthe role of the DSM.

Explaining rising regionalism and failing multilateralism

subscript. One country, country a, is a large ‘hub’ which trades with N smaller ‘spoke’countries, n of which are WTO members, indexed by a j subscript, and N−n of whichare non-members of the WTO, indexed by a k subscript (country a is itself a WTOmember). It is assumed that the pattern of trade is such that there is no inter-spoke trade.Whilst this is a restrictive assumption, it captures the extreme case of the situation thatis of interest here: when one country is more concerned about the completion of amultilateral trade agreement than the others. Allowing trade between the spoke coun-tries would moderate the results of this model quantitatively but not qualitatively andwould considerably increase the complexity of the following analysis.

In keeping with Maggi (1999), all bilateral trading relationships are assumed to beseparable such that trade policies in one relationship do not affect trade flows in otherrelationships, i.e. trade agreements do not cause trade diversion. This assumptionfocuses the analysis on the impact of multilateral versus bilateral bargaining bybypassing the possibility of the formation of strategic customs unions (CUs), which,whilst being relevant features of reality, are not of particular interest to the currentpaper. Consequently, as illustrated in Fig. 2 below, country a imports one unique goodfrom each spoke country (mJA in the case of country j and mKA in the case of country k)in exchange for one unique good (mAJ in the case of country j and mAK in the case ofcountry k). There are thus 2N goods, indexed by an l superscript.

All countries share the following utility function:

ui ¼X2Nl¼1

β illogcil ð3Þ

where: βil denotes country i’s preference for good l, and cil denotes country i’sconsumption of good l.

Fig. 2 Trade pattern between country a and its N trading partners

E. MacMillan

Each country is endowed with Eil of each good. Given the assumed pattern of trade,the endowment and taste parameters of the spoke countries equal 0 for all goods exceptthose that they exchange with country a. Furthermore, for simplicity, it is assumed thatall countries have equal preferences over the remaining goods, such that:

βal ¼1

2N; β jAJ ¼ β jJA ¼ βkAK ¼ βkKA ¼ 1

2:

The world endowment of each good is normalised to one such that the worlddistribution of endowments can be summarized as:

EaJA ¼ 1−EjJA; EjAJ ¼ 1−EaAJ ; EaKA ¼ 1−EkKA; EkAK ¼ 1−EaAK :

One further simplifying assumption is used to focus the analysis and to reduce thesize of expressions: the endowment combination between country a and country j isassumed to be the same as that between country a and country k. Thus:

EaAJ ¼ EaAK ¼ Ea; EjJA ¼ EkKA ¼ EN :

The specific utility function of each country is thus given as:

uaT ¼ n2N

log caAJ þ log caJAð Þ þ N−n2N

log caAK þ log caKAð Þ ð4Þ

uj ¼ 1

2log cjJA þ log cjAJ� � ð5Þ

uk ¼ 1

2log ckKA þ log ckAKð Þ ð6Þ

where uaT denotes the total utility of country a from trade with all N of its tradingpartners.

The following analysis is conducted for trade between country a and country j,however, at this stage, the results are analogous for trade between country a andcountry k.

The assumed preference structure implies that the autarchic relative price of goodmAJ, in terms of good mJA in each country is given as piAJ =PiAJ /PiJA=EiJA /EiAJ.The assumption that country a has a comparative advantage in good mAJ requires thatpaAJ < pjAJ, which implies that (1−EN)/Ea<EN /(1−Ea). Accordingly, the range of pos-sible endowment combinations is given as:

1 < EaAJ þ EjJA < 2 ð7ÞEach country levies a tariff at the rate τil on its import good. Note that, for

convenience, it is useful to write this as τil=Til−1 and to use Til=τil +1 when expressingtariffs. 5 Given that bilateral relationships are separable, country a sets tariffs with

5 Note that this implies Til = 1 under free trade.

Explaining rising regionalism and failing multilateralism

respect to country j to maximise utility subject to the following relationship-specificbudget constraint:

pWAJcaAJ þ TaJAcaJA ¼ pWAJEa þ TaJA 1−ENð Þ þ TaJA−1ð ÞiaJA ð8Þwhere pAJ

W is the world relative price of goodmAJ and iaJA is country a’s imports of goodmJA (hence (TaJA−1)iaJA represents tariff revenue). Expressions for country a’s con-sumption of goods mAJ and mJA are given as:

caAJ ¼ Ea − eaAJ ð9Þ

caJA ¼ 1−EN þ iaJA ð10Þwhere eaAJ denotes country a’s exports of good mAJ. Substituting (9) and (10) into (8)and rearranging yields the trade balance condition:

pWAJeaAJ ¼ iaJA ð11ÞGiven the assumed preference structure, utility is maximised by allocating equal

expenditures to each good such that: pAJW caAJ =TaJAcaJA. Substituting in expressions (9)

and (10) yields:

pWAJ ¼TaJA 1−EN þ iaJAð Þ

Ea−eaAJð12Þ

Substituting Eq. (12) into the trade balance condition given in Eq. (11) gives countrya’s offer curve:

Ea

eaAJ¼ TaJA 1−ENð Þ

iaJAþ TaJA þ 1 ð13Þ

Similarly, country j sets tariffs to maximise utility subject to the following budgetconstraint:

cjJA þ TjAJ pWAJcjJA ¼ EN þ TjAJ p

WAJ 1−Eað Þ þ TjAJ −1

� �ijAJ p

WAJ ð14Þ

Expressions for country j’s consumption of goods mAJ and mJA are given as:

cjJA ¼ EN − ejJA ð15Þ

cjAJ ¼ 1−Ea þ ijAJ ð16ÞSubstituting Eqs. (15) and (16) into (14) and rearranging gives the trade balance

condition:

pWAJ ijAJ ¼ ejJA ð17ÞAs in country a, utility is maximised by allocating equal expenditure to each good

such that: pAJWTjAJcjAJ =cjJA. Substituting in the expressions for consumption gives:

pWAJ ¼ EN − ejJAT jAJ 1 −Ea þ ijAJ

� � ð18Þ

E. MacMillan

Substituting Eq. (18) into the trade balance condition and rearranging gives countryj’s offer curve:

EN

ejJA¼ TjAJ 1−Eað Þ

ijAJþ TjAJ þ 1 ð19Þ

Utilising the trade balance conditions given in (11) and (17), and the fact that asingle world price prevails, each country’s offer curve can be solved for world prices:

country a : pWAJ ¼iaJAeaAJ

¼ TaJA 1−ENð ÞEa − TaJAeaAJ − eaAJ

ð20Þ

country j : pWAJ ¼ejJAijAJ

¼ EN

TjAJ 1−Ea þ ijAJ� �þ ijAJ

ð21Þ

From the trade balance conditions it is clear that country a’s exports must equalcountry j’s imports. Thus, solving expression (21) for imports and substituting intoEq. (20) according to eaAJ = ijAJ yields an expression for period 1 world prices in termsof endowments and tariffs:

pWAJ ¼TaJA þ Ea þ TaJAT jAJ 1−Eað ÞTjAJ þ EN þ TaJAT jAJ 1−ENð Þ ð22Þ

Substituting Eq. (22) into expressions (20) and (21) allows the following expressionsfor exports and imports to be derived:

eaAj ¼ ijAJ ¼ EaEN−TaJAT jAJ 1−Eað Þ 1−ENð ÞTaJA þ EN þ TaJAT jAJ 1−ENð Þ ð23Þ

ejJA ¼ iaJA ¼ EaEN − TaJAT jAJ 1−Eað Þ 1−ENð ÞTjAJ þ Ea þ TaJAT jAJ 1 −Eað Þ ð24Þ

Substituting Eqs. (23) and (24) into Eqs. (9), (10) yields expressions for consump-tion in terms of tariffs and endowments:

caAj ¼ Ea−EaEN−TaJAT jAJ 1−Eað Þ 1−ENð ÞTaJA þ EN þ TaJAT jAJ 1−ENð Þ ð25Þ

caJA ¼ 1−EN þ EaEN −TaJAT jAJ 1−Eað Þ 1−ENð ÞTjAJ þ Ea þ TaJAT jAJ 1−Eað Þ ð26Þ

cjJA ¼ EN −EaEN − TaJAT jAJ 1−Eað Þ 1−ENð ÞTjAJ þ Ea þ TaJAT jAJ 1−Eað Þ ð27Þ

cjAJ ¼ 1−Ea þ EaEN−TaJAT jAJ 1−Eað Þ 1−ENð ÞTaJA þ EN þ TaJAT jAJ 1−ENð Þ ð28Þ

Explaining rising regionalism and failing multilateralism

Utilising the expression for utility given in Eq. (4) the following expression can nowbe derived for the utility of country awith respect to all its trading partners (signified bythe subscript T) in terms of tariffs and endowments:

uaT ¼ n2N

logEa þ TjAJ 1−ENð Þ

1þ TjAJ 1−ENð Þ þ EN=TaJA

� �þ log

Ea þ TjAJ 1−ENð ÞTjAJ þ TaJATjAJ 1−Eað Þ þ Ea

� �� �

þ N−n2N

logEa þ TkAK 1−ENð Þ

1þ TkAK 1−ENð Þ þ EN=TaKA

� �þ log

Ea þ TkAK 1−ENð ÞTkAK þ TaKATkAK 1−Eað Þ þ Ea

� �� �ð29Þ

Similarly, the utility of country j in terms of tariffs and endowments (an analogousexpression exists for country k) is given as:

uj ¼ 1

2log

EN þ TaJA 1−Eað Þ1þ TaJA 1−Eað Þ þ Ea=TN

� �þ 1

2log

EN þ TaJA 1−Eað ÞTaJA þ TaJAT jAJ 1−ENð Þ þ EN

� �ð30Þ

3.2 Free trade outcomes

Free trade consumption levels are given by evaluating expressions (25)–(28) givenTaJA=TaKA=TjAJ=TkAK =1 to yield:

cFaAJ ¼ cFaJA ¼ 1

2Ea þ 1−ENð Þ ð31Þ

cFjJA ¼ cFjAJ ¼1

21−Ea þ ENð Þ ð32Þ

These expressions are utilised later on in the computation of NBPs.

3.3 Nash equilibrium tariffs

In this model, as in Mayer (1981), Dixit (1987) and Bagwell and Staiger (1999, 2002),governments intervene in trade for optimal tariff reasons and duly arrive at a prisoners’dilemma which motivates them to form trade agreements. Reaction functions aregenerated by maximising the expressions for utility given in Eqs. (29) and (30) withrespect to tariffs:

TaJA ¼ EN TjAJ þ Ea

� �TjAJ 1−Eað Þ 1þ TjAJ 1−ENð Þ� �

!12

ð33Þ

TjAJ ¼ Ea TaJA þ ENð ÞTaJA 1−ENð Þ 1þ TaJA 1−Eað Þð Þ� �1

2

ð34Þ

E. MacMillan

Solving Eqs. (33) and (34) simultaneously gives the Nash equilibrium tariffs asidentified by Kennan and Riezman (1988):

TNaJA ¼ TN

aKA ¼ EN

1−Ea

� �12

ð35Þ

TNjAJ ¼ TN

kAK ¼ Ea

1−EN

� �12

ð36Þ

Substituting Eqs. (35) and (36) into the expressions for consumption given inEqs. (25)–(28) gives Nash consumption levels:

cNaAJ ¼Ea þ Ea 1−ENð Þð Þ12

1þ Ea 1−ENð Þð Þ12 þ EN

EN1−EaAAJ

� �12

ð37Þ

cNaJA ¼ Ea þ Ea 1−ENð Þð Þ12Ea

1−EN

� �121þ EN 1−Eað Þð Þ12� �

þ Ea

ð38Þ

cNjAJ ¼EN þ EN 1−Eað Þð Þ12

EN1−Ea

� �121þ Ea 1−ENð Þð Þ12� �

þ EN

ð39Þ

cNjJA ¼ EN þ EN 1−Eað Þð Þ121þ EN 1−Eað Þð Þ12 þ Ea

Ea1−EN

� �12

ð40Þ

Equations (37)–(40) could be used to derive expressions for Nash equilibrium utilityin terms of endowments and tariffs. However, these, and subsequent expressions,would be too cumbersome to present here. Accordingly, utility from here on is simplygiven in terms of consumption. Thus, the utility achieved by countries a and j at theNash equilibrium is given as:

UNaT ¼ n

2NlogcNaAJ þ logcNaJA� �þ N − n

2NlogcNaAK þ logcNaKA� � ð41Þ

UNj ¼ 1

2logcNjJA þ logcNjAJ

� �ð42Þ

As is well established, Nash equilibrium utility is inefficient and can be improvedupon by mutual liberalisation via a trade agreement. However, such agreements can be

Explaining rising regionalism and failing multilateralism

formed either via the WTO or via a PTA. The following section details the distinction inthe current setting.

4 Bilateral versus multilateral trade agreements

In order to capture the benefits of multilateralism in this context it is assumed that dealsconducted within the WTO are cheaper to establish than bilateral deals. This assump-tion conforms with the literature discussed in section 2.2 regarding the myriad ways inwhich a central institution can mitigate the transaction costs associated with negotiatinginternational agreements. However, whilst in reality the WTO certainly confers suchbenefits on its members, negotiations conducted within it tend to take a considerablelength of time. Indeed, the last time a multilateral trade negotiating round was successfullyconcludedwas in 1994when theUruguay round came to an end. In contrast, over 160 PTAshave been notified to the WTO since 1995 (WTO 2007). In order to capture this aspect ofreality in the current model, it is assumed that WTO negotiations take two periods toconclude whereas PTA negotiations can be conducted within a single period. Accordingly,country a and the n other WTO member countries play the following two-period game:

In the first period countries a and the n other WTO members must decide whether tocommence multilateral negotiations or to form a web of PTAs. If bilateralism is optedfor, the game ends and the PTA outcome pertains in both periods. However, ifmultilateralism is chosen, the status quo, i.e. the Nash equilibrium, is maintained forthe remainder of the period and the game continues on to the second period. In thesecond period there are three possible outcomes: a successfully concluded WTOagreement; a web of PTAs; or the maintenance of the Nash equilibrium. Reversion toPTA-based negotiation is the threat point during WTO negotiations. The threat pointduring PTA negotiations is, as it was in the first period, maintenance of the Nashequilibrium. Of crucial importance is the fact that the veto power afforded by WTOrules means that reversion to PTA-based negotiation in a game between country a andany country j causes a reversion to PTA-based negotiation in all games betweencountry a and its trading partners (Fig. 3).

Note that the payoffs in Fig. 2 are given in terms of Vi, which represents inter-temporal utility. In the present scenario, discounting of the future would temper theoutcomes quantitatively but have no qualitative effect. Hence, to facilitate a parsimo-nious exposition, inter-temporal utility is simply given as the sum of the utilitiesobtained in each period:

V i ¼ ui þ Ui ð43Þ

where Ui denotes the utility obtained by country i in period 2.The superscripts refer to the outcome in each period, for example, the superscript

NW refers to Nash equilibrium in period 1 and a WTO agreement in period 2. In eitherperiod, the game between country a and the N−nWTO non-members is simply that ofbilateral PTA negotiation. The crucial question is whether an increasing WTO mem-bership augments or diminishes the welfare benefit of forming a multilateral tradeagreement, i.e. what is the effect of a rising value of n on Vi

NW−ViPP. This question isaddressed in the following sections.

E. MacMillan

4.1 Second period PTA bargaining

PTA negotiations are the outside option for all parties to multilateral negotiations. Thus,in order to compute the outcome of multilateral bargaining, it is first necessary toevaluate the outcome of the bilateral negotiations that would occur in the event of theircollapse. Utilising the approach of Mclaren (1997) and Epifani and Vitaloni (2006), theNash bargaining solution (NBS) involves countries going to free trade with a transfer,S, between the party that favours free trade the most and the party that favours it theleast. However, in order to capture the notion that the WTO yields benefits over andabove those obtained from bilateral bargaining, it is assumed that some portion of tradeis lost during PTA negotiations, which is not lost during WTO negotiations. Thus,countries cannot achieve free trade from a PTA; a portion, θ, of the difference betweenNash equilibrium and free trade levels of trade are lost. Accordingly, exports andimports under PTA-diminished free trade are given as:

eθFi ¼ eFi − eFi − eNi

� �θ ð44Þ

iθFi ¼ iFi − iFi − iNi� �

θ ð45ÞIn the second period PTA, Nash equilibrium is the outside option, thus the Nash

bargaining product (NBP) between countries A and j is given as:

NBPP ¼ UθFaj −

SP

PNAJP

NJA

� �1=2 −UNaM

!UθF

j þ SP

PNAJP

NJA

� �1=2 −UNj

!ð46Þ

Note that UajθF is the welfare that country a achieves from PTA-diminished free trade

with respect to country j holding trade with all other N−1 of its trading partnersconstant at the Nash equilibrium:

PTA

a, n

,PP PPan jV V ,NW NW

an jV V ,NN NNan jV V

1ST PERIOD

2ND PERIOD

a, n

Nash

WTO NashPTA

,NP NPan jV V

Fig. 3 The choice of trade agreement between WTO members

Explaining rising regionalism and failing multilateralism

UθFaj ¼ 1

2NlogcθFaAJ þ logcθFaJA� �þ N−1

2NlogcNaAK þ logcNaKA� � ð47Þ

UjθF is the utility that country j achieves from PTA-diminished free trade:

UθFj ¼ 1

2logcθFjJA þ logcθFjAJ

� �ð48Þ

Maximising the NBP with respect to SP yields:

SP ¼ 1

2UθF

aj −UNaT

� �−1

2UθF

j −UNj

� �ð49Þ

Given the assumption of symmetry between country a’s trading partners, the transferbetween country a and country j is of the same magnitude as that between country aand country k. Consequently, the utility obtained by all 3 country types can be given as:

UPaT ¼ UθF

aT−NSP ð50Þ

UPj ¼ UP

k ¼ UθFj þ SP ð51Þ

4.2 Second period WTO bargaining

WTO negotiations differ from PTA negotiations in several ways. Firstly, by assump-tion, they yield benefits over and above PTA negotiations; captured here by the absenceof the diminishing variable, θ, during WTO negotiations. Secondly, the outside optionduring WTO negotiations is the welfare that countries would obtain via PTA negoti-ations. Thus parties to the WTO bargain over the benefit of negotiating via the WTOrelative to negotiating a web of PTAs. Thirdly, given that bargaining is multilateral, theoutside option of country a during WTO negotiations is not simply a reversion to PTAnegotiations with respect to a single country, but a reversion to PTA negotiations withrespect to all countries. Consequently, all n WTO members that trade with country ahave significant bargaining power during WTO negotiations.

The multilateral NBP during WTO negotiations is given as:

NBPW ¼ U Fan− n

SW

PNAJP

NJA

� �1=2 −UPaM

!U F

j þ SW

PNAJP

NJA

� �1=2 −UPj

!n

ð52Þ

where UanF is the welfare that country a achieves from free trade with respect to all n

WTO members, holding trade with the N−n WTO non-member countries constant atPTA levels:

U Fan ¼

n2N

logcFaAJ þ logcFaJA� �þ N−n

2NlogcθFaAK þ logcθFaKA� �

− N − nð ÞSP ð53Þ

Maximising the NBP with respect to SW yields:

SW ¼ 1

1þ nU F

an−UPaM

� �−

1

1þ nU F

j −UPj

� �ð54Þ

E. MacMillan

The utility of WTO members following multilateral negotiations can thus be givenas:

UWan ¼ U F

an−nSW ð55Þ

UWj ¼ U F

j þ SW ð56ÞThe welfare of the N−n WTO non-members remains unchanged at that which they

achieve from PTA negotiations.

4.3 Bilateralism versus multilateralism in the first period

If WTO members decide not to participate in multilateral negotiations, the PTAoutcome will pertain in both periods such that inter-temporal welfare is given as:

VPPaM ¼ uPaM þ UP

aM ¼ 2UPaM ð57Þ

VPPj ¼ VP

k ¼ uPj þ UP

j ¼ 2UPj ð58Þ

If WTO member countries do decide to engage in multilateral negotiations, theymust accept that the bargaining process requires a period to complete; thus the Nashequilibrium pertains for the first period. Accordingly, the inter-temporal welfare ofcountries A and j from WTO negotiations is given as:

VNWan ¼ uNan þ UW

an ð59Þ

VNWj ¼ uNj þ UW

j ð60ÞNote that uan

N denotes the utility that country a achieves from remaining at the Nashequilibrium with respect to all nWTOmember countries, but negotiating PTAs with allN−n WTO non-members. This is given as:

uNan ¼n2N

logcNaAJ þ logcNaJA� �þ N−n

2NlogcθFaAK þ logcθFaKA� �

− N−nð ÞSP ð61Þ

5 Results

From the above analysis, it is clear that the utility of countries a and j are determined byendowments, the cost of a PTA, and the number of countries that are WTO members.Numerically simulating Vi

NW−ViPP for various values of n (n=1…50) yields multiple

different outcomes depending on the values chosen for endowments and θ. In order torestrict attention to scenarios more likely to correspond with reality, country a is assumednot to be smaller than country j in all simulations. This conforms with the notion thatcountry a is a large developed ‘hub’ that trades with multiple developing country‘spokes’. One important case is used as a benchmark and discussed in section 5.1 below:

Explaining rising regionalism and failing multilateralism

when country a chooses to negotiate multilaterally when WTO membership is low butchooses to negotiate bilaterally when membership expands too much. Section 5.2 thendiscusses how this benchmark changes with variations in endowments and θ.

5.1 Benchmark case: multilateralism collapses as WTO membership expands

The benchmark case is illustrated in Fig. 4 below:From Fig. 4 it is clear that, when country a is larger than country j, country a’s utility

from multilateralism relative to bilateralism is parabolic with respect to n. Hence, forcountry a, there is an optimal number of WTO member countries. If accessioncontinues past this point, the benefit country a obtains from negotiating multilaterallydeclines and eventually becomes negative. On the other hand, each country jbenefits from expanding WTO membership but at a declining rate.

The intuition underlying this result is as follows. Each acceding country shifts outthe multilateral bargaining frontier because of the utility it, and country a, gain fromnegotiating multilaterally rather than bilaterally. However, the total gains must then bespread amongst the existing membership plus the acceding country. Thus, as themembership grows, the extra gains brought by an acceding country must be spreadincreasingly thinly. For country a, and each country j, second period utility frommultilateralism is diminishing with respect to n. However, for country a, each accedingcountry entails a further loss in the first period relative to bilateralism because it mustendure the lower Nash equilibrium welfare for that period.

Given that, without country a’s consent, a WTO accord will not be reached, it isclear that, once WTO membership has expanded too much, multilateralism willcollapse and bilateralism will prevail; as appears to be occurring in reality.

Fig. 4 Multilateralism versus bilateralism for countries a and j given: Ea=0.9,EN=0.2,N=50,θ=0.9

E. MacMillan

Note that the y-axis intercept in Fig. 4 represents the relative value of multilateralismwhen there is only one spoke country. If the single-undertaking and consensusdecision-making principles were not applied, then this would be the utility country awould attain from the accession of each new WTO member. Given that utility is alsopositive for country j at the intercept, it is clear that, in this admittedly highly stylisedscenario, were negotiations within the WTO to be conducted purely bilaterally, thencountry a would embrace WTO negotiations and global welfare gains would result.

5.2 Other scenarios

As noted above, utility is dependent upon endowments, the cost of a PTA and thenumber of WTO members. Varying the cost of a PTA shifts the positions of the curvesvertically: a higher value of θ causes the curves to shift upwards; a lower value causesthe curves to shift downwards. It is thus possible to construct examples where multilat-eralism is always rejected by country a, always rejected by both countries, and alwaysaccepted by both countries. However, the basic intuition remains: multilateralism ismore likely to be rejected at higher levels of WTO membership.

Altering the relative sizes of countries a and j does not change the basic intuitioneither; it simply alters the optimum number of WTO members for country a. The closercountries a and j are in size, the lower the optimum membership for country a andhence the sooner it will reject multilateralism as membership expands.

6 Conclusions

By constructing a more realistic model of multilateral trade negotiations than haspreviously been ventured, the above analysis has highlighted that, under certaincircumstances, an expanding WTO membership can lead to a collapse in multilateralnegotiations because of the veto power afforded to member countries under WTOjurisprudence. The circumstances under which this situation arises are akin tothose currently prevailing in the WTO: when large ‘hub’ countries deal withsmaller ‘spoke’ countries; when WTO agreements yield benefits relative to a web ofPTAs; and, when the conclusion ofWTO negotiations is time-consuming.Moreover, thecurrent model shows that, were the veto power of countries to be removed,global welfare gains could be attained through the successful conclusion ofWTO negotiations.

It is important to highlight an important caveat: the pattern of trade assumed in thismodel negates the benefits of multilateralism as described by Maggi (1999). Whilst thisserved to highlight the deleterious effect of the single-undertaking and consensusdecision-making principles, it is necessary to temper advocacy of removal of theseprinciples with the realisation that the issue-linkages they facilitate can be beneficial.However, some halfway measures have been proposed. For example, Sutherland et al.(2004, p. 64) suggest that:

“a Member considering blocking a measure which otherwise has very broadconsensus support shall only block such consensus if it declares in writing, withreasons included, that the matter is of vital importance to it”.

Explaining rising regionalism and failing multilateralism

This proposal would dissuade countries from attempting to capture the gains thatother countries obtain from multilateralism relative to regionalism and would therebyassuage the pessimistic prognosis of the model in this paper. More extreme proposalshave been ventured that suggest that ‘variable geometry’ be introduced into the WTO,i.e. relaxing the single-undertaking restriction and thereby making the WTO an um-brella institution circumscribing a range of ‘plurilateral’ agreements, participation inwhich would be on a purely voluntary basis.6 According to proponents of this approach,variable geometry would allow coalitions of the willing to proceed unhindered by theobjections of others and thus impasse would be less likely to arise. Cornford (2004),Sutherland et al. (2004) and Lawrence (2006) note that variable geometry is not withouthistorical precedent in the multilateral trade negotiating system. For example, theTokyo round of the GATT established a number of ‘codes’ covering issues such as‘trade in civil aircraft’, the ‘arrangement on bovine meat’ and the ‘agreement ongovernment procurement’ (AGP) which were undertaken voluntarily. Lawrence(2006) suggests that a similar approach be adopted in the WTO as was utilised in theAGP. Thus, all members would be free to negotiate agreements but ratification wouldbe on a voluntary basis. To avoid free-riding, he suggests that non-signatories do notreceive the benefits of the agreement. The analysis in this paper suggests that such ideaswarrant consideration.

References

Abbot K, Snidal D (1998)Why States act through formal international organisations. J Confl Resolut 42(1):3–32Bagwell K, Staiger R (1999) An economic theory of GATT. Am Econ Rev 89(1):215–248Bagwell K, Staiger R (2002) The economics of the world trading system. The MIT Press, CambridgeBagwell K, Staiger R (2004) Backward stealing and forward manipulation in the WTO. NBER Working

Paper, No.10420Cornford A (2004) Variable geometry and the WTO: concepts and precedents. UNCTAD Discussion Paper,

No. 171Dixit A (1987) Strategic aspects of trade policy. In: Bewley TF (ed) Advances in economic theory fifth world

congress. Cambridge University Press, CambridgeEhlermann C-D, Ehring L (2005) Decision-making in the world trade organisation: is the consensus practice

of the world trade organisation adequate for making, revising and implementing rules on internationaltrade. J Int Econ Law 8(1):51–75

Epifani P, Vitaloni J (2006) GATT-think with asymmetric countries. Rev Int Econ 14(3):427–444Fearon JD (1998) Enforcement and international cooperation. Int Organ 52(2):269–305Hoekman BM, Kostecki MM (2001) The political economy of the world trading system: the WTO and

beyond, 2nd edn. Oxford University Press, OxfordHorn H, Mavroidis P (2001) Economic and legal aspects of the most-favoured nation clause. Eur J Polit Econ

17(2):233–279Hufbauer G (2005) Inconsistency between diagnosis and treatment. J Int Econ Law 8(2):291–297Jackson J (1998) The world trading system: law and policy of international economic relations. The MIT

Press, CambridgeKennan J, Riezman R (1988) Do big countries win tariff wars? Int Econ Rev 29(1):81–85Kennan J, Riezman R (1990) Optimal tariff equilibria with customs unions. Can J Econ 23(1):70–83Koremenos B, Lipson C, Snidal D (2001) The rational design of international institutions. Int Organ 55(4):

761–799

6 Authors subscribing to this approach include: Cornford (2004), Hufbauer (2005), Messerlin (2005), andLawrence (2006), etc. Moreover, whilst cognisant of the potential benefits of variable geometry, Sutherlandet al. (2004) temper their advocacy with concerns about the practicality of its implementation.

E. MacMillan

Lawrence R (2006) Rulemaking amidst growing diversity: a club-of-clubs approach to WTO reform and newissue selection. J Int Econ Law 9(4):823–835

Ludema R, Mayda A (2009) Do countries free ride on MFN? J Int Econ 77(2):137–150Maggi G (1999) The role of multilateral institutions in international trade cooperation. Am Econ Rev 89(1):

190–213Mayer W (1981) Theoretical considerations on negotiated tariff adjustments. Oxf Econ Pap 33(1):135–153Mclaren, J (1997) Size, sunk costs and judge bowker's objection to free trade. Am Econ Rev 87(3):400–420Messerlin PA (2005) Three variations on ‘the future of the WTO’. J Int Econ Law 8(2):299–309Saggi K (2009) The MFN clause, welfare and multilateral cooperation between countries of unequal size. J

Dev Econ 88(1):132–143Saggi K, Sengul F (2009) On the emergence of an MFN club: equal treatment in an unequal world. Can J

Econ 42(1):267–299Schott J, Watal J (2000) Decision making in the WTO. International Economics Policy Briefs, No.2. Available

online at: http://www.iie.com/publications/pb/pb.cfm?ResearchID=63Steinberg R (2002) In the shadow of law or power? Consensus-based bargaining and outcomes in the

GATT/WTO. Int Organ 56(2):339–374Sutherland P, Bhagwati J, Botchwey K, FitzGerald N, Hamada K, Jackson J, Lafer C, de Montbrial T (2004)

The future of the WTO: addressing institutional challenges in the new millennium. WTO Secretariat,Geneva

WTO (2007) World trade report 2007: six decades of multilateral trade cooperation. What have we learnt?WTO Secretariat, Geneva

Explaining rising regionalism and failing multilateralism