C. Causes and effects of pTas: Is it all about … · science focuses on the causes and effects of ... analysis of preferential trade agreements ... • Global production networks

WOrld Trade repOrT 2011

92

A vast literature in economics and political science focuses on the causes and effects of preferential trade agreements – and in particular on the way that border measures, such as tariffs, impact trade flows among countries both inside and outside such agreements. Often referred to as the “standard analysis of preferential trade agreements”, this literature is discussed in detail in Sections C.1 and C.2. However, many recent regional agreements have moved beyond border measures to include deeper forms of rules and institutions that can only be partly understood by the standard analysis of preferential trade. An examination of the economic motives – and the key issues – that lie behind these deeper integration agreements is discussed in Section C.3.

C. Causes and effects of pTas: Is it all about preferences?

II – tHe Wto AnD PReFeRentIAL tRADe AGReements

93

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

Contents 1. MotivesforPTAs 94

2. ThestandardeconomicsofPTAs 100

3. Goingbeyondthestandardanalysis 109

4. Conclusions 114

TechnicalAppendix:SystemiceffectsofPTAs 118

Some key facts and findings

• PTAs now cover a wider number of issues – beyond tariffs – and

involve more structured institutional arrangements.

• Global production networks increase the demand for deep

agreements since they provide governance on a range of regulatory

issues that are essential to the success of the networks.

• Deep integration agreements can complement rather than substitute

for the process of global integration.

• Economic theory needs to go beyond the standard trade-creation

and trade-diversion analysis of PTAs, which is about the impact of

preferential tariffs.


94

1. MotivesforPTAs

Economists and political scientists have identifiedseveral rationalesforpreferential tradeagreements–abriefoverviewofwhichisprovidedbelow.

(a) Neutralizingbeggar-thy-neighbourtradepolicies

Economistshavelongrecognizedthattradepolicycanhave “beggar-thy-neighbour” effects. That is,protectionist trade measures can be unilaterallyattractive but multilaterally destructive. Specifically,the beggar-thy-neighbour problem is based on theidea that tradepolicydecisionsofonecountryaffectthewelfareofanothercountrythroughaninternationalexternality (i.e. a cross-border effect). The economicliteraturehashighlightedtwomaineffectsassociatedwith trade policy: the terms-of-trade effect and theproduction relocation effect. These are discussed inmoredetailbelow.Independentlyofhowonecountry'strade policy affects its trading partners, a tradeagreement is ameansofneutralizingnegativecross-bordereffects.

The main logic of the terms of trade (or traditional)approach is that countries that have market power(i.e. that can influence their terms of trade) cannotresist the temptation to act non-cooperatively. Asnoted by Johnson (1953), each country sets tradepolicy in an attempt to improve its terms of trade(i.e. lower thecostsof its imports relative toexports)and increasenational income.1However, theresultingnon-cooperative (Nash) equilibrium is inefficient, aseach country's terms-of-trade-enhancing unilateralactions are cancelled out. More restrictive tradepolicies by all countries have little net effect on theterms of trade, but lead to a contraction of tradevolumeswhichreducesaggregatewelfare–asituationreferred to as a terms-of-trade-driven Prisoners’Dilemma(BagwellandStaiger,1999).

Theterms-of-tradeeffectmaynotbetheonlyrelevantexternality associated with trade policy. Trade policymayalsotrytoexpanddomesticproductioninasectorto the detriment of foreign production by changingrelative prices. This is referred to as the “productionrelocation effect” (Venables, 1987). Like a terms-of-trade-driven Prisoners’ Dilemma2, if all governmentschoose trade policies aimed at attracting moreproduction, no government actually succeeds. Inequilibrium, production does not relocate acrosscountries, but trade falls in response to the rise inrestrictive trade measures. To put it differently,countries are stuck in a production relocationPrisoners’Dilemma.

These non-cooperative situations can be avoidedthrough a trade agreement among countries whichencourages them to cooperate rather than to act

unilaterally.3Animportantquestioniswhethersuchanagreement should be at the regional or at themultilateral level. Studies by Bagwell and Staiger(2003) and by Ossa (2010) show that a multilateraltrade agreement based on simple rules that allowcountries to coordinate tariff reductions andreciprocate market access is the first-best option toneutralize negative (terms-of-trade or productionrelocation)externalities.

If a multilateral trade agreement such as the GATT/WTO is in place, there is no rationale for signing apreferential trade agreement (PTA)4 – and WTOmembers would have little incentive to form PTAs tosolvethesetypesofcoordinationproblems.5However,in the absence of multilateral trade cooperation,countries may seek a preferential agreement to limitcross-bordereffectsassociatedwithtradepolicy.

(b) Gainingcredibility

Aside fromavoiding the temptation toadopt “beggar-thy-neighbour”tradepolicies,preferentialagreementsmay also serve as instruments to stop governmentsfrom implementing“beggar-thyself”policies.By this itis meant that a government may choose to “tie itshands”andcommititselftotradeopennessthroughaninternational agreement in order to prevent futurepolicy reversal that might be convenient in the shortrun,butinefficientinthelongterm.Inotherwords,thegovernmentunderstands that anagreement canhelpit to make more credible policy commitments than itwouldotherwisebeabletomake.

Specifically, a government might sign a PTA to solvesome form of time-inconsistency problem.6,7 Thedifferent mechanisms through which a time-inconsistent trade policy may lead to inefficiencieshavebeenhighlightedinanumberofstudies(StaigerandTabellini,1987;Matsuyama,1990;Amin,2003).Inthese models, the government wants to usediscretionary trade policy to increase social welfare(for example, in response to an unexpected event, toallow temporaryprotection toan infant industry,etc.).However, theuseof tradepolicycanalter thenormalbehaviourofparticipants inaneconomysinceagentscan anticipate the policy change, and react to it inwaysthatwillreducethepolicy'simpactonthem.Thisimplies that the government will not be able to usediscretionary trade policy as originally intended,resultinginasociallyinefficienttradepolicy.

Similar credibility problems emerge when agovernment is exposed to political pressures fromdomestic interest groups lobbying for protection(Maggi and Rodriguez-Clare, 1998). The presence ofimport restrictions will reward import-competingproducers and will divert investments from othereconomicactivities.Thecostofthisdistortionmaybelarge in the long run, but in the short run domesticlobbying by the import-competing sector will prompt


95

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

the policy-maker to set high restrictions. In thesecircumstances, Maggi and Rodriguez-Clare (1998)identify two reasons why a government may want tocommittoaPTA:first,tominimizethecostlylong-termdistortions involved with protecting a politicallyorganized sector, where the country has nocomparative advantage and it is unlikely to gain it inthefuture;and,secondly,toavoidacostlydelayintheadjustmentprocessof thedecliningsector relyingongovernmentprotection.

These theoretical results contain a clear normativeimplication: governments should undertake bindingtrade policy commitments concerning their futurebehaviour.Atradeagreement,inadditiontofacilitatingpolicy cooperation as emphasized above, may haveprecisely this commitment role, as it reduces oreliminates the signatory governments' discretionarypower in setting tariffs, and raises the costs ofresorting to unilateral trade protectionism. Thisprovidesawelfare-improvingwaytoenforcedomesticcommitmentstoapolicyoftradeopenness.8

An important question is whether a PTA may providemore credibility than a multilateral treaty – in otherwords,wouldaWTOmemberchoosetosignaPTAtoimprovefurtherthecredibilityofitspolicyvis-à-vistheprivatesector.Onepossibilityisthatacountrymaybetoosmall inworldmarketsforothercountries tocareabout its GATT/WTO violations, whereas a countrythat has preferential access to that country has aparticular stake in making sure that this preferentialaccess ismaintained.Thisprovidesapossiblereasonwhyasmallcountryseekingtotieitshandsthroughatradeagreement–andthereby increaseitscredibilitywith itsownprivatesector–mightnaturally looktoaPTAinadditiontoGATT/WTOcommitments.

(c) Othereconomicmotives

There are several other economic reasons whycountries opt to form PTAs, some that mirror themotivesdiscussedaboveandothersthataresometimesreferredtoas“non-traditional”motives(FernandezandPortes, 1998). These are briefly reviewed below. Theyinclude, but are not limited to, increasing market size,increasing policy predictability, signalling openness toinvestorsandachievingdeepercommitments.

IncreasingmarketsizecanbeareasonforestablishingPTAs since it enables firms from signatory states toexploit economies of scale and to gain a relativeadvantageoverexcludedcompetingfirms. Inaddition,preferentialaccesstoa largermarketmay increaseacountry's attractiveness as a destination for foreigndirect investment (FDI). Both reasons are particularlyvalid for small economies, which may help to explainwhy these countries agree to make concessions onother more controversial issues, such as intellectualproperty rights or environmental standards, whennegotiatingPTAswithlargeeconomies.

Related to the time-inconsistency issues addressedabove, a trade agreement may also be signed toreduceuncertaintyonfuturetradepolicy,thussendingan important signal to investors. Since futureadministrations might have policy preferences thatdiffer from those of the current administration, agovernmentmaysignaPTAinanattempttolock-initspolicies (forexample, apro-open tradepolicy) and todiminish the likelihoodthat theymightbereversed. Inthis way, the government addresses not so much theissue of policy credibility as the issue of policypredictability(FernandezandPortes,1998).

A country with a reputation for protectionism mightfind it particularly valuable to signal itswillingness toshift towards a more liberal and business-friendlypolicy.Inthiscase,thepreciseprovisionsofaPTAarelessrelevant thandemonstratingto investors that thecurrentgovernment isopentobusiness.Alternatively,acountrymightwanttoenterintoaPTAtosignalthatitseconomy,oraparticularsector,iscompetitive.

Economicanalysisoftenoverlooksthesimplefactthattradepolicy isdecided inapoliticalenvironment,andgovernments may face incentives that differ fromsimple welfare considerations. However, some recenteconomicliteraturehasemphasizedtheroleplayedbyspecial interestgroups in tradepolicydetermination.9Simply put, interest groups lobby to influencegovernmentdecisionsand, inturn,governmentstradeoff the welfare effects of their trade policy choices(e.g. signing or not signing a PTA) with the politicalsupport of special interests. In this political context,the choice to sign a preferential agreement may bedriven by the interests of an organized lobby ratherthan by social welfare considerations (Grossman andHelpman,1995).10

AfinalargumentforsigningaPTArelatestotheneedto achieve a deeper form of integration which goesbeyondtraditionaltrade(i.e.border)measuressuchastariffs (Lawrence,1996).Thisdeeper integrationmayrequire institutions and levels of policy coordinationthatcanbemoreeasilyachievedat theregional thanat the multilateral level.11 This issue will be moreextensivelydiscussedinSectionC.3.

(d) Politicalmotives

The creation of PTAs cannot be fully understoodwithout considering the political context within whichthey are formed. Political science has providedadditionalexplanationsforwhystatesmightengageinPTAs, focusing in particular on the role of politicalintegration, domestic politics, forms of governments,institutions, diplomacy or the influence of power andideas. Some of the most important “political”argumentsforPTAsarediscussedbrieflybelow.

Preferentialtradeagreementshavelongbeenseenasplaying a key role in regional political integration.


96

Perhaps thebestmodernexamplewas the formationoftheEuropeanCommunity(EC) inthe1950swhich,at the time,was themost importantPTA in theworldand attracted considerable attention from politicalscientists. Initially, “functionalist” scholars, inspiredbythelogicofintegration,emphasizedtheimportanceofbureaucratic actors as key drivers of integration, aswellastheprocessbywhichnationalelitestransferredloyaltiestoasupranationallevel(Mitrany,1943;Haas,1958;SandholtzandZysman,1989).

Itwaspostulated that apolicy spill-over effectwouldincrementally drive integration from “low politics”(trade integration) to “high politics” (politicalintegration).This“functionalist”schoolofthoughtwaslaterchallengedbypoliticalscientistswhomarshalledempirical evidence that cast doubt on the extent ofspill-overs and helped explain the stagnation in theEuropeanintegrationprocess.Proponentsofan“inter-governmentalist” theory argued that nationalpreferences were more relevant in shaping the paceandcontentofpoliticalandeconomicintegration,andquestioned whether there had been a significanttransferofcontrol frommemberstatestoCommunityinstitutions(Hoffmann,1966;Moravcsik,1998).

To help explain the increasing number of tradeagreementselsewhere in theworld,political theoristsfirst attempted to apply the European integrationmodels. However, the limits soon became obvious.TradeintegrationoutsideEuropeproceededaccordingto different patterns and concomitant politicalintegration was lacking. Additional strategicexplanations emerged. These included a desire toincrease influence in international negotiations bypoolingresources(e.g.theCaribbeanCommunity),seeAndriamananjara and Schiff (2001), or the goal ofresistingthethreatofcommunisminSouth-EastAsia,by strengthening cooperation among like-mindedgovernments(e.g.theAssociationofSoutheastAsianNations)foranoverview,seeRavenhill(2008).Anotherstrategicmotiveforformingregionaltradeagreementswas to counteract the growth of other regionalarrangements. For example, Asia-Pacific EconomicCooperationwaswidelyseenasanattemptbytheUSto send a pre-emptive trade policy signal to theEuropean Community about the cost of building a“FortressEurope”.

Existing research has shown that democracies aremorelikelytoformPTAsamongthemselves(Mansfieldet al., 2002). One explanation is that democraticgovernments use trade agreements as a signallingdevice vis-à-vis domestic constituents that they areimplementingsensiblepolicies.Relatedresearchlooksat how governments calculate the political costs andbenefits of PTAs, and how voters hold their politicalleaders accountable. The work by Mansfield et al.(2007)suggeststhatacountry'sdecisiontoenterintoPTAsisrelatedtothenumberof internalvetoplayers(i.e. lawmakers or parliamentarians). In addition,

MansfieldandMilner (2010)showthat thenumberofvetoplayersinacountryaffectsthetransactioncostsof an agreement. As the number of veto playersincreases,ratificationbecomeslesslikely.

While veto players diminish the likelihood of enteringPTAs,theregimetype(democracy)affectstheratificationrate positively. Mansfield and Milner (2010) argue thatPTAs can serve as a strategic tool vis-à-vis voters. Inother words, PTAs can act as a credible signal thatgovernments can use to pursue trade objectivespreferred by a majority of voters rather than by specialinterests.Accordingtothisview,thespreadofdemocracysincethe1980s,especiallyacrossthecountriesofLatinAmerica,Asia,andCentralandEasternEurope,mayhelpexplaintheproliferationofPTAs.

ThedecisiontonegotiateandsignPTAsmayalsobeaffected by the extent to which countries use tradepolicy to reinforce wealth and empower relations. Ifgovernments distrust one another, they may formbilateral treaties in order to limit or to control thegrowth of other powers (e.g. to serve as counter-balances). Gowa and Mansfield (1993) and Gowa(1994) argue that trade integration stimulates tradeflowsbetweentwocountries,leadstoamoreefficientallocationofresourcesandthusfreesupresourcesformilitary use. The increasing wealth and power ofmember countries should be of concern to excludedcountries. An agreement between two countries maythus forceotherpairsofcountries to followsuit,withthe aim of retaining their current relative position(GowaandMansfield,1993).

Inasimilarvein, thedesignofPTAs isalso indicativeof power relations. Stronger states can more easilydictate the terms of agreements in a bilateral orregional context. Other diplomatic and foreign policyconsiderations may influence the decision to formPTAs. For instance, some states use PTAs to rewardalliesandtoreinforcekeyalliances.Inthisview,PTAsare an active part of foreign policy making (White,2005;Rosen,2004;Higgott,2004;Capling,2008).

PTAs might also serve as “diffusion mechanisms” –either directly, in the form of coercion, or moreindirectly, in the form of learning. For example, agrowing body of work treats the EU as a “conflictedpower”(MeunierandNicolaidis,2006),whichusesitsmarketpower(i.e.accesstotheEU'ssinglemarket)tocoerceweakerpowers, includingformercolonies, intoaccepting new types of trade arrangements (Farrell,2005)(forexample,EuropeanPartnershipAgreementswith the African, Caribbean and Pacific group ofstates).OthersconsiderthattheEuropeanCommunityprovidedanexample foreconomic integrationamongcountries in Latin America and Africa in the 1960s(Pomfret, 2001), demonstrating how the perceivedsuccessoftradearrangements“teach”otherstoadoptsimilarpolicies(Krueger,1997).


97

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

Finally, there may be a direct or indirect relationshipbetween the formation of PTAs and the multilateralsystem, either reflecting a lack of progress at themultilateral level or a strategy to improve states’leverage in the WTO. Gridlock or stagnation inmultilateral negotiations, for example, may createincentives for states to pursue preferential tradeliberalization, and encourage exporters to lobby theirgovernmentsforPTAs(forexample,seecasestudiesinCaplingandLow(2010),wherepolicycommunitiesnote both the “remoteness” and “slowness” of theWTO).Alternatively,statesmaysignPTAsinordertoincrease their bargaining power during multilateraltrade talks (Mansfield and Reinhardt, 2003). Thedrawn-out negotiations in the Uruguay Round, and inthecurrentDohaDevelopmentRound,mayexplainthecurrentproliferationofPTAs.

(e) WhatexplainsthegrowthofPTAs?

Changes in the underlying dynamic of traderelationshipsacrosstheglobemaypromptcountriestosign PTAs. Baldwin (1995) provided a model of theenlargement of Europe's economic integration whichrestedona“dominotheory”ofregionalism–i.e.wherethe potential loss of market share induces non-members to joinexistingPTAs, creatingaprocessofaction and reaction or contagion. Exporters in non-member countries push their governments to joinexisting PTAs or create new ones to counteract thepotential damage caused by preferential tradeliberalization(BaldwinandJaimovich,2010).Thereisaset of studies which find broad empirical support forBaldwin'sdominotheory–formationofPTAscreatesan incentive for outsiders to become members of anexistingPTAor to formnewPTAs (Egger andLarch,2008;BaldwinandJaimovich,2010;ChenandJoshi,2010). According to Egger and Larch (2008), theseresultsareparticularlyuseful to “predict” theprocessofregionalintegrationinEurope.

The political science literature also focuses on thecausal mechanisms behind the domino effect, inparticular how decision-makers and interest groupsreact todiscrimination.Pahre(2008)appliesthe ideaof a competitive spread of trade agreements to thenineteenthcentury.Mattli(1999)makesthisargumentwith respect to the enlargement of the EuropeanUnion,whileGruber (2000)doesso in thecontextofthe North American Free Trade Agreement (NAFTA).Inasimilarvein,Dür(2010)explainsthePTAssignedby the EU and the US in the 1990s and 2000s interms of competition for market access in emergingeconomies.Thisempiricalliteraturedoesnotdenytheimportance of factors other than potential tradediversion in explaining the growth of PTAs. Forexample, Manger (2009) argues that investmentdiscrimination as a result of the creation of NAFTAcontributed to Japan’s decision to conclude a tradeagreementwithMexico.

The concluding part of this section emphasizes theimportanceof “deep” integration–arrangements thatgobeyondextendingpreferentialtariffconcessionstoincludeareassuchas investment– inPTAformation.Furthermore, Section D assesses the relativeimportance of tariff liberalization and “deep”integrationinexplainingtherecentspreadofPTAs.

In the literature, the influence of existing PTAs onsubsequent PTA formations is often referred to as“endogenous regionalism”. Such “endogenousregionalism”,however,mayalsobe influencedby tradeliberalization at the multilateral level. For instance,Freund (2000) argues that as multilateral tariff levelsfall, the formation of PTAs, and hence the dominoeffect, is strengthened. This may be explained by theeffect of tariff reduction on competition, profits, andtariffrevenue.

Lowering tariffs enhancescompetition,which leads togreateroutput.Athighworldtarifflevels,thisefficiencyeffect is large and multilateral tariff reduction, whichhas a greater effect on competition than preferentialreduction, is better. However, lowering tariffs alsomeans smaller profits and less tariff revenue. At lowoverall tariff levels, theefficiencyeffect issmaller,butpreferential reduction is lesscostly–profitsand tariffrevenuefallbyless.Preferentialagreementseffectivelyallow members to divert part of the profit loss thatresults from lower tariffs to the third country whereoutputcontracts.Hence,thewelfaregainfromjoiningaPTAisgreaterthanthegainfromamovetoopentradewhentariffsarelow;thereverseistruewhentariffsarehigh.12 Empirical evidence confirms the aboveprediction. For example, Fugazza and Robert-Nicoud(2010)showthatreductionsintheUSmultilateraltariffofagivenproductintheTokyoandUruguayRoundsaresystematicallyassociatedwithlowerpreferentialtariffsforthatproduct,andwiththatproductbeingincludedinmorePTAsformedaftertheconclusionoftheUruguayRound.

Finally, there isanemerging literaturewhichprovidesa systematic explanation of the timing of PTAformations and enlargements since the late 1950susingeconometricdurationanalysis.ThishelpsexplainthepatternofPTA formationdescribed inSectionB.For instance,Bergstrandetal. (2010)13 identify threesystematic relationships between the “timing” of PTAevents and different economic characteristics.Specifically, natural tradingpartners (countriescloserto each other in terms of physical distance), pairs ofcountrieswithlargergrossdomesticproducts(GDPs),andpairsofcountrieswhoseeconomicsize issimilar,have a higher probability of forming a PTA – orenlarginganexistingPTA–soonerthancountriesthatdonotshare these threecharacteristics.14Liu (2010)drawssimilarconclusions.

Bergstrandetal. (2010)alsooutlineconditionsunderwhich PTAs create the greatest incentives for non-


98

BoxC.1:PtA case studies

The WTO periodically examines the national trade policies of its members through Trade Policy Reviews(TPRs). Themember being reviewed submits aGovernment Report that is published alongside the reportprepared by the Secretariat. These official statements present the government’s perspective on majordevelopmentsinthecountry’stradepolicy,includingthenegotiationandconclusionofPTAs.AlthoughthereisnodefinedstructuretotheGovernmentReports,theyoccasionallyprovideinsightintothemotivesbehindpreferentialagreements.

Therearecertainlimitationstothisanalyticalapproach.GiventhateachmemberdecideswhattoincludeintheGovernmentReports,someexplicitlyaddress themotivationbehindpursuingPTAs,whileothersavoidmentioningitaltogether.Furthermore,severalgovernmentstendtorepeatparagraphsfrompreviousTPRstoexplaintheirtradepolicywithoutdescribingmotivesthatarespecifictonewPTAinitiatives.Therefore,thissurveyofGovernmentReportsismostlyanecdotalandfarfromexhaustive.

A survey of Government Reports shows that PTAs are predominantly about securing preferential marketaccess and attracting investment, as these are the most commonly quoted motives. However, an array ofadditional motives is also mentioned, in particular the goal of addressing policy issues that go deeper orbeyondWTOrules(seeSectionDforcontentsofPTAs).ItalsoappearsthatPTAsaresometimesusedasameansofpromotingdeepercommitmentsinnewareas,withtheaimofeventuallyincorporatingthematthemultilaterallevel.

Forexample,theUnitedStatesstatedinitsGovernmentReportthatPTAs“challengethemultilateralsystemtokeeppacewith the interestsandneedsofmembers,andcontribute to theWTOsystemby introducinginnovation and strengthened disciplines”, and that “these agreements can become models for futuremultilateral liberalization in new areas, such as agriculture, services, investment, and environmental andlabourstandards”(WorldTradeOrganization(WTO),2008).

members to join existing agreements or to form newones. First, the closer a potential entrant is to a PTAthatanothercountryisalreadyamemberof,themorelikely that the two countries will form a PTA sooner,consequently enlarging the PTA. Second, the higherthe“intensityofregionalism”acountrypairfaces,themorelikelyitisthatthetwocountriesformorenlargean existing PTA sooner. Third, there is a “hump-shaped”relationshipbetweenthenumberofmembersof the nearest PTA and the likelihood of it enlargingsooner. At first, the probability that two countriesenlarge an existing PTA sooner increases with thenumber of members of the nearest PTA – reflectingdemandformembershipbypotentialentrants.Beyondacertainthresholdlevelofmembershipsize,however,this probability declines as the utility loss from anexpansion for the potentially “worst-off” existingmember15 prevents infinite enlargement.16 This isimportantsincethespeedofregionalismhasappearedto be “much slower” than the apparent growth indemand for membership by non-members suggests,giventhedominotheoryofregionalism.17

Overall, Bergstrand et al. (2010) show that therelationships suggested by the six economiccharacteristics described above are sufficient toexplain 62 per cent of the variation across 10,585pairsofcountriesand57yearsofthetimingof1,560PTAevents.Furthermore,themodel isabletopredictthe actual year of the PTA formation or enlargementbyacountry-paircorrectlyinnearly50percentofthe1,560 PTA events. Liu (2010) also emphasizes theimportance of certain political variables in explaining

the timing of PTA formation. For example, the authorshows that countries with similar polity scores,18 lackof political hostility and a shared colonial history aremorelikelytoformPTAs.

Based on answers provided by WTO members in theTradePolicyReviewsundertakenbytheWTOSecretariat,Box C.1 contains a short discussion of the motivesmentionedbyWTOmembersforwhytheysignPTAs.

The above sections have covered in depth thedeterminants of the formation of preferential tradeagreements. However, little mention has been made ofthose agreements that have been negotiated amongcountriesbuthaveneverbeenimplemented.Forexample,intheearly1990sdiscussionswerebeguntoestablishaFreeTradeAreaoftheAmericas(FTAA).Thisenvisioneda hemispheric-wide free trade area in the continent.However, the initiativehas largelyfallenbythewayside.One way to look at the motives of preferential tradeliberalization is that they provide a demand-sideexplanationofthecreationandenlargementofPTAbutassumesthatthereisanunlimitedsupplyofmembership.It is important though toalsoconsiderwhatconstraintsare operating on the supply-side of preferentialliberalization.InthecaseofenlarginganalreadyexistingPTA,forexample,thesupplyofnewmemberswouldbedetermined at the margin by the potentially worst-offmember(Bergstrandetal,2010).Hence,theremightbesituationsinwhichthedeterminantsofthedemandandthe supply of preferential liberalization membership areso dissimilar that an agreement will very unlikely bereached.Thisissuemeritsfurtherresearch.


99

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

Similarly, the Government Report of Mexico acknowledged that PTAs “establish important precedents insomeareas thatcouldbe included in futuremultilateralnegotiations”, and thatMexicowould “continue tonegotiate regional trade agreements insofar as they go beyond multilateral liberalization” (World TradeOrganization(WTO),1997).

Political motivations that go beyond trade policy are also expressed in the official statements. SeveralGovernmentReportsexplicitlydeclarethatPTAsaimtopromotedemocracyandpoliticalstability.PeaceandsecurityisalsosaidtobeadvancedthroughtradecooperationinPTAs.

IntheTPRontheEuropeanCommunities(EC),theECplacesparticularemphasisonthepoliticalcooperationdimensionof its respectiveagreements.Forexample, in its region-to-regionnegotiationswith theAndeanCommunityandCentralAmericancountries,theEC“aim[ed]toreinforcethepoliticalandeconomicstabilityofeachregion”(WorldTradeOrganization(WTO),2009b).

CommentingonitsPTAwiththeEC,Chilealsoassertsthattheagreement“coversnotonlytradeissues,butpoliticalandcooperationareasaswell. In thepoliticalarea, theagreementseeks topromote,disseminateanddefenddemocraticvalues”(WorldTradeOrganization(WTO),2003).

ThelinkagebetweenpoliticalstabilityandpeaceismoreevidentintheEC’sagreementswithneighbouringpartners:“TheEuro-MedagreementsconcludedwitheightMediterraneancountriescontinuetobethebasisfor intensifying bilateral and regional co-operation in support of an area of peace, stability and sharedprosperity”(WorldTradeOrganization(WTO),2004).

Similarly, the US Government Report argues that the Dominican Republic-Central American Free TradeAgreement(DR-CAFTA)“supportsregionalstability,democracyandeconomicdevelopment”contributingtothe“transformationofaregionthatwasconsumedbyinternalstrifeandborderdisputesjustadecadeago”(WorldTradeOrganization(WTO),2006).

In severalGovernmentReports, theslowpaceatwhichmultinationalnegotiationsarecurrentlyadvancinghasbeenusedasajustificationforseekingPTAs.

The Government Report of Chile admits that “the pace of multilateral discussions is not rapid enough ... arelativelysmalleconomy likeChilehasvery limitedcapacity toexertany influence in theresolutionof theseproblems.Bilateral initiativesarethereforeusefulasasupplementarywayofachievingsubstantialoutcomesmoreexpeditiouslythanwouldbepossibleatthemultilaterallevel”(WorldTradeOrganization(WTO),2009a).

Thecontagionordomino-theory,wherebytheconclusionofaPTAactsasacatalysttotriggerotherPTAs,alsoappearstobeacentralmotive.ThereisevidencethatcountriesareconsciousoftheeffectsPTAshaveonthirdcountriesandthemultilateralsystem.Somecountries,suchasMexico,havepursuedPTAswiththeexplicitgoalofencouragingothertradingpartnerstonegotiatesimilaragreements.Othercountries,suchasPakistanandJapan,havereactedtotheproliferationofPTAsbyconcludingthattheyhavenochoicebuttocreatetheirownnetworkofPTAs(despitebeinginitiallyopposedtopreferentialliberalization).

AfterconcludingitsfirstmajorPTA,MexicostatedinitsGovernmentReportthatNAFTA“isveryimportantforMexico,notonlyowingtotheparticipationofitsbiggesttradingpartner...butalsobecauseitgeneratedan incentive and interest among other trading partners for negotiating similar agreements” (World TradeOrganization (WTO), 1997). This has been a successful strategy, considering that Mexico went on toconcludePTAswiththeEC,theEuropeanFreeTradeAssociationandJapanwithinadecade.

Fearing being left out of the preferential liberalization taking place outside the multilateral negotiations,countries such as Pakistan are “cognizant of the proliferation of regional and bilateral Preferential TradingArrangements”andhavereasonedthat“manysucharrangementsplacePakistaniexportersatadisadvantagevis-à-vis theircompetitors. Inorder tocounter thesenegativeeffects,Pakistanhasbeenactively involved inseekingsucharrangementsonbilateralorregionallevel”(WorldTradeOrganization(WTO),2007).

Inits2000report,Japanremained“seriouslyconcernedthatsomeRTAshaveraisedtradebarrierstotradewithnon-membercountries,andthattheyhaveeffectivelyweakenedthefree,non-discriminatory,andopenmultilateral system formed under the WTO”. It clarified it did not “belong to any preferential regionalagreements”butthatasaresultoftheproliferationofPTAs“thepossibilityandthedesirabilityoffreetradeagreements[were]beingexaminedbyvarioussectors”(WorldTradeOrganization(WTO),2000).Twoyearslater,initsnextTPR,JapannotedthatithadbeguntopursuePTAs(WorldTradeOrganization(WTO),2002).


100

2. ThestandardeconomicsofPTAs

(a) AnoverviewoftheeconomiceffectsofPTAs

Thebasiceconomiceffectsofpreferentialagreementscan be illustrated in a simple model (Baldwin, 2009).Consideraworldcomposedofthreeidenticalcountriescalled Home, Partner and Rest of the World (RoW).Each country imports two goods from the other twonations, and exports one good to both destinations.Thetradepatternsofthismodeleconomyaredepictedin Figure C.1 below. Further assume that in an initialsituation,allcountriesimposeoneachotherthesame(non-discriminatory) tariff, referred to as the Most-Favoured Nation (MFN) tariff. In this scenario, thedomesticpriceishigherthantheborderpricefacedbythe twosuppliersand importsare lowercompared toopen trade. Importantly, however, the two suppliersshare equally the reduction in exports due to theimpositionofanMFNtariff.

Whataretheeffectsofapreferentialtradeagreement?Tohelpanswerthisquestion,consider thecasewhereHomeandPartnerformafreetradearea(oracustomsunion),so thatPartnerproducersgetduty-freeaccessintheHomemarket,andHomeproducersgetdutyfreeaccess in the Partner market (a complete graphicalanalysisiscontainedinTechnicalAppendixC.1).

Focusingfirstonthemarketforgood1,thegoodthatisimportedbytheHomeeconomy,thefollowingpriceand volume effects take place. The domestic pricefalls relative to the situation where there is a singleMFN tariff as the supply of the good in the Homeeconomy is increased, but now thereare twodistinctborder prices. The border price faced by Partner ishigher,asexportersnolongerfaceatariffintheHomemarket, while the border price faced by exporters inRoWislower,astheystillfaceatariffbutthedomesticprice in the Home economy is lower. As a result,exportsfromPartnerexpand,whileexportsfromRoWcontract.

AsthePTAisreciprocal, theeffectsdiscussedaboveonthemarketforgood1materializesymmetricallyforgood 2. The only difference, intuitively, is that in thismarket the Home economy is an exporter, whilePartner is the importer. Therefore, in this market,Home gains from a higher border price and greaterexports toPartner, whileRoW loses from thedrop inborder price and the reduction in its exports insector 2. Finally, the formation of a preferentialarrangementhasnoeffecton themarket forgood3,whereRoWistheimporter,asthatcountryisassumedtomaintainthesameMFNtariff.19

APTAhastwotypesofeffectsontheexportside.First,exporters in member countries gain from improvedmarketaccessasthetariffisremoved.Secondly,theseexporters also benefit from the fact that tariffdiscrimination reduces imports from RoW. The lattereffectissometimesreferredtoasthe“preferencerent”,as itwouldnotexist if tariff liberalizationwerecarriedoutinanon-discriminatoryfashion.20

On the import side, the preferential agreement hasambiguouseffectsonmembercountries.Considerthemarket for good 1, where the Home economy is theimporter (the effects on Partner for good 2 areanalogous). The formation of the PTA has offsettingvolume and price effects.21 The increased importsallow the Home economy to benefit from thereplacement of high-cost domestic production withmore efficient imports. The terms of trade (i.e. thepriceofexports relative to imports)ofHome improverelative to RoW and falls relative to Partner. Overall,whether themembersofaPTAgainor losedependson the level of the initial MFN tariff and on theelasticities of demand and supply (i.e. to what extentthe demand and supply of a product is sensitive tochangesinitsprice).

AfinalconsiderationrelatestothewelfareeffectofaPTA on non-members. As discussed above, RoWsuffersa reductionof itsexports to thePTAmembercountries. In addition, the non-member is hurt by anegative terms-of-trade effect, as the price of itsexports declines while the prices of its imports areunaltered. In other words, a preferential agreementcanbe interpretedasanegativeexternality thatPTAmembersimposeonnon-members.

(b) Tradecreationandtradediversion

The formal analysis of the economic impact of PTAsbegan with the work of Jacob Viner in the 1950s(Viner, 1950). He asked whether a PTA would makemember countries better off, and concluded that thiswasnotnecessarilyso.Whilehisapproachdisregardedsome of the effects discussed above, it had animportant and enduring effect on the academic andpolicy debate surrounding preferential agreements.22AreviewoftheVineriantheoryis,therefore,usefultounderstandmuchofthedebateonPTAs.

FigureC.1:the PtA diagram’s trade pattern

Home

RoW

Partner

Good 1

Good 2

Good 1 Good 2

Good 3 Good 3


101

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

In this theory, preferential liberalization has two maineffects– tradecreationand tradediversion–and thenet balance between the two determines whether aPTA increases welfare for its members. As tariffs ontradebetweenpartnersfall,somedomesticproductionis replaced by imports from more efficient producersfrom partners – thus resulting in trade creation andwelfare gains. But since the PTA also discriminatesagainst non-members, imports from partners replaceimports frommoreefficientoutsideproducersand themember countries end up paying more for the samegood. This second effect which harms members'welfare is known as trade diversion. The interactionbetween trade creation and trade diversion hasdominatedmuchof thesubsequent literatureonPTAsand regionalism. Box C.2 provides a simple graphicalanalysis to illustrate tradecreationand tradediversioneffects.

BuildingonViner'sinsightintotheuncertainimplicationsof PTAs' effect on welfare, Kemp and Wan (1976)

foundtheconditionsthatwouldmakeacustomsunion–aPTAwithacommonexternalpolicy–necessarilywelfare-improving. They concluded that a customsunionwillbewelfare-enhancing ifexternal tariffsareadjusted so as to leave world prices unchanged. Inother words, if tariffs are such that external trade isnot affected, any additional trade between membersmust be trade-creating and outsiders are not hurt. Inthiscase,thePTAisParetoimproving.23Thisgeneralprinciple has been extended to other forms of PTAs:free tradeareas (PanagariyaandKrishna,2002)andpartialliberalization(Neary,2011).Furthermore,Kempand Wan also found that it is possible to guaranteethat all members of a PTA are better off if countriescan compensate losing members through lump-sumtransfers.Even if in reality theexternal tariffsarenotfully adjusted and lump-sum transfers are not alwayspresent,theKemp-Wanlogicisimportantfromapolicyperspective because it proves that PTAs are notnecessarilybadforworldwelfare.24

BoxC.2:trade creation and trade diversion effects

Consider a world composed of three countries: Home, Partner 1 and Partner 2, trading a homogeneousgood.AssumeHomeisasmallcountrythattakesinternationalpricesasgiven,whilePartner1andPartner2arelargeeconomies,meaningthatHomecouldsatisfyitsentirenationaldemandforthegoodbyimportingfromeitherof them. IfHomehasnoPTA inplaceandapplies thesameMFNtariff tobothPartner1andPartner2,itwillgetallitsimportsfromthemostefficientcountry.

FigureC.2belowshows thesupplyanddemandcurves forHome.The free-tradepricesof thegood fromPartner1andPartner2arerepresentedbyPBandPC,respectively.NotethatPartner1isthemoreefficientproducer,asitiscapableofsupplyingtheproductatalowerpricethanPartner2.WhenHomeappliesthesametarifftobothcountries,thedomesticpricesincreaseequallyforbothandaredenotedbyPB

TandPCT.

Undertheseconditions,HomewouldimportsolelyfromPartner1,atthepriceofPBT,aquantityofthegood

givenbythesegmentD1–S1.

ConsiderfirstthecaseinwhichHomesignsaPTAwithPartner1.Insuchasituation,importsfromPartner1arenolongersubjecttotariffsandthedomesticpriceofthegoodfallstoPB.Atthisprice,HomewillimportfromPartner1 thequantityD2–S2.Tomeasure theneteffectof thePTAonnationalwelfare,onemustanalysehowconsumers,producersandthegovernmentareaffected.

FigureC.2:Home PtA with Partner 1: trade creation

P S

D

S2 S1 D1 D2

PCT

PBT

PC a c

PB

Q

d b


102

Since, inthiscase,HomeconcludedaPTAwiththemostefficientproducer,theagreementresultsinpuretradecreation.Thegainsoftradecreationaremeasuredbytheshadedtriangles“b”,whichrepresentsgainsin production efficiency, and “d”, which represents gains in consumption efficiency. Consumers in HomebenefitfromthePTAbecausethedomesticpriceofthegoodfallsandconsumptionrises.Thereby,consumersurplus increasesbyareasa+b+c+d.Producersurplus isreducedbythearea“a”.AsthepriceoftheproductonthedomesticmarketdecreasesthroughcompetitionfromPartner1,somedomesticproducerswillbeforcedtoreduceoutputorclosedownaltogether.Governmentalsolosesallofthetariffrevenuethathadbeencollectedonimportsoftheproductdepictedasarea“c”inFigureC.2.Thus,theoverallneteffectofthePTAfornationalwelfareispositivewithagainofb+d.

Now,consider thecase inwhich thatHomesignsaPTAwithPartner2 instead. In thiscase, thepriceofimportsfromPartner2fallstoPC,whichisbelowtheimportpricefromPartner1.Atthislowerprice,HomeimportsfromPartner2ratherthanPartner1.FigureC.3belowshowsthat,bygivingpreferentialaccesstotheleastefficientproducer,thePTAresultsintradediversion.

BeforesigningaPTAwithPartner2,HomewouldapplythesameMFNtarifftoallforeignproducersanditwouldimportfromthemostefficientcountry,Partner1,thequantityD1–S1atthepricePB

T.WhenHomeconcludes the PTA, the price of goods imported from Partner 2 falls to PC while imports from Partner 1remainatPB

T.Asaresult,HomewillimportonlyfromPartner2thequantityD2–S2atthepricePC.Onceagain,tomeasuretheneteffectofthisPTAonnationalwelfare,onemustanalysehowconsumers,producersandthegovernmentareaffected.

After signingaPTAwithPartner2, as in thefirst case, consumers inHomearebetteroff andconsumersurplusgainscompound to theareaa+b+c+d.Note thatwhile there isstill some tradecreation, theefficiency gains in production and consumption – triangles b and d – are smaller than in the previousscenario.Also,domesticproducerssufferareductioninproducersurplusequaltoarea“a”andgovernmentlosestariffrevenueequalto“c”.Themaindifferencebetweenthetwocasesisintheshadedarea“e”whichrepresents trade diversion. This shaded area is the amount of trade the PTA diverts away from the moreefficientproducer,Partner1,bygivingpreferentialaccess toPartner2. Inotherwords,Homesuffers thisefficiencylossandpaysahigherpriceforimportsbynotadoptingopentradetowardsallcountries.

Tocalculatenationalwelfare,onemustbalancetheefficiencygainsagainsttheefficiencyloss.InFigureC.3,it isclear that thearea “e” is larger thanb+d; thus thePTAwithPartner2hasanegativeneteffectonnationalwelfareinHome.However,thisisnotalwaysthecase.ItispossiblethataPTAistrade-diverting,butnotwelfare-reducing, if thegainsfromtradecreationare largerthanthe lossfromtradediversion–e.g. ife<(b+d).

FigureC.3:Home PtA with Partner 2: trade diversion

P S

D

S2 S1 D1 D2

PCT

PBT

PC

a b c

e

d

PB

Q


103

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

(i) The effects of PTAs in services

Up to this point, the analysis has focused on thewelfare effects of preferential liberalization in goodstrade. However, given the increasing importance ofservices in PTAs, it is useful to analyse the welfareimplicationsofservicesliberalization.Doestheformeranalysis also help us to understand the effects ofPTAsinservices?

The crucial difference between trade in goodsliberalizationandtradeinservicesliberalizationisthatPTAs in services do not involve tariff reductions butchanges to domestic regulations, and the removal ofrestrictions on the movement of foreign investment.Although protection in services sectors may assume

several forms, they can be grouped into threecategories: (i) variable cost increasing measures(“frictional barriers”); (ii) fixed cost increasingmeasures; and (iii) quantitative restrictions on thenumberof foreign serviceproviders.While regulatorymeasuresareoftennon-discriminatoryinnature,thereareexampleswherethisisnotthecaseandcountriesemploymeasuresthatde factoliberalizepreferentially.

The effects of PTAs in services are illustrated inBoxC.3.Thisanalysis isbasedontheworkofMatooand Fink (2002). Focusing on the first category ofservices protection, the authors study the trade andwelfare effects of discriminatory services tradeliberalization.

BoxC.3: the effects of PtAs in services

Considerathree-countrymodelsimilartotheoneinBoxC.2,butassumenowthattheHomeeconomycanimpose(discriminatory)frictionalbarriers.Thissituationcanberepresentedbyassumingthequalityoftheservice composed by a universal standard (U) which is equal across countries and a country-specificstandard(Vi).Ifaforeign-servicesupplierwantstoprovideaserviceintheHomecountry,ithastofacethecostofmeetingthespecificstandardinthedomesticcountry(Ci)sothevariablecost increasesbyCiVi. ItmayalsobethecasethattheHomecountrydoesnotaccepttheuniversalstandardcomponentprovidedbytheforeignsupplier.Underthesecircumstances,iftheforeignsupplierwantstosellinthedomesticcountry,ithastofaceanadditionalcostofCi(Vi+U),becauseitneedstoadapttoboththeuniversalandthecountry-specificstandard.

Given this framework, theanalysisofdiscriminatory regulation inservices trade follows thesame logicastradeingoods.AssumethattheHomeeconomyissmallandthattherearetwoforeigncountries(Partner1andPartner2,respectivelyindicatedbysubscriptsBandC)potentiallyexportingservices.Asintheprevioussection,assumethatPartner1isthemoreefficientproducer.SupposethattheautarchypricefortheserviceisP*and that,before recognition, foreignfirmshave tomeet theuniversal standard in theHomecountry.InitiallythevariablecostbyforeignfirmsinthedomesticmarketisCi(Vi+U)+Chome(Vhome+U).Whenthiscost is higher than P* (for both Partner 1 and 2), no trade occurs. But if Home recognizes the universalcomponent of quality by Partner 2 as equivalent to the domestic one, Partner 2 faces a reduction in itsvariablecost,nowCc(Vc)+Chome(Vhome+U).IfthiscostislowerthanP*,weobservetradeinservicesfromPartner 2 to the Home country (see Figure C.4). In this case, discriminatory recognition (liberalization) isnecessarilytradecreating.

Assumenowthatinitially,whentraderestrictionsapplytobothforeigncountries,CB(VB+U)+Chome(Vhome+U)<P*<Cc(Vc+U)+Chome (Vhome+U)onlyPartner1 sells its services in theHomeeconomy (seeFigureC.4).IftheHomecountryrecognizestheuniversalstandarduprovidedbyPartner2asequaltothedomesticone, itmaybethecasethat theonlyexportingcountry isPartner2and importsarehigher thanbefore.ThisistruewhenCc(Vc)+Chome(Vhome+U)<CB(VB+U)+Chome(Vhome+U)<P*.

ThewelfareeffectofthediscriminatoryliberalizationontheHomeeconomycanbeseeninFigureC.4:thereisagain inconsumersurplus(a+b+c+d)partiallyoffsetby loss inproducersurplus(a).An importantpointhereistounderstandtheroleoftheareac+e.Inthetraditionaltradeingoodscase,theareac+eisawelfarelossforHomesinceitrepresentsthefallingovernmenttariffrevenue.However,inthiscontext,theareac+e represents theadditionalcost thatPartner1had to facewhen itsupplied theHomeeconomy(CBUtimesthepre-recognitionvalueofimports).IfthiscostdidnothaveanyeffectontheHomecountry(forinstance,intheformofaregulatoryrent),theareac+edoesnotenterintothecalculationofthetotalHomecountry’swelfare.Ontheotherhand,ifashare(s)ofthecostsustainedbyPartner1constitutedaformofregulatoryrent,thenetwelfareeffectofservicesliberalizationintheHomeeconomyisb+c+d-s(c+e).


104

FigureC.4:effects of PtAs in services

S1 D1 D2Q

S2

CB(VB+U)+Chome(Vhome+U)

CCVC+Chome(Vhome+U)

CBVB+Chome(Vhome+U)

CC(VC+U)+Chome(Vhome+U)

P

P*

D

S

e

c dba

(c) Naturalpartners,accumulationandlocationeffectsofPTAs

TheeffectsofPTAsstudiedintheeconomicliteraturego well beyond the ones discussed in this section.Below, we briefly summarize three areas of researchthat provide additional insights into the welfareimplicationsofpreferentialagreements.

It is possible that the trade effects of a preferentialagreementdependontheeconomiccharacteristicsofPTA members themselves. In particular, if tradeagreements are more likely to be signed betweencountries that trade intensivelywitheachother,PTAsshould generally be expected to be trade creating.This idea is often referred to as the “natural tradingpartners”hypothesis.

Krugman (1991) shows that the costs of preferentialtrade agreements formed between “natural” tradingpartnersare likely tobe lower than forarrangementsbetween countries that do not trade heavily with oneanother. He models a world where countries arespread over many continents and where variations ininter-continental transport costs determine whetherthe formation of regional trading blocs are globallywelfare-improving. If inter-continental transport costsarehighenoughtoensurethatthebulkoftradetakesplaceregionallyintheabsenceofPTAs,theformationof “natural” trading blocs within a region is welfare-improvingasthegainsfromtradecreationarelikelytooutweightradediversion.25Thevalidityofthe“naturaltrading partners” hypothesis is discussed in theempiricalevidencesubsectionbelow.

The effects of PTAs are not necessarily limited totraditionaltradeeffects(i.e.theallocationofresourcesin participating economies). Specifically, preferentialagreementsmayinfluencewelfareofmembercountriesthrough accumulation (i.e. economies of scale) andlocationeffects(BaldwinandVenables,1995).

Thetradecreation, tradediversiondebatefocusesonthestaticeffectsofPTAs.However,itisreasonabletoexpectthatpreferentialagreementswillhavedynamicimplications (i.e. that change over time). Theaccumulation effect considers how a PTA affectsgrowth. Itdoes this throughchanges in the returnoninvestment in member countries determined bychanges in physical capital and human capital(management and technical expertise)or by changesin technology available to firms. In a sense, theredistributionofcapitalflowsaftertheconclusionofaPTAcanbeseenasinvestmentcreationanddiversion.If capital is internationally mobile, it is possible thatthere will be an increase in capital inflows within thePTAattheexpenseofnon-members.Inaddition,thereisawidebodyof literaturethatstudiestheeffectsoftrade on long-run growth (World Trade OrganizationWTO,2008).Thisareaofresearchgenerallydoesnotconsider theeffectsofpreferential tradeagreementsas opposed to non-discriminatory trade opening.However, some of the mechanisms through whichtrade affects growth (international knowledgespillovers, enhanced competition, etc.) apply to PTAsaswellastomultilateraltradeliberalization.26

The location effect looks at how the integration of acountry into a PTA may alter the distribution ofeconomic activity within thePTAand thereby lead toinequality among member countries. When tradebarriers are reduced, firms can alter their locationdecisions. This decision depends on the balancebetween production costs and the trade costs thatmust be incurred to supply different markets. On theone hand, locations where economic activity is moreconcentrated can be efficient in the presence ofexternal economies of scale that increase firms'productivity.Ontheotherhand,proximitytoconsumersreduces trade costs, particularly when trade policyrestrictionsareinplace.BaldwinandVenables(1995)findthatastradecostsdecline,havingcloseaccessto


105

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

consumers becomes less important. Thus, during aprocessoftradeliberalization,firmswouldbedrawnto“central” areas within the PTA. This agglomerationeffect may exacerbate regional inequalities betweenmembersofanagreement.

(d) EffectsofPTAs:theevidence

Severalstudiesexamine the impactofPTAsand testthe traditional theories on trade creation and tradediversion. While this literature is not conclusive, itsuggeststhattradediversionmayplayarole insomeagreements and in some sectors, but it does notemerge as a key effect of preferential agreements(FreundandOrnelas,2010).27

A first branch of the empirical literature analysesspecific agreements and, using differentmethodologies,reachesmixedconclusionsintermsofthe net welfare effects of PTAs. For example, a firstsetofstudiesfocusontheCanada-UnitedStatesfreetrade agreement (CUSFTA). Clausing (2001) findsevidence that the agreement increased US importsfromCanada,butdidnotdivertUSimportsawayfromotherUStradingpartners.Similarly,theCUSFTAstudyby Trefler (2004) confirms the finding that tradecreation outweighs the trade diversion effect. Incontrast, a study of NAFTA concludes that theagreementisoveralltradediverting(Romalis,2007).28Romalisuseschanges inEU tradeover theperiod tocapture the counterfactual (i.e. what would havehappenedintheabsenceoftheagreement),butfindsthatthewelfarecostsofNAFTAaresmall.

ChangandWinters(2002)evaluatethewelfareimpactoftheSouthernCommonMarket(MERCOSUR)fromadifferentperspective,lookingattheeffectthecustomsunion (between Argentina, Brazil, Paraguay andUruguay)hashadonexportpricestoBrazil.TheyfindthatArgentina'sexportpricesincreasedwhilethoseofexcluded countries have declined, suggesting theagreement istrade-divertingandthat ithashurtnon-members. Finally, Egger (2004) finds that joining aregional trading bloc does not exert any significantshort-termimpactontradevolumes,butthatthereisaconsiderable trade creation effect in the long-run.Hypothetically,removingtheEuropeanEconomicArea(EEA) would account for a 4 per cent reduction oftrade within the EEA. A similar estimate for NAFTAyieldsareductionin15percentofvolumetrade.

Anotherbranchoftheempiricalliteratureusesgravitymodelstoinferthetradeeffectsofanagreement.Thekey question is to what extent PTA partners trademore than would be predicted by standard bilateraltrade determinants (e.g. income, geographicalproximity,etc.).Magee(2008),forexample,usespaneldata for 133 countries in the 1980-1998 period andincludes several fixed effects to capture thecounterfactual: what would happen to trade if therewere no PTAs. He finds that the average impact of

PTAson tradeflows issmall–only3percent–andthat, on average, trade creation exceeds tradediversion. In contrast, an earlier gravity-model studycovering130countriesfrom1962to1996foundthatPTAs have generated a significant increase in tradebetweenmembers,oftenattheexpenseoftherestofthe world, suggesting evidence of trade diversion(Carrere,2006).

Finally, focusing on East Asia, Lee and Shin (2006)find thatPTAs in the regionare likely tocreatemoretrade among members without diverting trade fromnon-members.BaierandBergstrand (2007)estimatethe impactofPTAson tradeflows, takingaccountofthe “endogeneity"29problem– i.e. thepossibility thatcountriesjoinPTAsforunobservablereasonsthatmaybe correlated with the level of trade. They concludethat when taking into account the endogeneity of aPTA,thepositiveimpactoftheagreementonbilateralflowsbecomesstatisticallymorerobustandfivetimeslargerthaninestimatesthatdisregardtheendogenityproblem.30 Thus, it appears that countries generallyopt forwelfare improvingPTAswhen therearegainsfromliberalizingbilateraltrade.

Acharya et al. (2011) analyse trade creation effectsbothwithinthePTAandoutsideofthePTAforanumberof preferential trade agreements. They find strongevidenceofintra-PTAtradecreation,showingthatPTAsincreasethevalueoftradebetweenmembercountries(for 17 out of the 22 PTAs considered). On the otherhand, they do not find evidence of trade diversioneffects. Differently from other studies in this area,Acharya et al. (2011) also consider the possible tradecreationeffectoutsideofthePTA.MostoftheanalysedPTAsincreaseexportsfrommembercountriestonon-member countries. In particular, they find very strongand positive effects regarding MERCOSUR and theASEAN Free Trade Area, with an increase of exportsoutsideof thePTAby109percentand136percentrespectively.TradediversioneffectsoutsideofthePTAhave been found in a number of cases, including theCaribbean Community (CARICOM), the CentralEuropeanFreeTradeAgreement(CEFTA),theCommonMarketforEasternandSouthernAfrica(COMESA)andtheCloserEconomicRelationsFTAbetweenAustraliaandNewZealand.

Athirdapproachintheempiricalliteraturehasbeentotestthe“naturaltradingpartner”hypothesis(Krugman,1991). Also using a gravity model and concentratingontheAmericas,Frankeletal.(1995)seektoidentifytrade diversion by testing whether regional trade isgreater than could be explained by naturaldeterminants of trade, such as proximity and marketsize. They find that multiple PTAs with partialliberalization among neighbours within a continentwouldraisewelfare,andthatthissituationispreferableto a single continental free trade area. Thus, in theirview, the formation of trading blocs, such as NAFTAand MERCOSUR, among “natural trading partners” is


106

preferabletothefailedFTAA(FreeTradeAreaoftheAmericas).AnopposingviewisheldbyBhagwatiandPanagariya(1996),whoarguethatthevolumeoftradeandtransportcostcriteria,testedbyFrankeletal.,arenotsufficienttoensurethataPTAwillraisewelfare.

Addressing the points brought up by Bhagwati andPanagariya, Krishna (2003) uses detailed US tradedata to estimate the welfare effects of hypotheticalbilateral PTAs. He finds that neither geographicalproximitynortradevolumesaresignificantlycorrelatedwithwelfaregains,concludingthatthesearenotgoodindicators for the formationofPTAs,as the literaturesupporting the “natural trading partner” hypothesissuggests. Baier and Bergstrand (2004) study whichpairofcountrieswouldgainmostfromformingaPTAand whether these country-pairs are more likely tosignapreferentialagreement.Theydevelopageneralequilibriummodelwithasampleof53countries,usingdata from 1996. Testing for several variables thatpredict 85 per cent of the bilateral PTAs in theirsample,31 their results support the natural tradingpartnerhypothesis.32

(e) ThepoliticaleconomyofPTAsandexternaltariffs

SectionC.1makesthepoint that theviewsofspecialinterestgroupsmayweighheavilyongovernmentsandthat apolicymakermay signaPTA toaccommodatetheinterestsofpowerfullobbygroups.Inthispoliticalcontext, can inefficient PTAs be signed (or efficientones be rejected)? More precisely, under whatconditionswillatrade-distortingPTAbeendorsedbyagovernment?Twoinfluentialstudiesaddressingthesequestions reached a similar conclusion in that trade-divertingPTAsaremorelikelytobepoliticallyviable.33

TheworkbyGrossmanandHelpman(1995)providesthe basic structure for the so-called “new politicaleconomy” literature in trade. The key idea, which isembodied in all models discussed in this section, isthattheinteractionofgovernmentsintheinternationalarenaisatwo-levelgame(Putnam,1988). Inthefirststage, the policy preferences of a government areshapedbynationalwelfareconsiderationsandby thepolitically organized groups that represent differentindustrial sectors. In the second stage, governmentsnegotiateaPTAundertheconstraintsimposedbythedomestic political environment. The outcome of thisgameisthepoliticallyviablepreferentialagreement.

A PTA naturally requires the assent of bothgovernments involved. The question is, therefore,under what domestic conditions is such commonalityof purposes more likely? As lobby groups tend torepresent producers' interests, one needs tounderstand how a preferential agreement affectsproducers. Consider first a trade-diverting PTA(e.g. theonebetweenHomeandPartner2describedin Figure C.3). In this case, the price in the Home

marketfallsbyasmallmargin,andexportersinPartner2 gain from the high domestic price in the partner'smarket.Hence,domesticimport-competingproducersare hurt slightly and would weakly oppose anagreement, while exporters in the partner countrybenefit largely and strongly support the agreement.Consider next the case of a trade-creating PTA (e.g.theonebetweenHomeandPartner1, inFigureC.2).Thedomesticpricefallssubstantiallyasaresultoftheagreement, the domestic import-competing sectorsuffers larger losses while foreign exporters receivelittle benefit. In this scenario, domestic politicaloppositiontothePTAisstrong,whileforeignsupportismarginal.34

TheworkbyGrossmanandHelpman(1995) isbasedon the assumption that markets are perfectlycompetitive (i.e. no supplier has sufficient marketshare to affect prices). A question, therefore, ariseswhether results would be different under imperfectlycompetitivemarkets. Inanoligopolisticsetting,wherea small number of producers dominate the market,Krishna (1998) shows that it is still true that trade-diverting PTAs are politically viable, while trade-creating ones are not. Intuitively, trade diversionincreases the oligopolistic incomes (rents) ofproducers in the partners' economies and, therefore,creates political support for the agreement.Specifically, Krishna (1998) posits that a politicalrequirement for a PTA is that aggregate profitsincreaseinthepartners'economies.Iftradeisdivertedaway from third countries, it is more likely that firmsfrom within the agreement gain market share in thepartner's economy (to the disadvantage of third-marketcompetitors)andincreasetheirprofits.35

In brief, these earlier works conclude that theconditions needed for the political viability of a PTAmaycontradictthosethatensureitssocialdesirability.Thesestudies,however,donotconsider thatexternaltariffs(i.e.thetariffthatPTAmembersimposeonnon-members) may respond to the formation of apreferential agreement. For instance, Richardson(1993) first made the point that countries may havereason to lower their external tariffs after entering aPTA. Importantly, removing this assumption mayradically change the implications of these models.Intuitively,consideringthegraphinFigureC.2,ifHomelowers theexternal tariff toPartner1after signingaPTA with the less efficient Partner 2, it is entirelypossiblethatthePTAwillstillbetrade-creating.36

Ornelas (2005a: 2005b) revisits the Grossman-Helpman and Krishna theory, which deals with thesituationwheretheexternaltariffisallowedtochangeafteraPTAhasentered intoforce.Specifically, thesepapers allow tariffs on third countries to be set“endogenously”, that is, in a way that allows specialinterest groups to influence policy both before andafter an agreement is signed. Ornelas shows thatindependentlyofthestructureofmarkets(i.e.perfectly


107

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

competitive or not), welfare-decreasing preferentialagreements are unlikely to be politically viable.However,Ornelasshowsitisstillpossiblethatspecialinterest pressures may persuade governments not tosignsomepreferentialagreementsthatwouldimprovesocialwelfare.

Thestartingpoint for anaccurate characterizationofthesefindingsistoconsiderthepoliticaldeterminantsof external tariffs. The political demand for externalprotection is lower under a preferential agreement.AfteraPTAisformed,thedomesticimport-competingsector losesmarketshare to thepartners'producers.Inthisnewenvironment,anyincreaseinthedomesticprice thatmayresult froman increase in theexternaltariffbenefitsdomesticproducerslessthanitwouldifaPTAwasnotinplace.Thereasonisthattheexternalprotectiongrantedbythetariff“leaks”toPTApartnersand only partly benefits domestic producers.37 Putdifferently, the incentive of import-competing sectorstodemandprotection is stronger in theabsenceofaPTA, as their share of the domestic market is larger.Thisistruebothforperfectlycompetitiveproducersaswell as for oligopolistic firms. Moreover, the cost oflobbying is not changed under a PTA, as this stillreflects the cost of the external tariff to society atlarge.

The above reasoning has the following implications.First,aPTAweakens the impactofpoliticaleconomyforcesonexternaltariffsinequilibrium.AsthedemandforexternalprotectionfallsunderaPTAwhileitscostis unaltered, the external tariff is predicted to fall.Secondly, if preferential agreements destroyprotectionist rents, political support of organizedsectors cannot be a strong rationale for a PTA.Politicallyviableagreementsmust,therefore,bethosethatimproveaggregatesocialwelfare.

To some extent, these recent works on the newpolitical economy of preferential agreements shouldbe seen as complementary. Grossman and Helpman(1995) and Krishna (1998) focus on the decision tosignornotaPTA,buttheydonotexaminetheeffectthataPTAhasonexternaltariffs,whichisinsteadthefocusofOrnelas (2005a:2005b). If special interestscouldbothlobbytoinfluencethetraderegimedecisionaswellasthetariffformation,Ornelas'findingswouldbe qualified. In this scenario, trade-divertingpreferential agreements can be politically viable.However,thisnegativeoutcomeisnotaslikelyasonemightthink,asthepoliticalrentdestructioncausedbya PTA reduces governments' incentives to endorsewelfare-reducing agreements (Freund and Ornelas,2010).

Thenewpoliticaleconomyliteraturehasalsoraisedarelated but distinct question. A number of PTAs gowell beyond tariff arrangements and include “non-trade” issues, such as labour or environmentalstandards, provisions on intellectual property rights

and several other areas. As the next subsectiondiscusses more extensively, there are a number ofreasonsthatjustifythesedevelopments.Thequestionaddressed here is not on the economic rationale forsuch arrangements, but rather whether one shouldexpect external tariffs to fall when preferentialagreements encompass more than the lowering oftariffs.

Limão(2007)providesaneconomicmodelthatallowsan analysis of the importance of non-trade issues inPTAs,andtheireffecton incentives to lowerexternaltariffs. Specifically, he argues that, if preferentialagreements include non-trade issues rather than justtariff reductions,governmentsmaybemore reluctantto reduce external tariffs. The reason is that a PTAmaybevaluable toacountrypreciselybecause tariffreductionsencouragecooperationonothernon-tradeissues.However, inthiscase,agovernmentmayhavelittle appetite to reduce tariffs on third-countryimports, because a reduction in the external tariffswould lower the preference margin to partners andthusweakentheagreement.38

Ultimately,theeffectofPTAsonexternaltariffs isanempiricalquestion.However, the literatureappears tobediscordant.Inafirstsetofstudies,Estevadeordaletal. (2008) and Calvo-Pardo et al. (2009) find thatpreferentialagreementsinLatinAmericaandASEANcountries had the effect of reducing external tariffs.Specifically, they find that external tariffs declinefaster in those sectorswherepreferenceshavebeengrantedand that,contrary toprevailingopinion, thereis little evidence that preferences lead to higherexternal tariffs. In a second set of studies, Limão(2007)andKaracaovali(2008)showthattheoppositepattern emerges from an analysis of PTAs signed bytheUnitedStatesandtheEuropeanUnion.

While these contrasting empirical findings suggestthatmoreanalysisisneededinthisarea,theymaybeless controversial at a closer look. Specifically, thedifference in the sample of countries analysed mayexplainpartof thedifferences.PTAssignedbetweendeveloped and developing countries, such as thosesignedby theEuropeanUnionand theUnitedStateswith developing countries, may be more likely toinclude provisions that go beyond the lowering oftariffs than agreements between two developingcountries.Asthisisgenerallythecase(seeSectionB),itisnotsurprising,inlightofthetheory,tofindthatthePTAs between developed and developing countriestend to increase external tariffs, while agreementsbetweentwodevelopingcountriesarelikelytoreducethem.


108

(f) Rulesoforiginandtradediversion

(i) Rules of origin: a source of trade diversion

In PTAs which are not customs unions, membersmaintaintheirownexternaltariffs.Consequently,intheabsence of any rules, imports of particular productswould enter the country in the PTA with the lowestimportdutyontheiteminquestionandbere-exportedto other countries in the PTA. Hence, rules whichconfirm the true “origin” of the goods are required topreventsuchre-routingofgoods–or“tradedeflection”.For example, suppose the preferential tariff on theexportsofcountryAtocountryBiszero.Hence,whencountry A exports the good to country B, the latterneeds toensure that thegood reallydoesoriginate incountry A, and is not simply being re-routed viacountry A by some third country which does not havethesamedegreeofpreferencesincountryB.Empiricalevidencesupportsthishypothesisrelatingtotheroleofrules of origin (RoOs)39 in preventing trade deflection.Forinstance,Estevadeordal(2000)findsthatthehighertheabsolutespreadbetweenMexicanandUStariffstothirdparties,thehighertherestrictivenessbuiltintotheRoOsofNAFTA.Inreality,however,RoOsmaybeusedto protect certain favoured industries, thereby leadingto trade diversion or trade suppression (Krishna andKrueger,1995).

Considerthefollowingscenario.AssumeaproductionsharingnetworkbetweencountriesBandC,wherebycountry B exports a final good to country A usingintermediate goods from country C. Furthermore,assume that country A is a high-cost (relative tocountryC)producerofintermediategoodsusedintheproduction of this final good which is exported bycountryBtocountryA.Initially,countryAsignsaPTAwithcountryBandanotherPTAwithcountryC.Hence,agoodproduced inBwouldhavepreferentialaccessto A, as would a good originating in C. Under thenegotiated PTA, country A could impose stringentRoOsoncountryBwiththeresultthatthefinalproductthatcountryBexportstocountryAmaynotqualifyasoriginatingthere–perhapsbecausetheproportionofintermediategoodsfromCistoohigh.Hence,thefirmin country B can either continue to import theintermediate good from country C and not gainpreferential access to countryAor shift its purchaseof the intermediate good from C to A, in order tosatisfy the RoOs and obtain preferential access ontheirexportstocountryA.

Inotherwords,restrictiveRoOsmaymakeitprofitableforfirms incountryB toengage in “supplyswitching”by using a more expensive intermediate good eitherfromcountryAoradomesticfirm,i.e.restrictiveRoOsin final goods divert or supress trade in intermediategoods. Supply-switching strengthens the trade linkbetween countries A and B (hub-spoke), at theexpenseof tradebetweencountriesBandC (spoke-

spoke),i.e.countryAbenefitsbyusingRoOstoprotectexports of certain industries (Gasiorek et al., 2009).Furthermore, by influencing the sourcing ofintermediategoods trade,RoOsare likely to increasefirms'costsandhencehaveanadverseeffectonfinalgoods trade. This increase in cost strengthens the“spaghettibowl”effectofPTAsanalysedinSectionB.Hence, supply-switching – or the non-utilization ofpreferences,asa resultofRoOs– reduces the tradeliberalizingimpactofPTAs.Analysingimportdataforasampleofmore than150countriesduring theperiodfrom 1981 to 2001, Estevadeordal and Suominen(2008) find that restrictive product-specific RoOsencourage the trading of intermediate goods withinthe PTA (thereby leading to trade diversion) andundermineaggregatetradeflowsamongPTApartners.

Inasurveyof345firmsinfourLatinAmericancountriescarried out by the Inter-American Development Bank(IADB) in 2007-08, fewer than 10 per cent reportedhavingchanged theirsupplychain inorder toadapt torules of origin (Harris and Suominen, 2009). Thissuggests that most firms continue to import from thesame source as before, even if this means foregoingpreferentialaccesstotheirPTApartnercountrymarket.Among the multi-national corporations (MNCs) in thesample, however, about 75 per cent (ranging from 50percentinPanamatonearly90percentinColombia)described RoOs as an important factor in determiningwhere to invest inproduction facilities.However,whenasked whether investment in a subsidiary was madeexplicitly tomeetRoO requirements inoneormoreofthecountry'sPTAs,thefigurefallsto lessthan30percent40 (Harris and Suominen, 2009). This firm-levelevidencesuggeststhatforMNCs,whichrelyheavilyonflows of intermediate goods trade via productionnetworks, RoOs significantly affect investmentdecisions.Inparticular,firmsmayswitchtheirsourceofintermediategoods fromamoreefficient supplier inanon-member country to a less efficient supplier in amember country (where they establish productionfacilities),therebyresultingintradediversion.

(ii) Reducing such trade diversion: the way forward

The hypothetical scenario described above showedthat the final good originating in B has preferentialaccesstoA,asdoestheintermediategoodoriginatinginC.However, thefinalgood fromB,producedusingintermediate goods from C, which does meet rulesgranting originating status for B’s exporters to C,would not be eligible for preferential access. Such asystem of bilateral hub-spoke agreements withconstraining rules of origin is thus likely to enhancehub-spoketradeattheexpenseofspoke-spoketrade.

Gasiorek et al. (2009) have argued that thisdiscrimination, which protects the exports of certainindustries in country A and hence leads to tradediversion, can be resolved if country B signs a PTA


109

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

withcountryCandisthereafterallowedtoadditsownintermediateinputs(valueadded)withtheintermediateinputsfromcountryCindeterminingoriginatingstatuson theexportsof thefinalproductsold tocountryA.Thisistheprincipleof“diagonalcumulation”ofrulesoforigin. Under this arrangement, all participatingcountries agree bilaterally that in all PTAs concludedamongthemselvesmaterialsoriginatinginonecountrycanbeconsideredtobematerialsoriginatinginalltheother countries. This makes it easier to importintermediategoodsandstillsatisfytheRoOs.

Diagonalcumulationappliestotradebetweenthreeormore trading partners normally linked by PTAs withidentical RoOs. It builds on the concept of “bilateralcumulation” – materials originating in one country canbe considered as materials originating in the otherpartner country – which is a feature of all PTAs. Inaddition, there is the concept of “total cumulation”,which again applies to trade between three or morecountries,but involvesgreaterflexibility than “diagonalcumulation”. This is because it allows intermediateprocessingtobesplit inanywayamongallthepartiesto the PTA, provided that when added together, thecumulative processing is sufficient to meet the originrule.Inthecontextofourhypotheticalscenario,supposeforinstancethattheintermediategoodfromcountryCdoesnotqualifyasoriginatinginthatcountry.Withtotalcumulation,theproducerincountryBcancumulatethe

proportionofcountryC’svalueaddedtogetherwithitsownvalueaddedindeterminingoriginatingstatus.

Althoughtotalcumulationisrare,diagonalcumulationhas been used by some PTAs. The EU is a goodexample in this regard.BoxC.4providesanoverviewoftheEUexperienceinrelaxingRoOsinPTAs.

3. Goingbeyondthestandardanalysis

As shown in Section B and Section D, over the pastthree decades trade agreements have gone beyondbordermeasures,suchastariffs,andhave integrateda number of domestic policies and regulations,including intellectual property rights, productstandards,competitionandinvestmentpolicies.Thesedevelopmentsarenotinconsequential;oncetariffsareremoved, differing regulatory policies among nationsbecomemoresalient, creatingcomplexchallengesofaccommodation and coordination. Moreover, tradeopenness – along with the new forms of trade thattechnological development makes possible – createsnew pressures to reconcile divergent nationalpractices, and generates new forms of cross-borderpolicy effects (spillovers). These developmentsproducedemands forgovernanceand the ruleof lawthattranscendnationalborders.

BoxC.4: Lessons from the eu experience in relaxing rules of origin (Roos)

For theEU, the issueofmultipleRoOsbecame increasinglysignificant in the1990s,asagreementswereconcludedwithanumberofcountriesfromCentralandEasternEuropeandfromtheSouthMediterranean.ItbecameapparentthattheEU’s“spaghettibowl”ofcriss-crossingagreementswasrestrictingfirms’abilitytosourceintermediategoodsfromthecheapestsource,i.e.therewastradediversion(Gasioreketal.,2009).

Toaddressthisproblem,thePan-European(PANEURO)CumulationSystem(PECS)waslaunchedin1997.It established identicalprotocols forproduct-specificand regime-wideRoOsacross theEU’sexistingandfuturePTAs.ThisincludedarrangementswiththeEuropeanFreeTradeAssociation(EFTA)countries,datingfrom 1972 and 1973, as well as those forged in the 1990s and later – i.e. PTAs with several EasternEuropean countries, the Euro-Mediterranean Agreements, the Stabilization and Association Agreementswith Croatia and FYR Macedonia, as well as extra-regional PTAs with South Africa, Mexico and Chile(EstevadeordalandSuominen,2004).Hence, “diagonalcumulation”wasakeyprinciple introduced inpan-European rules. It enabled producers to use components originating in any of the participating countrieswithoutlosingthepreferentialstatusoffinalproduct.

Empirical evidence reveals that the harmonization of RoOs, via diagonal cumulation in the PECS, hasimpactedtradeflowssince1997.For instance,analysingthetextile industry,Augieretal. (2004)findthattradebetweennon-cumulatingcountriescouldbelowerbyupto50to70percent.Similarly,usingdataontradeflowsbetween38countries for threebaskets– trade inallgoods, trade in intermediategoods,andtradeinmanufacturedgoods–Augieretal.(2005)showthattradebetweencountriesthatbecamepartofthepan-Europeansystemofdiagonalcumulationwashigherrelativetotradewithothercountriesbyabout43 per cent between 1995 and 1999. In addition, they show that the introduction of the PECS in 1997increasedtradebetweenthespokesby7and22percent.However, theirmethodology isbasedonusingdummyvariablesinagravitymodeltocapturetheroleofcumulation.Hence,itispossiblethatthesevariablesarecapturingotherfactors.

Atthesametime,analysingdataontradeflowsbetween38countries,Gasioreketal.(2009)findthatthetradebetweennewlycumulatingcountries (following the introductionof thePECS in1997) risesbymorethantradebetweenthesecountriesandthirdcountriesforsomeselectedindustries.41


110

The following subsection looks at the new forms oftradeagreementsthatareemerging,usingtheconceptof “deep” integration (Lawrence,1996),andasks twomain questions. First, what are the motives behindthese agreements? Secondly, what determines thestructureofdeeperarrangements?Answers to thesequestionsareessentialtounderstandingtheeconomiccostsandbenefitsofdeeperintegration.

(a) Theconceptofdeepintegration

Trade agreements that deal mostly with bordermeasuresareoftendefinedas “shallow”agreements.On thedomesticside, theseagreementsaccordnon-discriminatorynationaltreatmenttoforeigngoodsandfirms (i.e. the same treatment that is accorded todomestic firms), but stop short of intervening indomestic economic policies beyond this requirement.In contrast, trade agreements that include rules ondomestic policies that fall “inside the border” arereferred to as “deep” agreements (Lawrence, 1996).There is no agreed definition of the scope of suchdeep agreements, and indeed the concept is widelyused to refer to any arrangement that goes beyondsimply extending preferential tariff concessions.However, thereareat least twodistinctdimensions–the“extensive”and“intensive”margins–toanydeeperintegrationagreement.

Thefirstdimensionreferstoincreasingthecoverageofan agreement beyond the lowering of tariffs (e.g. theharmonization of national regulations in financialservices).Mostdiscussionsofdeepintegrationfocusonthis dimension. The second dimension, the intensivemargin of deep integration, refers to the institutionaldepth of the agreement, such as the extent to which

certain policy prerogatives are delegated to asupranationallevelofgovernment(e.g.theformationofa customs or monetary union). These two dimensionsareoftenrelated.Thatistosay,extendingthecoverageof an agreement may also require creating commoninstitutionsandnew,moresophisticatedwaysofsharingsovereignty in order to administer it. The table belowprovidesaschematic(butnotexhaustive)pictureofthediverseformsofintegration.42

Like shallow integration arrangements, deeperagreements can be among advanced economies(North-North), advanced and developing economies(North-South), or just developing economies (South-South). Similarly, membership in deep integrationarrangements can be wide or narrow, ranging fromregional agreements involving several neighbouringcountriestobilateralagreementsbetweentwodistantpartners.43

(b) Whyisdeepintegrationgainingmomentum?

Deep economic integration and trade are intimatelyrelated (see Table C.1). Deep arrangements may benecessarytopromotetradeincertainsectorsoracrosseconomiesmorebroadly.Forinstance,harmonizationofcertain regulations may be a prerequisite for trade inservices or common competition policy rules may berequired toallowcomparativeadvantage tomaterialize(see Section D.2(b)). Conversely, trade liberalization –and the evolving structure of trade (for example, thegrowth of production networks) – can make the needfor deeper policy integration more pressing. In short,shallow and deep integration can be complementaryprocesses, as the first generates a demand for

TableC.1:shallow versus deep integration

Integration level type of PtA Features example

SHALLOWINTEGRATION

DEEPINTEGRATION

Freetradeagreement(FTA)Membersliberalizeinternaltradebutretaintheirindependentexternaltariffs

US-IsraelFTA

FTA+

AnFTAthatinadditionharmonizessomebeyondtheborderstandards(e.g.environmentalstandards)

NAFTA

CustomsUnion(CU)

Membersliberalizetradewithintheunionandadoptcommonexternaltariffsagainsttherestoftheworld

SACU

CommonMarket

EstablishmentofthefreemovementofallfactorsofproductionwithinthePTA,includinglabourandcapital

EU

MonetaryUnion

Establishmentofacommoncurrencyandcompletelyintegratedmonetaryandexchangeratepolicy

EuroArea

FiscalUnionEstablishmentofacommonfiscalpolicy

US

Note:ThedepthofintegrationofPTAsmightoverlapacrosstypesofagreementsincertaincircumstances.


111

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

governancethatthelattercanprovide.Thisrelationshipisunderscoredintheeconomicliterature.

A number of authors argue that markets need non-marketinstitutions(political,legalandsocial)iftheyareto function properly (Casella, 1996; Casella andFeinstein,2002;Padoa-Schioppa,2001;Rodrik,2000).These non-market institutions are essentially publicgoods that the market itself fails to provide. Othersmake the point that trade openness increases policyexternalities, rendering unilateral decision-makinginefficientcomparedwithcooperativedecision-making(BronerandVentura,2006;EpifaniandGanica,2006;BrouandRuta,2010;AntràsandStaiger,2008).

Insum,therelationshipbetweendeepintegrationandtradeworksbothways–inthesensethatonemaybethe cause and/or consequence of the other. Therelationship is also dynamic – in the sense that it islikely to develop over time. The remainder of thissection focuses on international production networkswhich exemplify the complementarity between tradeand governance that lies at the root of the currentproliferationofdeepagreements.

(i) International production networks and deep integration

Twenty-first century trade, as defined by Baldwin(2010), is a much more complex phenomenon thantradepriortotheearly1980s.44Thiscomplexityistheresultoftheincreasedroleofinternationalproductionnetworks in the global economy, which arecharacterized by the unbundling of stages ofproduction across borders. Increasingly, multinationalfirmsarenotonlydistributingmanufacturingstagestodecrease costs and exploit comparative advantages;they are also unbundling and outsourcing serviceswork,primarilyoffice tasks,makingglobalproductionnetworksevenmoresophisticatedandcomplex.

These new forms of international trade requirereconsiderationandreconceptualizationofpreferentialtrade. Most of the PTA models above assume thatcountries trade final goods and that producers areprotection seekers for these goods. However, theremight be some economic sectors, increasinglydependentonimportedintermediateinputs,thatseeklower levels of protection to reduce their productioncosts (Yi, 2003). Some empirical evidence suggeststhatwhencountrieshaveasignificantnumberoffirmsinvolvedinproductionnetworksthereismorepressureforunilateraltradeliberalization.45

Forsimilarreasons,countriesthatformpartofsupplychains involving multiple nations might be moreinclined tosignPTAswith their tradingpartners thanto unilaterally liberalize. As various stages ofproduction may take place in a number of differentcountries,theeffectsoftradebarriers,suchastariffsorothernon-tariffbarriers,onthecostofaparticular

stage of production is proportional to the number oftimes the product crosses other national boundaries.Inaddition,countriesmaysignPTAsinordertosecureor“lockin”tradingrelationships,thusreinforcingtheirpositionasthemainproviderofintermediateinputs.

Theoreticalconclusions regarding thewelfareeffectsofpreferentialtradeliberalizationalsochangewiththepresenceofproductionnetworks.Infact,internationalproduction sharing can mitigate the trade-diversioneffects of PTAs.46 The possibility of dividing up theproduction of final goods into various stages orcomponents alters the calculation of trade creationand tradediversionand,although theoutcome isstilluncertain, it leaves room for welfare-reducing PTAs,that trade only in final goods, to become welfare-improving PTAs, once members engage in trade ofpartsandcomponents.47

International production networks are not a newphenomenon, but their relevance is increasing inparticularregionsoftheworld(seeBoxC.5),andtheirpattern and composition has changed over time.Initially,countriesengaginginproductionsharingweremainlyrichcountries.48Fromthemid-1980s,however,production networks between developed anddeveloping countries started to increase (seeSectionB.3).

Is there any link between the recent growth ofproduction networks and the demand for deeperagreements? The theoretical and empirical literatureon FDI and offshoring highlights that despite thebenefitsofexploitingfactorpricedifferencesandnewtechnologicaldevelopments,thereareadditionalcostsofinternationalfragmentationofproduction–fromthemanagerial and logistic costs associated withmonitoring and coordinating international productionto learning about the laws and regulations that arerequired to do business in another country. Thesecostsmightbeparticularlyhighfordevelopingnationswhich are part of North-South production networks,and that may lack the kind of sophisticated businesslawsandtheproductandlabourregulationswhichrichcountriesusetoconsolidatetheirtradeinintermediategoods(Baldwin,2010).

Inthiscontext,theexpansionofproductionnetworks–and in particular of North-South production-sharing– should be related to the proliferation of deepagreementsaimedatfillingagovernancegapbetweencountries. Agreements that include provisions relatedto the institutional framework, competition policy, theproduct and labour-market regulations, infrastructuredevelopment,andotherareascouldmakeproduction-sharing activities more secure and less vulnerable todisruptionsorrestrictions(Yeats,2001).

This pattern can be observed in agreements such asNAFTAwhichnotonlyincreasemarketaccess,throughtariffreductions,butalsoincludedisciplinesthatreduce


112

therisks–andincreasetheprofitability–ofinvestmentin Mexico. Also the recent accession of easternEuropeaneconomiestotheEuropeanUnion,aswellassomeoftheeuro-Mediterraneanagreements,couldbepartlyexplainedasaresponsetothedemandfordeepintegration agreements associated with expandinginternationalproduction-sharing.

TheevolvingnatureoftradeagreementsinEastAsia,whereasignificantandgrowingshareofinternationalproductionsharingtakesplace,alsohighlightsthelinkbetween production networks and deep integration(seeSectionD.3 foramoredetailedanalysis). In thisregion, the growth of production sharing first tookplace through de facto economic integration.53However,more recentNorth-Southagreements, suchas Japan's economic partnerships with Malaysia,Indonesia, Thailand and Viet Nam, or ASEAN's pushfordeeperdisciplines,clearlyshowthat this region ismovingtowardsdeeperintegration.

Lawrence(1996)wasthefirsttohighlightthesystemicimplications of international production networks anddeep integration. With increased internationalcompetitionflowingfromreducedbarrierstotrade,theability to operate abroad – and to locate complex

production in the most cost-efficient regions –becomes increasingly important to firms'competitiveness. In order for cross-border productionnetworkstooperatesmoothly,certainnationalpoliciesneed to be harmonized across jurisdictions. Thisgeneratesademandfordeepformsofintegration.

The trade literature has largely failed to model theinteraction between international production networksand deep integration. One significant exception is therecent work by Antràs and Staiger (2008). They showthat the rise of offshoring creates new forms of cross-borderpolicyeffectsthatgobeyondthestandardtradepolicyexternalities,whengoodsareproducedinasinglelocation(i.e.theterms-of-tradeeffect).54Inthiscontext,theobjectiveoftradeagreementsismorecomplexthanthe standard theory would suggest, as negotiatingmarketaccessisnotsufficienttoaddressdistortionsofunilateral policy-making. An implication of this model isthat the changing nature of trade (from trade in finalgoods to trade in intermediate goods) is directlyresponsibleforthegrowingdemandfordeepagreementsthat can address these new cross-border effects.Specifically, externalities associated with productionoffshoring are different from those associated withtraditional market access, and cannot be easily

BoxC.5: Determinants of the regionalization of production networks

Standard elements of comparative advantage, such as variations in labour supply conditions, wages, orrelativefactorendowments,helpexplainnotonlytheproliferationofNorth-Southproductionnetworksbutalsotheregionalizationofsuchnetworks.StudiesbyAthukoralaandMenon(2010)ofEastAsia,forexample,showthateventhoughwagesinChina;HongKong,China;theRepublicofKorea;andChineseTaipeihavebeenrapidlyapproachingdeveloped-countrylevelsinrecentyears,wagesincountriessuchasMalaysia,thePhilippines,ThailandandVietNamremainlowerthan–orcomparableto–wagesinMexicoandcountriesonEurope’speriphery.

The role of distance is also important in explaining the regionalization of production networks. Severaleconomists have pointed out that despite technological advancements, distance still matters and certaincountries still suffer from geographic remoteness (Venables, 2001).49 In addition, there is evidence thatgeographicaldistanceremainsakeyfactorindetermininginternationaltransportcosts,especiallyshippingcosts,anddeliverytime(EvansandHarrigan,2005).Arguably,thesetypesofcostsareparticularlyrelevantforproductionnetworks,whereagoodcancrossbordersseveraltimesinthevariousstagesofproduction.

New geography models of economic agglomeration at the international level are also useful in explaining theregionalizationofproductionsharing.Accesstointermediategoodscreatesagglomerationofproduction,asfirmsgainfrombeingclosetocustomerandsupplierfirms.50Asmoreandmorefirmsmovetoacertainregion,theycreateademandforsuppliersofintermediategoodsandservices,reinforcingtheoffshoringattractivenessofthatregion for other firms in the industry and related fields. In addition, because production networks are formedaroundcentresofeconomicactivity,thedistancebetweentheseproductioncentresandtheperipherymatters.51

SchatzandVenables (2000)showthatmajoroutward investorscarryoutmuchof their investment,whichreliesheavilyonintermediategoodstrade,closetohome(theUnitedStatesinvestinginMexico;theEUinCentralandEasternEurope;Japan inAsia)and this trendcapturesan importantshareofFDIflows fromdevelopedtodevelopingcountries.52

InthecaseofEastAsia,AthukoralaandMenon(2010)findthattheregionhasbenefitedfroma“first-mover”advantage in hosting assembly operations of multinational corporations. Established companies haveattractedotherkeymarketplayersand, in turn,manyhaveupgraded the technologyemployedby regionalproduction networks and assigned greater global production responsibilities to local affiliates, reinforcingtheagglomerationeffects.


113

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

addressedwithgeneralrules,suchasnon-discriminationand reciprocity (Bagwell and Staiger, 2003). If thisargument is correct and the GATT/WTO system is notwelladaptedtohandlethesenon-marketaccessissues,countriesmightturntootheravailableinstruments,suchasPTAs,tosolvetheircoordinationproblems.

This presents the multilateral trading system with adifficult challenge. The recent wave of preferentialagreements may (at least in part) be an institutionalresponse to the new problems associated with thegrowth inoffshoring.On theonehand, this suggeststhat PTAs are efficiency-enhancing rather thanbeggar-thy-neighbour agreements.55 On the otherhand,PTAsmaymakeitmoredifficultfortheWTOtoperform its traditional role of providing reciprocalmarket access opening. In essence, the institutionalchallengefortheWTOistofindanapproachthatcanfacilitate the deeper integration that countries areseeking while at the same time upholding the coreprincipleofnon-discrimination.

(c) Thetrade-offsinvolvedindeepintegration

Unlike shallow integration, deep integration –regardless of the form it takes – requires commonpolicies and regulations among member countriesacross anumberof areas.56This raisesa completelydifferent set of questions. What are the costs andbenefitsofcommonpolicies?Whichcountriesshouldformadeepagreement?Whichpoliciesshouldremainin the national domain, and which should beharmonizedat–orassignedto–asupranationallevelof government? These questions are traditionallyaddressedinpubliceconomics,andhavegeneratedanextensive literature, mainly focused on fiscalfederalism,whichisbrieflyreviewedbelow.57

Economists have developed a simple principle tounderstandthecostsandbenefitsofcommonpolicies,knownastheOates'DecentralizationTheorem(Oates,1972). This theorem suggests that there is a basictrade-off between the benefits of common policies,which depend on the extent of cross-border policyspillovers,andtheircost,whichdependsontheextentof policy preference differences across membercountries.Forindividualcountries,thecostofcommondecision-making is that it moves the common policyaway from its preferred national policy (i.e. a loss innational sovereignty); the benefit is that policyspilloversareinternalized.

This basic principle sheds an important light on theremaining two questions – i.e. which countries andwhich policies should undergo deep integration.Regarding the first question, countries that havesimilarpolicypreferenceswouldbenefitthemostfromdeepintegration,asthiswouldlimitthepoliticalcostofintegration.Similarly,foracertainspectrumofnationalpolicy preferences, countries that are more

interconnected would also benefit more from deepintegration.Regardingthesecondquestion,countriesshould take common policy decisions in areascharacterized by large cross-border effects andmaintainnationalpolicyprerogativesinareaswithlowcross-border impacts (and where policy preferencesaredissimilar).

An interesting empirical issue is whether the fiscalfederalism theory can explain observed patterns indeep integration arrangements. First, the theorypredicts that countries sharing similar policypreferencesandgreater levelsof interconnectionarethe ones that should choose deeper over shallowintegration. While a direct test of this proposition ishard to verify, several deep PTAs are formed bygeographicallyclosemembers(theEUbeingaprimaryexample). To the extent that policy preferences arecorrelated with geographic location, this providesindirectevidenceinsupportofthetheory.

Secondly, the fiscal federalism theory states thatpolicies characterized by high cross-border spilloversand low heterogeneity of preferences for differentcountriesshouldbecentralized,whiletheprovisionofall other services should be decentralized. Alesina etal. (2005) contrast this benchmark with a set ofindicatorsthatmeasuretheroleoftheEUindifferentpolicy areas. They find that there is a partialinconsistency between the resulting allocation ofcompetencies to the EU and the Oates Theorem. Inparticular, their data suggest that the EU is active incertainareaswherecross-bordereffectsare lowandthat its intervention is too limited in some policydomainscharacterizedby largespill-oversandsimilarpreferencesacrosscountries.58

Three further issues are relevant to the debate ondeep integration: the welfare effects of deepagreements on member countries; the trade-offs ofbilateral North-South deep agreements; and thesystemiceffectsofdeepregionalarrangements.

As discussed in the preceding section, there is not asingle definition of deep integration agreements, asthis concept generally refers to any agreement thatgoesbeyondshallowarrangements.Asaresult,thereis not the same comprehensive analysis of theeconomic costs and benefits of deep integration asthere is forpreferential tariff liberalization.This isnotsurprising for two reasons: first, the effects of FTA-plusorcustomsunion-plusagreementsarelikelytobedifferentfromtheeffectsofstandardFTAsorcustomsunions. Like shallow agreements, deep agreementsreducethecostsoftrade,andthuscanbeexpectedtoincrease trade among members (Section D providesan empirical analysis of the trade effects of deepintegration). However, unlike shallow agreements,deep integration agreements may also providesupranational public goods (common rules, a stablemonetary system, etc.) that the markets or national


114

governments cannot offer. The welfare effects ofthese public goods can go well beyond the tradeeffects,andaremorecomplicatedtomeasure.

From the perspective of developing countries, deepintegrationwithadvancedeconomiesmaycreatecertainadvantagesanddisadvantages (BirdsallandLawrence,1999).Asregardsadvantages,for instance,developingcountries can import international regulatory systemsthat are “pre-tested” and represent “best practices”,withouthavingtopaythecostsofdevelopingthemfromscratch.Asregardsdisadvantages,developingcountriesmay be pressurized to adopt common rules which areinappropriate for their level of development, such ascertainenvironmentalandlabourstandards.Thisriskishigher the weaker the bargaining power of developingcountries vis-à-vis their advanced trading partners (orwhen policies and regulations are imposed rather thandevelopedcooperatively).Suchstandardscouldalsobeusedbyadvancedeconomiestoprotectvestedinterestsandtoclosemarketstopoorcountries.

In a model of regional integration where specialinterest groups can manipulate the decision-makingprocess, Brou and Ruta (2006) show that moreadvanced economies tend to be more politicallyorganized and exert a stronger influence on commonpolicies.Whiledeepintegrationcanstillbeaboonfordeveloping economies, the theory supports concernsthat the common policy will shift away from theinterestsofthelessdevelopedmember.

Whatarethesystemiceffectsofdeepintegration?Thereis a long-standing debate in the trade literature onwhether preferential agreements are friends or foes ofthe multilateral trading system. Although this debate isextensively reviewed in Section E, some preliminaryobservations are worth noting. First, deep integrationmay,insomecases,havetrade-divertingeffects.Facchiniand Testa (2009), in their work on common markets,showthatmobilefactorsofproductionaremorelikelytoexperience an increase in returns, while immobile onesaremorelikelytobemadeworse-offcomparedwiththestatusquo(i.e.nocommonmarket).Ifnoformofwealthtransferacrosscountriesispossible,acommonmarketispolitically viable – i.e. it would be supported by themedian voter in each member country – only if somefactorsremainedprotectedvis-à-vistherestoftheworldoncetheintegrationprocessiscompleted.

In an empirical study, Chen and Mattoo (2008) findthat regional harmonization of standards significantlyincreases intra-regional trade in affected industries,but that the exports of excluded countries decline.Thissuggeststhatfirmsintheexcludedcountriesarehurtmorebyanincreaseinthestringencyofstandardsthan by the scale benefit provided by integratedmarkets. In other words, standards harmonization inPTAscanbede factorestrictive.

Asecondimportantobservationisthattheprocessofdeep regional integration may be a complement torather than a substitute for the process of globalintegration. Deep agreements address behind-the-bordermeasuresthataremoredifficulttonegotiateatthegloballevel,becauseofthewidelydifferentpolicypreferences and needs among countries. Regionalgroupings may offer supranational public goods thatgovernments–aswellasmultilateralarrangements–sofarfailtosupply(e.g.redistribution,infrastructures),giving them an appropriate intermediate level role inintegration between the national and global levels(Padoa-Schioppa,2001).

4. Conclusions

This section has reviewed the main reasons forestablishingPTAsandwhattheconsequencesareforboth members and non-members. Much analyticalwork in the past has focused on shallow tradearrangements,suchasfreetradeareas,andthetrade-creation/trade-diversion effects of PTAs. Aspreferential agreements have evolved over time,however, the loweringof tariffs isno longer themainfocusofPTAs.Agreementsnowcoverawidernumberof issues – beyond tariffs – and involve morestructured institutional arrangements. Traditionaltheories about PTAs fail to explain these newdevelopments, both in terms of the causes andconsequences of “deep” agreements. In particular,traditional theories are silent on the relationshipbetween the growth of international productionnetworks and the formation of deeper policyarrangements among countries. While the abovediscussionhasshedsomelightonthecausesandthestructure of deep integration agreements – adiscussion that falls mostly outside the domain oftrade economics – there is clearly a need for furtherresearchinthisarea.


115

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

1 Theempiricalrelevanceofterms-of-tradeeffectsintradepolicyhasbeenthesubjectofarecentdebateintheempiricalliterature.Brodaetal.(2008)andBagwellandStaiger(2011)findevidenceconsistentwiththeviewthatgovernmentssetpolicytoexploitterms-of-tradegains.

2 Ingametheory,thePrisoners’Dilemmarepresentsasituationwherebeneficialcooperationdoesnotemerge.Inthegameitisassumedthatplayers(theprisoners)caneithercooperateornotandthatcooperationinvolveshigherjointwelfarethannon-cooperation.However,wheneverotherschoosetocooperate,eachplayeractingindividuallywillbebetteroffbydeviatingandchoosingnon-cooperation.Giventhatallplayersaretryingtomaximizetheirindividualwelfare,theonlyrationalequilibriumimpliestheinferiorsituationofnon-cooperation.

3 Asitiswellunderstoodinthetheoreticalliteratureandinthepracticeoftradepolicy,cooperationamongcountriescannotbeachievedintheabsenceofatradeagreement.Thereasonisthat,ifacountryunilaterallyreducesitstariff,thetradingpartnerswouldstillhaveanincentivetomaintainitslevelofprotection.A“tradewar”,ontheotherhand,isastable(Nash)equilibrium,asoncehighprotectionsareinplace,nocountryhasanincentivetoreduceitstariffunilaterally.

4 AsdiscussedinBagwellandStaiger(1998),PTAsmayevenposeathreattothefunctioningofthemultilateraltradingsystem.SeeSectionEforadiscussionoftherelationshipbetweenpreferentialandmultilateralagreements.

5 SectionC.3will,however,analysecaseswherepreferentialagreementsmayaddresscoordinationproblemsbeyondterms-of-tradeorproductionrelocationexternalities.

6 Timeinconsistencyarises,forexample,whenapolicydecisionisseparatedthroughtimefromitsimplementation,withtheresultthatforsomereason(e.g.organizedpoliticalopposition)theinitialpolicyintentionisnolongerfeasible.

7 Putsimply,atime-inconsistencyproblemreferstoasituationwherebyadecision-maker’spreferenceschangeovertimesothatwhatispreferredatonepointmightbeinconsistentwithwhatispreferredatanotherpointintime.

8 Whetheranagreementcanincreasetradepolicycredibilityandwhethercountriesarelikelytosignagreementstocommittheirtradepolicyareultimatelyempiricalquestions.StaigerandTabellini(1999)andTangandWei(2008)provideevidencethattheGATT/WTOincreasedcredibilityofpolicycommitments.Arcandetal.(2010)findthattheprobabilitythattwocountriessignaPTAislargerwhensuchagreementleadstocredibilitygains.

9 ThekeyreferenceinthelobbyingliteratureintradeisGrossmanandHelpman(1994).Severalstudieshavedocumentedtheroleoflobbyinggroupsininfluencingtradepolicyoutcomes.Forareviewofthisempiricalliterature,seeGawandeandKrishna(2003).

10 ThispoliticaleconomyliteratureismoreextensivelydiscussedinSectionC.2.

11 LevyandSrinivasan(1996)provideanexampleofthislogic.AparticularfeaturesomePTAshavethattheWTOsystemislackingisprivateagents’accesstodisputesettlementmechanisms.Inthemultilateralsystem,privatedisputantshavetorelyontheirgovernmentstoactontheirbehalfeventhoughtheultimateincidenceofthecostsandbenefitsofthesettlementfalllargelyonthem.Meanwhile,aPTAlike

theEuropeanUnionallowsprivatepartiesindirectaccesstodisputesettlementthroughtheEuropeanCourtofJustice.LevyandSrinivasan(1996)arguethatthisdifferenceinthedesignofdisputesettlementmechanismsmightbeamotiveforpreferringPTAs.

12 Naturally,thisargumentwouldonlyholdtruewhenMFNratesarepositiveandnon-negligible.WithzeroMFNrates,therewouldbenoscopeforusingPTApreferences(asexplainedinSectionB).

13 AnempiricalstudymotivatedbyaformalgeneralequilibriummodelofthedemandforandsupplyofPTAmembership.

14 Theserelationshipsbecomestatisticallyinsignificantwhensuchfixedeffectsarecontrolledfor.Dyadicvariablessuchasbilateraldistancearetime-invariantandhencenotde-meanedfollowingthedifferencingtransformation.

15 Mostagreementsrequireallexistingmemberstoadmitanewentrant.

16 Thisempiricalfindingisfacilitatedbythefactthatunlikeothermodels,Bergstrandetal.(2010)donotassumeaninfinitelyelasticsupplyofPTAmembership.

17 Thesethreerelationshipsarerobusttotheinclusionofcountrypairfixedeffectsintroducedviaatimede-meaneddifferencingtransformation.

18 Thisreferstoawidely-usedmeasureofthe“politicalregimecharacteristics”ofstates.Thepolityscoremeasuresthegoverningauthorityofstatesrangingfromfullyinstitutionalizedautocraciestofullyinstitutionalizeddemocracies.Statesarerankedona21-pointscalerangingfrom-10(hereditarymonarchy)to+10(consolidateddemocracy).Seehttp://www.systemicpeace.org/polity/polity4.htm.

19 Dependingontheassumptionsonpreferences,itwouldbepossibletohaveeffectsalsoonthemarketforgood3evenincaseRoWmaintainsthesamenon-discriminatorytariff.However,inthisdiscussionweabstractfromtheseadditionaleffects.

20 Inamodelwithmorethanthreecountries,theextentofthisrentcanbeshowntodependonthenumberofcountriesthathavepreferentialaccesstothemarketofthetradingpartner.Specifically,asthisnumberincreases,thepreferencerentdecreases,asituationreferredtointheliteratureas“preferenceerosion”.

21 Thenextsubsectionprovidesasimplegraphicalanalysisinthespecialcasewheretheimportingeconomyissmallanddoesnotaltertheworldprice.

22 SeeBaldwin(2009)foracriticalsurveyofVinerianregionalismandforadiscussionofthelimitsofthetraditionalgraphicalapproachpresentedinBoxC.2.

23 Inneoclassicaleconomics,aParetoimprovementischaracterizedbyanactionthatmakesatleastoneindividualbetteroffwithoutmakinganyotherindividualworseoff.Paretooptimalitydescribesasituationwherenofurtherimprovementstowelfarecanbemade.TheParetooptimumisindifferenttothedistributionalconsequencesoftheoutcome.

24 DixitandNorman(1980)haveshownthatintra-PTAcommoditytaxesandsubsidiesaresufficienttoobtainthesameresultwithoutlump-sumtransfers.

Endnotes


116

25 Schiff(1999)statesthatthevolumeoftradedoesnotnecessarilyprovideanobjectivemeasureoftheextenttowhichtradingpartnersare“natural”becausethevolumeoftradeisitselfaffectedbypolicy.Instead,Schiffproposestodefinecountriesas“naturaltradingpartners”iftheytendtoimportwhattheprospectivepartnerexports.

26 Forinstance,Bustos(2011)studiestheimpactofMERCOSURontechnologyupgradingbyArgentineanfirms.Sheshowsthattheincreaseinrevenuesproducedbytradeintegrationcaninduceexporterstoupgradetechnology.AnempiricaltestofthemodelrevealsthatfirmsinindustriesfacinghigherreductionsinBrazil’stariffsincreaseinvestmentintechnologyfaster.Similarly,thereisevidencethatNAFTAhadpositiveeffectsonproductivityandtechnologyadoptionfornewexportingfirms.Inparticular,LileevaandTrefler(2010)findthatlower-productivityCanadianplantsthatwereinducedbythetariffcutstostartexporting,increasedtheirlabourproductivity,engagedinmoreproductinnovation,andhadhighadoptionratesofadvancedmanufacturingtechnologies.

27 AsummaryofthemainfindingsisprovidedinAppendixTableC.1.

28 OtherstudiesanalysingthewelfareeffectsofNAFTAthroughageneralequilibriumapproachareBrown(1994);Brownetal.(1992);Brownetal.(1995);Cox(1994);Cox(1995);CoxandHarris(1992);Sobarzo(1992);Sobarzo(1994);Sobarzo(1995).

29 Inaneconometricmodel,avariableissaidtobeendogenouswhenthereisacorrelationbetweenthevariableandtheerrorterm,whichistheunexplaineddeviationofsampledatafromtheirunobservable“true”value.

30 Inarecentpaper,BaierandBergstrand(2009)provideevidenceofthetradeeffectofPTAsbyusingnon-parametricestimates.WhentheselectionintoaPTAisnotrandom,asshownbyBaierandBergstrand(2004),andsomenon-linearitiesexistbetweenco-variatesingravityequationandPTAdummies(seeFrankel,1997,andBradaandMendez,1985),parametricestimatorscanbebiased.Inthiscase,non-parametricestimatorsareneeded.Usingthiseconometrictechnique,theauthorsprovidemoreeconomicallyplausibleeffectsofPTAsontradecomparedtopreviousestimates.

31 ThelikelihoodofaPTAisshowntodependon:(i)geography(thecloserthetwocountriesaretoeachotherandthefurthertheyaretotherestoftheworld);(ii)income(thelargertheirGDPsandthesmallerthedifferencebetweentheirGDPs);and(iii)endowments(thelargertheirrelativefactorendowmentdifferenceandthewiderabsolutedifferencebetweentheirandtherestoftheworld’scapital-labourratios).

32 Bergstrandetal.(2010)findsimilarresultsconsideringthetimingofallPTAsbyusingadurationanalysis.

33 OtherstudiesincludeRichardson(1994)andPanagariyaandFindlay(1996).

34 Theprospectsforanagreementimproveifpoliticallysensitivesectorscanbeexcludedfromtheagreement(GrossmanandHelpman,1995).ThisisbecausesectorsthatanticipatelargelossesfromaPTA,andlobbyforrejection,maybeindifferentiftheagreementwouldnotaltertheprotectiontheyaregrantedfromthegovernment.Inotherwords,excludingsomesectorsmaybeawaytodiffusepoliticaloppositiontoanagreementandimprovethechancesofachievinganaccordthatisbothpoliticallyviableandwelfareimproving.

35 TheworkbyKrishna(1998)hasalsoimportantimplicationsfortheregionalismversusmultilateralismdebate,asitimpliesthatpoliticallyfeasiblePTAsarelikelytohindermultilateraltradeopening.ThisissuewillbefurthertakenupinSectionE.

36 ThiswouldbethecaseifpBT,theborderpricefacedby

producerslocatedin1thatsellintheHomemarket,islowerthanpC,thepriceatwhichproducerslocatedin2cansellinHome.

37 Thoseanalysesarerestrictedtonon-cooperativemultilateralsettings(i.e.whereamultilateraltradeagreementsuchastheGATT/WTOisnotinplace).Ornelas(2008)studieshowtheformationofPTAsaffectsexternaltariffsandglobalwelfareinacooperativemultilateralenvironment.ThismodelshowsthatthecomplementaritybetweenexternalandpreferentialtariffsfoundintheliteraturediscussedinSectionC.2(e)generalizestothecasewherecooperationatthemultilaterallevelissignificant.

38 OtherworksthathavemadeasimilarpointontheroleoftradepreferencesininducingcooperationinotherpolicydomainsareJackson(1997);PerroniandWhalley(2000);andWorldBank(2000).

39 HereafterreferredtoasRoOs.

40 ThisisaffectedbytheMNCsoperatinginChile,ofwhich53percentrespondedthattheRoOshadbeenthedecidingfactor.Intheotherthreecountries,lessthan20percentofMNCsreportedRoOsasthedeterminingfactor.

41 Theauthorscontrolforothervariablesthatchangedbetweenthepre-1997andpost-1997periods,aswellasforunobservablepair-specificfactors.

42 NotethatTableC.1doesnotnecessarilyimplyalinearprogressionbetweendifferentstagesofintegration.Forinstance,acustomsunioncanbeformedevenintheabsenceofFTA+harmonizationsoramonetaryuniondoesnotnecessarilyimplythatacommonmarkethasbeenpreliminarilyestablished.

43 SeeSectionB.1fordataandafurtherdiscussion.

44 Systematicempiricalanalysisoftheinternationalfragmentationofproductionismissingduetolackofdata.However,recenteconomicliteraturehighlightsthreemajortrends.First,bothmerchandiseandservicesoffshoringhasrapidlyincreasedinthelasttwodecades.Second,althoughinternationaloutsourcingofintermediategoodsisquantitativelymoreimportant,servicesoffshoringhasbeenincreasingatafasterpaceinrecentyears.Third,thesetrendshavebeenwidespreadacrosssectorsandtypesofinputs(Helpman,2006).

45 SeeLipson(1982);Cantwell(1994);Chengetal.(2000);ArndtandKierzkowski(2001);ChengandKierzkowski(2001);Ando(2005);andBlanchard(2005).

46 SeeArndt(2004a,2004b).

47 Potentialcostsavingsfromintra-productspecializationmaybeloweredbyrestrictiverulesoforigininthecaseofafreetradearea.

48 SeeGrunwaldandFlamm(1985).

49 Inaddition,studiessuchasAndersonandvanWincoop(2004)havealsoshownthat,followingrecentwavesofliberalization,non-tariffbarrierstotradelikeshippingcostshavebecomemorerelevant.


117

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

50 SeeFujitaetal.(2001)foratheoreticalanalysisofclusteringattheinternationallevel.

51 Severalempiricalpapersusinggravitymodelsshowthatthereisapositiverelationshipbetweenproximitytointernationalcentresofeconomicactivityandpercapitaincomelevels(Hummels,1995;Leamer,1997).

52 HorizontalFDI,ontheotherhand,isstilldeterminedmostlybymarketsizeandtheseinvestmentflowsarecharacterizedbybeingbetweendevelopedeconomies.

53 ThelackofadeepAsianregionaltradeagreementhasbeencompensatedwithotherwaysofliberalizationsuchasbilateralinvestmenttreaties(BITs),which,accordingtoUNCTAD,increaseddramaticallyduringthe1990s,andunilateralliberalizationandpro-businessreformspromotedbyemergingmarketstoattractFDI.Inaddition,thereisalsoevidencethatseveralcountriesinEastAsiahaveconcentratedtheirpublicresourcesonthedevelopmentofeconomicinfrastructuresthatfacilitateproduction-sharing(AndoandKimura,2005;Ando,2005).

54 IntheAntràsandStaiger(2008)model,finalgoodsproducersandinputsuppliersarelocatedindifferentcountries.Contractsareincompleteandinvestmentsarerelation-specific.Inthiscontext,governmentshaveanincentivetousetradepolicybeyondterms-of-tradeeffects,asitaffectstheconditionsofex postbargainingbetweenforeignsuppliersanddomesticproducers.Thisisattherootofthenewcross-borderspillovereffectcreatedbytheriseinoffshoring.

55 Beggar-thy-neighbourisanexpressionineconomicsdescribingpoliciesthatseekbenefitsforonecountryattheexpenseofothers.

56 Commonpoliciesandregulationsareseenhereastheresultofinternationalcooperation.Analternativeisthatonecountrythathasahigherbargainingpowerimposesitspolicyandregulatoryframeworkontheother(possiblyinexchangeformarketaccessorasaformofhegemonicimposition).Thelattercaseisbrieflydiscussedbelow.

57 Forasurveyofthisliterature,seeOates(1999).Ruta(2005)andAlesinaandSpolaore(2005)provideextensivediscussionsoftherelatedpoliticaleconomyliteratureondeepintegration(i.e.theformationofinternationalunions).

58 TheOatesTheoremisbasedontheassumptionthatgovernmentshavenopoliticalmotivationsandmaximizesocialwelfare.Alargebodyofliteraturehasrevisitedthisprincipleinmodelsthataccountforpoliticalmotivationsofgovernments(AlesinaandSpolaore,1997;BoltonandRoland,1997;BesleyandCoate,2003;AlesinaandSpolaore,2005;Alesinaetal.,2005;Lockwood,2008;BrouandRuta,2006).ThesepoliticaleconomymotivationscanexplainthedeparturefromOates’normativetheoryandtheobservedallocationofcompetenciesintheEU(Ruta,2010).


118

ThisappendixfocusesonthesystemiceffectsofPTAs− that is, on the consequences of preferentialarrangements for members and non-members. Theapproach used is based on a graphical analysis anddrawsontheworkofBaldwinandWyplosz(2004).

Suppose that initially there is open trade across allcountries. Under these conditions, Home imports thequantityMatapriceofPFTdefinedbytheequilibriumof the import supply (MS) and import demand (MD)curves inHome (seeAppendixFigureC.1).Note thatM is the sumof theexport quantities fromRoW (XR)andPartner(XP)givenbytheintersectionoftheopentrade price line PFT , with each country export supplycurve shown as points 1 and 2 in the diagram,respectively.

IfHomemoves from free trade toapplyingauniformMFN tariff to all countries, the imposition of such atariffshiftstheimportsupplycurveuptoMSMFN.AsaconsequenceofthetariffT,thedomesticpriceforthegoodatHomerisestoP'andthequantityofimportsisreduced to M'. Meanwhile, the new border price forcountriesexportingtoHomeisgivenbyP'–T.Atthislower price, producers from RoW and Partner arewilling to supply less and exports are reduced to X'RandX'P,respectively.

After Home and Partner conclude a PTA, one ofHome's import suppliers gets duty-free access whilethe rest still pay T. Therefore, the new import supplycurve in Home, given by MSPTA, will lie between theoriginalopentradeandMFNsupplycurves(Appendix

AppendixFigureC.1: open trade and mFn tariffs

TPFT

P’

MD

2

XSp

X’p Xp Partner exports

Border price

Partner

MSMFN

M = XP + XR Home importsM’

MS

Domestic price

Home

X’R XR RoW exports

P’-T

1

XSR

Border price

RoW

AppendixFigureC.2: PtA price and quantity effects

T

T

P’

P’’

MD

XSp

X’p X’’p Partner exports

Border price

Partner

MSMFN

M’’ Home importsM’

MSPTA

MS

Domestic price

Home

X’’R X’R RoW exports

P’-TP’’-T

P*

XSR

Border price

RoW

Pa1

Technical appendix: Systemic effects of pTas


119

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

FigureC.2).MSPTAisnotastraightlinebecausethereis a thresholdpricebelowwhichonlyproducers fromPartner will be willing to export. The tariff preventsRoW firms from exporting until the domestic price atHome rises above the price marked Pa. This is sobecausewhenHome'sdomesticpriceisbelowPa,theborderpricefacedbyRoWexportsisbelowtheirzero-supplypricemarkedasP*.Consequently,Partnerfirmshave an effective “monopoly” over the access toHome's market up to the quantity denoted by thepoint1.Afterthispoint,firmsfromRoWwillalsosupplyimportstoHomeandMSPTAresumesitsnormalslope.

In the post-PTA equilibrium where MSPTA meets MD,Home will import the quantity M'' and the newdomestic price is P'', which is lower than the MFNdomesticpriceP'.ThePTA'simpactonborderpricesismore complex. For Partner-based producers,liberalizationmeans that their borderprice rises fromP' – T to P'', Home's new domestic price. For RoW-basedproducers,however, theborderprice falls fromP'–TtoP''–T.AwaytounderstandthiseffectistothinkthatRoWfirmsmustcuttheirborderpricesothatthey can enter Home's market and be competitive(be able to sell at a domestic price of P'') after thetariff T is added to their exports. As a result of thischange in border prices, Partner exports increase toX''PwhilethosefromRoWfalltoX''R.

The change in Home's import composition wheregoodsfromPartnerarefavouredoverthoseofRoWisknownastradediversion.Inotherwords,discriminatoryliberalization induces Home to switch some of itspurchases to import suppliers who benefit from thePTA and away from suppliers from nations that wereexcluded.ThePTAhasdistortedpricesignalssothatHome consumers are not aware that Partner goodsmay actually cost more than those from RoW. Homeconsumers ignore theborderpriceofgoodsandonlyobserve thedomesticpriceP'',which is thesame forimportsfromanysource.

To measure the welfare effects of the PTA, we mustevaluate the impact it has on the foreign exportingcountries (Partner and RoW) and on the importingcountry(Home).TheseeffectsareshowninAppendixFigure C.3. It is straightforward that the tradeagreement has favoured Partner as it experiences apositive border price effect (from price P' – T to P'')andapositivetradevolumeeffect(fromquantityX'PtoX''P).Thus,Partner'sgainsarecapturedbytheshadedareaD.TheoppositeistrueofRoWasitexperiencesequal but negative effects. RoW loses from the PTAbecause it facesa lowerborderpricefor itsgoodsatP'' –Tand its trade volumealso falls to thequantityX''R.TheselossesarecapturedbytheshadedareaE.

The PTA has more ambiguous welfare effects onHomeasithascreatedapositivetrade-volumeeffectbut also some conflicting terms-of-trade effects thatstemfromthedifferentiated(discriminatory)post-PTAborder-pricesPartnerandRoWface.By lowering thedomesticprice,preferentialliberalizationhasincreasedimportsfromM'toM'',leadingtoagaininconsumptionmeasured by the shaded area A. The positive trade-volume effect that has led to an efficiency gain inconsumptioncanbeseenasthetradecreationeffectofthePTA.Inotherwords,thePTAhascreatedtradebyallowingHometoaddthe importquantityM''–M'thatwasnotpresentbeforetheagreement.

Turning to the price effects of the PTA, HomeexperiencesanimprovementintermsoftradeagainstRoW as imports from this country have becomecheaper. Thus, Home imports a quantity of X''R fromRoW at a lower cost and gains from this change inborder price (the shadedareaB). TheareaBcanbeseen as a production efficiency gain, as producersfromRoWhave tobecomemoreefficient tocompeteinHome'smarketwhilefacingalowerborderprice.Onthe other hand, Home experiences a deterioration interms-of-trade against Partner as imports from thiscountryhavebecomemoreexpensiveafterthePTA.

AppendixFigureC.3:Welfare effects of preferential liberalization

P’P’’P’’

MD

XSp

X’p X’’p Partner exports

Border price

Partner

M’’ Home importsM’X’’R

Domestic price

Home

X’’R X’R RoW exports

P’-TP’’-T

P’-TP’’-T

P*

XSR

Border price

RoW

E

D

B

C

A


120

ThehikeintheborderpriceaffectsthequantityM'–X'Rand yields a loss to Home equal to the shaded areamarked C in the diagram. Since we have assumedPartnerandRoWtobeidentical,andthereforethereisnotamoreefficientproducer,weconcludedthatunderopentradeHomeimportedanequalamountfrombothcountries (50-50 share). After the PTA, however,

imports from Partner are favoured and represent alarger share of Home's imports. Thus, a portion ofareaCcaptures the trade-diversioneffectof thePTA,namely theamountof imports thathavebeendivertedawayfromRoW'soriginalshareinHome'smarket.Thenet welfare effects of the PTA on Home are given by(A+B)–C,whichmightbepositiveornegative.

AppendixTableC.1: empirical findings on trade creation and trade diversion

Authors Data and methodology trade creation trade diversion

Romalis(2007) CGEapproachontradeflowsbetweentheUnitedStates,Canada,Mexicoandtherestoftheworldintheperiod1989-1999.ThepaperfocusesonCanada-USFreeTradeAgreement(CUSFTA)andNorthAmericaFreeTradeAgreement(NAFTA)

EvidenceoftradecreationonlyfortradeflowsinvolvingMexico

EvidenceoftradediversionbyCUSFTAandNAFTA

Trefler(2004) CGEapproachonCanadianimportsfromUSandtherestoftheworldintheperiod1989-1996.ThepaperfocusesonNAFTA

NAFTAraisedCanadianimportsfromtheUnitedStates

NAFTAloweredCanadianimportsfromtherestoftheworld

Clausing(2001) CGEapproachonUSimportsfromCanadaandtherestoftheworldbetween1989and1994.ThepaperfocusesonCUSFTA

ThetariffliberalizationbyCUSFTAwasresponsibleforUSD21increaseinUSimportsfromCanadabetween1989and1994

Thereisnoevidenceoftradediversion

SoloagaandWinters(2001)

Gravitymodelonbilateralimportsfor58countriesfrom1980to1996.ThepaperfocusesontheEuropeanUnion(EU),EuropeanFreeTradeArea(EFTA),AssociationofSoutheastAsianNations(ASEAN),GulfCo-operationCouncil(GULFCOOP),NAFTA,CentralAmericanCommonMarket(CACM),LatinAmericanIntegrationAssociation(LAIA),AndeanCommunity(ANDEAN),SouthernCommonMarket(MERCOSUR)

AllthePTAsinvolvingLatinAmericancountrieshaveapositiveeffectonintra-bloctrade

TradediversioneffectforEUandEFTA

BaierandBergstrand(2007)

Gravitymodelonbilateraltradeflowsfor96countriesfrom1960to2000

PTAincreasestradebetweentwomembercountriesbyabout100percentonaverageafter10years

-

Frankeletal.(1995) Gravitymodelonbilateraltradeflowsfor63countriesovertheperiod1965-1990.ThepaperfocusesonEastAsiaEconomicCaucus(EAEC),Asia-PacificEconomicCo-operation(APEC),EuropeanCommunity(EC),EFTA,NAFTA,MERCOSURandANDEAN

PTAsboosttradebetweenmembercountries(exceptionsareEFTAandNAFTAwhichdonothavesignificanteffectontradeflows)

-

LeeandShin(2006) Gravitymodelonbilateraltradeflowsfor175countriesfrom1948to1999

JoiningaPTAraisesintra-bloctradeby51.6percent

PTAmembers'tradewithnon-membersrisesby6.5percent

Carrere(2006) Gravitymodelonbilateralimportsfor130countriesfrom1962-1996.ThepaperfocusesonEU,ANDEAN,CACM,LAIA,MERCOSUR,NAFTAandASEAN

Thereisevidenceoftradecreationeffectfor5outof7PTAsanalysed

Theincreaseinintra-regionaltradeiscoupledwithareductioninimportsfromtherestoftheworldin6outof7PTAsanalysed

Egger(2004) GravitymodelonbilateralexportsforOECDcountriesfrom1986to1997.ThepaperfocusesonEU,EFTAandNAFTA

Strongevidenceoftradecreationeffect

-

Magee(2008) Gravitymodelonbilateraltradeflowsfor133countriesfrom1980to1998

ThelongrunimpactofaPTAisestimatedtobean89percentincreaseintradeflows

Noevidenceoftradediversion

SilvaandTenreyro(2006)

Gravitymodelonbilateralexportflowsfor136countriesin1990

Strongevidenceoftradecreation

-


121

C. C

Au

se

s A

nD

eFFe

Cts

oF P

tAs: Is

It ALL A

Bo

ut

P

Re

FeR

en

Ce

s?

AppendixTableC.1: empirical findings on trade creation and trade diversion (continued)

Authors Data and methodology trade creation trade diversion

GhoshandYamarik(2004)

Gravitymodelonbilateraltradeflowsfor186countriesovertheperiod1970-1995

PTAmembershipraisesintra-bloctradeby39percent

PTAmembershiplowerstradeoutsidetheblocby6percent

BaierandBergstrand(2009)

Non-parametricestimationsonbilateraltradeflowsfor96countriesovertheperiod1965-2000

AveragelongruneffectofPTAsontradeflowsis100percent

-

Aitken(1973) Gravitymodelonbilateraltradeflowsfor12countriesovertheperiod1951-1967.ThepaperfocusesonEFTAandEEC

PositiveeffectofPTAsonbilateraltrade

-

Bergstrand(1985) Gravitymodelonbilateraltradeflowsfor15countriesforyears1965,1966,1975and1976.ThepaperfocusesonEFTAandEEC

PTAshadapositiveeffectonbilateraltrade

-

Acharyaetal.(2011) Gravitymodelonbilateraltradeflowsfor179countriesovertheperiod1970-2008

TheimpactofPTAsonintra-PTAtradeispositivefor17outof22PTAsanalysed.PTAsalsoincreaseimportsandexportsfrommembercountriestonon-membercountriesby20percentand21.5percentonaverage

Intra-PTAtradediversionhasbeenfoundin3outof22PTAsanalysed;5PTAslowertheextra-PTAexportsfrommembertonon-membercountries

C. Causes and effects of pTas: Is it all about … · science focuses on the causes and effects of ... analysis of preferential trade agreements ... • Global production networks

Documents