Page 1
1
TheDomesticPoliticsofPreferentialTradeAgreementsinHardTimes
EdwardD.Mansfield HelenV.MilnerDepartmentofPoliticalScience DepartmentofPoliticsUniversityofPennsylvania PrincetonUniversityPhiladelphia,PA19104 Princeton,[email protected] [email protected]
PreparedfortheannualmeetingoftheAmericanPoliticalScienceAssociation,SanFrancisco,CA,September3‐6,2015.
AbstractWepresentatheoryofwhysomecountriesnegotiatetradeagreementsduringeconomicdownturns.Wearguethatpoliticalleaderscangainfromsuchagreementsbecauseofthesignalstheysendtotheirpublics.Publicsarelesslikelytoblameleadersforbadeconomicconditionswhentheyhaveimplementedsoundeconomicpolicies,suchassigningandimplementingagreementsdesignedtoliberalizeandexpandtrade.Leadershaveparticularreasontoseekthistypeofinsuranceiftheycompeteforofficeinacompetitivepoliticalenvironment.Themoredemocratictheirpoliticalsystemis,themoretheycangainfromimplementingtradeagreements.Weevaluatethisargumentbyanalyzingallpreferentialtradeagreements(PTAs)ratifiedbycountriessince1951.WefindthatdemocraticcountriesareespeciallylikelytoratifyPTAsduringhardeconomictimes.
Page 2
2
Introduction
Thefinancialcrisisof2008‐2009stimulatedwidespreadfearsthatprotectionism
wouldriseacrosstheglobe.Observersexpressedconcernthatthisepisodewould
resembletheGreatDepression,whichwasaccompanied,ifnotdeepenedby,theglobal
spreadofprotectionism(Kindleberger1973).Indeed,thereceivedwisdomisthat
economicdownturnsfrequentlypromptcountriestoraisetradebarriers(Conybeare1983;
Cassing,McKeown,andOchs1986;BoharaandKaempfer1991;BagwellandStaiger2003).
Surprisingly,however,thecrisisgeneratedlittleprotectionismandthevolumeoftrade
recoveredquicklyafterthedownturn.TheleadersoftheG20pledgedeachyearfrom2008
to2011thattheywouldresistthetemptationtoraisetradebarriers.1AnOECDstudy
foundthattheseleaders’effortshadsucceededinavertingthegrowthofprotectionismand
otherstudieshavearrivedatasimilarconclusion(OECD2010;Bown2011;
VandenbusscheandViegelahn2011;Kee,Neagu,andNicita2013;Gawande,Hoekman,and
Cui2014).
Notonlydidheadsofstateworktoavoidaspikeinbarrierstooverseascommerce
duringthefinancialcrisis,leadersintheAmericas,Europe,andEastAsiawerenegotiating
tradeagreementsduringthisepisode,mostnotablytheTrans‐PacificPartnership(TPP),
theTransatlanticTradeandInvestmentPartnership(TTIP),andtheRegional
1Seehttp://www.nytimes.com/2008/11/16/washington/summit‐text.html?pagewanted=all;http://www.international.gc.ca/trade‐agreements‐accords‐commerciaux/agr‐acc/wto‐omc/pledge‐engagement.aspx?view=d.
Page 3
3
ComprehensiveEconomicPartnership(RCEP).Thisdevelopmentwasnocoincidence.
Rather,asweargueinthispaper,itreflectsabroadertendencyforleadersofdemocratic
countriestoformtradeagreementsduringhardeconomictimes.
Morespecifically,wearguethatdemocraticleadersmaychoosetonegotiateand
implementtradeagreementsbecausetheyyieldpoliticalaswellaseconomicbenefits.
Suchagreementscanhelpthemretainofficeinthefaceofeconomicdownturns,especially
incountrieswithcompetitivepoliticalsystems.Indemocracies,leadershavetobe
concernedthatcitizenswillholdthemresponsibleforthedownturnandvotethemoutof
office.Policiesdesignedtoliberalizeandpromotetheflowoftradesignaltovotersthatthe
downturnwasnottheproductofrent‐seekingorincompetence,butinsteadwasdueto
circumstancesbeyondtheleader’scontrol.
Wetestthisargumentbyanalyzingtheformationofpreferentialtradeagreements
(PTAs),whichareabroadclassofinternationalinstitutionsthatincludecommonmarkets,
customsunions(CUs),freetradeareas(FTAs),andeconomicunions.Theseagreements
havemarkedthegloballandscapeforcenturies,buthaveproliferatedespeciallyrapidly
overthepasthalf‐century.PTAsaredesignedtofostereconomicintegrationamong
member‐statesbyimprovingandstabilizingtheaccessthateachmemberhastotheother
participants’markets.Theseagreementsarethusasourceofliberalizationamongthe
contractingparties,althoughtheyalsodiscriminateagainstthirdparties(Freundand
Ornelas2010).Consistentwithourargument,wefindthatdemocraciesexperiencinghard
timesareparticularlylikelytoformPTAs.DuringtheperiodsinceWorldWarII,
Page 4
4
democraticcountrieshavedisplayedagreatertendencytoestablishtheseagreementsthan
theirnon‐democraticcounterparts,andtheyhavebeenmostlikelytoenterPTAsduring
downturnsinthebusinesscycle.
ATheoryofPTAsinHardTimes
ItiswidelyrecognizedthatPTAsinfluencetheflowoftradeandinvestment,butwe
arguethatnationalleadersoftenconcludetheseagreementsforpoliticalaswellas
economicreasons(MansfieldandMilner2012).WealsoarguethatleadersenteringaPTA
realizedomesticpoliticalbenefitsthataredifficulttoobtainthroughunilateraltradepolicy
measuresalone.Arationalgovernmentwillonlyformanagreementwithothercountriesif
theexpectedbenefits—bothdomesticandinternational—ofdoingsoexceedthecostsof
negotiatingandratifyingtheagreement.Tradeagreementsgenerateimportantbenefits
forbothgovernmentsandthepublic.Theycanconveyinformationtothepublicand
interestgroupspreferringfreetradeaboutthenatureandactivitiesofleaders.Such
informationcancontributetopoliticalsupportforleaders,helpingthemretainoffice.
Leadersmaypreferdifferentlevelsofprotection,basedontheweightsthatthey
assigntothebenefitsofrentsversussocialwelfare(Gawande,Krishna,andOlarreaga
2009).Promotingsocialwelfareyieldspoliticalsupportamongthepublicandfreetrade
interestgroupsthathelpslengthenaleader’stenureinoffice.Thepublicandthesegroups,
however,generallycannotbecertainofwhatbalancebetweenrentsandsocialwelfarea
governmenttrulydesires.Theyneedsomekindofreassuranceaboutthemotivesand
Page 5
5
actionsofthegovernment.Thepublicalsohasheterogeneouspreferencesabouttrade
policy;someindividualspreferhighlevelsofprotection,whereasotherspreferfreertrade.
Weassumethatthemedianmemberofsociety,whocommandstheattentionofleaders,
preferssomepositiveleveloftradebarriers(thatis,sheisnotcommittedtoafreetrade
policy).Sincetradebarrierscreaterentsforcertaininterestgroups,office‐holdersmay
seektoraisebarriersbeyondthelevelpreferredbythemedianindividualtoextractthese
rents.Thepublic,whichdoesnotgainfromtheserentsandprobablylosesbecauseofthem,
doesnotknowtheextentofgovernmentrent‐seekingsincecitizensdonotknowtheir
leaders’exacttradepreferencesorpolicies.Governmentswouldliketolimittheamountof
protectiontheyfurnishsinceithurtstheeconomyinthelongrunand,totheextentthat
voterscastballotsbasedoneconomicconditions,damagestheirreelectionprospects.
Butgovernmentsmaybetemptedtoprovideprotectiontocertaindomesticgroups.
Rentsforgovernmentsrisewiththelevelofprotection.Althoughleadersmaynotdesireas
muchprotectionaskeyinterestgroupsdemand,theymaybetemptedbytherentsthat
accruefromfurnishingprotection.Facedwithspecialintereststhatdemandprotection,
particularlywhentheeconomysours,leadersneedtofindwaystoreassurethemass
publicthattheyarenotbeingcapturedbyprotectionistinterestsandthattheyaremaking
soundforeigneconomicpolicy.Increasingtradebarriersmaywinleaderssupportfrom
someimport‐competinginterests,butdoingsocanalsoantagonizepro‐tradeinterest
groupsaswellasthegeneralpublic,whichwillbeharmedifprotectionismcontributesto
Page 6
6
slumpingeconomicgrowth.PTAsprovideamechanismforleaderstomanagesuch
societalpressures.2
Othershavemadesimilarargumentsabouttheroleoftradeagreementsand
domesticpolitics(MaggiandRodriguez‐Clare1998and2007;StaigerandTabellini1999;
Mitra2002).MaggiandRodriguez‐Clare(1998),forexample,arguethatgovernmentsmay
formtradeagreementstoprovidecrediblecommitmentsvis‐á‐visthepublicandinterest
groups.Theyshowthatgovernmentsfaceatimeinconsistencyproblem.Manyinterest
groupsdemandprotectionisttradepolicy,especiallyduringeconomicdownturns.
Althoughgovernmentsfinditdifficulttoresistthesedemands,protectioncauses
investmentdistortions,whichharmthegovernmentpoliticallyinthelongerrunby
reducingefficiencyandgrowth.Governmentsthenusetradeagreementstomakepolicy
commitmentsthatarecredibleandpreventinterestgroupsfrompressingforheightened
tradebarriersinthefuture.
Forleaders,concludingatradeagreementcanhelptoreassurethepublicthatthey
aremakingsoundforeigneconomicpolicy.Leaders,however,alsoworryaboutthe
domesticcostsinvolvedinratifyingagreements.Balancingthesetwoforcesisacentral
partofadecisionmaker’scalculationaboutwhethertosignaPTA.Acountry’sregimetype
iscrucialinthisregard.
2WerecognizethatPTAsmayhaveimportanteconomiceffectsandthatleadersmaybemotivatedbyadesiretoachieveeconomicbenefits.Ourpoint,however,isthattradeagreementsalsohaveimportantdomesticpoliticaleffectsthathavenotbeensufficientlyappreciated.
Page 7
7
DemocracieshavegreaterpoliticalincentivestoenterPTAsthanothercountries.
Thefree,fair,andregularelectionsthatarethehallmarkofdemocraciesmotivateleaders
tosignsuchagreements.Leadersinvarioustypesofpolitiesarecaughtbetweenthe
pressuresofspecialinterestgroupsandthepreferencesofvoters.Somespecialinterests
pressforpolicies—suchasprotectionisttradepolicies—thatbenefitthembutadversely
affecttheoveralleconomy.Headsofstatemaywanttosatisfytheseinterestgroupsin
exchangeforbenefitslikecampaigncontributionsorothersourcesofpoliticalsupport.But
givingintoallinterestgroupdemandswouldhaveveryharmfuleconomicconsequences
andcouldimperiltheirholdonoffice.
Leaders,atthesametime,haveahardtimeconvincingthepublicthattheywillnot
accedetospecialinterestdemands.AsMaggiandRodriguez‐Clare(1998)pointout,
governmentsfaceatimeinconsistencyproblemvis‐à‐visinterestgroups.Theywouldlike
tobeabletoresistsomeprotectionistdemands;butwhensuchdemandsarise,
governmentsareusuallybetteroffgivingintoeachgroupthatpressesforprotection.The
publicandfreetradeinterestgroupsknowthisandareharmedbygovernmentrent‐
seeking(MaggiandRodriguez‐Clare1998and2007).Theycanthreatenactiontolowerthe
incumbentgovernment’sprobabilityofretainingoffice.Buttheyalsofaceaninformational
problem.Membersofthepublicmaynotknowthepreferencesofortheexacttradepolicy
chosenbythegovernment,andthustheycannotdistinguishperfectlybetweenadverse
exogenouseconomicshocksandtheextractivepoliciesofleaders.Aneconomicdownturn
couldbecausedbyeitherhighlyprotectionistpoliciesoranexogenousshock,suchasa
Page 8
8
globalrecessionoraninternationalcrisis.Bothevents,forexample,mightincreasethe
pricethatthepublicpaysforgoodsandservices,andthusdampenthepublic’spolitical
supportforthegovernment.
Leaders,inturn,wouldliketofindawaytodemonstratetothepublicthatpoor
economicperformanceisnottheresultofextractivepolicies,therebyreducingthe
domesticpoliticalcoststhattheyface.Whiletheycouldchoosetounilaterallylowertrade
barriers,doingsoistimeinconsistent.Leaderscanreducebarriers,buttheyandthepublic
knowthatfuturespecialinterestdemandsforprotectionmaywellbemet.Soheadsof
governmentmustfindotherwaystoreassurethepublicthattheywillnotengagein
excessiveprotectionism.
Onewayofdoingsoisbyenteringintoaninternationaltradeagreement,whichis
bothavisiblecommitmenttorestrictprotectionismandaninstitutionalreassurancetothe
publicandfreetradeinterestgroupsthatarelativelyopentradepolicyhasbeenadopted.
Theagreementcommitsparticipatingcountriestoaleveloftradebarriersbeloweach
government'sidealunilaterallevelanditservesasamonitoringmechanism.Other
member‐statescanuseaspectsofthetradeinstitution(suchasthedisputesettlement
mechanismincludedinvarioustradeagreements)tosignaltoeachparticipating
government’ssocietyifitstradebarriersriseabovetheagreeduponlevel.Theagreement
ispublicandthereforeprovidesinformationthatdomesticgroupscanusetomonitortheir
leaders.Themonitoringthataninternationaltradeagreementprovidescanhelppolitical
leadersovercometheirreassuranceproblem.
Page 9
9
Thus,enteringintoatradeagreementcanbolstersupportforagovernment,even
whentheeconomyexperiencesadownturn.Whenelectionstakeplaceinthefaceof
adverseeconomiccircumstances,citizensmayblameincumbentsandvotethemoutof
office.Assuch,chiefexecutivesneedtofindwaystoreassurethepublicandother
domesticgroupsthateconomicdownturnsarebeyondtheircontrolandarenotsimplyan
outgrowthofleadersbucklingundertothedemandsofprotectionistgroups.3
PTAsprovidesuchapoliticalreassurancemechanism.Theseagreementsallow
leaderstocommittolowertradebarriersandsignalvotersthatleaders’tradepoliciesdid
notdirectlycausehardeconomictimes.Inturn,theseleadersaremorelikelytoremainin
officesincevotershavereasontoviewthemascompetenteconomicstewards,evenduring
recessions.Themoreelectoralcompetitionthatexists,themorethatleadershavetoworry
aboutbeingejectedfromofficeandthegreatertheirneedtoreassurethepublic.Hence,we
arguethatdemocraticgovernmentsshouldbemorelikelytosigntradeagreementsthan
othergovernments.
Forautocracies,thecalculationsdiffer.Interestgrouppressuresforprotectionism
inautocraciesvestleaderswithanincentivetoresistenteringPTAsthatreducetherents
3Mitra(2002)buildsontheanalysisconductedbyMaggiandRodriguez‐Clare(1998),butdemonstratesthatthecommitmentproblemforpoliticiansismoregeneralthantheyposit.ThedemandforaprecommitmenttoanFTAdoesnothavetobedrivenbythepossibilityofcapitalmisallocationalone,asMaggiandRodriguez‐Clare(1998)argue,orbythepossibilityoforganizationalcostsarisingintheexpectationofprotection.Demandforsuchanagreementcanoccurwhengovernmentsorinterestgroupsfaceresourcecostspriortolobbyingbecauseoftheactionstakenintheexpectationofsuccessfullobbying.Mitrashowsthattheinabilityofgovernmentstocommitunilaterallytoresistprotectionistpressuresbyinterestgroupscreatessubstantialcostsforgovernments.Undercertainconditions,thesecostscandrivegovernmentstoseekinternationaltradeagreements.
Page 10
10
theycanprovidetosupporters.Equally,autocratshavelessneedtoreassurethepublic
thattheyarecompetenteconomicdecisionmakerssinceelectoralcompetitiondoesnot
determinetheirfate.Consequently,autocratshavelessincentivetoenterintotrade
agreementsthantheirdemocraticcounterparts.
OurargumentfocusesonpoliticalleadersbecausetheyinitiateandratifyPTAs.But
interestgroupsmayalsoplayanimportantrole.Indeed,manytheoriesoftradepolicythat
examinedomesticpoliticsfocusoninterestgroupssincetheyhavetheresourcesand
coherencenecessarytoovercomecollectiveactionproblemsandexertinfluenceon
politicians(GrossmanandHelpman1994;Gawande,Krishna,andOlarreaga2009;Manger
2009).Someinterestgroupspreferfreetrader;othersaremoreprotectionist.Fromour
standpoint,opponentsoftradeopennessaremostimportantwhentheyareabletoexert
influenceasvetoplayersinthenegotiationandratificationofPTAs.Wethereforeaccount
forsuchvetoplayersinourempiricalanalysis.However,whileinterestgroupshelpto
shapethedemandforPTAs,theydonotcontrolthepoliticalprocessthroughwhichthese
agreementsareinitiatedandratified.Attheendoftheday,politicalleadersmustinitiate
andadvancetheseagreements,andtheirincentivestodosoareourprimaryfocus.
Ourargumentassumesthatthemedianmemberofthepublicisnotstrongly
protectionist.Weassumethatthemedianmemberofsocietydoesnotopposetrade
barrierscompletely,butthatshealsosupportstradeenoughthatleaderswanttopursue
tradeagreements.PTAsrarelyeliminateallbarrierstotrade;moretypicallytheylower
somebarriersandliberalizetheeconomyinotherways.Butdoesthepublicdesiresuch
Page 11
11
exposuretotheinternationaleconomy?ThePewResearchCenterGlobalAttitudesSurvey
examinedattitudestowardtradeinalargenumberofcountriesin2002,2007,and2008.
Inallofthesecases,anoverwhelmingmajorityofrespondentsfeltthattradewasgoodfor
theircountry.Othersurveys,suchastheLatinobarometerandAfrobarometer,alsoshow
majoritysupportforfreetradeacrossthoseregions(MansfieldandMilner2012:31‐32).
Insum,publicopinionpollsincountriesinEurope,LatinAmerican,andAfricaprovide
supportfortheclaimthatthemedianmemberofthepublicviewstradefavorably,
providingsomejustificationforourassumptionthatthepublictendstobemore
supportiveofopentradethanofprotectionism.
Anotherclaimwerelyonisthatsomemembersofthepublicandcertaininterest
groupsareawareofthetradeagreementstheirgovernmenthassignedandgenerallyview
suchaccordsfavorably.Publicopiniondatasuggeststhat,inmanycountries,thereisa
publicawarenessofandafavorableattitudeaboutinternationaltradeaccords.Manyofthe
samesurveysnotedabove,aswellastheInternationalSocialSurveyProgramme,which
havebeen conductedinnumerouscountriesovertime provideevidencethat largemajority
ofthosesurveyedhadheardquiteabitaboutthemajorPTAstheircountrywasinvolvedin,
andmostrespondentsfeltthatbeingamemberofthatPTAbenefitedhisorhercountry
(MansfieldandMilner2012:31‐32).Furthermore,weknowthattheinstitutionssetupby
tradeagreementscanhelptransmitinformationtodomesticgroupsaboutgovernments’
behavior.ManyPTAshavebodiesthatreportaboutthesignatories’behavioroverregular
intervals(Martin2000:chap.7).Thelegalizeddisputeprocesses,forinstance,often
Page 12
12
associatedwithinternationalinstitutions,suchastheWorldTradeOrganization’s(WTO’s)
disputesettlementmechanism,alsoplayanimportantroleintransmittinginformation
aboutthepoliciesofmembergovernmentstopreviouslyuninformedsub‐subnational
actors,suchasthevotingpublic.
Ourargumentalsoemphasizeshowinternationalagreementscanhelpleaders
reducethepotentialforlostpoliticalsupportbyreassuringthepublicinadvanceabout
theirintentions.Butdoessocietycareaboutwhetherthegovernmenthassignedand
abidedbyPTAs?Recentpublicopinionresearchclaimsthatitdoes.Herrmann,Tetlock,and
Diascro(2001),forexample,suggestthatvotersvaluetradeagreementsandbelievethey
areneededtosupportanopentradingsystem,implyingthatleadersmaypayapolitical
priceforviolatingtherulesofsuchinstitutions.
RegimeType,theBusinessCycle,andPTAs
Wehavearguedthatvotersinademocracyconsiderthestateoftheeconomywhen
goingtothepolls.Thus,governmentsarelikelytobepenalizedwhentheeconomy
performsbadly.Votersassumethatsuchdownturnsareatleastpartlyattributabletothe
policiesenactedinresponsetointerestgrouppressures.Assuch,governmentsfacea
credibilityproblem;votersaremorelikelytoremovethemfromofficeinbadeconomic
times,eveniftheydidnotgiveintospecialinterestdemandsandover‐protectthe
economy.Leadersthereforeseekwaystodemonstratetothepublicthattheyarenot
Page 13
13
overlysolicitoustospecialintereststhatdemandprotection.Onewaytodemonstratethis
istosignatradeagreement.
Inpoliticalsystemswherethepubliccannotvoteleadersoutofoffice,thisproblem
islesssevere.Insystemswithcompetitiveelections,bycontrast,theproblemisacute.The
moreleaders’fortunesdependonthevotingpublic,themoreincentivestheywillhaveto
findmechanismstoreassurethepublicthattheyhavenotgivenintospecialinterest
demandsandthushurttheeconomy.Consequently,themoredemocraticacountryis,the
greatertheincentiveforleaderstomakeacrediblecommitmenttoanopentradepolicy
andhencethemorelikelytheyaretosigninternationaltradeagreements.
Thisdynamicisespeciallypronouncedduringhardeconomictimes,whenleaders
areoftensuspectedofhavingchosenpoliciesthatfavoredspecialinterestsandcontributed
tothedownturn.LeadersthusseekmembershipinPTAsduringdipsinthebusinesscycle
todemonstratethattheyarenotoverlyinfluencedbyprotectionistinterests.Forthechief
executiveofcountriesmarkedbycompetitivepoliticalsystems,thesepressuresare
especiallypronounced.Thus,weexpectdemocraciestorespondtoeconomicdownturnsby
initiatingandratifyingPTAsevenmorefrequentlythantheydoingoodtimes.
Ourdata,whicharedescribedingreaterdetailbelow,revealnumerouscaseswhere
ademocracyratifiedtradeagreementsduringeconomichardtimes.Japan,forinstance,
ratifiedaPTAwithSingaporein2002(Japan’sfirst)inthefaceofaneconomicdecline.
IsraelsignedagreementswithBulgariaandRomaniain2001,duringaneconomicslump.
Zambia,anewdemocracythathadjusthelditsfirstmulti‐partyelectionsindecades,joined
Page 14
14
thebothSouthernAfricanDevelopmentCommunity(SADC)andtheAfricanEconomic
Communityduringaneconomicdownturnin1992.From1991to1993,Switzerland
concludedalargenumberofPTAswithEastandCentralEuropeancountries,aswellasone
withIsraelundertheEFTAumbrella,inaperiodwhenitexperiencedpooreconomic
performance.Finally,aswementionedearlier,manyofthecountriesnegotiatingand
formingtradeagreementsduringtheGreatRecessionweredemocracies.
Ourargumentemphasizestheeffectsofthebusinesscycleindemocracies,butthat
ishardlytheonlyfactorsguidingtheestablishmentofPTAs.Domesticinterestgroups,
internationalpolitics,andeconomicfactorshavealsobeenlinkedtoPTAformationandwe
trytoaccountfortheseinfluencesinourempiricalanalysis.
AnEmpiricalModelofPTAFormation
Ouranalysiscentersonestimatingthefollowingmodel:
PTARatificationij=0+1Democracyi+2GDPi+3(Democracyi×GDPi)+4
VetoPlayersi+5ExistingPTAij+6Tradeij+7GDPi+8Disputeij+9Allyij+
10FormerColonyij+11Contiguityij+12Distanceij+13Hegemony+14
Post‐ColdWar+15GDPRatioij+16%DyadsRatifyingPTA+17Global
BusinessCycle+β18GATTij+ij
Page 15
15
TheDependentVariable
IntheprocessofconcludingaPTA,countriesnegotiate,signtheagreement,and
thenbringithomefordomesticratification.Indemocracies,governmentsoftenrequire
(formalorinformal)legislativeratification.Inothercountries,ratificationmayinvolvethe
legislatureorthesupportofotherindividualswithintheregime.Inanycase,governments
havetosatisfyvetoplayersbeforetheagreementcanberatifiedandratificationisthefinal
stageintheprocessofformingaPTA.
Ourdependentvariable,PTARatificationij,isthelogoftheoddsthatstateiratifiesa
PTAinyeartwithstatej,whereweobserve1ifthisoccursand0otherwise.Ouranalysis
coverstheperiodfrom1951(forvariablesmeasuredinyeart‐1and1952forvariables
measuredinyeart)to2010(forvariablesmeasuredint‐1and2011forvariables
measuredint).Weaddressreciprocalagreements,whichinvolvepolicyadjustmentonthe
partofallmembers,andexcludearrangementswhereonestateunilaterallygrantsanother
countrypreferentialaccesstoitsmarket.Sinceweareinterestedintheformationof
preferentialagreements,theobservedvalueofPTARatificationijis1inyearswhenstates
initiallyratifyanewPTAorwheniorjjoinsaPTAtowhichtheotherstateisalreadya
party,butnotinsubsequentyearswhentheagreementisinforce.
Iftheexactyearofratificationcouldnotbedetermined,werelyonthedatethat
stateisignedthePTAwithstatej.Sincemostagreementsareratifiedrelativelysoonafter
theyareformedandsincewearemissingratificationdatesinlessthan30percentofthe
Page 16
16
casesofPTAformation,thisisareasonableapproach.Sincestatesiandjneednot—and,
indeed,oftendonot—ratifyatradeagreementinthesameyear,ourunitofanalysisisthe
annual“directeddyad.”Consequently,foreachdyadineachyear,thereisoneobservation
correspondingtostateiandasecondobservationcorrespondingtostatej.Forexample,in
thecaseoftheUnitedStates‐Mexicandyadin1985,weincludeoneobservationwherethe
U.S.isiandMexicoisj,andasecondobservationwhereMexicoisiandtheU.S.isj.Each
monadicvariable,isincludedinthismodelonlyonce,forthecountrylistedasiineach
particularobservation.Becausefocusingondirecteddyadsdoublesthenumberof
observationsinthesample,therebyproducingstandarderrorsthataretoosmall,we
clusterthestandarderrorsovertheundirecteddyad.
TheIndependentVariables
Ourprimaryindependentvariablesaretheregimetypeofeachnation‐stateand
fluctuationsinthebusinesscycle.First,Democracyiindicateswhetherstatei’spolitical
regimetypeisdemocraticornotinyeart.Tomeasureregimetype,werelyonawidely‐
usedindexconstructedbyGurr,Jaggers,andMoorethatrangesfrom1forthemost
autocraticcountriesto21forthemostdemocraticstates,aswellasdatadrawnfromthe
PolityProject(Gurr,Jaggers,andMoore,1989;JaggersandGurr1995;MarshallandJaggers
2009).4
4WeusedthePolityIVdata,generatedin2011.
Page 17
17
Inthefollowinganalysis,weusethreedifferentthresholdsfordemocracy:(1)a
scoreof20‐21ontheaforementionedindex,(2)ascoreof16‐21,and(3)ascoreof12‐21.
AlmostallOECDcountriesarecodedasdemocraticbasedonthefirstthreshold(the
exceptionsaretheCzechRepublic,Estonia,Mexico,SouthKoreaandincertainyears,
BelgiumandFrance).Sotooareanumberofsmallercountries.In2010,forexample,
CostaRica,Estonia,Lithuania,Mauritius,Mongolia,Taiwan,Trinidad,andUruguaywere
democraticbasedonthemostrestrictivedefinition.Usingthesecondthresholdadds
variouscountriestothedemocraticranks;dependingontheyearinquestion,these
countriesincludeArgentina,Brazil,Burundi,Georgia,India,Liberia,Nepal,Pakistan,South
Africa,Turkey,Ukraine,andalloftheOECDcountriesthatdidnotcrossthefirstthreshold.
Thethirdthresholdgeneratesamuchmoreheterogeneoussetofcountries—
includingcontemporaryAlgeria,Cambodia,Iraq,Nigeria,Russia,SriLanka,andThailand—
someofwhichfalloutsidetheboundsofwhatmostobserverswouldconsiderdemocratic.
OurargumentimpliesthatthetendencyfordemocraciestoenterPTAsduringhardtimes
shouldbemostpronouncedifcountrieshavethemostfullyformeddemocraticinstitutions
andleastpronouncedifcountrieshavelessdemocraticinstitutions.Bycomparingresults
acrossthesethreethresholds,wecanassessthisimplication.
Second,tomeasurethebusinesscycle,weincludeGDPi,thepercentagechangein
thegrossdomesticproduct(GDP)ofstateifromyeart‐1toyeart.5Wealsoincludethe
5GDPdataaretakenfromthePennWorldTables(Heston,Summers,andAten2011)andareexpressedinconstantU.S.dollars.
Page 18
18
interactionbetweenDemocracyiandGDPitoaddressourargumentthatdemocraciesare
especiallylikelytoestablishPTAsduringeconomicdownturns.Ourdatarevealthaton
averageittakescountriesabout349days(205daysisthemedian)toratifyan
internationaltradeagreementafterithasbeensigned.Sinceleadersusuallyhavesome
abilitytomanipulatethetimingofthedomesticratificationprocess,weassumethatthis
lessthanoneyearperiodalsocorrespondstothetimeoftheeconomicshocktheyare
experiencing.
WealsoincludeavarietyofvariablesthathavebeenlinkedtoPTAformationin
priorresearch,manyofwhichmightbeassociatedwithacountry’sregimetype,its
businesscycle,orboth.Tobegin,thereisamplereasontoexpectthatinterestgroups
influencethenegotiationandratificationoftradeagreements.Weaccountfortheireffect
byexaminingthenumberofvetoplayersinacountry.Theseactorshaveinstitutional
capacitytoaffectwhetheranexecutiveisabletopasstradelegislation.Thenumberofveto
playersaffectsthetransactioncoststhatthegovernmentbearswhenratifyingaPTA.
Thesecostsaregreaterincountriesmarkedbyalargenumberofvetoplayers,whichin
turnreducestheincentivesforleaderstotrytonegotiateandratifyPTAs.Consequently,
theoddsofastateenteringapreferentialarrangementarelikelytodeclineasthenumber
ofvetoplayersrises.6
6Onvetoplayers,seeHenisz(2000and2002)andTsebelis(2002).
Page 19
19
WethereforeincludeVetoPlayersi,whichindicatestheextentofconstitutionally
mandatedinstitutionsthatcanexercisevetopoweroverdecisionsinstateiaswellasthe
alignmentofactors’preferencesbetweenthoseinstitutionswithinthestate(Henisz2000
and2002).7Thismeasureiscontinuousandrangesfrom0to1.WhenVetoPlayersiequals
0,thereisacompleteabsenceofvetoplayersinstatesi.Highervaluesindicatethe
presenceofeffectivepoliticalinstitutionsthatcanbalancethepoweroftheexecutive.Note
thatDemocracyiandVetoPlayersiaremeasuredinyeartbecausepriorresearchindicates
thatthesevariablesshouldhaveacontemporaneouseffectonPTAratificationandthereis
littlereasontoworrythatourresultswouldbecompromisedbyanysimultaneitybias.
Afterall,itseemshighlyunlikelythatthedecisiontoformaPTA,muchlesstheratification
ofsuchanagreement,wouldinfluenceeitherastate’sregimetypeorthenumberof
domesticvetoplayers(MansfieldandMilner2012).Theremainingvariablesinourmodel
aremeasuredinyeart‐1.
ExistingPTAijindicateswhetherstatesiandjarealreadymembersofthesame
PTA(s).Participatinginatradeagreementislikelytoaffectastate’sproclivitytocreateor
joinanotherarrangementwiththesamepartner.Tradeijisthelogarithmofthetotalvalue
oftrade(inconstant2000USdollars)betweenstatesiandj.8Variousobserversarguethat
7Weusethemostrecentversionofthesedata,whichwereupdatedin2012.Heniszhasdevelopedtwomeasuresofvetopoints,onethatincludesthejudiciaryandonethatdoesnot.WeusethelattermeasuresincethereislittlereasontobelievethatthejudiciarywouldinfluencethedecisiontoenteraPTA.However,ourresultsarequitesimilarwhenweusethealternativemeasure.
8Weadd.001toallvaluesoftradesincesomedyadsconductnotradeinparticularyearsandthelogarithmofzeroisundefined.
Page 20
20
increasingeconomicexchangecreatesincentivesfordomesticgroupsthatbenefitasa
resulttopressgovernmentstoenterPTAs,sincethesearrangementshelptoavertthe
possibilitythattraderelationswillbreakdowninthefuture(Nye1988).Moreover,
heightenedoverseascommercecanincreasethesusceptibilityoffirmstopredatory
behaviorbyforeigngovernments,promptingfirmstopressfortheestablishmentofPTAs
thatlimittheabilityofgovernmentstobehaveopportunistically(Yarbroughand
Yarbrough1992).9
Besideseconomicrelationsbetweencountries,economicconditionswithin
countriesarelikelytoinfluencePTAformation.Particularlyimportantinthisregardisa
state’seconomicsize.Largestatesmayhavelessincentivetoseektheexpandedmarket
accessaffordedbyPTAmembershipthantheirsmallercounterparts.Wethereforeanalyze
GDPi,thelogarithmofstatei’sgrossdomesticproduct(inconstant2000USdollars).
Inaddition,politicalrelationsbetweenstatesmayinfluencewhethertheyjointhe
samePTA,independentoftheirrespectivedomesticpoliticalstructures.Cooperationalso
dependsontheextentofdifferencesinpreferencesbetweencountries’leaders.Thefurther
apartarethesepreferences,thelesslikelyiscooperation.Toaccountforthese
preferences,weincludeanumberofvariablesthatmeasuretheforeignpolicydifferences
betweenstates.Militaryhostilitiesbetweenstatessignallargedifferencesinpreferences
betweencountriesandmaydiscourageeconomiccooperationandthustheirpropensityto
9NotethatweusetheInternationalMonetaryFund’sDirectionofTradeStatistics(variousyears)asthemainsourceforthetradedata.WedeflatethesedatausingtheUSGDPdeflator.
Page 21
21
signPTAs.Similarly,political‐militarycooperationmaypromoteeconomiccooperation
(Mansfield1993;Gowa1994).Disputeijiscoded1ifstatesiandjareinvolvedinadispute,
0otherwise.Thoughmanystudiesofpoliticaldisputesrelyonthemilitarizedinterstate
disputes(MIDs)dataset(Jones,Bremer,andSinger1996;GhosnandPalmer2003),these
datadonotextendbeyond2000.Toanalyzethelongestpossibletimeframe,wetherefore
usethePRIOdataoninterstatearmedconflict,whichcoverstheperiodfrom1951to
2011.10Allyijequals1ifstatesiandjaremembersofapolitical‐militaryalliance,0
otherwise.WecodethisvariableusingtheATOPdata(Leedsetal.2002).11Further,since
previousresearchhasfoundthataformercolonialrelationshipbetweeniandjincreases
thelikelihoodthattheywillenterthesamePTA,weincludeFormerColonyij,whichequals1
ifstatesiandjhadacolonialrelationshipthatendedafterWorldWarII,0otherwise
(MansfieldandReinhardt2003;MansfieldandMilner2012).12
10Weusev4‐2008ofthedatafrom:http://www.prio.no/CSCW/Datasets/Armed‐Conflict/UCDP‐PRIO/.Theirdataincludes4typesofconflict:(1)extra‐systemicarmedconflictoccursbetweenastateandanon‐stategroupoutsideitsownterritory;(2)interstatearmedconflictoccursbetweentwoormorestates;(3)internalarmedconflictoccursbetweenthegovernmentofastateandoneormoreinternaloppositiongroup(s)withoutinterventionfromotherstates;and(4)internationalizedinternalarmedconflictoccursbetweenthegovernmentofastateandoneormoreinternaloppositiongroup(s)withinterventionfromotherstates(secondaryparties)ononeorbothsides.Type3conflictsweredropped.WekepttheotherthreetypesandexpandedthedatasothatallpossibledyadsbetweenthecountriesonsideAandthoseonsideBwerecreated.Datathatdidnothaveanindependentcountryasoneofthesideswerethendropped.ThesethenshouldbealldyadicconflictsintheUppsaladata.SeeGleditschetal.(2002)andHarbometal.(2008).
11FortheATOPdata,weuseversion3.0,specificallytheatop3_0ddyr.dtafilewhichisthedirecteddyaddataset.Becausethedataonlygoto2003,datafor2004werefilledinwith2003data.SeeLeedsetal.(2002).
12DataonformercolonialrelationsaretakenfromKurian(1992).
Page 22
22
GeographicproximityisanotherimportantinfluenceonPTAformation.Statesoften
enterPTAstoobtainpreferentialaccesstothemarketsoftheirkeytradepartners.These
partnerstendtobelocatednearby,sincecloserproximityreducestransportationcostsand
otherimpedimentstotrade.Weintroducetwovariablestocapturedistance.Contiguityijis
adummyvariablethatiscoded1ifstatesiandjshareacommonborderorareseparated
by150milesofwaterorless.Distanceijisthelogarithmofthecapital‐to‐capitaldistance
betweeniandj.Itisusefultoincludebothvariablessincesomestateshavedistantcapitals
(forexample,RussiaandChina)yetshareborders,whileotherstatesdonotshareborders
butareinrelativelycloseproximity(forexample,BeninandGhana).13
Inaddition,systemicconditionsarelikelytoaffecttheprospectsofPTAformation.
Sincethereisevidencethatdeclininghegemonycontributestotheproliferationof
preferentialarrangements,weincludeHegemony,theproportionofglobalGDPproduced
bythestatewiththelargestGDP(inoursample,theUSforeachyear).Thisvariable
thereforetakesonthesamevalueforeachcountryinyeart‐1.WefurtherincludePost‐
ColdWar,whichequals0from1950to1988and1thereafter,toaccountforthespikein
PTAsaftertheBerlinWall’scollapse.Wealsoexaminewhetherpowerdisparitiesinfluence
theestablishmentofpreferentialarrangementsbyincludingGDPRatioij,whichisthe
naturallogarithmoftheratioofthecountryGDPsforeachdyad.Incomputingthis
variable,thelargerGDPisalwaysinthenumerator;henceanegativesignonthecoefficient
13DataondistanceandcontiguityaretakenfromCEPII’sgravitydataset(Head,Mayer,andRies2010).
Page 23
23
ofthisvariablewouldindicatethatagreaterdisparitybetweenthecountriesdecreasesthe
likelihoodofratification.
Becausevariousobservershavearguedthatthespreadoftradeagreementshas
beenmarkedbydiffusion,weaddthepercentofalldyadsinthesystemthatratifiedaPTA
inyeart‐1,%DyadsRatifyingPTA.Sinceagivencountry’sbusinesscycleuslikelytobe
affectedbytheglobalbusinesscycle,weincludeGlobalBusinessCycle,ameasuredrawn
fromtheOECD’sCompositeLeadingIndicatorsdata.Thisvariableiscreatedbyassessing
trendsinvariousaspectsofeconomicoutputtoidentifyupturnsordownturnsinthe
internationaleconomy.WerelyontheOECDcompositemeasurethatidentifiesannual
change(fromJanuarytoJanuary)inthebusinesscycle.BecausetheGeneralAgreementon
TariffsandTrade(GATT)andWTOrecognizeandattempttogoverntheestablishmentof
PTAs,membersoftheseglobalinstitutionsmayalsobedisproportionatelylikelytoenter
preferentialarrangements(MansfieldandReinhardt2003).Toaccountforthispossibility,
weintroduceGATTij,whichequals1ifstatesiandjarebothmembersoftheGATTineach
yearpriorto1995oriftheyarebothmembersoftheWTOinyearsfrom1995on,and0
otherwise.14
Initially,weestimatethemodelwithregionalfixedeffects,usingeightregional
categoriesidentifiedbytheWorldBank,sinceitiswidelyarguedthattheprevalenceof
PTAsvariesacrossregions.Wethenestimatethemodelwithcountryfixedeffectsforstate
14DataaretakenfromtheWTOwebsite:http://www.wto.org.
Page 24
24
iandjtoaccountforanyunobservedheterogeneityacrosscountriesinPTAformation.15
Finally,ijisastochasticerrorterm.
DescriptivestatisticsforallofthesevariablesarepresentedinTable1.Thesample
inthefollowinganalysesiscomprisedofallpairsofstatesduringtheperiodfrom1951to
2011.Becausetheobservedvalueofthedependentvariableisdichotomous,weuse
logisticregressiontoestimatethemodel.Testsofstatisticalsignificancearebasedon
robuststandarderrorsclusteredonthedyadtoaddressanypotentialproblemswith
heteroskedasticityorthedirecteddyadresearchdesign.Toaccountfortemporal
dependenceintheformationofPTAs,weincludeasplinefunctionofthenumberofyears
thathaveelapsed(asoft)sinceeachdyadlastformedaPTAwithknotsatyears1,4and7,
assuggestedbyBeck,Katz,andTucker(1998).Inthefollowingtable,however,the
estimatesofthisfunctionareomittedtoconservespace.
Results
InTable2,wereporttheestimatesofourmodel.Thefirsttwocolumnsshowour
resultswhenDemocracyiiscoded1forstatesthatscore20or21onthePolityregimetype
15Werelyonfixedeffectsforcountriesratherthandyadsbecausealmost85percentofthedyadsinoursampledonotformaPTAduringtheperiodweanalyzeandaredroppedfromthesamplewhenthemodelisestimatedwithdyad‐specificfixedeffects.Thereisnoreasontoexpectthatthesmallsetofcountry‐pairsusedtoestimatethemodelwithdyad‐specificfixedeffectsarerepresentativeofallcountry‐pairs.Consequently,thatestimationtechniquerisksgeneratingresultsthataremisleading,whichiswhyvariousobserverscounselagainstusingdyadicfixedeffectsinanalysesofdataassparseasours(BeckandKatz2001;King2001).Theuseofcountry‐specificfixedeffectsavoidsthisproblem.
Page 25
25
index,thethirdandfourthcolumnsshowstheresultswhenitiscoded1forstatesthat
score16orgreater,andthefinaltwocolumnsshowourresultswhenthisvariableiscoded
1forstatesthatscore12orhigher.Theestimatedeffectsofregimetypeundervarious
macroeconomicconditionsdrawnfromourdataandbasedonthecountryfixedeffects
specificationareshowninFigures1‐3.Regardlessofhowstringentadefinitionof
democracythatweuse,democraciesaremuchmorelikelytoformPTAsduringdownturns
inthebusinesscyclethanduringperiodsofeconomicgrowth.Thelikelihoodofanon‐
democracyjoiningatradeagreement,incontrast,ismuchlesssensitivetothebusiness
cycle.
Equally,democraciesaremorelikelytoestablishPTAsthanothertypesofregimes,
butthistendencyismostpronouncedduringhardtimes.Finally,bothofthesetendencies
becomeincreasinglypronouncedasweimposeincreasinglystrictdefinitionsofdemocracy.
Amongstatessufferingan8percentreductioninGDP,forexample,democraciesaremore
thanfourtimesaslikelytoaccedetoaPTAasnon‐democraciesifwedefinedemocraciesas
statesthatscore20or21onthePolityindex,roughly55percentmorelikelyifwedefine
democraciesasstatesthatscore16orgreateronthisindex,andabout15percentmore
likelyifwedefinedemocraciesasstatesthatscore12orgreateronit,basedonthefixed
effectsspecification.
Asexpected,theoddsofratifyingaPTAalsoriseasthenumberofvetoplayersfalls.
Ineachmodel,theestimatedcoefficientofVetoPlayersiisnegativeandstatistically
significant.Tofurtheranalyzetheeffectsofthisvariable,wecomparethepredicted
Page 26
26
probabilityofstateiformingaPTAwhenithasfewvetoplayers—whichwedefineasthe
10thpercentileinthedata—tothepredictedprobabilitywhenithasmanysuchplayers—
whichwedefineasthe90thpercentileinthedata,holdingconstanttheremainingvariables
inthemodel.BasedontheresultsinthesecondcolumnofTable2,astatewithfewVeto
Playersiisabout20percentmorelikelytoratifyaPTAthanonewithmoreVetoPlayersi.
Thisfigurevariessomedependingonhowdemocracyisdefined,buttheseresults
nonethelessreinforcethepointthatdomesticpoliticsplaysanimportantroleinshaping
thedecisiontoentertradeagreements.
Notsurprisingly,however,variouseconomicandinternationalfactorsarealso
importantinthisregard.StatesthattradeextensivelytendtoformPTAs;ineachcase,the
estimatedcoefficientofTradeijispositiveandstatisticallysignificant.Increasingthemean
ofTradeijbyonestandarddeviationincreasesthepredictedprobabilityofratificationby
over15percent,holdingconstanttheremainingvariablesinthemodel.Further,basedon
thecountryfixedeffectsspecification,thereisevidencethatlargercountriesarelesslikely
toenterPTAsthantheirsmallercounterpartssincetheestimatedcoefficientsofGDPiare
negativeandsignificant.
Turningtothesystemicvariables,thereisevidenceofthediffusionofPTAsandthat
theoddsofratifyingsuchanarrangementroseintheColdWar’saftermath.Theestimated
coefficientofPost‐ColdWarispositiveandstatisticallysignificantineachinstance.Sotoo
isthecoefficientof%DyadsRatifyingPTA,whichindicatesthatPTAformationtendsto
Page 27
27
clusterovertime.Statesmaybeeitherstrategicallyconditioningtheirbehavioronwhat
theircounterpartsdoorsimplyfollowingtheherd.
PTAsarealsoespeciallylikelytoformasahegemonicpowerrises.Theestimated
coefficientsofHegemonyinTable2arepositiveandstatisticallysignificant,indicatingthat
theoddsofratifyingapreferentialarrangementriseastheportionoftheworld’soutput
accountedforbytheleadingeconomyincreases.IncreasingthemeanvalueofHegemony
byonestandarddeviationyieldsalmosta30percentriseinthepredictedprobabilityof
ratification.Thisresultdiffersfromwhatwefoundinearlierresearch,perhapsbecause
thecurrentanalysiscoversalongertimeframe(MansfieldandMilner2012).
Inaddition,alliances,GATT/WTOmembership,andexistingPTAmembership
promotetheratificationofpreferentialarrangements.Asexpected,alliesaremorelikelyto
formpreferentialagreementsthanotherstates.However,political‐militarydisputeshave
littleeffectonPTAs;theestimatedcoefficientsofDisputeijarepositive,butarenot
statisticallysignificant.Thatmembersofthemultilateralregimearemorelikelytoform
PTAsthanotherstatesmightseemsurprisingatfirstblushsincethisinstitutionwas
intendedtocombatregionalismandbilateralism.However,theGATT’sArticleXXIVmade
specificprovisionsforsuchagreementsandPTAshaveflourishedamongmembersofthis
regime.Italsomightseemsurprisingthatcountriesthatalreadyparticipateinthesame
PTAaremorelikelytoformanotheronethanstatesthatarenotPTApartners.Butin
2005,forexample,1,126countrypairswerepartiestotwopreferentialagreements;415
pairstothreePTAs;82pairstofourPTAs;27dyadstofivePTAs;andthreepairstosix
Page 28
28
PTAs.In1976,forinstance,PapuaNewGuineaandAustraliainkedabilateralagreement,
followedbybothcountriesjoiningtheSouthPacificRegionalTradeandEconomic
CooperationAgreement(SPARTECA)in1980.SingaporeandNewZealandsigneda
bilateralagreementin2000,afterwhichbothcountriesenteredtheTrans‐PacificStrategic
EconomicPartnershipAgreementin2005.In1997,theGreaterArabFreeTrade
Agreement(GAFTA)wassigned.Amongthememberswerethreecountries(Morocco,
Tunisia,andLibya)thatwerealsointheArabMaghribUnion,aswellassixmembers(Iraq,
Egypt,Syria,Yemen,Kuwait,andtheUnitedArabEmirates)thatwerealreadyjoinedunder
theCouncilofArabEconomicUnity(CAEU).
TheresultsalsoshowthatcontiguousstatesareunlikelytoformPTAs.In
combinationwiththeobservedeffectsofDistanceij,thissuggeststhatPTAsaremostlikely
toformbetweenstatesthatarenearbybutdonotshareaborder.Finally,whilemany
observersassumethatPTAsareformedbetweenalarge,richcountryandasmall,poor
one,ourresultsindicateotherwise.ThecoefficientestimateofGDPRatioijisnegativeand
statisticallysignificant,implyingthatgreaterimbalancesinnationalincomediscouragethe
ratificationofPTAs.Sincecountriesthatareequallypowerfulmaybebetterableto
concludeagreementsthatinvolvereciprocalconcessions,thisresultmaynotbethat
surprising.ButtheideathatmostsmallcountriesareforcedintoPTAswithlargerones
againsttheirwilldoesnotseemtobeborneout(Gruber2000).Finally,theestimated
coefficientofGlobalBusinessCycleisnegativeandstatisticallysignificant,indicatingthat
Page 29
29
PTAsaremorelikelytoformduringeconomicdownturnsthanduringupswingsinthe
internationaleconomy.
Anumberofrobustnesschecksprovidefurthersupportforourclaims.Tobegin,we
replacethePolitymeasuresofdemocracywithanotherwell‐knownmeasuredevelopedby
Przeworskiandhiscolleagues(2000),andupdatedbyCheibub,Gandhi,andVreeland
(2010).Doingsohasnobearingonourfindings.NextwereplacethePennWorldTables
dataonGDPandthechangeinGDPwithdatadrawnfromtheWorldBank'sWorld
DevelopmentIndicators.AkeydifferencebetweenthesedatasetsisthattheWDIdatado
notstartuntil1960soalltheyearsbetween1950and1960aredroppedwiththis
variable.16Thischangedoesnotalterourmainresults.
Next,wewanttomakesurethatourresultsdonotsimplyreflecttheEuropean
Community/EuropeanUnion(EU),whichiscomprisedofdemocraciesthathaveformed
manytradeagreementssinceWorldWarII.WethereforestartedbydroppingallEuropean
countriesandthendropallpairsinwhicheithercountryisanEUmember.Neitherof
thesetestsyieldsanychangestoourcoreresults.Inaddition,ourresultsarevirtually
unchangedwhenweestimatethebasemodelusingarareeventslogitspecification,which
accountsforthefactthattheformationofaPTAisanuncommonoccurrence(Kingand
Zeng2001).Equally,althoughwethinkaoneyearlagismostappropriategiventhe
averagetimebetweenthesigningandratificationofatradeagreement,wealsomeasured
16ThecorrelationbetweentheWDIandthePennWorldTablesmeasureisonly0.80.ThismaybebecausethePennWorldTablesdataareinconstantinternationaldollarswhiletheWDIdataareinUSdollars.WepreferthePennWorldTablesdataduetothelongertemporalcoverage.
Page 30
30
thepercentagechangeinGDPovertwo‐yearandfive‐yearintervals.Inbothcases,no
matterwhichdemocracycutoffweuse,ourfindingsstillhold.
Further,weanalyzethefirstPTAthatagivenpairofcountriesformsanddropany
subsequentagreementthattheyjoinfromthesample.Wealsoexplorewhetherourresults
changeifwecodethedependentvariablewhenPTAsaresignedbymember‐states,rather
thanwhentheyareratified.Noneofthesetestsyieldanychangestoourresults.
Finally,weaccountforseveraldomesticvariablesthatmightaffecttheobserved
influenceofregimetypeandthebusinesscycleonPTAformation.First,weincludevarious
measuresofthetimingofnationalelections,butfindnoevidencethattheyinfluenceeither
tradeagreementsortheeffectsofregimetypeandthebusinesscycleonsuchagreements.
Second,weexaminethepartisanorientationofgovernment.Interestingly,wefind
preliminaryevidencethatright‐winggovernmentsarelesslikelytojoinPTAsthanleft‐
winggovernments,andthatleftistgovernmentsindemocraticcountriesareespecially
likelytoenterPTAs.Nonetheless,introducingpartisanshipinourstatisticalmodeldoes
notaltertheeffectsofregimetypeandthebusinesscycle.Insum,ourresultsseemtobe
veryrobust.
Conclusions
Thereceivedwisdomisthateconomicdownturnspromoteprotectionism.Itis
thereforesurprisingthateconomichardtimessometimesleadpoliticalleaderstonegotiate
andratifytradeagreementsthatreducetradebarriersandpromotetradeamongthe
Page 31
31
members.Wehavearguedthatdomesticpoliticalincentivesexistfordemocraticleaders
toestablishtradeagreementsduringhardeconomictimes.Ofcourse,domesticpoliticsis
notthesolefactorshapingPTAs;wehavefoundthatawidevarietyofeconomicand
internationalpoliticalvariablesalsoexertastronginfluenceoftheestablishmentofthese
agreements.ButthedomesticpoliticallogicofPTAshasbeenunderstudiedand
underappreciatedtodate,agapthatwehaveaimedtohelpfillinthisstudy.
PTAssignaltothepublicthatacountry’sleaderisnottotallycapturedby
protectionistspecialinterests.Hence,wheneconomictroublesarise,votersandpro‐trade
interestgroupsarelesslikelytoblametheleaderforthem.Establishingtradeagreements
thattendtoliberalizeandexpandoverseascommerceindicatestothepublicthatitsleader
isnotexploitativeorincompetentandthathardeconomictimesshouldbeattributedto
sourcesbeyondhisorhercontrol.Underthesecircumstances,leadershaveabetter
chanceofretainingofficeinthefaceofbadeconomictimes.Leadersrealizepoliticalgains
asaresultandthegeneralpublicalsobenefitsfromfreertrade.
Whereleadersfacegreaterpoliticalcompetition,theseconsiderationsare
particularlyimportant.Inmoredemocraticsettings,leadersaremoreconcernedwithhow
thepublicreactstobadeconomictimesandthusmorelikelytoenactpoliciesthatreassure
thepublic.Tradeagreementshelptoprovidesuchreassurance.Wethereforeexpectthat
leadersinmoredemocraticpoliticalenvironmentswillbemorelikelytonegotiateand
ratifyPTAsinbadtimesthanotherwise.
Page 32
32
Autocratsareoftenlesssusceptibletothepoliticalconsequencesofeconomic
downturns(BuenodeMesquitaetal.2003).Hence,theyhavelessreasontopursuetrade
agreementsingeneralandduringbadtimes.Indeed,duringdipsinthebusinesscycle,they
mayavoidmakingagreementsbecausethedistributionaleffectsofdoingsomayharm
theirsupporters.Autocratsoftendependonthesupportofthemajorsectorsofthe
economyandmay,inturn,heavilyprotectthesesectorstogeneratepoliticalsupport.
Reducingtradebarriersinbadtimesmayunderminethissupportandthusjeopardize
theirholdonpower.Consistentwiththisobservation,wefindthatthebusinesscyclehas
relativelittlebearingonwhennon‐democraciesentertradeagreements.
Ourresearchsuggestssomegoodnews.PTAsdolesstopromotewelfarethan
unilateralormultilateraltradeliberalization;butwiththeWTOstrugglingtoadvance
multilateralliberalizationandthedifficultiesthatmanycountriesfaceunilaterally
liberalizingforeigncommerce,PTAsmaybethebestwaytokeeptheglobaltradingsystem
open.Moreover,itisgoodnewsthatdemocracyhasbeenspreadinggloballyandthat
democraticleadershavepoliticalreasonstoresistprotectionisminthefaceofbad
economictimes.
Page 33
33
Table1.Descriptivestatistics.Obs. Mean Std.Dev. Min. Max.
PTARatification 1021530 0.009 0.094 0 1
Democracy≥20 1021530 0.259 0.438 0 1
Democracy≥16 1021530 0.472 0.499 0 1
Democracy≥12 1021530 0.533 0.499 0 1
ΔGDP 1021530 0.019 0.072 ‐0.656 1.191
ΔGDP×Democracy≥20 1021530 0.003 0.019 ‐0.199 0.291
ΔGDP×Democracy≥16 1021530 0.007 0.032 ‐0.310 0.296
ΔGDP×Democracy≥12 1021530 0.009 0.036 ‐0.337 0.315
VetoPlayers 1021530 0.227 0.218 0 0.720
ExistingPTA 1021530 0.106 0.308 0 1
Trade 1021530 ‐2.622 4.978 ‐6.908 13.016
GDP 1021530 17.319 2.041 12.161 23.298
Dispute 1021530 0.000 0.021 0 1
Ally 1021530 0.106 0.308 0 1
FormerColony 1021530 0.007 0.082 0 1
Contiguity 1021530 0.020 0.140 0 1
Distance 1021530 8.735 0.763 2.349 9.901
Hegemony 1021530 0.223 0.033 0.165 0.315
GDPRatio 1021530 2.436 1.794 0.000 11.077
Post‐ColdWar 1021530 0.598 0.490 0 1%DyadsRatifyingPTA 1021530 0.008 0.009 0 0.038
GlobalBusinessCycle 1021530 2.963 2.098 ‐4.529 7.579
GATT 1021530 0.471 0.499 0 1
Page 34
34
Table2.EstimatedeffectsofregimetypeandthebusinesscycleonPTAratification,1951‐2011.
Democracy≥20 Democracy≥16 Democracy≥12RegionFEs CountryFEs RegionFEs CountryFEs RegionFEs CountryFEs
Democracy 0.545** 0.768** 0.305** 0.301** 0.129** ‐0.043(0.039) (0.056) (0.036) (0.047) (0.036) (0.048)
ΔGDP 0.232* 0.468** 0.274* 0.529** 0.346** 0.629**(0.115) (0.125) (0.125) (0.133) (0.126) (0.135)
Democracy×ΔGDP
‐4.774** ‐5.706** ‐1.883** ‐2.258** ‐1.820** ‐2.256**(0.532) (0.563) (0.306) (0.344) (0.275) (0.308)
VetoPlayers ‐0.521** ‐1.019** ‐0.619** ‐1.096** ‐0.384** ‐0.755**(0.071) (0.096) (0.085) (0.105) (0.085) (0.108)
ExistingPTA 0.133* ‐0.170** 0.133* ‐0.157** 0.130* ‐0.155**(0.054) (0.056) (0.054) (0.056) (0.054) (0.056)
Trade 0.034** 0.056** 0.035** 0.056** 0.034** 0.055**(0.003) (0.004) (0.003) (0.004) (0.003) (0.004)
GDP ‐0.010 ‐0.346** 0.003 ‐0.343** 0.000 ‐0.358**(0.011) (0.051) (0.011) (0.051) (0.011) (0.051)
Dispute 0.429 0.259 0.465 0.263 0.452 0.249(0.257) (0.264) (0.257) (0.265) (0.256) (0.265)
Ally 0.308** 0.848** 0.300** 0.842** 0.303** 0.845**(0.054) (0.052) (0.054) (0.052) (0.053) (0.052)
FormerColony ‐1.739** ‐1.959** ‐1.764** ‐1.959** ‐1.754** ‐1.956**(0.400) (0.436) (0.400) (0.435) (0.400) (0.435)
Contiguity ‐0.596** ‐0.734** ‐0.607** ‐0.733** ‐0.611** ‐0.732**(0.062) (0.061) (0.062) (0.061) (0.062) (0.061)
Distance ‐1.041** ‐1.116** ‐1.040** ‐1.115** ‐1.041** ‐1.114**(0.050) (0.044) (0.050) (0.043) (0.050) (0.043)
Hegemony 10.739** 8.848** 10.491** 8.352** 10.344** 7.995**(0.573) (0.822) (0.577) (0.829) (0.578) (0.831)
Post‐ColdWar 0.964** 1.134** 0.931** 1.100** 0.936** 1.138**(0.042) (0.046) (0.042) (0.046) (0.042) (0.046)
GDPRatio ‐0.174** ‐0.201** ‐0.173** ‐0.200** ‐0.173** ‐0.200**(0.009) (0.010) (0.009) (0.010) (0.009) (0.010)
%DyadsRatifyingPTA
44.733* 44.355** 44.139** 43.447** 44.319** 43.460**(1.111) (1.156) (1.115) (1.160) (1.112) (1.157)
GlobalBusinessCycle
‐0.113** ‐0.114** ‐0.112** ‐0.113** ‐0.112** ‐0.113**(0.006) (0.006) (0.006) (0.006) (0.006) (0.006)
Page 35
35
GATT 0.182** 0.212** 0.179** 0.211** 0.193** 0.231**(0.029) (0.043) (0.029) (0.043) (0.029) (0.043)
Constant 0.933* 7.693** 0.840 8.251** 0.934* 8.807**
(0.456) (1.364) (0.456) (1.370) (0.455) (1.374)
N 1033945 1021530 1033945 1021530 1033945 1021530Clusters 29394 28598 29394 28598 29394 28598Log‐likelihood ‐42778.87 ‐40009.48 ‐42855.35 ‐40099.8 ‐42879.23 ‐40112.61Note:Entriesarelogisticregressionestimateswithrobuststandarderrorsclusteredbydyadinparentheses.Statisticalsignificanceisindicatedasfollows:**p<0.01;*p<0.05.Alltestsofsignificancearetwotailed.
Page 36
36
Figure1.Predictedprobabilityofdemocraciesandnon‐democraciesratifyingaPTAundervariousdomesticeconomicconditions,1951‐2011,wheredemocracyisdefinedasaPolityscoreof20orgreater.
.00
2.0
04
.00
6.0
08
.01
.01
2P
rob
abili
ty o
f ra
tific
atio
n
-.08 -.06 -.04 -.02 0 .02 .04 .06 .08 .1Change in GDP
Polity<20 Polity>=20
PTA ratification probabilityPolity>=20
Note:Dashedlinesrepresent95%confidenceintervals.
Page 37
37
Figure2.Predictedprobabilityofdemocraciesandnon‐democraciesratifyingaPTAundervariousdomesticeconomicconditions,1951‐2011,wheredemocracyisdefinedasaPolityscoreof16orgreater.
.00
3.0
04
.00
5.0
06
.00
7P
rob
abili
ty o
f ra
tific
atio
n
-.08 -.06 -.04 -.02 0 .02 .04 .06 .08 .1Change in GDP
Polity<16 Polity>=16
PTA ratification probabilityPolity>=16
Note:Dashedlinesrepresent95%confidenceintervals.
Page 38
38
Figure3.Predictedprobabilityofdemocraciesandnon‐democraciesratifyingaPTAundervariousdomesticeconomicconditions,1951‐2011,wheredemocracyisdefinedasaPolityscoreof12orgreater.
.00
3.0
035
.00
4.0
045
.00
5.0
055
Pro
bab
ility
of
ratif
ica
tion
-.08 -.06 -.04 -.02 0 .02 .04 .06 .08 .1Change in GDP
Polity<12 Polity>=12
PTA ratification probabilityPolity>=12
Note:Dashedlinesrepresent95%confidenceintervals.
Page 39
39
ReferencesBagwell,Kyle,andRobertW.Staiger.2003.ProtectionandtheBusinessCycle.Advancesin
EconomicAnalysis&Policy3(1):Art.3.Beck,Nathaniel,andJonathanN.Katz.2001.ThrowingOuttheBabywiththeBathWater:A
CommentonGreen,Kim,andYoon.InternationalOrganization55(2):487‐95.Beck,Nathaniel,JonathanN.Katz,andRichardTucker.1998.TakingTimeSeriously:Time‐
Series‐Cross‐SectionAnalysiswithaBinaryDependentVariable.AmericanJournalofPoliticalScience42(4):1260‐88.
Bohara,Alok,andWilliamH.Kaempfer.1991.ATestofTariffEndogeneityintheUnitedStates.AmericanEconomicReview81(September):952‐60.
Bown,ChadP.(ed.)2011.TheGreatRecessionandImportProtection:theRoleofTemporaryTradeBarriers.Washington,DC:WorldBank.
BuenodeMesquita,Bruce,AlastairSmith,RandolphM.Siverson,andJamesD.Morrow.2003.TheLogicofPoliticalSurvival.Cambridge,MA:MITPress.
Cassing,James,TimMcKeown,andJackOchs.1986.ThePoliticalEconomyoftheTariffCycle.AmericanPoliticalScienceReview80(3):843‐62.
Cheibub,JoséAntonio,JenniferGandhi,andJamesRaymondVreeland.2010.DemocracyandDictatorshipRevisited.PublicChoice143(1‐2):67‐101.
Freund,Caroline,andEmanuelOrnelas.2010.RegionalTradeAgreements.AnnualReviewofEconomics2(1):139‐166.
Gawande,Kishore,PravinKrishna,andMarceloOlarreaga.2009.WhatGovernmentsMaximizeandWhy:TheViewfromTrade.InternationalOrganization63(3):491‐532.
Gawande,Kishore,BernardHoekman,andYueCui.2014.GlobalSupplyChainsandTradePolicyResponsestothe2008Crisis.WorldBankEconomicReview(advanceaccessJanuary23,2014):1‐27.
Ghosn,Faten,andGlennPalmer.2003.CodebookfortheMilitarizedInterstateDisputeData,Version3.0,fromhttp://cow2.la.psu.edu
Gleditsch,NilsPetter,PeterWallensteen,MikaelEriksson,MargaretaSollenberg,andHavardStrand.2002.ArmedConflict1946‐2001:ANewDataset.JournalofPeaceResearch39(5):615‐37.
Gowa,Joanne.1994.Allies,Adversaries,andInternationalTrade.Princeton,NJ:PrincetonUniversityPress.
Grossman,GeneM.,andElhananHelpman.1994.ProtectionforSale.AmericanEconomicReview84(4):833‐50.
Gruber,Lloyd.2000.RulingtheWorld:PowerPoliticsandtheRiseofSupranationalInstitutions.Princeton,NJ:PrincetonUniversityPress.
Gurr,TedR.,KeithJaggers,andWillH.Moore.1989.PolityII:PoliticalStructuresandRegimeChange,1800‐1986(Vol.9263).AnnArbor,MI:Inter‐UniversityConsortiumforPoliticalandSocialResearch.
Page 40
40
Harbom,Lotta,HåvardStrand,andHåvardMokleivNygård.2008.UCDP/PRIOArmedConflictDatasetCodebookVersion4‐2008.UppsalaConflictDataProgram;CentrefortheStudyofCivilWars,InternationalPeaceResearchInstitute.Uppsala,Oslo.
Head,Keith,ThierryMayer,andJohnRies.2010.TheErosionofColonialTradeLinkagesafterIndependence.JournalofInternationalEconomics81(1):1‐14.
Henisz,WitoldJ.2000.TheInstitutionalEnvironmentforMultinationalInvestment.JournalofLaw,EconomicsandOrganization16(2):334‐64.
_____.2002.TheInstitutionalEnvironmentforInfrastructureInvestment.IndustrialandCorporateChange11(2):355‐89.
Herrmann,RichardK.,PhilipE.Tetlock,andMatthewN.Diascro.2001.HowAmericansThinkAboutTrade:ReconcilingConflictsamongMoney,Power,andPrinciples.InternationalStudiesQuarterly45(2):191‐218.
Heston,Alan,RobertSummers,andBettinaAten.2011.PennWorldTableVersion7.0.CenterforInternationalComparisonsofProduction,IncomeandPrices.UniversityofPennsylvania.Philadelphia,PA.
InternationalMonetaryFund(IMF).2010.APolicy‐Driven,MultispeedRecovery.WorldEconomicOutlookUpdate.Washington,DC:IMF.Retrievedfromhttp://www.imf.org/external/pubs/ft/weo/2010/update/01/pdf/0110.pdf
_____.(variousyears).DirectionofTradeStatistics(DOTS).Wahington,DC:InternationalMonetaryFund.
Jaggers,Keith,andRedR.Gurr.1995.TrackingDemocracy'sThirdWavewiththePolityIIIData.JournalofPeaceResearch32(4):469‐82.
Jones,DanielM.,StuartA.Bremer,andJ.DavidSinger.1996.MilitarizedInterstateDisputes,1816–1992:Rationale,CodingRules,andEmpiricalPatterns.ConflictManagementandPeaceScience15(2):163‐213.
Kee,HiauLooi,CristinaNeagu,andAlessandroNicita.2013.IsProtectionismontheRise?AssessingNationalTradePoliciesduringtheCrisisof2008.ReviewofEconomicsandStatistics95(1):342‐46.
Kim,In‐Song.2013.PoliticalCleavageswithinIndustry:FirmLevelLobbyingforTradeLiberalization.WorkingPaper.DepartmentofPolitics.PrincetonUniversity.
Kindleberger,CharlesP.1973.TheWorldinDepression.Berkeley,CA:UniversityofCaliforniaPress.
King,Gary.2001.ProperNounsandMethodologicalPropriety:PoolingDyadsinInternationalRelationsData.InternationalOrganization55(2):497‐507.
King,Gary,andLangcheZeng.2001.LogisticRegressioninRareEventsData.PoliticalAnalysis9(2):137‐163.
Kurian,GeorgeThomas.1992.EncyclopediaoftheThirdWorld(4thed.ed.Vol.1‐3).NewYork:FactsonFile.
Leeds,Brett,JeffreyRitter,SaraMitchell,andAndrewLong.2002.AllianceTreatyObligationsandProvisions,1815‐1944.InternationalInteractions28(3):237‐60.
Maggi,Giovanni,andAndresRodriguez‐Clare.1998.TheValueofTradeAgreementsinthePresenceofPoliticalPressures.JournalofPoliticalEconomy106(3):574‐601.
Page 41
41
_____.2007.APolitical‐EconomyTheoryofTradeAgreements.AmericanEconomicReview97(4):1374‐406.
Manger,Mark.2009.InvestinginProtection:ThePoliticsofPreferentialTradeAgreementsbetweenNorthandSouth.NewYork:CambridgeUniversityPress.
Mansfield,EdwardD.1993.EffectsofInternationalPoliticsonRegionalisminInternationalTrade.InKymAndersonandRichardBlackhurst,eds.,RegionalIntegrationandtheGlobalTradingSystem,pp.199‐217.NewYork:HarvesterWheatsheaf.
Mansfield,EdwardD.,andHelenV.Milner.2012.Votes,Vetoes,andthePoliticalEconomyofInternationalTradeAgreements.Princeton,NJ:PrincetonUniversityPress.
Mansfield,EdwardD.,andEricReinhardt.2003.MultilateralDeterminantsofRegionalism:TheEffectsofGATT/WTOontheFormationofPreferentialTradingArrangements.InternationalOrganization57(4):829‐62.
Marshall,MontyG.,andKeithJaggers.2009.PolityIVProject:PoliticalRegimeCharacteristicsandTransitions,1800‐2009.fromIntegratedNetworkforSocietalConflictResearch(INSCR)Program,andCenterforSystemicPeace(CSP)http://www.systemicpeace.org/polity/polity4.htm.
Martin,LisaL.2000.DemocraticCommitments:LegislaturesandInternationalCooperation.Princeton,NJ:PrincetonUniversityPress.
Milner,HelenV.1988.ResistingProtectionism:GlobalIndustriesandthePoliticsofInternationalTrade.Princeton,NJ:PrincetonUniversityPress.
Mitra,Devashish.2002.EndogenousPoliticalOrganizationandtheValueofTradeAgreements.JournalofInternationalEconomics57(2):473‐85.
Nye,JosephS.1988.NeorealismandNeoliberalism.WorldPolitics40(2):235‐251.OrganisationforEconomicCooperationandDevelopment(OECD).2010.Tradeand EconomicEffectsofResponsestotheEconomicCrisis.Paris:OECD.Przeworski,Adam,MichaelE.Alvarez,JoseAntonioCheibub,andFernandoLimongi.2000.
DemocracyandDevelopment:PoliticalInstitutionsandWell‐BeingintheWorld,1950‐1990.NewYork:CambridgeUniversityPress.
Rose,AndrewK.2012.ProtectionismIsn'tCounter‐Cyclic(anymore).NBERWorkingPaper18062.NationalBureauofEconomicResearch.Cambridge,MA.
Staiger,RobertW.,andGuidoTabellini.1999.DoGATTRulesHelpGovernmentsMakeDomesticCommitments?EconomicsandPolitics11(2):109‐44.
Tsebelis,George2002.VetoPlayers:HowPoliticalInstitutionsWork.Princeton,NJ:PrincetonUniversityPress.
Yarbrough,BethV.,andRobertM.Yarbrough1992.CooperationandGovernanceinInternationalTrade:theStrategicOrganizationalApproach.Princeton,NJ:PrincetonUniversityPress.