Peace through Trade or Free Trade?
Author(s): Patrick J. McDonald
Source: The Journal of Conflict Resolution, Vol. 48, No. 4 (Aug., 2004), pp. 547-572
Published by: Sage Publications, Inc.
Stable URL: http://www.jstor.org/stable/4149808
Accessed: 15/10/2010 11:42
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=sage.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of
Conflict Resolution.
http://www.jstor.org
Peace through Trade or Free Trade?
PATRICKJ. McDONALD
Departmentof Government
Universityof Texasat Austin
This study argues that a subtle shift in the primary independent variable of the commercial peace
literature-from tradeto free trade-provides an opportunityto respondto the some of the strongestcriticisms of this researchprogram.Free trade, and notjust trade,promotes peace by removing an important
foundationof domestic privilege-protective barriersto internationalcommerce-that enhancesthe domestic power of societal groups likely to support war, reduces the capacity of free-tradinginterests to limit
aggression in foreign policy, and simultaneouslygeneratespolitical supportfor the state often used to build
its war machine. A series of statisticaltests demonstratesthat higher levels of free trade,ratherthan trade
alone, reducemilitaryconflict between states. Moreover,contraryto conventionalwisdom, these arguments
suggest how the puzzlingcase of WorldWarI may confirm,ratherthancontradict,the centralclaims of commercial liberalism.
Keywords: commercialliberalism;free trade;peace; tradepolicy
The debateover whetherandhow internationalcommercealtersthe foreign policy of
states, and in particularthe decision for war, has gained renewed prominence in the
fields of internationalsecurityand internationalpoliticaleconomy. Despite substantial
empiricalsupportfor the propositionthatincreasinglevels of cross-bordereconomic
flows-defined eitherin termsof tradeor capitalmovements-decrease the probability of conflict, scholarshaveyet to approacha consensus concerningthe precise nature
of this link. A numberof explanations have been proposed. Trade promotes peace
through communication and transnationalties that increase understandingamong
societies and the potentialfor cooperation.While expanding an internationalweb of
commerce throughspecialization,trademakes war less likely by increasingthe costs
of severing such economic links. Interdependencemakes conflict less likely because
of its efficiency over conquest in acquiring resources necessary for growth and
prosperity.
AUTHOR'S NOTE:Previousversions of this studywere presentedat the 2002 AnnualMeetings of the
Midwest Political Science Association and the 2002 Annual Meetings of the American Political Science
Association. I would like to thank David Bearce, Dale Copeland,Tim Frye, Ed Mansfield, John Mueller,
Dave Rowe, KevinSweeney,andparticipantsat the Researchin InternationalPolitics Seminarat Ohio State
for commentson earlierversionsof this study.Thankyou to the ChristopherH. Browne Centerfor International Politics at the Universityof Pennsylvaniafor financialsupportof this researchandto Michael Hiscox
andScott Kastnerfor sharingtheirdata.The dataused for this studyareavailableat www.yale.edu/unsy/jcr/
jcrdata.htm/.
OFCONFLICT
Vol.48 No.4, August2004547-572
JOURNAL
RESOLUTION,
DOI:10.1177/0022002704266117
O 2004 Sage Publications
547
548
JOURNALOF CONFLICTRESOLUTION
Yet these hypotheses remain subject to two key theoretical challenges. First, by
imposing costs on states that withdrawfrom the internationaldivision of labor, the
expansionof commerce may instead possess an indeterminateeffect on the outbreak
of conflict. On one hand, it may deter a dependentpartnerfrom using force. At the
same time, the recognitionof this weakness on the partof other states may embolden
potentialadversaries(e.g., Morrow 1999; Gartzke,Li, and Boehmer 2001). Second,
most explanationsoften rely on a series of assumptionsconcerning the natureof relations between state and society thatare rarelydiscussed or accountedfor in empirical
tests (Stein 1993; Barbieriand Schneider 1999; Mansfield and Pollins 2001; Schneider and Schulze 2003; Simmons 2003). A recent trendin the interdependenceliteraturehas begun to addressthis lattershortcomingby exploringthe interactiveeffects of
trade and democratic institutions on foreign policy and the outbreak of conflict
(Weede 1995; Papayoanou1999; Gelpi and Grieco 2003).
This study buildson this recentworkby exploringanothersource of domestic variation-the extentto which governmentsregulateinternationaltrade.I arguethata subtle refinementin the logic of commercial liberalism, namely, a shift from aggregate
tradeflows to the level of free tradeas the crucial independentvariable,provides the
opportunityto push the debate forwardon a numberof fronts.
Theoretically,this shift enables the incorporationof importantaspects of statesociety interactionsthatshapeany link between tradeandconflict. Althoughit is yet to
be fully integratedinto the commercialpeace debate, standardtradetheory illustrates
thatinternationalcommerceincreasesthe aggregateincome of an economy and simultaneouslyaltersthe relativedistributionof income across society. Groupsthatsee their
incomes decline from internationaltrade, namely, import-competing sectors, are
unlikely to lobby the state for a pacific foreign policy that promotesexpandingtransnationaleconomic ties. Moreover,the stateis not a neutralarbiterin the domestic battle over commercial and foreign policies. It can use economic regulationto co-opt
societal supportfor its public policies, includingthose thatlead to war.A focus on free
tradeor the extent to which states regulatecommerce in response to societal demands
shifts theoreticalattentiontowardthe domestic level of analysis and allows me to generate hypotheses linking these distributionalconsequences of commerce to peace.
This shift also carriesimportantempiricalimplications.Most of the literaturerelies
on bilateraltradeto gross domestic product(GDP) ratios to operationalizesuch concepts as the relativedependence of an economy on tradeand test the claims of commercialliberalism.'Here I add more directmeasuresof the level of regulatorybarriers
on tradeto standardstatisticalmodels of conflict. Their inclusion allows me to separateout the respectiveeffects of free tradeand tradeon conflict while comparingthe
domestic explanation presented here with alternativehypotheses more commonly
referredto in the literature.
Morebroadly,this studyarguesthata neglectedversionof commercialliberalismrootedin standardtradetheoryand the classical writingsof Cobden (1868, 1870) and
Schumpeter(1919/1951)-Isheds new light on how internationalcommercegenerates
1. For exceptions to this, see Mansfield, Pevehouse, and Bearce (1999-2000); Mansfield and
Pevehouse (2000); and Polachek, Robst, and Chang (1999).
McDonald / PEACE THROUGHTRADEOR FREE TRADE?
549
peace between states. Free trade, and not just trade, promotes peace by removing
an importantfoundation of domestic privilege-protective barriersto trade-that
enhances the domestic power of societal groups likely to support war, reduces the
capacity of free-tradinginterests to limit aggression in foreign po.'cy, and createse
mechanismby which the state can build supportivecoalitions for war.A series of statistical tests supportsthese claims by showing that lower regulatorybarriersto trade
were associated with a reductionin militaryconflict between states duringthe postWorldWarII era.
The rest of this study proceeds as follows. First, I outline the classical liberalroots
and the contemporaryliteraturethatexplores the hypothesis that free tradepromotes
peace. Second, I discuss how these insights find supportin contemporarytradetheory
and neoinstitutionaleconomics, both of which illustratehow societal and state interests for peace and warcan be shapedby exposureto the global economy. Third,I present a series of statistical tests to support the link between protection and conflict.
Finally,I conclude anddiscuss how these argumentssuggest an importantresponseto
one of the standardcriticisms of commercial liberalism, namely, that globalization
failed to preventthe outbreakof WorldWarI.
INTERDEPENDENCE AND PEACE:
MOVING BEYOND RICARDO
An extensive base of empiricaltests across a numberof researchdesigns-including differences in the operationalizationof the independentand dependentvariables,
the temporaldomain under study, and the unit of analysis-support the conclusion
that internationalcommerce promotes peace among states (e.g., Polachek 1980;
Domke 1988; Mansfield 1994; Oneal and Russett 1997, 1999; Russett and Oneal
2001; Gartzke,Li, and Boehmer 2001; for an exception, see Barbieri2002). The contemporarydebate has traditionallyrelied on four variantsof the broaderhypothesis
thattradepromotespeace. The firsthas been labeledthe opportunitycost or deterrence
model. Because conflict or even the threatof it tendsto disruptnormaltradingpatterns,
potentiallylargeeconomic costs will deterdependentstates from using militaryforce
to solve their political conflicts (Polachek 1980). A second mechanism that I call
here the "efficiency argument"compares the relative costs of acquiring productive
resources.As commerce grows, the incentives for plunderor conquest decrease simply because it is a more costly means of generatingeconomic growth (Rosecrance
1986). Third, a sociological hypothesis concentrateson how trade helps to increase
contact and communication across societies. By building a broader cosmopolitan
identity across societies, tradedisplaces national loyalties and competitive relations
between governments that generate military conflict (e.g., Deutsch et al. 1957).
Fourth,drawing on bargainingmodels, some scholars argue that internationalcommerce provides an importantsignaling mechanism that can help states achieve a
negotiatedcompromise shortof war duringa crisis (e.g., Morrow 1999; Gartzke,Li,
and Boehmer 2001).
550
JOURNALOF CONFLICTRESOLUTION
Despite substantialempirical evidence for these hypotheses, scholars have challenged these claims for failing to elaboratea model of domestic politics thatlinks commerce, societal interestsin tradeand peace, and the state with the creationof a pacific
foreign policy (Stein 1993; Barbieriand Schneider 1999; Mansfieldand Pollins 2001;
Schneiderand Schulze 2003; Simmons2003). This oversightis particularlysignificant
given the concentrationof liberalinternationalrelationstheoryon the behaviorof individuals and state-society interactionsto explain foreign policy behavior (Moravcsik
1997).
The neglect of domestic politics stems in partfrom a failureto incorporateimportant revisions to standardtradetheory into the commercial peace debate (Schneider
and Schulze 2003). Most of the literaturedrawson Ricardo(1821/1973) and the principles of comparativeadvantage,which demonstratesthat economies increase their
aggregate consumption possibilities (or economic growth more generally) through
specializationand trade.Accordingly,the standardhypotheses focus on these aggregate welfare benefits to explain any link between commerce and peace. For example,
this claim helps motivatethe opportunitycost hypothesis. Fearingthe aggregatecosts
of economic disruption, societies will lobby the governments for a pacific foreign
policy.
By concentratingon Ricardo'svaluableinsight, the literatureoverlooks how trade
based on comparativeadvantagealso creates societal coalitions opposed to further
integration (e.g., Rogowski 1989). Both the Heckscher-Ohlin and Ricardo-Viner
frameworks demonstrate that economic integration redistributesincome within a
domestic society.2Because some groups see their incomes decline from international
trade,they are unlikely to supportopen markets.Such a conclusion casts doubton the
proposition that trading sectors in society will always win out among competing
domestic interestsandpossess the political capacityto constrainan aggressiveforeign
policy.' Consequently,littleresearchhas been conductedon how these domesticdistributionalconsequencesof commercemay also shapethe domestic politics of decisions
for war and peace.4
Moreover,the process by which these trading interests translatepreferences for
maintainingan open economy into foreign policy outputs that include both an open
tradingsystem and a more pacific orientationin foreign policy must also account for
the structureof domestic institutionsthat mediate societal conflict, aggregate these
economic interests,and determinepolicy. The opportunity-costhypothesis implicitly
adopts a pluralistmodel of domestic politics that fails to acknowledge that the struc2. The original studiesof HeckscherandOhlin anda review of theircontributionsto tradetheorycan
be found in Flam and Flanders(1991). On the Ricardo-Vinerframework,see Jones (1971) and Samuelson
(1971).
3. Similarly,workon political or state-ownedfirmssuggests thatall enterprisesdo not respondto the
profit motive that drives the opportunitycost hypothesis (Eggertsson 1990). Firmsand societal groupsare
hypothesizedto lobby the state for peace, fearingthe adjustmentcosts following economic disruptionsdue
to militaryconflict.If mostdomesticfirmsarepubliclyownedandpossessalternative
goals,suchas maximizing employment,they may be less reactiveto such priceshocks andconsequentlyless willing to pressure
the state for peace.
4. Inan importantstudyexamininghow globalizationplayeda centralrole in the originsof WorldWar
I, Rowe (1999) provides an exception to this by disaggregatingsocietal interestswith respect to tradeand
war accordingto relativefactorendowments.
McDonald / PEACE THROUGHTRADEOR FREE TRADE?
551
tureof domestic institutionscan privilege portionsof society preferringa closed trading system and thatthe statepossesses an independentcapacityto arbitrateamong and
shape these interests. One strandof the interdependenceliteraturehas responded to
this shortcomingby examininghow democracyconditionsthe effects of commerceon
conflict (Brawley 1993; Papayoanou 1999; Gelpi and Grieco 2003). The ability of
commerceto promotepeace may be restrictedto democraticstatesbecause the groups
most hurtby interruptionsin commerce can successfully lobby their governmentsin
these polities for more peaceful foreign policies. Whereasregime type provides one
means to characterizestate-societyinteractions,it may not captureall of the variation
acrosseconomies. Historically,import-competingsectorshave been able to lobby and
obtain protectionistlegislation in democracies.5This possibility suggests the need to
explore the independenteffects of protectionon conflict.
THE SECOND-IMAGE DYNAMICS
OF A COMMERCIAL PEACE
This section argues that a neglected fifth variantof the commercial liberalismmotivated by the writings of Cobden (1868, 1870), Schumpeter (1919/1951), and
standardtrade theory--offers importantinsights into how internationalcommerce
shapesthe domestic politics of war.These classical scholarsconnect the domestic politics of internationalconflict with the domestic distributionalconsequence of commercial policy to explain how trade promotes peace. The elimination of protective
commercial policies empowers societal groups most opposed to war and constrains
the abilityof governmentsto redistributethe costs of waronto groupsoutside its ruling
coalition. Together,these twin pressuressuggest that the adoption of free tradepolicies andnot simply an increasein aggregateeconomic integrationshouldbode well for
peace.
Both Cobden (1868, 1870) and Schumpeter(1919/1951) built these explanations
froma broaderliberalmodel of war,which holds thatthe externalcosts andbenefits of
waging warare inseparablefrom its domestic costs andbenefits.6Justas a stategoes to
warto extend its influence in the internationalsystem or eliminate a threatto its security,war simultaneouslycreatesopportunitiesfor a governmentto redistributeincome
towardits political supportersand solidify its domesticposition. Forexample, Cobden
(1868, 1,44-45; 1870, 2,429) arguedthatwarsand"warscares"allowed governments
to postpone domestic reforms that would necessarily expand individual liberty and
5. The United States standsout as an importantexample to this. Tariffswere the primarysource of
public revenuesuntilWorldWarI. For a discussion of the relationshipbetween tariffsand taxes in the U.S.
case, see fHansen(199tY).
6. Summarizinga liberaltheory of war,Michael Howard(1978, 31) writes,
By the end of the eighteenthcenturya complete liberaltheoryof internationalrelations,of war and
peace, had thus alreadydeveloped.... Accordingto this doctrine,mankindwould naturallylive in a
stateof perfectharmonyif it were not for the vested interestsof governments.... The whole "warsystem"was contrivedto preservethe powerandthe employmentof princes,statesmen,soldiers,diplomats,andarmamentsmanufacturers,andto bindtheirtyrannyever morefirmlyuponthe necks of the
people.
552
JOURNALOF CONFLICTRESOLUTION
limit the role of governmentin domestic life. To preventwar, societies thereforeneed
mechanismsto monitorand punish their government'seffort to utilize foreign policy
for domestic political gain.7
The next step linking tradeand peace focused on the domestic distributionalimplications of commercialpolicy. Behind much of the clamorfor free tradein the classical
literaturewas a strong opposition to monopolies in the domestic economy. Because
tariffstendedto shield noncompetitivesectors and shift the distributionof wealth in a
society towardthese groupsandawayfromconsumers,the political motivationbehind
free tradewas just as often domestic as it was international,thatis, to promotepeace.8
By removingtariffsand encouragingfree trade,a transformationof the domestic balance of power would necessarily empower broaderelements of society and simultaneously erode regulatoryprotectionfor merchants.9
The consequences of free tradeon the domestic distributionof power then shaped
the domestic distributionalimplicationsof war.Just like restrictivecommercialpolicies, the costs of war generally fall on the poorerelements of society who possess no
interestin conflict. The eliminationof traderestrictionsunderminedthe ability of the
stateandprotectedsectorsof the economy or the domestic groupsmost responsiblefor
war to shift the burdensof public finance onto disorganizedmembersof society who
were benefitingmost fromopen internationalmarkets.In this second-imagevariantof
the commercialpeace hypothesis,the abilityof commerce to promotepeace depended
crucially on trade's ability to alter the distributionof domestic political power. Free
tradeand not necessarily tradewas the key to peace.
These classical argumentsfind supportin standardtradetheory.While increasing
the aggregateincome of an economy, tradesimultaneouslyaltersthe domestic distributionof income. This possibility links internationalcommerce with domestic distributional issues that often lie at the heartof a liberal theory of conflict and creates a
foundationto understandthe processes by which these economic pressuresshape foreign policy. In what follows, I build on the insights of Cobden (1868, 1870) and
Schumpeter(1919/195 1) to identifya series of mechanismsemanatingfromboth society and the state wherebyfree tradeenhancesthe prospectsfor peace between states.
SOCIETAL PRESSURES FOR WAR AND PEACE
By focusing on the aggregatebenefits of commerce, most variantsof commercial
liberalismrisk assumingthatall individualswithin society receive income gains from
7. The democraticpeace literaturehas extensively studied how one such mechanism, competitive
elections, constrainsthe state's ability to go to war.
8. Forexample, in one of his many speeches duringthe anti-Corn Law campaign,RichardCobden
(1870, 1, 78) writes,
The single and undisguisedobject of the League is to putdown commercialmonopoly;butthatcannot be done by saddlinguponourbacksa fixed dutyon corn.... The Corn-lawis the greattreeof Monopoly,underwhose banefulshadowevery otherrestrictionexists ... The sole object of the League
is to putan end to andextinguish,at once andforever,the principleof maintainingtaxes for the benefit of a particularclass.
9. Cobden(1868, 1, 186) wrote, "Thelaws for the encouragementof tradeare direct and important;
and their tendency is to destroy the privileges of the nobles, by raising up a middle class."
McDonald/ PEACE THROUGHTRADEOR FREE TRADE?
553
tradeand thatthese same individualswill lobby theirgovernmentto enact a peaceful
foreign policy to realize these economic goals. As the politicalinfluence of the beneficiaries of commerceandtradeexpandin relativelyequalproportion,then largeraggregate tradeflows should promotepeace (Domke 1988, 48).
These mechanisms linking the economic interests of society to preferences over
foreign policy decisions on warandpeace need to be refinedfor severalreasons. I have
alreadydiscussed how revisions in standardtradetheory illustratethat all factors or
sectors in an economy do not receive income gains from trade. Consequently,economic interestscan lead some societal groups to favor the closure of nationalmarkets
to internationaltrade.Forexample, scarce factorsof productionare likely to lobby for
protection to reduce imports and prevent the erosion of their income (Stolper and
Samuelson 1941). This propositionsimilarlyimplies thateconomic interestscan lead
certaingroupsto supportany policy, includingthe use of militaryforce, which reduces
commerce and importpenetration.'0At the very least, protectionistinterests are less
likely thanothergroupsto lobby the statefor peace when conflict threatensto interrupt
trade.
Sectorsrelyingon protectionmay even actively supportaggressiveforeign policies
for two reasons. First,by reducingimportsand foreign competition, militaryconflict
may create income gains for these sectors by expandingtheir share of domestic markets. Second, militaryexpansioncan also provideconcentratedincome gains to these
groupsby enlarginga protecteddomestic marketthroughconquestand the integration
of anothereconomy. Similarto the standardargumentslinking imperialistintereststo
conflict, an importantdistinctionseparatesthis claim from more traditionalones. The
extent to which economic interests rely on regulatoryprotection from the state to
remainprofitableplays a critical role in their supportof militaryforce for economic
expansion. Sectors thatdo not rely on the state for protectionwhile survivingin internationalmarketscan generatenew marketsfor simple efficiency reasons.Theirgoods
penetratenew overseas marketsbecause they are cheaperthan those of international
competitors.Given that the use of military conflict may carry the risk of additional
costs to these sectors, they should be unlikely to supportsuch a policy if it were to
achieve the same outcome (of new overseas markets)they could achieve without the
use of force. At the same time, sectorsthatareless competitivein internationalmarkets
may be morewilling to riskrecoveringany costs of warin new marketsthatthey otherwise could not acquirewithoutthe use of force." These possibilities demandthatany
10. Militaryconflict can reducecommercefor a numberof reasons.Normalshippinglanes can be shut
down or become subjectto attack,therebyincreasingtransportationandinsurancecosts. Governmentsmay
impose sanctionsor blockadeenemy tradingpartners.For a study showing thatmilitarizeddisputes reduce
tradebetween countries,see Gowa and Mansfield (forthcoming).
11. The links here between protectionistgroups and supportfor militaryconflict differ from that of
Schneiderand Schulze (2003), who arguethat import-competingindustriespreferlower levels of military
hostilities undertimes of increasingeconomic integration.They notethatimport-competingsectors face two
costs from militaryconflict. The first comes from higher taxes imposed for militaryoutlays necessary to
defend the state from greaterexternal threats.The second flows from a decline in productiondue to the
destructionof industrialinfrastructure,a reductionin importedinputs, and more costly supply routes (p.
109). I thinkthreefactorsaccountfor my differenthypothesis.First,by focusing on the demandfor new public revenuescreatedby interstatehostilities, they overlook the state's capacity to distributethese taxation
burdensaway fromsome groups.Forstatesthatfund a largeportionof theirpublicrevenuesfromtariffs,it is
554
JOURNALOF CONFLICTRESOLUTION
claim thatcommerce reduces militaryconflict must account for the relative political
strengthor veto capacity of societal groups unlikely to supportliberal commercial
policies and peace for economic reasons.12
Apartfromunderstandingthe role of protectionistinterestsin decisions to use military force, one must also examine the influence of their societal opponents-those
favoringpeace for economic reasons-in this domestic struggle.The primarybeneficiaries of free trade face significant organizationalhurdles in their lobbying efforts
pursuingbothfree tradeandpeace. Studiesof the domestic politics of commercialpolicy (e.g., Alt and Gilligan 1994) illustratethatgains from trade,whetherin the form of
reducedpricesfor consumergoods or new exportmarkets,areoften diffused throughout society. This diffusion reducesthe willingness of these groupsto undertakecostly
lobbying activity for the relativelysmall benefits of greateropenness to trade.At the
same time, the costs of economic integration,whetherin the form of factory closures
or job losses, are often more concentratedin specific industriesor firms. Organizational advantagesthus help those groups most hurtby internationaltradeto overcome
the collective action problemand mobilize supportfor protectivepolicies thatinsulate
their sectors from internationalcompetition.
Similar organizationaldifficulties may reduce the effectiveness of societal constraintson a government'sefforts to use militaryforce as an instrumentof foreign policy (Gowa 1999). The aggregatebenefitsthatstem from avoidingthe costs of warmay
be spreadevenly across society to preventindividualsfrom undertakingthe costly lobbying activity necessary to restraina governmentfrom using force. These twin possibilities suggest that the income gains from trade may not always provide sufficient
incentives to preventconflict if their beneficiaries possess relatively weak domestic
influence.
To understandthe conditions when tradeactivates these societal interests to produce peace, we need to assess the relativepolitical strengthof these groups.Justas battlefield outcomes reduceambiguitiesaboutthe prewarbalance of militarypower and
resolve between states (Fearon 1995; Wagner2000), tradepolicy outcomes can indicate the relative balance of political strength between societal groups fighting a
domestic battle over commercial policy. Greaterlevels of protectionin an economy
indicate that import-competingsectors have successfully lobbied the state to enact
regulatorybarriersand possess relativelymore political influence than the beneficiaries of tradewithin the domestic political game. The level of protectionin an economy
thusprovidesone meansto assess the relativestrengthof domestic economic interests
in favorof waror peace. As the size of protectedsectors increases,the domestic politireasonableto expect that import-competingindustrieswill be shielded from these fiscal burdensbecause
highertariffsreduceimports.Cobden(1870) criticized the CornLaws for this very reasonbecause they created income gains for the landedaristocracyby driving up the price of grain. Second, some of the costs in
declining production,such as longer supply routes, should also reduce importsand thus expand domestic
marketshareof import-competingindustries.Third,they do not incorporatethe possibility thatmilitaryhostilities will providenew income opportunitiesfor some domestic producersthroughthe conquest and integrationof new markets.
12. Given findings that link democracyto peace, Weede (1995, 528-29) argues that protectionist,or
rent-seeking,groups generateconflict indirectlyby constrainingthe ability of free tradeand prosperityto
generatedemocracy.
McDonald/ PEACE THROUGHTRADEOR FREE TRADE?
555
cal influence of these groupson the state and the potentialfor militaryconflict should
also increase. Conversely,smallerlevels of protectionindicate that free tradelobbies
likely to oppose military aggression possess relatively more domestic political
influence.
THE STATE'S REGULATORY ROLE OVER
THE ECONOMY AND CONFLICT
Althoughprotectedinterestscan push a society towardwar to serve theirown economic interests,the presenceof economic regulationcan also increasethe potentialfor
war by increasingthe state's ability to build supportivecoalitions that strengthenits
domestic power. Because the state possesses a monopoly on coercion and the consequent ability to define the basic structureof propertyrights, and in particularmonopoly rights, it can restrict entry into domestic markets and regulate the terms of
exchange between buyersand sellers. This market-makingfunctionallows the state to
generatetangibleassets and then sell these rights to economic groups in exchange for
eitherrevenueand/orpolitical support(Stigler 1971; GrossmanandHelpman 1994).
Industriesthat rely on protection to remain profitable are in many senses "captured"by the state and more likely to supportits entire range of domestic and foreign
policies. Because the rightto sell goods in regulatedmarketsis generally not transferable to anothereconomy, it can be regardedas a specific asset (Williamson 1985). If a
firmwishes to locate to anothereconomy, it necessarilyloses this rightand mustrepurchase it from anothergovernment.The relativebargainingposition of the state vis-4'vis these sectors increasesas producersbecome dependenton the governmentfor tariff protectionand regulatedfactor markets(Levi 1988, 37). The sale of these regulations enables the stateto co-opt supportfor its policies, includingthe decision for war.
As the protectedsectors of the economy increase in size, a governmentcan drawon a
largerpool of society to supportthe use of militaryforce against other states. In this
second set of mechanisms linking protectionand conflict, a governing elite, and not
protectedsocietal interests,providesthe initial impetus for conflict."
For example, both the Frenchand Germangovernmentsused these capacities to
build coalitions in supportof a more aggressive foreign policy before WorldWarI. In
France,the need for approvalfrom the Foreignand FinanceMinistriesbefore floating
the loans of foreigngovernmentsin the Parismoney marketallowed the governmentto
use the economy's vast financial reserves to shape balance-of-powerdiplomacy in
Europe(Viner 1951).14 In Germany,the reinstitutionof a broadseries of tariffsreversing the Caprivireformscreatedthe financial and political means to unify agricultural
and industrialinterestsbehind the policy of Weltpolitik(Berghahn 1993, 38-55).
In summary,protectionincreases the likelihood of war throughtwo complementarymechanisms.The firstconcentrateson how competingsocietal interestsover eco13. By focusing on the state'scapacityto buildsocietal supportfor war,this mechanismrecognizes that
statecan go to warfor any numberof reasons,includingpurelysystemic pressures.It insteadconcentrateson
a government'scapacity to build supportivecoalitions so such a policy can be enacted.
14. Forexample,the Frenchgovernmentused access to its capitalmarketas a meansto pressureRussia
into buildingstrategicrailwaysin Polandin 1913 (Spring 1988;Stevenson 1996, 323-26). It was hoped that
quickerRussian mobilizationwould slow a Germanoffensive into France.
556
JOURNALOF CONFLICTRESOLUTION
nomic integrationmay shape foreign policy decisions. As the domestic influence of
protectionistinterests grows, the capacity of consumers and exportersto lobby the
state and producea peaceful foreignpolicy declines. Moreover,protectionistinterests
may reduce opportunitiesfor peace by actively supportingmilitaryexpansion. Second, the sale of economic regulations enhances the state's independentcapacity to
build supportivecoalitions for policies that may include war.
FREE TRADE AND PEACE: AN EMPIRICAL TEST
The previoussection arguedthatthe level of free tradein an economy providesthe
crucial link in understandinghow tradepromotes peace. This theoreticaldistinction
between trade and free trade possesses importantempirical implications. By often
offering multiple causal mechanisms as the foundation for empirical investigations,
the literaturehas neglectedthe possibility thateach of these hypothesesdemandsseparate tests for its validation.This problem is most apparentwhen comparingclassical
liberalreferencesto "freetrade"with the sociological or opportunitycost variants.By
focusing on the quantityof transnationalcontacts, the sociological hypothesis suggests thatall commerce promotespeace and thatthe properspecificationof the independentvariableis simply the total tradeof a state, either with its partnerin a dyad or
the rest of the world. The opportunitycost hypothesis predictsinsteadthat a measure
of trade'srole, relativeto total nationalincome, serves as the best operationalization
for the pacifying effects of commerce.On the otherhand,a focus on free tradeand the
size of the protectedsectorin the domestic economy suggests thatthe primaryconcept
of interestis the state's abilityto shapethis commerce throughsuch regulationsas tariffs or importquotas. This section tests how two measuresof protectionare relatedto
the onset of internationalconflict.
MEASURING PROTECTION
The quality of tradepolicy indicatorshas long been a source of controversyin the
economic growth literature(e.g., Edwards 1993, 1998; Leamer 1988; Rodrik 1995;
Hiscox and Kastner2002). A numberof measureshave been suggested, includingthe
ratioof totaltradeto nationalproduct,the ratioof duties to total imports,black market
premiumsfor foreignexchange,subjectiveevaluationsof tradepolicy orientation,and
residualsfrom models predictingaggregatetradeflows. Because statespossess a wide
variety of instruments,including tariffs, quotas, subsidies, and quality controls, to
shape tradeflows, measurementsthatrely on any one of these tools may poorly reflect
the aggregatelevel of regulation.Moreover,protectionis often industryspecific, complicatingefforts to create a measureof protectionfor an entireeconomy. Given these
problemsand the numberof tradepolicy indicators,one recentstudy (Edwards 1998,
384) notes that "attemptsto constructa single indicatorof tradeorientationmay be
futile" and suggests thatempiricalstudies of the effects of openness on growth must
shift to assessing the robustnessof findings across multiple indicatorsof openness.
McDonald / PEACE THROUGHTRADEOR FREE TRADE?
557
This debate has particularrelevance for empirical studies of trade and conflic
because of the latter'sreliance on trade intensity ratios, such as imports and export
dividedby GDP,to assess commercialintegration.Althoughwarningof the dangersor
relying on any single indicatorto assess openness, it also suggests thattradeintensity
ratios(such as tradedividedby GDP) may be relativelypoorindicatorsof political bar
riers to trade.Leamer (1988, 147-49) notes that in traditionalsmall country models
tradeis a functionof resourcesupplies, internationalprices, technology,tastes, natura
barriersto trade,and artificialbarriersto trade.Consequently,without controlling fo
all these sources, high dependence ratios (total trade divided by GDP) may simply
reflect dramaticallydifferent factor endowments instead of low political barriersto
trade.
To measureprotectionin tests of the commercial peace hypotheses, I rely on two
indicatorsof openness. The first is the ratio of a country's customs duties to its tota
imports.15 This indicatoroffers a number of conceptual benefits over total trade to
GDP ratios.First,the depthof tariffprotectionprovidesan indicatorof the quantityor
free tradein an economy. As tariffsincrease,the quantityof free tradeshould decrease
Fewergoods are likely to enteran economy duty free as customs revenueconstitutesa
largershare of total imports. Instead of assuming that a strong negative correlatior
between tarifflevels and tradeexists, this indicatorrecognizes thatthereare a numbel
of costs capturedin the price of a tradedgood, including inputcosts, transportatior
costs, insurance,foreign exchange contracts,and tariffsamong others. All these costs
can affectthe priceof a tradedgood andconsequentlythe size of aggregatetradeflows
A measureof tariffsprovidesone meansto isolate the componentof the total price ofa
good thatreflects governmentintervention.
Second, as alreadydiscussed, protectionlevels can measurethe relative domestic
strengthof groups within society thatbenefitor are hurtby free trade.Higher barriers
to tradesuggest thatimport-competingsectors have successfully lobbied the stateand
purchasedregulationsfrom the governmentthatredistributeincome towardthem. Or
the otherhand,lower barriersindicatethatfree tradelobbies andconsumershave beer
more successful in defeating protectionistinterests.
Third,the inclusion of these measuresalso providesan opportunityto test different
hypotheses within commercialliberalism.Bilateraltotal tradeto GDP ratios, labelec
DEPENDin the following tables,arenormallyused to operationalizesuch concepts as
the extentof transnationalties across societies and the relativedependenceof an economy on trade.Consequently,these aggregatetraderatios test the opportunitycost anc
sociological variantsof commercial liberalism. The inclusion of measurementsfor
protection levels along with DEPEND in statistical tests allows me to compare the
domestic version of commercialliberalism with these alternativeexplanations.
Despite these advantages,the use of tariffs to measureprotectionpossesses some
shortcomingsthat suggest using additionalindicatorsto ensure the robustnessof any
15. Data from the WorldBankDevelopmentIndicators are availablefrom 1970 to 2001. The World
Bank (2003) defines importduties in the following manner:"Importduties compriseall levies collected or
goods at the point of entry into the country.They include levies for revenuepurposesor importprotection
whetheron a specific or ad valorembasis, as long as they arerestrictedto importedproducts.Data are showr
for centralgovernmentonly."
558
JOURNALOF CONFLICTRESOLUTION
empirical conclusions. The quantity of import tariffs cannot capture the extent to
which states use nontariff barriers to shield noncompetitive domestic producers.
Moreover,the imposition of prohibitivetariffs may not be reflected in measures of
protectionthatrely on customs revenue.If largeenough, tariffscan eliminatethe trading of foreign goods in a domestic economy. A governmentwould then not collect any
customs revenue, and a measure of protection relying on import duties may then
remainunchanged.
To respond to these potential shortcomings, the following statistical tests also
employ a second measure of protection.Leamer (1988) suggests that the residuals
from econometric models of trade can measure protection. By modeling wellacceptedcovariatesof trade,such as factorendowments,distance, GDP,and development, unexplainedvariancecan be attributedto political barriersto trade.In a recent
study, Hiscox and Kastner(2002) offer a slightly different approach.They utilize a
gravity model of tradeand predictthe size of bilateraltradeflows between a pair of
states (or dyad) given the GDP of each tradingpartnerand the distance between the
economies. They then add a series of dummy variablesfor each country-yearin the
sample to this baseline model.'6Because the gravity model has been used to estimate
normalor naturalpatternsof tradeamong nations,the coefficients on the country-year
variablesmeasurecountry-specificdeviations from a baseline "free-trade"state in a
year. This deviation then proxies for the size of political barriersto trade that a
governmenterects in a year.
Hiscox and Kastner(2002) arguethatthis approachto measuringprotectionoffers
multiple advantagesover previous attempts.First, it extends coverage of protection
levels both spatially and temporally.Second, by tracinga series of changes in trade
policy across a numberof countries,they illustratehow theirmeasurerespondsto the
implementationor eliminationof tariffand nontariffbarriers.'7Finally,they also demonstrate that their indicatorcorrelatesreasonably well with alternativemeasures of
protection,includingcustoms collected as a portionof imports.'
The need for employing new measuresis underscoredby theirrelativelysmall correlationwith the predominantmeasurementin the commercialpeace literature,total
tradeover GDP ratios. Hiscox and Kastner(2002, 36) reporta bivariatecorrelation
between theirmeasureof protectionandthe tradeintensityratioof -.21 in a sample of
82 statesfrom 1960 to 1992. In my samples, the correlationsbetweendyadic measurements of duties and aggregatetrade levels divided by GDP (labeled DEPENDL)are
16. Thus,the numberof dummyvariablesequals the numberof countriesin the sample (82) multiplied
by the numberof years in the sample (33) minus 1, or 2,705.
17. Hiscox and Kastner(2002, 3-4) write,
Ourmeasurediffersmarkedlyfromthe mostcommonly used indexesof traderestrictionsin a variety
of importantways and cases. In particular,it does not understatepolicy openness in economies less
predisposedto tradefor naturalreasonshavingto do with geographyandresourceendowments,butit
also does not overstate policy openness in countries that favor non-tariffforms of protection over
tariffs.
18. Hiscox and Kastner(2002) reportthe bivariatecorrelationbetween importduties and the gravity
estimation of protection as .45 in their sample. This differs from the bivariate correlation reported as shown
in Table 1 because of dyadic measurement of protection that my sample employs. This distinction is discussed shortly.
McDonald/ PEACE THROUGHTRADEOR FREE TRADE?
559
relatively small. In the sample using import duties as a measure of protection, the
bivariatecorrelationwith total trade to GDP ratios is -.19. In the sample using the
Hiscox/Kastnerindicator,the bivariatecorrelationis -. 14. These low correlationssuggest that these different operationalizationsare measuring different concepts and
againunderscorethe need to controlfor protectionin tests of the commercialpeace.19
RESEARCH DESIGN
To test the hypothesis that greaterlevels of protection increase the probabilityof
interstateconflict, I adopt standardconventions in the empiricalliteratureon the liberalpeace.The dyad-yearserves as the unitof analysis in a series of statisticaltests that
comparehow protectionaltersthe likelihood of conflict in a year between two states.
All interstatedyads in the internationalsystem provide the spatial domain of cases,
whereasthe temporaldomainis a functionof how protectionis measured.When using
dataon importduties, the cases spanfrom 1970 to 2000. Data utilizingthe Hiscox and
Kastnerindicatorare availablefrom 1960 to 1992.
The dependentvariableis the onset of a new militarizedinterstatedispute between
dyad membersi andj.?2 This variable,MIDON, takeson a value of I in the first year of
a new dispute.
Protectionor openness is operationalizedin two ways. Both rely on the weak-link
hypothesis,which assertsthatthe least constrainedmemberof a dyad drives the conflict potentialof the dyad (Dixon 1993).2' Withrespectto protectionbarriers,the state
that is least constrainedpossesses higher barriersto trade. Societal opponents to the
income losses stemming from tradeinterruptionsare relatively weaker in this economy than if there were lower barriers.Moreover, larger protected sectors suggest
greatersocietal pressuresfor war and/ora largerpool of societal supportfrom which
the governmentcan draw if it chooses to pursuewar.The first indicatorof protection
measuresthe proportionof customs revenuedivided by total importsin the state pos19. These indicatorsfor protectionalso have relatively low correlationswith democracy, the other
primaryliberal peace variable. The bivariatecorrelation between the WorldBank indicator and democracy is -.34; andthe bivariatecorrelationbetweenthe Hiscox and Kastnerindicatoranddemocracyis-. 12.
20. Militarizedinterstatedisputes (MIDs) are "historicalcases in which the threat,display or use of
militaryforce shortof warby one memberstate is explicitly directedtowardsthe government,official representatives,official forces, property,or territoryof anotherstate"(Jones, Bremer,and Singer 1996, 168). The
source of this data is the revised MID data version 3.0 and derived from Eugene 3.03 (Bennett and Stam
2000).
21. The weak-linkhypothesisseems particularlyapplicableto tests of the commercialpeace hypothesis. Highprotectionlevels in one statenotonly generategreatersocietal pressuresfor war,they also generate
greatersupportfor hostile policies towardit by a dyadic partner.Although protectivebarriersin one state
propup import-competingsectors within the domestic economy, they simultaneouslycan injureexporting
interestsin the othereconomy. Given this grievance,these exportersmay providea base of societal support
for a governmentthatwishes to retaliateandpursuea moreaggressive foreignpolicy againstthe protectionist state. The rising tensions between Germanyand Russia in the monthspriorto July 1914 illustratethese
dynamics. The tariffs enacted to protect Prussianlandownersand induce them to supporting Weltpolitik
simultaneouslyexertedsignificantcosts on Russianagrarianinterests.AlthoughRussia'srelativelandabundance suggested that it should have been exportinggrain to Germany,the Germangovernmentsubsidized
exports of grain to Finland and western Russian provinces. By the summer of 1914, common interests across
Prussianand Russian agriculturalgroups had eroded, and both were clamoring for retaliatorytariffs and
pushing their respective governments to adopt a harsher policy line against the other (Spring 1988).
560
JOURNALOF CONFLICTRESOLUTION
sessing the greatersuch ratioin the dyad. A similarlogic follows for datafrom Hiscox
and Kastner (2002). They present protection scores as deviations from the sample
maximum,or free trade,statein theirdataset-the Netherlandsin 1964. Largervalues
for a statein a year indicategreaterdeviationsin importsfrom a gravitymodel prediction and subsequentlyof more regulatorybarriersto imports. For both indicatorsof
protection, PROTECTHmeasures the score of the state in the dyad that possesses
higherbarriersto trade.This value is measuredin period t - 1 to accountfor potential
endogeneity effects between conflict and protection. I expect PROTECTHto be
positively relatedto militaryconflict between states.
I also include the more standardmeasure of economic integration in the commercial peace literature. DEPENDLis the lower proportion of total dyadic trade
(imports plus exports)divided by state i's GDP or total dyadic tradedivided by state
j's GDP.22As bilateraltradeconstitutesa largerportionof the moreweakly constrained
member's GDP, the likelihood of conflict within that dyad should decrease. Like
DEPENDLis measuredin period t - 1.
PROTECTH,
To ensure the robustnessof any significant relationshipbetween protection and
conflict, I add a series of control variablesto the baseline regressionmodel. Because
extensive supporthas been offered for the propositionthat democraticstates are less
likely to go to war with each other (Oneal and Russett 1997), I control for the level of
democracywithin a dyad. Drawingon the Polity 4 dataset, a democracyvalue-ranging from -10 to 10--for each state is constructedby subtractingthe autocracyscore
fromdemocracyscore of regimes (Jaggersand Gurr 1995). Employingthe weak-link
hypothesis, I include a control variable,DEMOCRACYL,which is the lower democracy score of the states within a dyad.
Because largeeconomies have greaterpools of resourcesfrom which to build their
militaries,they may possess strongercapabilitiesandbe morelikely to enterinto a military conflict with another state. Moreover, because of the positive relationship
between openness and economic growth, openness may influence conflict through
this wealth effect. Instead of constraininga government'sforeign policy, economic
liberalizationmay foster economic growth and expand the total pool of resources on
which the state can rely to build its war machine. I control for these possibilities by
including the relative size of the largesteconomy, GDPH,within the dyad.23The primary source for this economic variable is version 6.1 of the Penn World Tables
(Heston, Summers,and Aten 2002). Its value is measuredin thousandsof 1996 U.S.
dollars at time t - 1.
Changesin short-termeconomic conditions may also alter a state's propensity to
initiate a conflict with another state. The diversionary-war hypothesis suggests
that governments may deliberately initiate a conflict with another state to shift
societal attention away from lingering domestic troubles accompanying political
scandalsand downturnsin the domestic economy (Levy 1989). Consequently,I conthe percentagechange in per capita
trol for this possibility by including GROWTHL,,
22. The data sources for this variable are Oneal and Russett (1999); Heston, Summers, and Aten
(2002); and the InternationalMonetaryFund (2003).
23. I again draw on the logic of the weak-link hypothesis and assume that the largereconomy faces
fewer resourceconstraints.
McDonald / PEACE THROUGHTRADEOR FREE TRADE?
561
GDP measuredin 1996 U.S. dollarsover a 3-yearperiod priorto time t, in the baseline
model specification.24
Criticsof the liberalpeace hypotheses and its democraticvariantin particularargue
that similar political interests, or the absence of motivations for war, explain peace
withindyads (Gartzke1998; Gowa 1999). I controlfor these argumentswith two variables. First,I code a dummyvariable,ALLY,thattakes on a value of 1 when the members of a dyad are membersof the same alliance (Gibler and Sarkeesforthcoming).25
Second, INTERESTS measures the weighted global S score (Signorino and Ritter
1999) for alliance portfoliosimilaritybetween the two states in the dyad. Both of these
variablesare measuredat time t.
Numerous realist argumentssuggest that power is the primarydeterminantof a
state's influence and willingness to use force in internationalpolitics. I control for
these possibilities in two ways. The firstfocuses on differences in capabilitiesas a predictor of military tensions (Kugler and Lemke 1996). Preponderanceproponents
arguethatbecause the dyadic balance of capabilities between two states increasingly
favorsone side, peace shouldbe morelikely. Takenfrom the Correlatesof War(COW)
project, CAPRATIOis the naturallog of the ratio of the stronger state's capability
index to the weakerstate's index (Singer,Bremer,and Stuckey 1972). Additionally,as
an absoluteindicatorof capability,I include MAJORPOWER,a dummy variablethat
takeson a value of I when eithermemberof the dyad is defined as a greatpower by the
COW project.
I also includetwo variablesto controlfor geographicconditionsthatmightcontribute to militarydisputes. As states become closer to each other,they have more opportunities for interactions that can become conflictual. First, a dummy variable,
CONTIGUITY,takeson a value of I when both membersof a dyadaregeographically
contiguous by land. Second, given that the sample under investigation includes all
potentialdyads in the internationalsystem, many of which may have limited diplomatic interactionsand consequently diminished opportunitiesfor conflict, I include
the logged distance in miles between capitalcities, defined as DISTANCE,to control
for this possibility. As the distance between two states increases, the likelihood of
conflict between them should decrease.
EMPIRICAL RESULTS
Given the dichotomousdependentvariables,I estimatedthe following model with
logistic regressionand the Beck, Katz, and Tucker(1998) correctionfor time-series,
cross-sectional analysis with a binarydependentvariable.26
24. Again, as standardin the literature,such a specification entails the adoption of the weak-link
hypothesis.The source of this data is version 6. I1of the Penn WorldTables (Heston, Summers, and Aten
2002).
25. The data for joint alliance membership,contiguity,majorpower status,capabilityratio, and distance was generatedfrom version 3.03 of the Eugene project(Bennett and Stam 2000).
26. Because observationsare likely to be temporallydependentand lead to standarderrorsthatunderestimatethe varianceof a coefficient, a naturalspline functionof the numberof years since i andj were last
engaged in a militarizeddisputewas included.The values of these coefficients are not reportedin the table,
but the base of the function is always statistically significant. Additionally, robuststandarderrors were
562
JOURNALOF CONFLICTRESOLUTION
MIDONij= B0 + B, x PROTECTH+ B2 x DEPENDL+ B3 x DEMOCRACYL+ B4 x ALLY
+ B, x CONTIGUITY+ B6 x CAPRATIO+ B7 x GROWTHL+ B8 x INTERESTS
+ B9 x GDPH+ B10x DISTANCE+ B,, x GREATPOWER+ eij.
The results can be seen in Table 1. The first three models utilize importduties as the
operationalizationfor PROTECT;and the final three regressions use the Hiscox and
Kastnerindicator.Models 1 and4 serve as the baseline regressionsand include all the
independentvariables.Models 2 and 5 drop DEPENDL;and PROTECTHis dropped
from the thirdand sixth models.
With the exception of the positive and sometimes significantcoefficient on ALLY,
the results on the control variablesconform to standardexpectations across the six
regressions.As the democracyscore for the more weakly constrainedmemberof the
dyad increases, the probabilitythat a new dispute breaksout between the two states
decreases.The positive and significantcoefficient on contiguityindicatesthatgovernments engage in moredisputesagainststates with which they sharecommon borders.
This link between geographyand conflict is furthersupportedby the negative coefficient on distance.As distance increases,new militarydisputes are less likely to break
out between dyadic partners.The negative sign on CAPRATIOindicates that larger
power disparitiesbetween states decrease conflict. Not surprisingly,one can also see
thatdyads that possess at least one great power engage in more militarydisputes.
Turningto the firstrow of the table,one sees how free tradeshapes the likelihood of
militaryconflict within a dyad. The consistently positive and statisticallysignificant
indicatesthatthe probabilityof a new militarydisputeincreasesas
sign on PROTECTH
the level of protectionincreases.Conversely,the likelihoodof militaryconflict within
a dyad decreases as the level of free trade increases (that is, protectiondecreases).
More important,as shown by models 1, 2, 4, and 5, these results hold across both
operationalizationsof tradebarriers.Given the hypothesizedrole playedby protection
levels in understandinghow societal interests and institutions mediate between the
pressuresof globalizationand foreignpolicy, these resultsdirectlysupportthe domestic version of commercial liberalismpresentedhere.
Moreover, if one turns to the coefficient on DEPENDL,one can compare this
domestic explanationwith more standardones in the literature.Because DEPENDL
has been offered as an indicator both of the degree of mutual trade dependence
between states and the scope of transnationalties across societies, the negativecoefficients on DEPENDLoffer tentativesupportfor these hypotheses. As pairs of states
become more dependenton bilateraltradeflows between them, they are less likely to
engage in militaryconflict with each other.
If one comparesmodels with and without PROTECTH,
one sees how the omission
of protection levels can alter conclusions concerning the path by which commerce
generatespeace. The opportunitycosts and sociological variantsof commercialliberalism predictthatthe coefficient on DEPEND should be negative and significant. As
shown in models 3 and 6, these claims find supportin both of the samples createdby
estimated by clustering on each dyad. I also estimated this model with a general estimating equation (GEE)
and a random effects logit. These alternative procedures did not yield any different conclusions on the rela-
tionship between protection,aggregatetradeflows, and militaryconflict.
TABLE 1
Baseline Dyadic Models: Logistic Regressionwith Splines (Not Shown)-Dep
Model
1
PROTECTH
DEPENDL
DEMOCRACYL
ALLY
CONTIGUITY
CAPRATIO
GROWTHL
INTERESTS
GDPH
DISTANCE
GREATPOWER
CONSTANT
(I
0.024***
(0.008)
-22.244t
(15.184)
-0.036**
(0.015)
0.483**
(0.233)
2.011***
(0.343)
-0.146**
(0.072)
-6.790***
(2.059)
-0.995**
(0.373)
2.3 x 10-***
(6.9 x 10")
-0.425***
(0.140)
0.769**
(0.322)
0.054
(1.179)
2
3
0.026***
(0.008)
-0.043***
(0.015)
0.504**
(0.236)
1.942***
(0.334)
-0.117*
(0.071)
-7.177***
(1.989)
-1.076***
(0.379)
2.1 x 10-***
(7.6 x 10-)
-0.412***
(0.136)
0.665**
(0.314)
-0.066
(1.147)
-27.603*
(15.969)
-0.046***
(0.014)
0.456*
(0.234)
2.046***
(0.344)
-0.146**
(0.074)
-6.180***
(2.027)
-0.983***
(0.375)
2.4 x 10-***
(7.1 x 10-)
-0.407***
(0.133)
0.642*
(0.341)
0.345
(1.136)
4
0.053***
(0.009)
-20.582
(17.871)
-0.020t
(0.013)
0.406t
(0.259)
1.460***
(0.327)
-0.219***
(0.058)
-5.673***
(1.449)
-1.075***
(0.342)
3.8 x 10(7.8 x 10-0.709***
(0.160)
0.976***
(0.272)
0.451
(1.297)
0.
TABLE 1 (continued)
-4
Model
1
n
Log-likelihood
87,180
-1,358.290
2
87,180
-1,360.920
3
87,180
-1,365.091
4
92,354
-1,556.294
SOURCE:For PROTECT,WorldBank (2003) in models 1, 2, and 3; Hiscox and Kastner(2002) for models 4, 5, and
NOTE:Robuststandarderrorsclusteredon dyad are listed in parentheses.
tp ? .10 (one-tailedtest). *p < .10. **p < .05. ***p < .01 (two-tailedtests).
McDonald/ PEACE THROUGHTRADEOR FREE TRADE?
565
the differentoperationalizationsof protection.However,if one moves from model 3 to
model 1, one sees thatthe additionof a variablefor tradepolicy shrinksthe coefficient
on DEPENDLand reduces its statisticalsignificance below the .10 level (two-tailed
test). Comparingmodels 4 and 6, this effect is even more dramaticin the sample createdby the Hiscox and Kastnerindicator.WhereasDEPENDLis negativeand statistically significantwhen PROTECTis not included,the size of its coefficient shrinksdramatically and is no longer statistically significant when adding PROTECTto the
model. Not only do these results cast doubt on the opportunitycost and sociological
variantsof commercialliberalism,they simultaneouslysupportthe domestic explanation offeredhere,becausethe sign on PROTECTH
is always positive andsignificant.
The relativerole of tradeandtradebarrierson conflict can also be examinedby addto the baseline
ing an interactionvariablethat multipliesDEPENDLand PROTECTH
equation.This variablechecks the possibility that the respective effect of these variables on the outbreakof militaryconflict depends on the value of the other variable
(Friedrich1982). If this is the case, the coefficients and standarderrorson DEPENDL
are conditioned by the value of PROTECTH.
For example, as suggested by the arguments of Cobden (1868, 1870) and Schumpeter(1919/1951), larger bilateral trade
flows may only enhancethe prospectsfor peace between states thatalreadyhave relatively low levels of importbarriers.
I explore this possibility in models I and 2 of Table2. Here I add anothervariable,
x DEPENDL,to test for these interactioneffects. In these regressions,the
PROTECTH
coefficient on DEPENDLillustratesits effect on the probabilityof a dispute when
PROTECTH
equals 0.27 As shown in the first model, when import tariffs are used to
operationalizeprotectionlevels, the coefficient on DEPEND, is negativeand statistically significant.Fordyads in which neithergovernmenterects any importbarriersto
trade,largertradeflows reducethe probabilityof militaryconflict between two states.
x DEPENDLshows
However,the positive and significantcoefficient on PROTECTH
As the
thatthese negativeresultswill not be maintainedfor all values of PROTECTH.
latterreacheshigherlevels, the tendencyof importduties to enhance the likelihood of
conflict overcomesthe inhibitingrole of DEPENDL.DEPENDLonly remainsnegative
reachesa value of 4.56. Forthe currentsamandsignificant(p <. 10) untilPROTECTH
ple, slightly more than9% of the total cases (7,878) meet this condition. In short,the
ability of greaterbilateraltradeflows to reduce militaryconflict is restrictedto dyads
in which both governmentshave adoptedfree tradepolicies.
Moreover,aggregatetradeflows do not condition the relationshipbetween protective barriersand militaryconflict to the same extent. In model 1 of Table 1, the coefficient on PROTECTHis positive and significant. Even when bilateraltrade is absent
between the two membersof a dyad, higher levels of regulatorybarriersstill increase
27. Ina standard
say,Y=Bo+ BIX,+ B2X2,theeffectsof X, andX2areestimatedrespecregression,
tively by the size and the directionof the coefficients, B, and B2.The effect of X, on Yis independentof the
This
existsbetweenthetwovariables.
levelorvalueof X2.Thisis notthecaseif a conditional
relationship
yieldingY= Bo+ B,1X+
possibilitycanbe examinedby addinganothervariable,XIX2,to theregression,
B2X2+ B3XIX2. Now, the effect ofX2 on Ychangesas the level ofXi changes. One can see this by rearranging
terms:Y= Bo + BIXI + (B2+ B3X,)X2.Now, the effect of X2on Yis estimatedby the "new"coefficient B2 +
B3Xi. When X1 equals zero, this effect of X2 is reducedto the coefficient B2.
566
JOURNALOF CONFLICTRESOLUTION
TABLE 2
InteractionTests-Dependent Variable,MIDON:
Models 1 and 2, Logistic Regression with Splines (Not Shown)
Model 1
PROTECTH
DEPENDL
PROTECTHx DEPENDL
DEMOCRACYL
ALLY
CONTIGUITY
CAPRATIO
GROWTHL
INTERESTS
GDPH
DISTANCE
GREATPOWER
CONSTANT
N
Log-likelihood
0.016*
(0.009)
-54.445**
(21.625)
6.053***
(1.943)
-0.032**
(0.015)
0.456**
(0.230)
1.967***
(0.339)
-0.128*
(0.072)
-6.503***
(2.042)
-0.873**
(0.381)
2.4 x 10"***
(7.0x 10 H)
-0.420***
(0.136)
0.764**
(0.315)
0.012
(1.156)
87,180
-1,350.984
Model 2
0.050***
(0.010)
-82.318
(83.435)
2.079
(2.563)
-0.019t
(0.013)
0.397t
(0.259)
1.449***
(0.329)
-0.213***
(0.059)
-5.616***
(1.452)
-1.049***
(0.348)
3.7 x 10(8.0 x 10')
-0.707***
(0.160)
1.011***
(0.282)
-0.519
(1.296)
92,354
-1,555.547
SOURCE:For PROTECT,WorldBank (2003) for model 1; Hiscox and Kastner(2002) for model 2.
NOTE:Robuststandarderrorsclusteredon dyad are listed in parentheses.
tp ? .10 (one-tailed test). *p < .10. **p 5 .05. ***p ? .01 (two-tailed tests).
the likelihoodof militaryconflict. The positive and significantvalue on PROTECTH
x
DEPENDLshows thatthese effects of protectionon conflict only increaseas bilateral
tradegrows. Although the size of the effect of PROTECTH
on conflict changes across
values of DEPENDL,it remains statistically significant for all observed values of
DEPENDL.
The second model of Table 2 tests for these interactioneffects while using the
Hiscox and Kastnerindicatorfor regulatorybarriersto trade.As previously illustrated
in the baseline regression,the links between aggregatetradeflows and peace areagain
relativelyweak. The coefficient on DEPENDLis negativebut neverachieves standard
levels of statisticalsignificance for any observedvalue of PROTECTH.
is
PROTECTH
McDonald/ PEACE THROUGHTRADEOR FREE TRADE? 567
positive and statistically significant for more than 99.6% of the observed values for
DEPENDL.28
Next, I would like to turnto the substantiveresults of these findings to underscore
these differencesbetween free tradeand tradeon peace. Table3 displays the predicted
probabilitythata new militarydisputeoccurs between two states in a given year under
four differentcircumstances.In the baseline examples derived from the results found
in models 1 and 4 of Table 1, all the independentvariables are held at their means
except the primary liberal peace variables-PROTECTH, DEMOCRACYL,and
DEPENDL.These values are takenat the 10thpercentilefor each variableto indicate
"illiberal"scores, that is, dyads with large import barriersto trade, low democracy
scores, and low levels of bilateraltraderelativeto GDP.The bottomthreescores identify how the predictedprobabilityof a new dispute changes as the value for the specified liberalpeace variablegoes from its 10thpercentileto its 90th percentile.29Shown
in the second line, a shift from a highly autarkicdyad (customs duties make up nearly
27%of total imports)to a free tradingdyad (customs duties make up slightly less than
5% of total imports) decreases the likelihood of observing a new military dispute
between a pairof statesby morethan40%.")A similarshift in DEMOCRACYfrom its
10th percentile (-9) to its 90th percentile (9) decreases the likelihood of conflict by
47%. The substantiveeffects of DEPENDLarecomparativelyquite small. A shift from
its 10th percentile score (0) to its 90th percentile score (0.0012) decreases the
probabilityof conflict by only 3%.
These relativeeffects become even more pronouncedwhen using the Hiscox and
Kastnerindicatorfor protection.A shift from a protectionistdyad with a score at the
10thpercentile(48.37) to free-tradingdyad at the 90th percentile(25.8) decreasesthe
probabilityof a new disputeby nearly70%. A comparableshift in the democracyscore
decreases the probabilityof conflict in a dyad by 30%. The effects of aggregatetrade
flows on conflict are relativelysmall, decreasing the probabilityof a new dispute by
only 2%. Two importantconclusions flow from this. First, the tendency of protective
tradepolicies to increasemilitaryconflict is both statisticallyand substantivelysignificant. Second, the level of free tradeexertsa largereffect thanaggregatetradeflows on
the outbreakof peace within a dyad.
The combination of these results strongly supports the claim that free trade
enhances the prospects for peace. Although aggregate levels of internationalcommerce aredrivenby price shifts thatdependon manyfactors,includingartificialbarri28. The coefficient on PROTECTHcontinuesto increasein size butbecomes statisticallyinsignificant
when DEPENDLexceeds a value of .0257. However,only 336 of 92,354 cases meet this condition.
29. A shift fromthe I0th percentileto the 90th percentiledemonstratesa shift fromthe"illiberal"to the
"liberal"end of the continuum.Values for DEPEND and DEMOCRACYwill thus increase as they move
from the I0th to the 90th percentiles,whereasvalues for PROTECTwill decrease. Accordingly,the relative
risks in Table3 will all be negativebecause they indicatehow the predictedprobabilityof MIDONchanges
as a dyadbecomes moreliberalalong threedimensions-democracy, bilateraltradeflows, andtradepolicy.
30. These substantivemeaningof these value can be mademoreconcreteby comparingthe scores of a
protectioniststate with a more liberal one. Before it initiatedeconomic reformsin the early 1990s, India's
lowest level of importduties in the currentsample was 25.7 in 1980. From 1977 until 1993, Japan'shighest
level of importduties was 4.1 in 1993. Holding all other variablesat theirmeans, the replacementof India
(protectionscore of 25.7) with Japan(protectionscore of 4.1) in a dyad reduces the probabilityof conflict
from .00139 to .00084, or approximately65%.
568
JOURNALOF CONFLICTRESOLUTION
TABLE 3
SubstantiveEffects of LiberalPeace
Variableson PredictedProbabilityof MIDON
TableI Models (Logistic Regression with Splines)
1 (ImportDuties)
Xi
1. BASELINE
PROTECT
DEMOCRACY
DEPEND
2. PROTECTat 90th percentile
3. DEMOCRACYat 90th percentile
4. DEPEND at 90th percentile
p
Change
4 (GravityModel Estimates)
X,
.00112
.00098
.00184
Change
.00220
.00189
26.87
-9.0
0.0
4.88
9.0
0.0012
p
-40.8
-46.7
-2.6
48.37
-9.0
0.0
25.8
8.0
0.0012
.00067
.00156
.00215
-69.6
-29.4
-2.4
NOTE:In baselinecase, PROTECT,DEMOCRACY,andDEPEND possess "illiberalvalues"(10th percentile), whereasall other independentvariablesheld at theirrespective means. Examples 2, 3, and 4 illustrate
howpredictedprobabilityof MIDONchangesas it movesfrom 10thto 90thpercentileon liberalpeacevariables.
ers to trade,transportationcosts, relativefactor endowments, and tastes, the secondimage version of commercial liberalismexplicitly focuses on how only one of these
inputs-the regulatorybarriersimposed by the state-is likely to affect the foreign
policy and conflict propensity of states in the internationalsystem. These findings
illustratethe importanceof rethinkingthe measurementdemandscreatedby exploring
the differentcausal mechanismsof the commercial peace between states. Free trade
and trade alone may not always exert similar effects on the likelihood of conflict
between states.
CONCLUSIONS AND IMPLICATIONS
This study has critiqued the interdependenceliteraturefor neglecting important
componentsof the classical literaturethatlink tradeand war.Scholarssuch as Cobden
(1868, 1870) and Schumpeter(1919/1951) have recognized that although internationalcommercemay offer the potentialto establishmutualties of dependenceamong
statesthatmake warless likely, free tradealso simultaneouslytransformedthe domestic distributionof power by eliminatingeconomic regulationsthat strengthenedsocietal groups most likely to supportwar.Solidly groundedin the foundationsof liberal
theorythatfocus on how individualincentives and domestic institutionsalterthe foreign policy of states,such argumentsdistinguishtradefrom "free"tradeandrecognize
that domestic interestsand institutionsfilter the effects of commerce on peace. Free
tradereduces militaryconflict in the internationalsystem by underminingthe domestic politicalpowerof intereststhatbenefitfromconflict andby limitingthe state'sabil-
McDonald/ PEACE THROUGHTRADEOR FREE TRADE? 569
ity to enact commercial policies to build domestic coalitional support for its war
machine.
These argumentshave importantimplicationsfor a criticalcase in the globalization
and peace debate. Realists and liberals alike have characterizedthe global economy
priorto WorldWarI as an open tradingsystem that fostered interdependenceamong
economies. Combined with the outbreakof war in 1914, realists have offered this
descriptionof the prewareconomy as powerful, disconfirmingevidence of the commercial peace hypothesis (e.g., Waltz 1979; Mearsheimer1991; Copeland 1996).
These claims rely in parton an assumptionthatincreasingtradeflows indicate liberalcommercialpolicies. Wasthis trueduringthis eraof globalization?Economic historiansnow arguethatthis eraof globalizationwas drivenlargelyby a sustainedreduction in transportationcosts and not by a reductionin protectiveregulations(O'Rourke
andWilliamson 1999). In fact, a globalizationbacklashled to an increasein protection
in Europeafter 1879. Apartfrom wide-rangingtariffson the continent,Germanyand
France,two of the key playersin global capitalmarkets,possessed capitalcontrolsthat
allowed them to funnel domestic savings towardtheir political allies and away from
potentialenemies. Withthe exception of GreatBritain,the greatpowers all shapedthe
dynamics of the labor marketthroughconscription.Russia, the great power perhaps
most responsible for upsetting this period's military balance of power (Stevenson
1996), received nearly two-thirdsof its public revenues from tariffsand state-owned
assets (Apostol, Bernatzky,and Michelson 1928, 43, 61, 222). In short,the pre-World
WarI global economy was not an archetypalliberaleconomic order.
This era may insteadprovideanotheropportunityto comparethe relativestrength
of variantsof the commercialpeace hypothesis. In this case, the third-imageversions
(opportunitycost, efficiency, and sociological variants)and second-image arguments
provideverydifferentpredictions.It is clear thatthe Europeangreatpowers were each
other'sprimarytradingpartnersand thus economically dependent.3 According to the
opportunitycost hypothesis, these mutualdependencies should have preventedwar.
The outbreakof war in 1914 thus seems to contradictthese claims. On the otherhand,
by pointingto the rise of tariffsand economic regulationsin the decades leading up to
July 1914, the second-image commercial peace hypothesis discussed here would
insteadpredictmilitaryconflict. The impositionof agriculturaltariffsin Germanycreated an opportunityfor the state to wed agriculturaland industrialinterests behind
Weltpolitik.Whereas naval constructionpoisoned relations with the British, but the
tariffsproduceda stronganti-Germancoalition within Russia pressuringthe government for a more aggressive foreign policy againstGermaninterests.The outbreakof
war thus suggests that the variantof commercial liberalism emphasized here sheds
importantlight on a criticalcase for liberalinternationalrelationstheoryand may help
to clarify the precise mechanismslinking commerce and conflict.
3 1. Forexample, Ripsmanand Blanchard(1996-97) discuss how Germanywas dependenton imports
of food and raw materials.Consequently,they also relied on British goodwill to make sure shipping lanes
remainedopen. Spring (1988, 82) notes that Russia was dependenton its tradinglinks with Germany.He
writes, "In 1913 45 percentof Russianexportsand 50 percentof her importswere with Germany.No other
great power was so dependenton one market."
570
JOURNALOF CONFLICTRESOLUTION
REFERENCES
Alt, JamesE., andMichaelGilligan. 1994. The politicaleconomy of tradingstates:Factorspecificity,collective actionproblems,anddomestic political institutions.JournalofPolitical Philosophy2 (2): 165-92.
Apostol, PaulN., MichaelW. Bernatzky,andAlexanderM. Michelson. 1928. Russianpublicfinance during
the war.New Haven, CT: Yale UniversityPress.
Barbieri,Katherine.2002. Theliberal illusion: Does tradepromotepeace ? Ann Arbor:Universityof Michigan Press.
Barbieri,Katherine,and GeraldSchneider.1999. Globalizationand peace: Assessing new directionsin the
study of trade and conflict. Journal of Peace Research 36 (4): 387-404.
Beck, Nathaniel, Jonathan M. Katz, and Richard Tucker. 1998. Taking time seriously: Time-series
cross-section analysis with a binary dependent variable. American Journal of Political Science 42
(4): 1260-88.
Bennett,D. Scott, andAllan Stam.2000. EUGene:A conceptualmanual.InternationalInteractions26:179204.
Berghahn,V. R. 1993. Germanyand the approachof war in 1914. 2nd ed. New York:St. Martin's.
Brawley,Mark. 1993. Regime types, markets,and war. ComparativePolitical Studies 26 (2): 178-97.
Cobden, Richard. 1868. Thepolitical writings of RichardCobden, vols. 1-2. London:William Ridgway.
. 1870.Speeches on questionsof public policy, vols. 1-2, ed. John BrightandJames E. ThoroldRogers. London:Macmillanand Co.
Copeland,Dale C. 1996. Economic interdependenceand war:A theoryof tradeexpectations.International
Security20 (4): 5-41.
Deutsch, Karl,et al. 1957. Political communityand the NorthAtlantic area: Internationalorganizationin
the light of historical experience. Princeton,NJ: PrincetonUniversityPress.
Dixon, WilliamJ. 1993. Democracyandthe managementof internationalconflict. Journalof ConflictResolution 37 (1): 42-68.
Domke, William. 1988. Warand the changing global system. New Haven, CT: Yale UniversityPress.
Edwards,Sebastian. 1993. Openness, tradeliberalization,and growth in developing countries.Journal of
EconomicLiterature31 (3): 1358-93.
. 1998. Openness, productivityand growth:Whatdo we really know'?EconomicJournal 108:38398.
Eggertsson,Thrdiinn.1990. Economicbehaviorandinstitutions.Cambridge:CambridgeUniversityPress.
Fearon,James D. 1995. Rationalistexplanationsfor war.InternationalOrganization49 (3): 379-414.
Flam, Harry,and M. June Flanders. 1991. Heckscher-Ohlintrade theory.Cambridge,MA: MIT Press.
Friedrich,RobertJ. 1982. In defense of multiplicativeterms in multiple regression equations.American
Journal of Political Science 26 (4): 797-833.
Gartzke,Erik. 1998. Kant we all get along? Opportunity,willingness, and the origins of the democratic
peace. AmericanJournal of Political Science 42 (1): 1-27.
Gartzke,Erik,Quan Li, andCharlesBoehmer.2001. Investingin the peace: Economic interdependenceand
internationalconflict. InternationalOrganization55 (2): 391-438.
Gelpi, Christopher,and Joe Grieco. 2003. Economic interdependence,the democraticstate, and the liberal
peace. In Economic interdependenceand international conflict: New perspectives on an enduring
debate,ed. EdwardD. Mansfieldand BrianM. Pollins,44-59. Ann Arbor:Universityof MichiganPress.
Gibler,Douglas M., and MeredithSarkees.2004. Measuringalliances:The Correlatesof WarFormalInterstate Alliance data set, 1816-2000. Journal of Peace Research41 (2): 211-222.
Gowa,Joanne. 1999. Ballots and bullets: Theelusive democraticpeace. Princeton,NJ:PrincetonUniversity
Press.
Gowa, Joanne, and EdwardD. Mansfield. Forthcoming.Alliances, imperfect markets,and major-power
trade.InternationalOrganization.
Grossman,Gene M., and ElhananHelpman. 1994. Protectionfor sale. AmericanEconomicReview84 (4):
833-50.
Hansen,John Mark. 1990. Taxationand the political economy of the tariff.InternationalOrganization44
(4): 527-51.
McDonald/PEACE THROUGHTRADEOR FREE TRADE?
571
Heston, Alan, RobertSummers,and BettinaAten. 2002. Penn WorldTableversion 6. 1. Centerfor International Comparisonsat the Universityof Pennsylvania.http://pwt.econ.upenn.edu.
Hiscox, Michael J., and Scott L. Kastner.2002. A general measureof tradepolicy orientations:Gravitymodel-based estimates for 82 nations, 1960 to 1992. Workingstudy, HarvardUniversity,Cambridge,
MA.
Howard,Michael. 1978. Warand the liberal conscience. New Brunswick,NJ: RutgersUniversity Press.
InternationalMonetaryFund.2003. Direction of tradestatistics. CD-ROM.Washington,DC: International
MonetaryFund.
Jaggers,Keith,and RobertGurr.1995. Trackingdemocracy'sthirdwave with the Polity IIIdata.Journalof
Peace Research 32 (4): 469-82.
Jones, Daniel M., StuartA. Bremer,and J. David Singer. 1996. Militarizedinterstatedisputes, 1816-1992:
Rationale,coding rules, and empiricalpatterns.Conflict Managementand Peace Science 15 (2): 163213.
Jones, Ronald. 1971. A three-factormodel in theory,trade,and history.In Trade,balance ofpayments, and
growth, ed. J. Bhagwati, Ronald Jones, Robert A. Mundell, and Jaroslav Vanek, 3-21. Amsterdam:
North-Holland.
Kugler,Jacek,and Douglas Lemke, eds. 1996. Parityand war: Evaluationsand extensionsof The WarLedger. Ann Arbor:Universityof Michigan Press.
Leamer,EdwardD. 1988. Measuresof openness. In Tradepolicy issues and empirical analysis, ed. Robert
E. Baldwin, 147-200. Chicago: Universityof Chicago Press.
Levi, Margaret.1988. Of rule and revenue.Berkeley: Universityof CaliforniaPress.
Levy, Jack S. 1989. The diversionarytheoryof war: A critique. In Handbookof war studies, ed. Manus I.
Midlarsky,259-88. Boston: Unwin Hyman.
Mansfield, EdwardD. 1994. Power,trade, and war. Princeton,NJ: PrincetonUniversityPress.
Mansfield, EdwardD., and Jon C. Pevehouse. 2000. Tradeblocs, trade flows, and internationalconflict.
InternationalOrganization54:775-808.
Mansfield, EdwardD., Jon C. Pevehouse, and David H. Bearce. 1999-2000. Preferentialtradingarrangements and militarydisputes. SecurityStudies 9:92-1 18.
Mansfield, Edward D., and Brian M. Pollins. 2001. The study of interdependenceand conflict: Recent
advances,open questions,anddirectionsfor futureresearch.Journalof ConflictResolution45 (6): 83459.
Mearsheimer,JohnJ. 1991. Back to the future:Instabilityin Europeafterthe cold war. In Thecold war and
after: Prospectsfor peace, ed. Sean M. Lynn-Jones,141-92. Cambridge,MA: MIT Press.
Moravcsik,Andrew. 1997. Takingpreferencesseriously:A liberaltheoryof internationalpolitics. International Organization51 (4): 513-54.
Morrow,James D. 1999. How could tradeaffect conflict? Journal of Peace Research 36 (4): 481-89.
Oneal,JohnR., and BruceRussett. 1997.The classical liberalswere right:Democracy,interdependence,and
conflict, 1950-1985. InternationalStudies Quarterly41 (2): 267-94.
. 1999. Assessing the liberalpeace with alternativespecifications:Tradestill reducesconflict. Journal of Peace Research36 (4): 423-32.
O'Rourke,Kevin H., and JeffreyG. Williamson. 1999. Globalizationand history: Theevolutionof a nineteenth-centuryAtlanticeconomy.Cambridge,MA: MIT Press.
Papayoanou,Paul. 1999. Powerties: Economic interdependenceand war Ann Arbor:Universityof Michigan Press.
Polachek, Solomon W. 1980. Conflict and trade.Journal of Conflict Resolution24:55-78.
Polachek, Solomon W., John Robst, and Yuan-ChingChang. 1999. Liberalism and interdependence:
Extendingthe trade-conflictmodel. Journal of Peace Research36 (4): 405-22.
Ricardo,David. 1821/1973. Theprinciples of political economy and taxation.London:Dent.
Ripsman,NorrinM., and Jean-MarcF. Blanchard. 1996-97. Commercialliberalism underfire: Evidence
from 1914 and 1936. SecurityStudies 6 (2): 4-50.
Rodrik,Dani. 1995.Tradepolicy andindustrialreform.InHandbookof developmenteconomics, vol. 3B, ed.
Jere Behrmanand T. N. Srinivasan,2925-82. Amsterdam:Elsevier.
572
JOURNALOF CONFLICTRESOLUTION
Rogowski, Ronald. 1989. Commerceand coalitions: How trade affects domestic political alignments.
Princeton,NJ: PrincetonUniversityPress.
Rosecrance,Richard.1986. Therise of the tradingstate: Commerceand conquestin the modernworld.New
York:Basic Books.
Rowe, DavidM. 1999. Worldeconomic expansionandnationalsecurityin pre-WorldWarI Europe.International Organization53 (2): 195-231.
Russett,Bruce, and John R. Oneal. 2001. Triangulatingpeace: Democracy, interdependence,and international organizations.New York:Norton.
Samuelson,Paul A. 1971. Ohlin was right.Swedish Journal of Economics 73 (4): 365-84.
Schneider,Gerald,and GuntherG. Schulze. 2003. The domestic roots of commercialliberalism:A sectorspecific approach.In Globalizationand armed conflict, ed. Gerald Schneider,KatherineBarbieri,and
Nils PetterGleditsch, 103-22. Lanham,MD: Roman & Littlefield.
Schumpeter,Joseph A. 1919/1951. The sociology of imperialisms.In Imperialismand social classes, ed.
Paul M. Sweezy, trans.Heinz Norden, 1-130. New York:Augustus M. Kelley, Inc.
Signorino,CurtisS., andJefferyRitter.1999.Tau-bor not Tau-b:Measuringthe similarityof foreign policy
positions. InternationalStudies Quarterly43:1 15-44.
Simmons,Beth. 2003. Pax Mercatoriaandthe theoryof the state. In Economicinterdependenceand international conflict:Newperspectiveson an enduringdebate,ed. EdwardD. Mansfieldand BrianM. Pollins,
31-43. Ann Arbor:Universityof MichiganPress.
Singer, J. David, StuartBremer,and John Stuckey. 1972. Capability distribution,uncertainty,and major
powerwar, 1820-1965. In Peace, war,and numbers,ed. BruceRussett, 19-48. Beverly Hills, CA: Sage.
Spring,D.W. 1988. Russia and the coming of war.In Thecoming of the First WorldWar,ed. R. J. W. Evans
and HartmutPogge von Strandmann,57-86. Oxford, UK: Clarendon.
Stein, ArthurA. 1993. Governments,economic interdependence,and internationalcooperation.In Behavior, society, and internationalconflict, ed. Philip Tetlock, Jo Husbands,RobertJervis, Paul Stern, and
CharlesTilly, 241-324. New York:Oxford University Press for the National ResearchCouncil of the
National Academy of Sciences.
Stevenson,David. 1996.Armamentsand the comingof war: Europe,1904-1914. Oxford,UK: Clarendon.
Stigler, George J. 1971. The theory of economic regulation.Bell Journal of Economics and Management
Science 2 (1): 3-21.
Stolper,WolfgangF., and PaulA. Samuelson. 1941. Protectionand real wages. Reviewof EconomicStudies
9 (1): 58-73.
Viner,Jacob. 1951. Internationaleconomics. Glencoe, IL: Free Press.
Wagner,Harrison.2000. Bargainingand war.AmericanJournal of Political Science 44 (3): 469-84.
Waltz, KennethN. 1979. Theoryof internationalpolitics. New York:McGraw-Hill.
Weede, Erich. 1995. Economic policy and internationalsecurity: Rent-seeking,free tradeand democratic
peace. EuropeanJournal of InternationalRelations 1 (4): 519-37.
Williamson,Oliver E. 1985. The economic institutionsof capitalism. New York:Free Press.
WorldBank. 2003. Worlddevelopmentindicators.CD-ROM. Washington,DC: WorldBank.