Keeping the Peace after Secession: Territorial ... - University of Kent

2 downloads 0 Views 4MB Size Report
tries, with consequences ranging from the peaceful interstate relations following the breakup ..... India, Ireland, Palestine, and the former Yugoslavia, a group of antisecession authors ..... apparently irrelevant dyads (e.g., Estonia-Kazakhstan).
Keeping the Peace after Secession: Territorial Conflicts between Rump and Secessionist States Author(s): Jaroslav Tir Reviewed work(s): Source: The Journal of Conflict Resolution, Vol. 49, No. 5 (Oct., 2005), pp. 713-741 Published by: Sage Publications, Inc. Stable URL: http://www.jstor.org/stable/30045150 . Accessed: 01/02/2012 07:23 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp 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

Keepingthe Peace after Secession TERRITORIAL CONFLICTS BETWEEN RUMP AND SECESSIONIST STATES

JAROSLAVTIR Departmentof InternationalAffairs Universityof Georgia

Secession is an attemptto resolve a domesticallybased territorialdisputeby dividinga country'shomelandterritoryinto new,secessionist (e.g., Eritrea)andrump(e.g., Ethiopia)states.Yet,the secession maynot have resolvedthe originaldisputeto the states'satisfaction.In the aftermathof a secession, the leaderof the rumpstateis motivatedto use forceby the benefitsof retaking(some of) the landlost to the secessioniststate, while the secessionist state's leaderis motivatedby the benefits of acquiringeven more land. The peaceful versus violent secession process furtheraffects whetherthese desires escalate into the use of force. The results-based on the examinationof the consequencesof all twentieth-centurysecessions--reveal thatethnicallybasedterritorialdisputesplay a muchgreaterrole in conflict onset thando theireconomicallyor strategically based counterpartsand thatpeaceful secessions lead to peaceful relations. Keywords: secession; partition; territorialdispute; territorialchange; internationalconflict

One of the characteristicsmarkingthe end of the cold war has been an apparent explosion in the numberof demandsfor full-fledgedethnicself-determination,thatis, independence.Several states have been brokeninto smaller,ethnically based countries, with consequences ranging from the peaceful interstate relations following the breakup of Czechoslovakia, to the interventions of Serbia-Montenegro and Croatia in Bosnia, to the apparent ability of the division to end hostilities between SerbiaMontenegro and Slovenia. The vast range of experiences has helped fuel the debate over the desirability of acquiescing to the secessionist movements' desire to form their own countries out of the lands of existing states. The debate has separated the scholarly community into two diametrically opposed camps: those who view the policy as a disastrous, unacceptable course of action that only leads to future violence (e.g., Horowitz 1985; Etzioni 1992; Kumar 1997) and those who see it as the only policy that

AUTHOR'SNOTE:I thankPaul Diehl, Bear Braumoeller,Dina Zinnes,JohnVasquez,Doug Stinnett, and Doug Giblerfor theirhelp in the developmentof this article,as well as Paul Huthand ToddAllee for sharing their territorialdispute data. The data and Stata do-files used for the analyses can be found at www.yale.edu/unsy/jcr/jcrdata.htm. JOURNAL OFCONFLICT RESOLUTION, Vol.49 No. 5, October2005713-741 DOI:10.1177/0022002705279426 © 2005SagePublications 713

714

JOURNALOF CONFLICTRESOLUTION

is able to preventfuturearmedconflict (e.g., Kaufmann1996; Tullbergand Tullberg 1997). Because muchof the debatehas been markedby each camppromotingits own viewpoint, there is currentlylittle consensus on why interstaterelationsin the aftermath of state breakupare sometimes peaceful (e.g., Estonia-Russia)and in other instances violent (e.g., Ethiopia-Eritrea). In this work, we referto these and similarcases of statebreakupas secessions (or partitions).More specifically,we define secession as an internallymotivateddivision of a country'shomeland(i.e., noncolonial) territorythat results in the creationof at least one new independent(i.e., secessionist) state(e.g., Eritrea)-with full sovereign rights and legal recognitionby the internationalcommunity-and leaves behind the now territoriallysmallerrumpstate(e.g., Ethiopia).'We see this type of separationas distinctfroma de facto controlof territoryby a rebelgroup(e.g., the TurkishRepublic of NorthernCyprus)or internalgovernmentrestructuring,even if it results in broad autonomy-because no new state has been created-and from decolonization(e.g., Angolan independence)-because the new countriesare not formedout of the imperial state'shomelandterritory.Althoughmost attentionhas been paidto the post-cold war ethnic secessions, secessions are neitherrecent nor exclusively ethnic phenomena. For instance,Sweden andNorway ended theirunion in 1905, andcountrieshave been divided for ideological (e.g., China-Taiwan)as well as economic (e.g., Singapore-Malaysia)reasons. We thus strive for a parsimoniousapproachthat applies to ethnicandnonethniccases alikeas well as to a timeperiodsignificantlylongerthanthe post-cold war era. We tacklethe issue of secession's aftermathfromthe territorialdisputeperspective because, in their drive for self-determination,secessionists disagree with the central governmentover who should have sovereigncontrolover not only which people but also over what land. Hence, the disagreementis territorialin nature,and what commences as a territorialdisputewith domesticoriginsleads-if the countryis dividedto the creationof new internationalboundaries.The new boundariesseparatetwo or more internationallyrecognized countriesthat formerlybelonged to the same state. Forthe sakeof brevity,we referto these rumpandsecessionist statesas statesinvolved in secessions (SISs). The aim of this workis to investigateone of the consequencesof the bordercreation,thatis, the circumstancesunderwhich the secession attenuatesthe underlying territorialdispute to the extent that subsequent armed confrontations between SISs over the new boundariesare avoided.2 More formally,the focus of this projectis on the causes of international-levelmilitarizedconflict over territory(conflict for short),takingplace between the rumpand secessionist states-or between secessionist states in the case of multilateralsecessions-that formerlyconstituteda single country.The studyaims to answerwhy some secessions are followed by boundaryconflicts between SISs while othersarenot. By 1. This use of the termsecession is consistentwith Wood's (1981) view thatit representsa substate unit's demandfor formalwithdrawalfrom the state on the basis of claims to independent,sovereignstatus. Secession is thus a subsetof separatism,which involves all aspectsof political alienationthatwould reduce the centralgovernment'scontrolover the region. 2. Domestic-level consequencesof state breakupare examinedby Tir (2005).

Tir/ KEEPINGTHEPEACEAFTERSECESSION

715

providingan explanationthatcould accountfor bothpeacefulandviolent postdivision outcomes, we seek to inform the pro- versus antisecession debate. Furthermore,by adoptingthe territorialdisputeperspective,we aim to extendthe territorialconflict literature'scoverageto this largelyignoredsubjectarea.In partdrivenby Holsti's (1991) observationthat most interstateconfrontationsover the past few centurieshave had territorialorigins, the territorialdispute approachhas become one of the dominant ways of studying the causes of armedconflict in the field of internationalrelations. Among many others,the approachis favoredby researcherssuch as Vasquez(1993), Huth (1996), Huth and Allee (2002), and Senese and Vasquez (2003). Yet, this researchvein generally focuses on attemptsto alter internationalboundarieswhile largelyignoringthe consequencesalterationsproduce(Diehl 1999). Of the systematic efforts focusing on the consequences of territorialchanges (Goertz and Diehl 1992; Tir 2003), only Tir (2002) examines specifically secessions. Tir is able to show systematicallythat secessions are followed not only by conflictualbut also by peaceful outcomes. Yet, the aim of that study is not to explain the variationsin the outcomes eithertheoreticallyor throughmultivariateanalysis.Moreover,due to datalimitations, the author'sabilityto takeinto accountthe consequencesof secessions occurringatthe end of the cold warwas quitelimited.This articlethereforenot only buildson previous studies-by developing and testing a theoreticalexplanationthat accounts for the variance in secessions' aftermath-but also helps bridge the traditional divide between internationaland ethnic/domesticconflict research. In terms of policy making, we seek to identify the circumstancesunder which dividing a countrymay-or may not-be beneficial. The extantscholarlyadvice has been to either allow secession in every situation (e.g., Mearsheimer 1993, 1998; Mearsheimerand Van Evera 1995, 1996, 1999) or never (e.g., Etzioni 1992; Kaldor 1996; Kumar1997). Perhapsreflective of such contradictoryadvice, policy makers havebeen inconsistentin the mannerthey have dealtwith ethnic secessionist conflicts (e.g., allowing Bosnia, Croatia,and Slovenia to secede but not allowing secessions of Kosovo, Montenegro,or RepublikaSrpska).Knowing when to divide countriesand when not to do so is importantbecausethereexist manypotentialdecision pointsin the future when the division or preservationof a (multiethnic) state will have to be addressed (e.g., Indonesia, Russia, Rwanda, Israel, Sudan, Iraq, Afghanistan, Sri Lanka,Nigeria,Congo, Cyprus,Spain,Canada).The internationalcommunityis inevitably involved in such decisions because it ultimatelydetermines-by extending or denying recognition-whether a secessionist region becomes a new country. The internationalcommunityalso decides withinwhatbordersto recognize the new countries.The commonly acceptedrule in recentrecognitionshas been to makethe former internal,administrativebordersthe new internationalborders.Are the policy makers makinga mistakeby not tacklingpotentialpostsecessionterritorialdisputes?Is therea particularaspectof the dispute(e.g., ethnicallybasedterritorialdispute)thatshouldbe resolved at the expense-if necessary-of otheraspectsof the dispute(e.g., economically based territorialdispute)?Should the more powerful countrybe favoredin the ensuingdistributionof the land?Wouldit be betterif the internationalcommunityrecognized new states before the secessionists take up arms againstthe centralgovernment (or before the lattertriesto suppressthe movementviolently), or is the secession

716

JOURNALOF CONFLICTRESOLUTION

processirrelevant?These areall importantpolicy-makingquestionsthatarein need of systematicallyderivedanswers. The remainderof the studyis organizedas follows. In the next section,we presenta model linkingsecessions with futurepeace andconflict oversecession-createdboundaries. In lieu of a literaturereview, extantworks are consultedin the developmentof the model and associatedhypotheses.The researchdesign and empiricalresults sections follow. We close with policy-makingand researchimplications.

SECESSIONS AND SUBSEQUENT MILITARIZED CONFRONTATIONSOVER LAND: A MODEL We startby observing that SISs (i.e., rumpand/orsecessionist countries)may be dissatisfied with how the secession distributedthe land between them. Throughthe secession, the rumpstatehas lost some of the territoryit previouslycontrolledto the secessionist stateandmay wanta portionor all of it back. Conversely,the secessionist state may not be satisfied with how much land it has received and may desire even more of the rump state's land.3Finally, the secessionist state may set its sights on another secessionist state's territory, as the Armenia-Azerbaijan dispute over illustrates.In all these scenarios,the secession was the resultof an Nagorno-Karabakh attemptto deal with a domestic-level territorialdispute;however, one of its consequencesmay havebeen the onset of international-levelterritorialdisputesin which the SISs seek to redistributethe land in ways more favorableto them. Priorresearchindicates thatin territorialdisputesituations,the use of force is quitelikely: both Vasquez (1993) andHolsti (1991) findthatcountriesaremorelikely to use force to acquireland than to deal with any otherissues (e.g., trade,ideology).4We aim to ascertainunder what circumstancesthe territorialoutcome of the secession will be challengedby the use of force. Webuilda theoreticalmodel in which we presupposethatin chargeof SISs aretheir respectiveleaders,who areoffice holdersinterestedin improvingtheirpositionsinternationally(e.g., greaterinfluence, prestige) and domestically (e.g., extendedtenure, promotionof theiragenda,a positiveplace in the country'shistory).Domestic support is assumed to be importantto both democratsand autocrats.All leadersrequirethe backing of certain domestic groups to remain in office; Bueno de Mesquita and Siverson(1995) referto these key constituenciesas the selectorate.The model ties the leader'sdesiresfor influence andretentionof office to the factorsrelatedto the secession to explainthe leader'swillingness to use force in the secession's aftermath.These 3. For example,in the recentEthiopia-Eritreawar,the two countriesdisagreedabouttheircommon boundary.Some observerseven arguedthatEthiopiadid not simply want more borderlandbut wantedthe whole Eritreaback. 4. In contrast,Brooks(1999) arguesthatterritorialcontrolis becomingless importantto statesin the age of tradeandtechnologicaladvancement.Yet Liberman(1996) notes thatterritorialconquestis useful if the conqueroris willing to exploit the land beyond what tradingcould accomplish.We add thatterritorial controlis the only policy alternativethatgoes beyondprovidingat best partialcontrolover the valuableland resources.Finally,territorycan be valuedin termsof meaning(Newman 1999;GoertzandDiehl 1992), so the value of controllingit remainsunaffectedby modernadvancements.See below.

Tir/ KEEPINGTHEPEACEAFTERSECESSION

717

factors are postsecession territorialdisputes, the secession process, and the level of potentialantagonismbetween certainSISs. The decision to fight to regaina lost piece of land (from the rump state's perspective)or to extend the newly acquiredterritory even further(fromthe secessionist state'sperspective)is influencedin partby whether andhow muchvalue the landin questionpossesses. (Re)acquiringthe landthatcan be used to boost the country'srelativepoweror economic clout bringsthe leadera greater internationalinfluence. Domestically, the leader can increase his or her popularity among those portions of the selectoratethat are concerned with the country's economic and defensive abilities and its internationalstatus. The leader can also reach beyondthese narrowerconstituenciesby attemptingto (re)takethe landthe citizens in generalconsiderintegralto theirnationalidentity.5Furthermore,the peaceful versus violent process by which the secession took place presentsthe leader with potential constraintsagainstandopportunitiesfor using force to challengethe secession's territorialoutcome.Finally,we arguethatthe relationshipbetween rumpand secessionist statesmaybe moreconflictpronethanthe interactionbetweentwo secessionist states. TERRITORIAL DISPUTES

From the rump state's perspective,the secession has given the secessionist state land that the rump state may value and want back. If the lost land is valuableto the rumpstate,by retakingthe land,the leaderof the rumpstatehopes to obtainthe previously describedbenefits. The rumpstatemay value the lost land along one or both of the following dimensions;in turn,each manifestationof each dimensionrepresentsan alternatepathbetween the secession and subsequentconflict. First,the land in question may have tangible (Newman 1999) or intrinsic(Goertz and Diehl 1992) value, manifestedin strategicor economic terms.Land'sstrategicvalue arisesfromits characteristicsand/orlocation. Losing a high groundor an impenetrableswampor desert may makethe countryeasierto invadeandthusundermineits defensiveability.Losing a piece of landcontainingresourcessuch as ore deposits,ports,and so on undermines the rump state's economic-and, by extension, military-capability. The desire of countriesto pursuepower is one of the cornerstonesof the realist school of thought, andat least some realistsview the role of territorialcontrolas crucialto a state'spower (e.g., Spykman1944;Mackinder1919;Morgenthau1948; GochmanandLeng 1983). Controlof importantstrategicand economic lands can thus be regardedas contributing to a country'spower base, so the rumpstatehas an incentiveto contest the loss of the landvaluedalong the tangibledimension.This agendais expectedto resonateparticularlywell among the portion of the selectoratethat is concernedwith or would directlybenefit from the acquisitionof such lands (e.g., business enterprises,defense contractors,border-regionpopulation). Second, the leader may want to use force to reclaim the lost land because it possesses meaning to the country's citizens. Such land is thought to have intangible 5. Note thatleadershope to derivethe relevantbenefits:they do not have sole controlover the outcomes (e.g., remainingin office), as they also dependon otherfactors(e.g., electorallaws, political opponents).Thus, the benefitsderivedfrom territorialconquestdo not guaranteethe achievementof the desired goals.

718

JOURNALOF CONFLICTRESOLUTION

(Newman 1999) or relational(GoertzandDiehl 1992) value. Valueas meaningmanifests itself in forms representingadditional paths to territorialconflict: ancestral homeland (e.g., Kosovo to Serbs), a religiously significant site (e.g., Jerusalem),or landinhabitedby ethnicbrethren(e.g., Hungariansin westernRomania).Importantly, scholars(Touval1972; Vasquez1983; GoertzandDiehl 1992; Diehl 1999) arguethat intangibilityintensifiesthe willingness to fight for the land.In termsof the model, the acquisition of such lands is likely to appeal to a potentially larger portion of the selectorate,broadeningthe leader'spopularityandlegitimacyandgiving him or heran additionalincentive to use force. Yet, because of currentdata limitations,we cannot test for the impactof the nonethnicaspects of intangiblevalue; we thus focus on the ethnic componentof intangiblevalue, althoughsimilar argumentshold for its other manifestations.From the citizens' perspective,the foe-controlledland inhabitedby theircoethnics is perceivedas integralto the nationalidentityand is thereforeseen as (1) personal,(2) indivisible, and (3) unsubstitutable(Gottman1973; Sack 1986). (1) The people feel as if theyhavea personalstakein the landpopulatedby theircoethnics. Acquisitionof these landsbringsto the largerportionof the citizenrya greatersense of rewardthandoes, say, a coal mine because the rewardfrom tangiblyvaluableland is, somewhatironically,seen as more abstract.Hence, the people will be more likely to rewardthe leader who acquiresthe coethnics' territory.(2) Having anotherpartyin control of even a portionof this land can lead to feelings that the state is not whole because it does not encompass the entire nation. In contrast,the coal mine could be dividedor shared.(3) Anotherpiece of territorycannotsubstitutefor the originalone because the emotional attachmentsto the coethnics' land cannot be replicatedelsewhere (e.g., evacuatingthe remainingSerbs from Kosovo and resettlingthem in Serbia properis seen as unacceptable).In contrast,a differentsource of coal could be found. Furthermore,the attachmentstend to persistfor generations,meaningthatold disputescan be revisitedeven afterprolongedperiodsof peace. Due to reasons(2) and (3), the controlof landspopulatedby ethnicbrethren-and of intangiblyvaluedlands in general-tends to be an all-or-nothingventure,so peaceful compromise,division, andland sharingaredifficultif not impossible.The benefits-drivenleaderwill have an easier time convincingthe generalpopulationthatmilitarizedactionis the only alternativeofferingaccess to the desiredland.In contrast,it is both comparativelyeasierto compromiseovertangiblyvaluedlandsandharderto convincethe broaderselectorate thatfighting is necessary;hence, a violent approachis expected to be less beneficial for the leader. From the perspectiveof the leader of the secessionist state who is contemplating whetherto use force to redrawthe secession-createdmap,the key differenceis thathis or her countryhas not lost land due to the secession. Instead,to receive the benefits of territorialconquest,the leaderfocuses on obtainingeven more land;the targetof the leader's desire can be eitherthe rumpstate or anothersecessionist country.An argument could be made thatthe leader'sopportunityto profitfrom additionalterritorial gains may be limited.Throughthe secession thatestablishedit as a new, independent country,the secessionist state has alreadyreceived large swaths of territory;this has decreasedsubstantiallythe totalvalue of the landthe secessionists wantedpriorto the secession. In turn, this would make the leader's willingness to become involved in

Tir/ KEEPINGTHEPEACEAFTERSECESSION

719

futureconflict over territorylow. Yet this may not be the case in two situations.First, through the secession, the secessionist state obtained only a portion of the land it hopedto acquire,as indicatedby territorialclaims the secessionist statemakesagainst otherSISs afterthe secession. Second, the secessionist stategainedonly a segmentof the landoccupiedby the ethnicbrethren.Because this territoryis likely to be perceived as indivisible (Newman 1999; Diehl 1999), partialgains arenot likely to decreasethe value of the remainingterritoryto a greatextent.In eithercase, the leadercan profitif the secession failed to resolve the territorialdisputethatmotivatedthe secession to the country'ssatisfaction.Thus, similarto the rumpstate'sleader,the secessionist state's leaderis motivatedto use force by the benefitsof conqueringterritorythe constituents find valuable.The above discussion leads to the following hypotheses: disputeswithone anotherwill be morelikelyto Hypothesis1: SISsinvolvedin territorial experiencemilitarized disputes. conflictthanSISswithno territorial Hypothesis2: SISswithterritorial disputesinvolvinglandsoccupiedby ethnickinwill be thanSISswithterritorial morelikelyto experiencemilitarized conflictoverterritory disputesinvolvingtangiblyvaluedland. Extantliteratureoffers preliminarysupportfor these expectations.Findings from the territorialconflict literatureconfirmthatthe presenceof territorialdisputespositively affects the probabilityof conflict (e.g., Senese and Vasquez2003; Huth 1996; Hensel 2000). Moreover,the roles of intangible-including ethnic-factors' intensifying influence are supportedby Goertz and Diehl (1992), Vasquez(1983), Bremer (1992), and Huth (1996). However,apartfrom Goertzand Diehl (see also Tir 2003), none of these works examines these issues in the context of territorialdisputes that resultfrom alterationsof internationalboundaries.While Goertz and Diehl consider this context,they do not investigatethe consequencesof secessions, andtheiranalysis is limitedto whetherthe statethathas lost landis consideringto recoverit by the use of force; that is, the additionalterritorialambitionsof the state that has alreadygained land(i.e., the analogueof the secessionist state)arebeyondthe scope of theirwork.6 The (ethnic) secession literaturedoes not, unfortunately,offer much systematic evidence aboutthe impactof territorialdisputeson futureconflict; yet, a host of case studies opposing secession develops an argumentabout futureconflict based on the territory'sethnic value. Mindful of problems following secessions in areas such as India,Ireland,Palestine,and the formerYugoslavia,a groupof antisecessionauthors (e.g., Hachey 1972; Fraser 1984; Horowitz 1985; de Silva and May 1991; Etzioni 1992; Brown 1993; Posen 1993; McGarryand O'Leary 1993; Kaldor 1996; Kumar 1997) arguesthatSISs arelikely to push for violent postsecession territorialrevisions becauseit is impossiblefor the secession to createethnicallyhomogeneousstates.The domestic problems associated with the poor treatmentof minoritieswill create tensions between the states, as each state seeks to protectits ethnic brethrenwho have been "left"in the otherstate.The nationalistleaderswill pushfor the redrawingof bor6. This is similarto Huth's(1996) theoreticalapproach,which accommodatesonly one challengerto the territorialstatusquo. Here,both the rumpand secessionist statescan challenge the postsecession status quo.

720

JOURNALOF CONFLICTRESOLUTION

dersby force, thatis, for addressingthe secession-created,ethnicallybased territorial disputes. Even the secession's supportersacknowledge these dangers. To ensure postsecessionpeace, Kaufmann(1996, 1998) andTullbergandTullberg(1997) advocate (forced) population transfersthat would eliminate territorialdisputes. (Note, however,thatthis may simply shift the value of the territorytowardanotherintangible value-that is, ancestral-dimension.) Even if this logic appearscompelling,it entails at least two shortcomings.The ethnic territorialdisputeissue is assumedto both exist in every case andbe the sufficientcause of conflict. In contrast,we allow for the possibility thatethnicallybased territorialdisputesare not ever-presentand thatotherfactors may tempertheirpronenessto escalate into the use of force.7 SECESSION PROCESS

We furtherarguethathow the secession was conductedmay constrainthe leadersof SISs against using force to revisit the secession-createdboundaries.The two basic processes consideredare peaceful and violent secessions. Vasquez(1993) maintains thatterritorialwarscan be avoidedif disputantsagreeon how to allocatethe contested land (see also Kocs 1995;Huth 1996). Applyinghis logic to the secession context,this means that an agreed-on secession either preemptsor rendersfutureterritorialdisputes so minor thatthey are not worthfighting over. Vasquez'slogic of agreement's benefits can be expanded:the partiesreceive benefits such as land, securityguarantees, or evenjust a decreasein uncertaintyover theirfutureinteractions.The threatof the loss of benefits and the reputationalcosts of breakingthe agreementdecreasethe utility of futureconflict. If, in contrast,the secession results from violent confrontations, the boundaryis imposedand,at best, preemptsthe futureterritorialdisputeonly in the eyes of one of the states.8 Vasquez(1993) arguesfurtherthatmutuallyacceptableborderscan be negotiated bothbefore andafterviolent confrontations(i.e., a postwarsettlement).We agreewith the initialpartof the logic suggestingthatpeaceful secessions would bringpeace. We, however,rejectthe implicationthatpostsecessionistconflict settlementsbringpeace9 andinsteadmaintainthatsecessions precededby violence createincentivesfor future confrontations,regardlessof whetheran agreementhas been signed. Peaceful secessions constrainthe SISs' leadersagainstfutureuses of force. At the time of the secession, the leadersusuallysign an agreementindicatingthattheyaresatisfied with the distributionof landbetween theircountries.The leaderwho uses force 7. Althoughthe focus of the secession literatureis predominantlyon the ethnicvalueof the territory, Schaeffer(1990) notesthe significanceof strategiclandsin makingthe argumentthatsecessions createrivalries overlandcontrolthatmay escalatethe competitionto the nuclearlevel, as in the MiddleEastandIndian subcontinent.In addition,in their studies of the Israelipursuitof expandedborders,both Newman (1999) and Vanzo(1999) arguethatIsraelhas had an interestin obtainingstrategicallyandeconomicallyvaluable territory. 8. If the conflict's winnerwas not able to conquerall the lands it sought, it too has an incentiveto revisitthe secession-createdboundary.SequentialIsraelienlargementis an exampleof such unilateral"settlement,"showing that the initial, secession-relatedterritorialgain was insufficient. 9. Likewise, Werner(1999) reportsthatwar-endingtreatieshave no effect on the durationof subsequentpeace.

Tir/ KEEPINGTHEPEACEAFTERSECESSION

721

to revisitthe outcome of a peaceful secession risks bringingan arrayof possible costs (e.g., damagedinternationalreputation,difficulty in securingtradearrangements)to the country.The leaderwho attemptsto rally the country'speople againstthe foe will thusbe hard-pressedto find much supportand may even be perceivedas jeopardizing the country'ssafety and reputationfor personalgain. Yet a peaceful, agreed-onsecession is not a guaranteeagainstfutureterritorialconflict. Preferencesover land distributioncan change-due to a change in the country's leadershipor generalpolitical climate-and the leadercan actuallyreceive domestic supportfor renegingon an agreementthathas become unpopular.In some cases, the secession-relatedboundariesmay have been unpopularfrom the startbecause the leaderssigning the associatedagreementdid not reflect the population'spreferences or becausenonviolentcoercionwas appliedagainstthe leaderat the negotiatingtable. Both states can fall victim to this problem:the futurerump state could be pressured into relinquishinglandit did not wish to give up, while the would-be secessionist state could be pressuredinto scaling down its territorialambitions.In fact, the leaders of SISs areoftenpressuredby the internationalcommunityto acceptthe previouslyinternal, administrativeborders as the new internationalboundaries.In sum, peaceful, agreed-onsecessions presentconstraintsbut not guaranteesagainstthe use of force. In contrast,the leadersareconstrained,to a lesser extent,againsttakingmilitarized action to revisit the outcome of a violent secession. In the rumpstate, the leadercan rally the people against the secessionist state by claiming that the loss of land was unjustandthatthe countrymustwardoff the possibility of futuresecessions by showing thatno such actionwill stand.In some cases, no agreementfollowing the confrontationhas been signed,meaningthattherearefew reputationalandloss of side benefits costs. In postconflict agreementcases, the leaderof the rumpstate can claim thatthe agreementwas made under duress and is hence null and void. Given the preceding armedconfrontation,he or she is more likely to be believed by the constituentsand internationalobserversthan a leader of a rump state who lost land peacefully. The imposed agreementthus fails to constrainthe leader.For the secessionist state, it is possible thatits violent secession victorywas incomplete.Its leadersdesireeven more of the rumpstate'sland,andthey recognizethe opportunityto derivebenefitsby pushing further.If therewas a postconflictagreement,the secessionist statecan claim that the conflict-basedagreementis illegitimatebecause it was signed underbattlefieldor third-partypressure.10 In additionto the lack of restraints,the violent breakupmay actuallyencouragethe use of force in the futurefor at least threeinterrelatedreasons.First,violent secession involves the mobilizationof armedforces on both sides. Demobilizationmay not be rapid due to lingering, fighting-relatedtensions. The presence of mobilized armed forces makes it easier for the leadersto use this instrumentin dealing with the other state. Second, violent secession elevates military leaders to the position of power, 10. Even in the cases of internallymotivatedsecessions thatwere ultimatelyenabledby the internationalwardefeat(e.g., thebreakupof Austria-Hungaryfollowing WorldWarI), statesinvolvedin secessions (SISs) are facing a similarincentivestructure.In both scenarios,the secession is violently imposed against the rumpstate's wishes, while the leadersof the secessionist state may be dissatisfiedwith how much land was given to them because thirdpartiesdeterminedthe new borders.

JOURNALOF CONFLICTRESOLUTION

722

especially in the secessionist state.Such leadersare,arguably,morepronenot only to rely on uses of force but also to engenderfeelings of mistrustand insecurityin the rumpstate.This may serve to increasethe statureof militaryleadersin the rumpstate, creatinga similarpronenessto use force. Finally,the secessionist fighting helps constructemerging,international-levelbilateralrelationson the basis of enmity.This creates suspicionon both sides, makingfutureuses of force appearmore appropriateand justifiable to the selectorate(see Lustick 1993). We arguefurtherthata similarexpectationholds for the relationshipbetween two secessionist states following a multilateralsecession. The above logic implies that a peaceful departureof secessionist regions indicates that the would-be secessionist states accept the proposed territorialdivision-both vis-a-vis one anotherand the rumpstate.In contrast,violence betweenthe would-besecessionistregionspriorto or during the secession and over domestic-level sovereignty issues is an indication of seriousproblems.The would-berumpstate-with its preponderantpoliticalandmilitarypower-is expected to have preventedthe secession-mindedregions from fighting one anotherwhile they all still formedthe same country.If the supposedlydominant force was unable to prevent the violence, it is likely that once this factor is removed,the postsecessionrelationswill be problematic.A case in point is the Armenia-Azerbaijanconflict, which startedpriorto the dissolutionof the Soviet Union and grewto a full-scale territorialwarafterthe two republicsbecameindependent.Linking this observationto the model, the presecessionviolence betweentwo would-besecessionist stateslowers the thresholdat which the leaders'use of force is deemed acceptable by theirrespectiveconstituents.The leaderscan arguethatthe boundaryvis-a-vis the other secessionist statehas been imposed on the countriesagainsttheirwill. This argumentis more believable if the two secessionist groups have fought each other immediatelypriorto or duringthe statebreakupprocess." In sum, we expect thatin both the rump versus secessionist and the secessionist versus secessionist states contexts, the following hypothesis will hold: Hypothesis3: SISs aremorelikelyto experiencefutureconflictoverterritorywithone anotherif theywereengagedin a violent-as opposedto a peaceful-secession. The territorialconflict literatureoffers somewhatmixed supportfor the hypothesis. GoertzandDiehl (1992) find thatpeaceful alterationsof bordersareno betterin terms of futureconflict preventionthantheir violent counterparts.In contrast,Tir's (2003) results-based on examining boundarymovements between two existing statesshow that the peaceful approachis beneficial. Furthermore,both Huth (1996) and Kocs (1995) find thatlegally defined (which, by definition,meanspeacefully agreedon or, in their language, "settled")bordersare rarelyfought over. Yet none of these works offers a directtest of the hypothesis because they do not examine specifically the process of boundarycreationvia secessions. 11. Note, however,thatpeaceful presecessionrelationsarenot a guaranteeof futurepeace. The prior peace may have been a peace imposed by the preponderantpower(e.g., Russia).

Tir/ KEEPINGTHEPEACEAFTERSECESSION

723

The broaderinternationalconflict literatureimplicitly supportsthe hypothesis. Applying the reciprocityargument(e.g., Goldsteinand Freeman1990; Rajmairaand Ward1990) to secession suggeststhatif violence was used to promptthe secession, the futureuse of nonviolenttechniquesto retakethe landwill be viewed as a sign of weakness on the rump state's part and is thus avoided. Unlike reciprocity,in which two states direct similartypes of behaviortowardone another,reinforcement(e.g., Most andStarr1980; KirbyandWard1987) involvesrelyingon a previouslysuccessfulpolicy. In the secession context,this would meanthatif the secessionist stategainedland throughviolence, thenviolence appearsas a fruitfulpolicy for dealingwith additional territorialambitionsagainstboth the rumpand other secessionist states. Finally,in the contextof ethnicsecessions, Kaufmann(1996, 1998) proposesa diametricallyopposedexpectation:violent secessions arelikely to be followed by peaceful relations.He arguesthatdividing countriesis the only real solution to interethnic civil wars. The conflict producesethnic "unmixing,"a situationin which people flee and seek protectionamong their own kin. By creatingseparatecountries, secession finalizesandformalizesthe homogenizationprocessas well as dividesthe belligerents andallows themto live on theirown andin chargeof theirown affairs.This will allegedly create few interethnicproblems within the SISs. Internationally,the SISs will havelittlereasonto intervenein each other'sminorityaffairs,so the leaderswill not be able to benefit from the uses of force. Althoughthe authordoes not explore the converseformof the argument,it would suggestthatpeaceful secessions do not entailethnic unmixingand thereforeopen up the possibility for the mistreatmentof minorities andsubsequentmilitaryinterventions.Sambanis(2000) puts Kaufmann'sassertionto empiricaltest but finds thatsecessions areno betteror worse in endingcivil warsthan other solutions. Although important,this finding does not speak directly to our hypothesisbecause examiningthe consequencesof peaceful secessions is beyond the scope of Sambanis'swork. Also contradictingthe hypothesis are the secession's opponents, such as Hachey (1972), Fraser(1984), Horowitz (1985), de Silva and May (1991), Etzioni (1992), Brown (1993), McGarryand O'Leary (1993), Posen (1993), and Kumar(1997). By arguingthat all secessions will inevitablyfail to preventfutureconflict because new internationalborderscannotbe drawnin mannersthatcreateethnicallyhomogeneous states,they implicitly statethatthe secession process is irrelevant.The cases of Palestine, Ireland, and India are often cited in support, even though each involved presecessionviolence. Only a small segmentof the secession literaturesupportsthe hypothesis.Tullberg andTullberg(1997; see also Gurr1993; Tir2002) criticizeKaufmann(1996) by arguing that people who flee conflict zones will want to returnto reclaim their land. (In termsof ourmodel, this meansthatviolent secessions provideincentivesfor leadersto seize upon the landreturnissue and use force againstthe othercountryto bettertheir politicalstatus.)Instead,they maintainthatpeaceful secessions-preceded by a referendumon whetherto keep the countrytogether-are the only way to avertfuturevio-

724

JOURNALOF CONFLICTRESOLUTION

lence. Furthermore,Maoz (1989) confirmsour expectation,even thoughhe relies on an analysis of decolonization and on a different conflict-producingreasoning. He arguesthatnew, decolonizing statesthatgained statehoodthroughviolence areunder pressureto establishthemselvesby showing a willingness to use force. If they areperceived as weak or reluctant,their reintegrationmay be attempted.In contrast,states that gained independence peacefully are under no such pressure because their statehoodis viewed as more legitimate. RUMP-SECESSIONIST STATE INTERACTION

Whetherforce is used to challengethe territorialoutcomeof the secession may also dependon the type of countriesinvolved.The secession is likely to createmoretension between the rumpand secessionist states thanbetween two secessionist states. Each new secessionist statemustdismantleits union with the rumpstate,while the dissolution vis-a-vis other secessionist states is more or less automatic;in fact, some secessionist statesmay even cooperatewith one anotherto facilitatetheirdeparturefromthe union (e.g., the Baltic republics,Slovenia-Croatia).The rump-secessionistrelations are,in contrast,not off to as good of a start.The rumpstatemayharborresentmentover havinglost landto the secessionist state,while the secessioniststatemaybe suspicious of the intentionsof the rumpstate-which may include rolling back the secessionand of the rump state's patronizingbehavior(e.g., Russia's "we know best" attitude towardthe former Soviet republics). In the rump state, the resentmentwill make it politically less controversialfor the leaderto use force to deal with issues-including the distributionof territory-that arisewith the secessionist state.Similarly,the secessionist state's leader can arguethat the use of force is appropriatebecause the state needs to send a signal of determinationand strengthin dealing with the rump state. These dynamicslowerthe thresholdfor using force to deal with the territorialoutcome of the secession. In short: Hypothesis 4: Interactions betweenrumpandsecessioniststateswillbe moreconflictprone thaninteractions betweentwo secessioniststates. The territorialconflict and secession literaturesdo not explore this possibility. A potentiallyrelevantargumentis madeby the prospecttheory(Levy 2000). According to this view, the leaders will adopt a more risk-acceptantbehavior-which includes loweringthe thresholdfor the use of force-when theyoperatein the domainof losses. The rumpstate'sleadershiparguablyoperatesin the domainof losses because this is the statethathas lost landthroughthe secession. The secessioniststate'sleader,in contrast,may be operatingin the domainof gains and adopta morerisk-averseapproach in ordernot to gamble away the state's independence.Despite the similaritybetween this and our argument,note thatthe predictionsdiffer somewhat:we predictthatboth the rumpand secessionist states will be likely to use force in their dealings with one another.Aftera considerationof the theoreticallinkagesbetweenthe key variablesand postsecession territorialconflict, we turnto the researchdesign issues.

Tir/ KEEPINGTHEPEACEAFTERSECESSION

725

RESEARCH DESIGN OPERATIONAL DEFINITIONS

Secession and relevantSIS dyads. We definedsecession as an internallymotivated divisionof a country'shomeland(i.e., noncolonial)territorythatresultsin the creation of at least one new independentstate-with full sovereignrightsandlegal recognition by the internationalcommunity-and leaves behindthe now territoriallysmallerrump state.Small andSinger's(1982) definitionof the stateis used to determinewhetherthe secessionistregionqualifiesas a full-fledged state,while the Tiret al. (1998) list of territorialchanges helps us determinewhetherthe new states have been createdout of homelandterritories.According to these coding rules, the Germanand Koreandivisions at the end of WorldWarII areexcludedbecausethey do not representcases of an Yettwo grayerareacases associatedwith decolointernallymotivatedstatebreakup.12 nization-the divisions of IndiaandPalestine-are included.Consistentwith the definition employed, the Muslim populationin Indiaand the Jewish populationin Palestine werepursuingstatesseparatefromtheirneighborsas the colonialrulewas nearing its end, thus indicatingan internalmotivationfor the division. Arguably,these arenot so muchcases of decolonizationbutcases of stateformationout of the homelandterritory of the state that the neighboringgroups wanted to form after British departure. This is clearerin the Indiancase, while the Palestiniancase is more debatable.A portion of Palestinebecame Israel, while the remainderwas incorporatedinto Jordan,a state emerging from the neighboring and likewise British mandate territory of Transjordan.These cases are analyzed in just about every prior study of secession's aftermath,so their omissions would renderthis work less capableof speakingto this literature.In any case, the inclusion of these cases has little effect on the results. A relevantdyad is composed of a pair of states thatbelonged to the same unified statepriorto the secession (e.g., Ethiopia-Eritrea).This rule is sensible for secessions thatproduceonly one secessionist state,but in cases with multiplesecessionist states (e.g., the breakupof the Soviet Union), matching all possible pairs produces some apparentlyirrelevantdyads (e.g., Estonia-Kazakhstan).An alternativepossibility is to consideronly those dyadsthataredirectlycontiguous(i.e., includeEstonia-Latviaand Estonia-Russiabut not Estonia-Kazakhstan).This solution, however,ignores potential ties thatthe once-dominantrumpstatemay haveto the territoryof a noncontiguous secessionist state.For instance,Russia may lay claim to partsof Kyrgyzstanbased on the presence of a sizable Russian minority.As a sensible compromise,we adoptthe politically relevant dyad (PRD) concept (Maoz and Russett 1993) commonly employed in the internationalrelationsresearchas our main approach:in this study, politicallyrelevantdyads are all contiguousdyads emergingfrom the same stateplus all rump-noncontiguous-secessioniststatepairs.The PRDs resultingfrom all twentieth-centurysecessions are listed in the appendix.To ensure that the results are not a 12. AlthoughAustria-Hungary'sbreakupwas promptedby its defeat in WorldWarI, we includethis case because the division of the countryhad a strong internalimpetus:the breakupof the state has been sought by its many ethnic groupsfor some time.

726

JOURNALOF CONFLICTRESOLUTION

functionof the dyadselectionrule,we rerunthe analyseson all andon only contiguous dyads. Dependent variable. The dependentvariableis the onset of fatal militarizedconflict over territorythattakes place between two SISs. To operationalizethis concept, we consultGhosn,Palmer,andBremer(2004), who providea listing of all militarized interstatedisputes(MIDs)'3through2001 andnotebothwhetherthey involvedcasualties and were fought over territory.14 To evaluatehypothesis 1, we use all fatal MIDs listed, butfor the otherhypotheses,we use only those instancesin which the fatalMID concernedterritorialcontrolfor at least one of the disputants.The dataarearrangedon a yearly basis, so the interactionwithin each dyad is trackedfor every year afterthe secession; dyad-yearis thereforethe unit of analysis. To verify the robustnessof the findings, both nondirected(testing for conflict involvement)and directed(testing for conflict initiation) designs are employed. The relatedfatal territorialMID rates are about 3 percentand 1.5 percent,respectively. Territorialdispute.We rely on HuthandAllee's (2002)15data,which note the presence of territorialdisputesandreportwhetherthe disputedlandis valuedalong a set of mutuallynonexclusiveindicatorsrepresentingthe land'sethnic,economic, andstrategic values.The territory'seconomic andstrategicvaluesrepresentdifferentmanifestations of the tangibledimensionof the land'svalue.Indicatorsof intangiblevalue other thanethnicmanifestationareunfortunatelynot availablefromthis or otherdatasets at this time. We thus employ the ethnic value as the sole indicatorof intangiblevalue.16 Initially,we attemptedto include all three dimensions of value for each partyin our statistical equations, but this created multicollinearityproblems because territorial claims are often reciprocated.We thus had to modify our original design by simply notingthe existenceof a particulartype of territorialdisputewithinthe dyad,insteadof assigningchallengerandtargetcategories.We feel comfortablewith this compromise because the initial results suggested that the dyad members'behavior mirrorseach other (see Tir 2001) and because the directed dyad results-in which we trackthe behavior that the challenger directs toward the target-closely resemble the nondirectedfindings. 13. A militarizedinterstatedispute(MID) is definedas "aset of interactionsbetweenor amongstates involvingthreatsto use militaryforce, displaysof militaryforce, or actualuses of militaryforce... these acts must be explicit, overt, nonaccidental,and governmentsanctioned"(Gochmanand Maoz 1984, 587). 14. Analyzing only fatal MIDs eliminates less serious confrontations(e.g., fishing disputes) and, becausenonfataldisputesin theThirdWorldoftengo unreported,reducesthe Westernbias in the dataset. 15. To Huth's(1996) list, coveringthe periodfrom 1918 to 1995, we addterritorialdisputesoccurring in the earlyandlate partsof the twentiethcenturyusing similarcoding criteria.The sourcesfor the extension and additionalcoverage of the data set include Allcock (1992), the New YorkTimes,Facts on File, World Almanacand theBookof Facts,Keesing'sArchives,CIA's WorldHandbook,Encarta,EncyclopediaBritannica, and so forth. 16. Huth(1996, 256-7) defines a disputedpiece of land as economically valuedif it containsnatural resourcessuch as oil, iron ore, uranium,and so on; strategicallyvalued if it is in close proximityto major shippinglanes, chokepoints,militarybases, attackroutes,andso forth;andethnicallyvaluedif it contains"a minority[that]speaksthe same languageandsharesthe same ethnicbackgroundas the largestethnicgroup within the challenger"state;see the source for full definitions.

Tir/ KEEPINGTHEPEACEAFTERSECESSION

727

Secessionprocess. The Tiret al. (1998) list of territorialchanges codes a change as violent if there was organized,violent conflict in the year precedingthe change. The coding rule shows that twenty-eight (out of sixty-six, about42 percent) SISs parted underviolent circumstances. Rump-secessioniststate interaction.Using Tir et al.'s (1998) list, we distinguish dyads composed of rump and secessionist states from dyads made up of two secessionist states. Controlvariables.In additionto the factorsthatmay contributeto the leader'swillingness to use force consideredin the model, severalotherinfluences may temperor exacerbatethe chances that the leader actually does so. First, whether a leader uses force to reconfigurethe new boundarymay be determinedin partby the country'srelative abilityto (re)takethe land;in fact, losing a warcan cost the leaderhis or her office (Bueno de MesquitaandSiverson 1995). The importanceof power distributionis well documentedin the internationalrelationsresearch,even though there is a disagreement about which distributiontype--balance (e.g., Morgenthau1948; Waltz 1979; Mearsheimer1993, 1998; Mearsheimerand VanEvera 1995, 1996, 1999) or preponderance(e.g., Organskiand Kugler 1980; Bremer 1992)-is best for keeping peace. We rely on the Correlatesof Warmaterialcapabilities for states' yearly composite measuresof power.Following the commonpractice(e.g., RussettandOneal2001), in the nondirectedanalyses, we note the dyad's relativepower configurationby taking the naturallogarithmof the strongerto weakerstate'spowerratio.The powerratiocalculationis adjustedslightlyin the directedanalysesby measuringthe ratiobetweenthe initiator'sandtarget'spowerlevels. Using the originaloperationalizationproducedno appreciablechange in the findings. Second, extant literaturehas shown that democracies tend to interactpeacefully with one another(e.g., Russett and Oneal 2001). If a leader of a democraticcountry wantsto challenge anotherdemocracyover the outcome of a secession, he or she will face normativeand/orstructuralimpedimentsopposing successful militarymobilization againstthe targetcountry.Therefore,the leader's agenda of territorialconquest againstanotherdemocracyis unlikely to generatemuch support.To distinguishpairs of democracies from mixed or nondemocraticpairs, Polity data (Jaggers and Gurr 1995) are used. Third,the anarchicalnatureof the internationalenvironmentdictatesthatbeing on good termswith countriessharingcommonsecurity interestsis especially important. Acting againstsuch a countrybringsto the leaderinternationalanddomesticcosts and consequently rendersit more detrimentalto pursue the path of violence to acquire land. To assess the presence of common security interests, we use Singer's (1995) alliance data. Fourth, the role of geographic proximity between the members of the dyad is importantbothbecausethe country'sabilityto projectits force dropsoverdistanceand because issues closer to the homelandare perceivedas more salient (Vasquez 1995). Yet the readershould note thatthis well-documentedfactor(e.g., Bremer 1992) may be less influentialin the postsecession context.Most secessions producenot only con-

728

JOURNALOF CONFLICTRESOLUTION

tiguous states but also states with a sharedhistory.Dispute salience may hence not decrease as sharply over distance. Furthermore,because the rump state exercised influence over the entire presecession territory,it has an in-depthknowledge of the secessionist state'sterrain.This, in turn,makes it easier for the rumpstateto act militarilyover distance.We use the contiguityby land measure(Stinnettet al. 2002). The datafor all of the abovecontrolvariablesareobtainedfromBennettandStam(2000). Fifth, we also controlfor the potentialimpactof leadershipchange. A new leader may be more willing to challengeviolently the secession-createdboundarybecause it was eitheragreedto by or imposedon someone else. Forthe generalargumentlinking leadershipchange to conflict, see Chiozza and Goemans (2003). Finally, a dummy variableidentifiesdyadsresultingfrom the disintegrationof the Soviet Union. It controls for potentialinfluences relatingto this uniquehistoricalevent thatcontributeda sizable portionof cases to our sample. METHOD OF ANALYSIS

To evaluate the hypotheses, we rely on the Cox regression (Cox 1975; BoxSteffensmeierand Jones 2000) with robuststandarderrorsand repeatfailures.17This method of analysis is appropriatefor longitudinaldatawith a binarydependentvariable whose valueis almostalwayszero (Beck 1998). In addition,it relaxesthe assumptions (1) of durationindependence,(2) of independencebetweenconflicts, and(3) that the dataare not censored,which are made by othercommonly used methods such as logit (Long 1997). Relaxing the formerand latterassumptionsis particularlyimportantbecause our dataset is composed of both older andrecentsecessions, which have had an unequal opportunityto experience conflict. In terms of interpretingthe Cox coefficients,we observethe impactof independentvariableson the hazardrate(probability of failure [i.e., conflict] at a given point in time, given survivalup to thatpoint) ratherthanthe usual,directimpacton the probabilityof conflict. Aside fromthe technical difference,the substantiveinterpretationof the coefficients is actuallysimilarto logit.18

EMPIRICAL RESULTS We begin by reporting,in Table 1, the resultsof nondirectedanalysesthatfocus on SIS dyad members'pronenessto become involvedin fatal MIDs with each other.In additionto the observationthat all statisticalmodels are highly significant,we note thatmost of the resultsreportedare consistentwith the hypotheses--even when controllingfor othercorrelatesof conflict. Startingwith hypothesis 1, the territorialdis17. Robuststandarderrorsaccountfor the fact thatobservationsfrom the same dyad areinterrelated. The repeat-failuresetup is used because one confrontationdoes not exhaust the possibility of future confrontations. 18. Cox regressionwas chosen overothereventhistorytechniquesbecauseit is nonparametric,meaning that it relies on fewer and weaker assumptionsaboutthe shape of the durationdistributionthan other techniques(e.g., Weibull,exponential;see Box-Steffensmeierand Jones 2000).

TABLE 1

Cox Regression Estimates of Fatal Postsecession Militarized Conflict In

Variable Territorialdispute Strategicterritorialdispute Economic territorialdispute Ethnic territorialdispute Violent secession Secessionist-only dyad Relative power Allies Dyadic democracy Contiguity New leader Soviet Union dummy 2(df) n

Model 1: MIDs, PRDs Only 3.040***

3.629*** -1.226*** -0.269** 0.480* -0.829 0.558 -0.322 1.353* 48.80***(9) 1,433

Model 2: TerritorialMIDs, PRDs Only

Model 3: TerritorialMIDs, All Dyads

T Con

(0.929)

(1.219) (0.459) (0.126) (0.331) (0.783) (0.886) (0.375) (0.934)

-0.654 0.474 3.136** 3.160*** -1.075* -0.291* 0.681** -0.769 -0.468 -0.130 1.056 97.25***(11) 1,433

(0.645) (0.926) (1.303) (1.106) (0.726) (0.189) (0.331) (0.830) (1.076) (0.333) (0.951)

-0.645 0.411 3.243*** 3.728*** -1.179** -0.282* 0.703** -0.795 -0.390 -0.143 1.086 118.82***(11) 2,174

(0.643) (0.945) (1.267) (1.112) (0.711) (0.193) (0.330) (0.843) (1.159) (0.333) (0.950)

-0.3 0.8 3.7 3.2 -0.9 -0.2 0.8 -0.9

0.0 1.3 105.2 1,223

NOTE:Cell entriesreportcoefficients androbuststandarderrors(in parentheses).Unit of analysis is a nondirecteddyad-year. secessions (SISs). MID = militarized interstatedispute;PRD = politically relevantdyad. *p < .10, one-tailed. **p < .05, one-tailed. ***p < .01, one-tailed.

729

730

JOURNALOF CONFLICTRESOLUTION

pute coefficient is significantand positive in model 1, meaningthatthe relationships between SISs arestronglyinfluencedby the presenceof unresolvedborderissues. The finding that the existence of territorialdisputes significantlyincreases the hazardof fatalMIDs helps extendthe territorialdisputeresearchinto the postsecession context. To put this finding into perspective,a look at the datareveals thatterritorialdisputes arepresentin about88 percentof the SIS dyads. Given thatwe arguedthatsecessions result from territorialdisputes with domestic origins means that secessions' success rateis about 12 percent.This figurecan be comparedto anotherprocessthatresultsin the creation of new borders,namely, decolonization. Applying the Huth and Allee (2002) datato boundariescreatedby decolonizationin the twentiethcenturyreveals that they are disputedat a rate of 34 percent.The secession's proclivity to generate territorialdisputesis thereforequite high. Yet, SIS leaders could be making territorialdemandswithout a resolve to follow them up with uses of force. Hence, it is importantto assess the risk to peace posed by the frequentpresenceof territorialdisputes.Territorialdisputesescalateto fatalMIDs in about24 percentof the SIS dyads.This escalationrateis notablybetterthanfor the decolonization-relatedterritorialdisputes (43 percent).In sum, despite the relatively high percentageof SIS dyads with postsecession territorialdisputes,the escalationto territorialMIDs is farfrom inevitableand low in comparisonto the cases of decolonization. This finding contradictsthe statementsof secessions' opponents,who claim that all secessions are likely to lead to conflict over the new borders. To determinewhich aspectsof territorialdisputesareparticularlyproblematicand to evaluatethe empiricalaccuracyof hypothesis2, we directour attentionto model 2. In this model, the dependentvariableis a fatalMID fought over territorialissues, and the territorialdisputesaredifferentiatedaccordingto theirunderlyingcharacteristics. The hypothesisreceives strongsupport:the ethnicvaluecoefficientis not only positive and significant,butin its importance,it also easily surpassesthe insignificantstrategic and economic (i.e., tangible)value coefficients. The leadersof rumpand secessionist states find more rewardin fighting for territoriesthat would unite lands occupied by ethnic kin than in fighting for lands thatwould increasetheircountries'economic or militarymight. As arguedabove, fighting for ethnically valued lands is expected to have a broaderpopularappeal.Furtherincreasingthe importanceof fightingfor ethnically valuedlandis the fact thatthe secession itself has helpedelevatethe statusof this issue on the country'sagenda.Given that most twentieth-centurysecessions involve separationof ethnic groups means that SISs tend to define themselves along ethnic lines. As a consequence,the objectiveof bringingall the kinsmenunderone roof, so to speak, providesthe leaderswith a significantpotentialpayoff. A closer look at the data reveals that ethnically based territorialdisputes among SISs arethe most commontype:they occurin about80 percentof the dyads,while the economically and strategicallybased territorialdisputesarepresentin 52 percentand 36 percentof cases, respectively.A comparisonto decolonizationrevealssome differences in the natureof territorialdisputes:landis most often contestedon the economic basis (23 percent),somewhatless frequentlyon the basis of ethnicity(20 percent),and quite rarelyon the strategicbasis (5 percent).Yet the greaterfrequencyof ethnically basedterritorialdisputesin the postsecessioncontextdoes not translateinto an unusu-

Tir/ KEEPINGTHEPEACEAFTERSECESSION

731

ally gravethreatto peace:only about26 percentof such disputesescalate,which is less thanone-halfof the ratefor similardisputesin the decolonizationcontext(56 percent). The strongsupportfor hypothesis2 indicatesthatthe findingsfrompriorterritorial conflict research(e.g., Goertz and Diehl 1992; Huth 1996) hold in the postsecession contextwhile discreditingcommonrealistargumentsthatpower-related(i.e., strategic and economic) territorialdisputesare the focus of states'interactions.In termsof the secession literature,the resultsshow thatthe debateappropriatelyfocuses on the most important culprit of postsecession conflict: ethnically based territorialdisputes. Hence, both opponents(e.g., Horowitz 1985; de Silva and May 1991; Etzioni 1992; Posen 1993; McGarry and O'Leary 1993; Kumar 1997) and proponents (Mearsheimer1993, 1998; MearsheimerandVanEvera 1995, 1996, 1999; Kaufmann 1996, 1998) accuratelyidentifythis issue as important.Yet, again,the opponentstend to overstatethe gravityof the problemposed by the presence of such disputes. The behavioranalogousto the findingsfor the firsttwo hypothesescan be foundin the cases of the Serbo-Montenegrinand Croatianinterventionsin Croatiaand Bosnia in the early 1990s. The rumpstate(Serbia-Montenegro)fought the secessionist states (CroatiaandBosnia) with the goal of bringingthe Serb-inhabitedlands in both countriesunderits control.Croatiahad a similaragendaof expandingits bordersover Croatian-inhabitedpartsof Bosnia. In fact, the interventionsconcentratedon the poorer and less-developedregions of Croatiaand Bosnia, suggesting only a peripheralinterest in acquiringeconomically or strategicallyvaluablelands. The Serbs concentrated much of their initial war effort on the eastern portion of Bosnia, while Croatians focused on Herzegovina;both areasare poor in resourcesand industry-two characteristicsfor which centralBosnia is actuallywell known. Likewise, most of the Serb war effort in Croatiawas focused on the lands inhabitedby ethnic Serbs and not on economically or strategicallyimportantareas. Examplesof secessionist statesattackingtheirrumpcounterpartsalso can be easily found. In both the 1908 Bulgariansecession from the OttomanEmpireand the 1919 Finnishsecession from the Soviet Union, the secessionist countriesfought lateron to acquireadditionalterritoriesinhabitedby theirethnicbrethren;the same landsheld little economic or strategicimportance.Finlandjoined WorldWarII-and did so on the Axis side, despitethe fact thatit was the only democracyto do so-exclusively to fight the Soviet Union and push its boundaryeastward.Similarly,Bulgariaparticipatedin the First (successfully) and Second (unsuccessfully) Balkan Wars to acquire even more of the Ottoman-controlledlands inhabitedby Bulgarians. Next, we turnto hypothesis3. Both models 1 and 2 show thatthe violent secession coefficientis significantandpositive. The findingthatviolent secessions significantly increasethe hazardof futureconflict is consistentwith the hypothesisandrelatedtheorizing:peacefulsecessions makeit morecostly anddifficultfor the leadersto use force thando violent secessions. An SIS dyadwhose memberspartedpeacefullyhas experienced futurefatalconfrontationsoverlandin only about5 percentof the cases; for violent secessions, this figure skyrocketsto 43 percent.In the decolonizationcontext,we also find thatpeacefully establishedbordersareless likely to be fought over (15 vs. 31 percent);this finding is consistentwith Maoz's (1989) argumentthat statesemerging from peaceful decolonizationare more peace loving than those that had to fight for

732

JOURNALOF CONFLICTRESOLUTION

their independence.Yet, importantly,the secession process variableproduces more extremeoutcomesthandoes its decolonizationcounterpart.The differencein the hazardthata violently-as opposed to peacefully-created borderwill be fought over in the secession contextis muchgreater(8.6 times) thanis the case for the decolonization context (2.1 times). This means that the stakes relatingto how the secession is performedarequitelarge.In fact, the resultspresentedin Table 1 indicatethatin termsof substantiveimpact,the process variablerivalsthe ethnicterritorialdisputevariableas the most influential.Sweden-NorwayandIndia-Pakistancases, for example,illustrate thatthe secession process has substantialinfluence on the qualityof futurerelations. The former,peaceful secession has producedpeaceful relationsover the past century, while the latter,violent secession has resultedin numerousconfrontations,including full-fledged wars. Although the finding that peaceful secessions lead to peaceful relationsmay not soundall thatsurprising,this findingis quiteimportantin the contextof territorialconflict research.Despite the expectationthatpeaceful alterationsof bordersare beneficial, GoertzandDiehl (1992) fail to show this empirically.In fact, this is only the second systematic empirical study (the exception is Tir 2003) to show that peaceful alterationsof bordersare beneficial and the first study to show this specifically for secessions. The findingsextendthe territorialconflict literature'sargumentscontending thatpeacefully agreed-onbordersare rarelyfought over (Kocs 1995; Huth 1996) to the contextof secessions. The findingsalso havebearingon the secession literature. We find evidence contradictingthe expectationsthatall secessions lead to futureconfrontations,regardlessof the secession process (e.g., Horowitz 1985; de Silva and May 1991; Etzioni 1992; Posen 1993; McGarryand O'Leary 1993; Kumar1997), or that violent secessions lead to peaceful relations (Mearsheimer 1993, 1998; Mearsheimerand Van Evera 1995, 1996, 1999; Kaufmann1996, 1998). Instead,we offer first systematic empiricalevidence supportingTullbergand Tullberg's(1997) and Gurr's(1993) arguments. The final hypothesis,hypothesis4, arguesthatrump-secessionistrelationswill be more adversarialthan relationsbetween two secessionist states. This will affect the thresholdat which the leadersfind it profitableto use force. The hypothesisfinds some supportin four out of the five models presentedin Table 1. The results suggestthough by no means strongly-that the main axis of territorialconflict lies between rumpandsecessionist states.We investigatebelow whetherthe prospecttheory-based explanationthattherumpstatesarethe likely initiatorsis responsiblefor the finding. Before turningto the control variables,we assess the robustnessof our findings. Recallingthe discussionon the topic of which sets of SIS dyadsto includein the analyses, we rerunmodel 2 using all dyads (model 3) andonly contiguousdyads (model 4). In model 5, we dropthe controversialIsrael-Jordancase; similarresults(not reported) obtain if the India-Pakistancase is dropped. The new findings resemble the ones reportedfor model 2 closely. As an additionalcheck of the findings'robustness,we use the directeddyaddesign. This design allows one to go beyond analyzingconflict involvementand observeconflict initiation.Assuming no preemptivestrikes,much of our logic suggests that the statedissatisfiedwith the secession-createdborderswill be the one to rely on the use of

Tir/ KEEPINGTHEPEACEAFTERSECESSION

733

force. Table 2 presentsmodels 6 through 10, which follow the setup of the first five models and evaluatethis portionof the argument.The first three hypotheses remain applicablein the (territorial)conflict initiation context, although some changes are notable.First,the economic territorialdisputecoefficient is now significant,suggesting thatthese disputesmay providethe leaderswith sufficientpayoff to use force. Nevertheless, the earlierfinding that the ethnically based territorialdisputes provide the greatestpayoff to the leaderstands.In model 7, the lattervariablehas abouta 1.5 times (or 50 percent)greaterimpacton the hazardof fatalconflict initiationthanthe former. Second, we noted earlierthatthe process variablehad a slightly largerinfluence than the ethnic territorialdispute variable.In the initiationcontext, the differenceis more discernible.The hazardof a fatal territorialMID being initiatedis 2.1 times (or 110 percent)greaterif the secession was violent thanif the potentialinitiatordisputesthe target'sterritorialholdings on an ethnic basis. In termsof the final hypothesis,we investigatewhetherthe prospecttheory-based explanationfor the greaterincidence of conflict between rumpand secessionist states is accurate.The related argumentmaintainsthat-all other things being equal-the leaders of rump states will be more likely to risk using force to alter boundaries becausethey areoperatingin the domainof losses. Accordingly,we changethe related variableto reflect whetherthe countryinitiatinga fatal (territorial)MID is the rump state. The incorrectcoefficient signs in every model indicatethatthe prospecttheory interpretationfinds no support.The reason for the lack of supportmay be that the secessionist statealso operatesin the domainof losses. It is possible thatthe secessionist state'sconstituentsandleadersee the disputed,rumpstate-controlledlandas rightfully theirs.The inabilityto controlthe landis thusinterpretedas a loss thattook place duringthe establishmentof the secessionist state.Both statesthereforehave a similar psychological perspectivetowarddesiredlandscontrolledby the foe. In sum, we find more convincing evidence to supportthe original explanation,arguingthat the main axis of conflict lies between rumpand secessionist statesratherthanthe prospecttheory-based alternative. We finally turnto the resultsfor the controlvariables.Tables 1 and 2 revealthatthe control variables do not perform as well as the key explanatoryvariables in the postsecession context. The relative power and dyadic democracy coefficients come the closest to meetingourexpectations,even thoughthey aresignificantonly by the weakest of standardsand not in every model. Ratherthan challenging the wellestablisheddemocraticpeace findings, the lacklusterperformanceof the democracy variablemay be a reflectionof too little varianceon this variable,as few SIS pairsare democratic.The alliance variableresults, to some extent, resemble the contradictory argumentsfoundin the literatureover whetherallies aremoreor less likely to fight one another(e.g., Maoz 2000). Furthermore,we alreadypointedout thatthe effect of proximity may be mutedin the postsecession context, andthis is what we find. The results aresimilarfor the leadershipandSoviet Union controls.Changeof leadershipdoes not appearto provide an additional impetus to challenge the border and-despite the momentousnatureof the Soviet Union's collapse-the relatedSISs do not show much evidence of acting differentlythan SISs emerging from the breakupsof other states.

TABLE 2

734

Cox Regression Estimates of Fatal Postsecession Militarized Conflict

Variable Territorialdispute Strategicterritorialdispute Economic territorialdispute Ethnic territorialdispute Violent secession Rump state Power advantage Allies Dyadic democracy Contiguity New leader Soviet Union dummy 2(df) n

Model 6: MIDs, PRDs Only 2.667***

Model 7: TerritorialMIDs, PRDs Only

Model 8: TerritorialMIDs, All Dyads

(0.737) -0.789 (0.584) 1.342** 2.018*** (0.723) 4.254** (1.806) -0.660 (0.537) 0.101* (0.072) 0.347 (0.341) (0.780) -1.062* -0.308 (0.782) (0.437) -0.169 (1.254) 0.966 78.52***(11) 2,899

-0.775 (0.746) 1.332** (0.588) 2.061*** (0.714) 4.544*** (1.850) -0.643 (0.530) (0.074) 0.103* 0.358 (0.342) (0.778) -1.062* -0.210 (0.881) -0.185 (0.431) 0.964 (1.231) 95.78***(11) 4,389

T Cont

(0.622)

3.846** (1.603) (0.632) -0.604 0.071 (0.068) (0.298) 0.089 -0.968 (0.773) (0.725) 0.277 -0.293 (0.580) 0.363 (1.141) 101.48***(9) 2,899

-0.4 1.4 2.1 3.8 -0.7 0.1 0.3 -1.1

-0.0 1.2 78.6

NOTE:Cell entriesreportcoefficients androbuststandarderrors(in parentheses).Unit of analysis is a directeddyad-year.Eac cessions (SISs). MID = militarized interstatedispute; PRD = politically relevantdyad. *p < .10, one-tailed. **p < .05, one-tailed. ***p < .01, one-tailed.

Tir/ KEEPINGTHEPEACEAFTERSECESSION

735

TABLE3

Summaryof Key Findings ViolentBreakup

Peaceful Breakup

Boundaryis contested on the High hazardof fatal militarized Medium hazardof fatal militarized interstateconflict over the basis of ethnicity interstateconflict over the boundary boundary Boundaryis not contested on Medium hazardof fatal Low hazardof fatal militarizedinterthe basis of ethnicity militarizedinterstateconflict state conflict over the over the boundary boundary

POLICY MAKING AND RESEARCH IMPLICATIONS Secession is of coursebutone way policy makerscan deal with domestic-leveldisagreementsover the degree of self-rulethatsubstategroups should have; alternatives includepolicies such as culturalor territorial(butshortof full sovereignty)autonomy, consociationalor other power-sharingarrangements,and so on. Yet the aim of this projecthas not been to comparethe benefitsof variousoptions-that importanttaskis left to futureresearch-but to determinewhat makes one particularoption, namely secession, likely to succeed or fail. Perhaps the most important insight is that postsecessionistrelationsbetween SISs are dominatedby the factorsassociatedwith secessions and that the standardconflict reductionapproaches(e.g., Mearsheimer's [1993, 1998] insistence on creatingthe balancedpower distributionbetween SISs) may not be all thathelpful.This meansthatpolicy makersneed to focus on how to distinguishbetweencountrieswhose division is likely to lead to a lastingpeace versusthe ones in which secession would lead to a disasterin futureinterstaterelations.To simplify the presentationof the above results,we focus on the two variablesthatemerged as most important:ethnicallybased territorialdisputesand secession process.19Table 3 presentsa gridestimatingthe relativedangersof postsecession conflict based on the two factors.Policy makerscould use this schematicto determinewhethersecessions may be an appropriatepolicy choice for a countrydealing with internalsovereignty problems. If thereis a consensus on how to divide up the country'slandthatwould not lead to the creationof ethnicallybasedterritorialdisputesbetweenSISs andif the division can be conductedpeacefully, there is little reason to stand in the secession's way. More problematicare the cases in which there is a disagreementover the location of the postsecession boundaryand/orin which the centralgovernmentor secessionists use force to impose theirwill. Yet,in at least some of these situations,a proactiveapproach can be helpful.First,the currentconventionalwisdom has been to acceptthe old internal bordersas the new internationalboundaries.This strategyhardly optimizes the 19. Note, however,that the currentproject does not addressthe potentialcausal links between the severityof pre- and postsecession violence. A futurestudy could explore whethersecessions promptedby full-fledged civil wars set the stage for futurefull-scale territorialwarsbetween SISs.

736

JOURNALOF CONFLICTRESOLUTION

chances for future peace, especially if ethnically based territorialdisputes are left behind. Changing the location of the boundary to resolve these disputes-at the expense, if necessary,of creating strategicallyand, to a lesser extent, economically based disputes-can help. Second, if it is not possible to resolve impendingethnically based territorialdisputes, policy makers should keep in mind that peacefully conductedsecessions can negatethe problematicinfluenceof these disputes.Forinstance, Czech Republic and Slovakia separatedpeacefully and laterpeacefully alteredtheir border-showing that such disputes can be resolved constructively given the precedentof peaceful interactions. Disputants therefore stand to gain much from peaceful secessions. Resolving potentialborderissues will be more productiveif the protagonistscan negotiatethe new boundaries. Unilaterally imposed, violent "solutions"are unlikely to induce peace except in those rarescenariosin which a violent secession does not leave behind an ethnicallybasedterritorialdispute(e.g., Slovenia-Yugoslavia).Morecommonly,the combatmay createthe boundarythatgives the winneradditional-but unimportantlands, while the loser may have lost key lands and remainin possession of territories thatit cares little about.A peaceful secession could avertsuch scenariosandthe need to reexaminethe bordersin the future.Also, peaceful,agreed-onsecessions createdisincentivesagainstthe futureuse of force, throughreputationalandloss of side benefits costs. Forbothreasons-and also to avoid the loss of humanlife in combat-peaceful secessions are preferredto their violent counterparts. Giventhatfew stateswillingly give up land,the criticalquestionthenbecomes how a secession can occurwithouthavingthe disagreementsover landcontrolbetweenthe centralgovernmentandsecessionistsescalateto the pointof organizedviolence. From the policy-makingstandpoint,internationaleffortscould be used to ensurea peaceful breakup.This canbe done by offeringmediationefforts,puttingpressureon the adversariesto work theirproblemsout peacefully,and not waitinguntil violence occurs. It often takes a while before secessionist desires evolve into full-fledged violence, but the internationalcommunityhas been so incapableof facingthe prospectandrealityof statebreakupthatit has often failed to react in time. For instance,the initial phase of the disintegrationof Yugoslavia-between the election of independence-mindedleaders in SloveniaandCroatiaandthe declarationsof independencethatled to the onset of war-took well over a year (March1990 to June 1991). In general,the lack of timely reactiongives the centralgovernmenta green light to eitherignore the secessionists' frustrationsor to repressthe groupfurther.Faced with such prospects,the secessionists may feel thatthey have little otherchoice but to press for a violent secession. This chain of events has ultimatelyled to secessionist violence and, in some cases, to the violent breakupof the state. As our results show, the violent secession scenariois an invitationfor futuredisaster.The advice to policy makersis to take a more proactive approachto addressthe grievancesof secessionists who aredemandingthe establishment of theirown state.The internationalcommunityholds the power of recognition, and it can use it as leverage againstboth the centralgovernmentthat is unwilling to compromise(i.e., threatenrecognition)and the secessionists who are makingunreasonabledemands(i.e., withholdrecognition).If pressureis put on the centralgovem-

Tir/ KEEPINGTHEPEACEAFTERSECESSION

737

ment to resolve the grievancespeacefully, the secession demandmay be dropped.If the demandis not dropped,conductingthe secession peacefully is preferable. Fromthe scholarlyperspective,the questionof how to achievepeaceful secessions shouldbe pursuedin futureresearch.Groundworkfor relatedstudieshas alreadybeen laid. Young's(1997) examinationof the pathto peaceful secessions revealsthatat least ten mutuallyinterlockingconditionsneed to be present;amongthem, the centralgovernment'sconsentto let the secessionist region withdrawis the absoluteminimalnecessary conditionthatneeds to be met. Yet Young'sresearchdesign (i.e., investigating only the cases of peaceful secession) preventshim from determiningwhat conditions may be sufficientfor a peaceful secession. That is, the full arrayand complete influence of potentiallyrelevantvariablesneeds to be investigatedby including cases of both peaceful and violent secessions. Additionalinsights are offered by Heraclides's (1997) study of seventy separatistwars. She finds that fighting for secession is not a very effective strategy,as only five of these wars resultedin independence.Although this studydoes not tell us directlyhow secessions can takeplace peacefully,combining its insightswith those fromYoungsuggests thatdecentralizedstatepowercan be helpful in two ways. First,by avoidingdominanceof stateinstitutionsby anothergroup,it reducesthe chancesthatthe prospectivesecessionistgroupwill feel thatit has no other way of improvingits situationthanto push for a violent secession. Second, decentralized power reduces the centralgovernment'scapabilitiesto fight; the governmentis thus more likely to seek resolutionthroughcompromisethanthroughviolence and a no-secession-at-any-costattitude.Furthertheoreticaldevelopmentthatwould include additionalkey variables(e.g., internationalsupportfor the secessionist cause, presence of secessionistcoethnicsin a neighboringstate,majorpowersupportfor the preservationof the state, geographicaldistributionof the pro- and antisecession groups, economic inequalities,historyof strife)andsystematicevaluationsof relatedhypotheses are necessary steps that the researchon secessions needs to pursue. That is, the importanceof buildingon Heraclides'sandYoung'sstudiesis only highlightedby the currentproject: the process of breakupnot only has immediate consequences on whether breakup-relatedviolence will erupt but also sets the stage for long-term interactionsbetween SISs.

738

JOURNALOF CONFLICTRESOLUTION

APPENDIX List of Politically Relevant Dyads Produced by Twentieth-Century Secessions Year 1903 1905 1908 1913 1919 1919 1919 1919 1919 1919 1919 1919 1919 1919 1919 1919 1919 1921 1921 1922 1947 1948 1949 1960 1961 1965 1971 1990 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991

Dyad Members Colombia Sweden Turkey Turkey Austria Austria Estonia Germany Hungary Latvia Poland Poland Soviet Union Soviet Union Soviet Union Soviet Union Soviet Union China Turkey GreatBritain India Jordan China Mali Egypt Malaysia Pakistan South Africa Armenia Armenia Croatia Estonia Georgia Kyrgyzstan Kyrgyzstan Latvia Latvia Lithuania Moldova Russia Russia Russia Russia Russia Russia Russia Russia Russia

Panama Norway Bulgaria Albania Czechoslovakia Hungary Latvia Poland Czechoslovakia Lithuania Latvia Lithuania Estonia Finland Latvia Lithuania Poland Mongolia Yemen Ireland Pakistan Israel Taiwan Senegal Syria Singapore Bangladesh Namibia Azerbaijan Georgia Slovenia Latvia Azerbaijan Kazakhstan Uzbekistan Belarus Lithuania Belarus Ukraine Armenia Azerbaijan Belarus Estonia Georgia Kazakhstan Kyrgyzstan Latvia Lithuania (continued)

Tir/ KEEPINGTHEPEACEAFTERSECESSION

739

APPENDIX (continued) Year 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1992 1992 1993 1993 1993

Dyad Members Russia Russia Russia Russia Russia Tajikistan Tajikistan Turkmenistan Turkmenistan Ukraine Uzbekistan Yugoslavia Yugoslavia Croatia Yugoslavia Czech Republic Ethiopia Yugoslavia

Moldova Tajikistan Turkmenistan Ukraine Uzbekistan Kyrgyzstan Uzbekistan Kazakhstan Uzbekistan Belarus Kazakhstan Croatia Slovenia Bosnia Bosnia Slovakia Eritrea Macedonia

REFERENCES Allcock, J. B. 1992. Borderand territorialdisputes.3rd ed. Detroit,MI: LongmanGroup. Beck, N. 1998. Modeling space and time: The event historyapproach.In Researchstrategiesin social sciences, edited by E. Scarbroughand E. Tanenbaum.Oxford,UK: OxfordUniversityPress. Bennett,D. S., and A. C. Stam. 2000. EUGene: A conceptualmanual.InternationalInteractions26:179204. Box-Steffensmeier,J. M., and B. S. Jones. 2000. Timingand political change: Event history modeling in political science. Ann Arbor:Universityof MichiganPress. Bremer,S. 1992. Dangerousdyads:Conditionsaffectingthe likelihoodof war.Journalof ConflictResolution 36:309-38. Brooks, S. G. 1999. The globalizationof productionandthe changingbenefitsof conquest.Journalof Conflict Resolution43:646-70. Brown, M. E., ed. 1993. Ethnic conflict and internationalsecurity. Princeton,NJ: PrincetonUniversity Press. Bueno de Mesquita,B., andR. M. Siverson. 1995. Warand the survivalof political leaders:A comparative study of regime types and political accountability.AmericanPolitical Science Review 89:841-55. Chiozza,G., andH. E. Goemans.2003. Peace throughinsecurity:Tenureandinternationalconflict. Journal of ConflictResolution47:443-67. Cox, D. R. 1975. Partiallikelihood.Biometrika62:269-76. de Silva, K. M., and R. J. May. 1991. Internationalizationof ethnic conflict.New York:St. Martin's. Diehl, P. F. 1999. Territorialdisputes. In Encyclopediaof violence, peace, and conflict, vol. 1, edited by Lester Kurz.San Diego: Academic Press. Etzioni, A. 1992. The evils of self-determination.ForeignPolicy 89:21-35. Fraser,T.G. 1984.Partitionin Ireland,India,andPalestine:Theoryandpractice.New York:St. Martin's. Ghosn, F, G. Palmer,and S. Bremer.2004. The MID3 data set, 1993-2001: Procedures,coding rules, and description.ConflictManagementand Peace Science 21:133-54. Gochman,C. S., and R. J. Leng. 1983. Realpolitik and the road to war. InternationalStudies Quarterly 27:97-120.

740

JOURNALOF CONFLICTRESOLUTION

Gochman,C. S., and Z. Maoz. 1984. Militarizedinterstatedisputes, 1816-1976: Procedures,patterns,and insights. Journalof ConflictResolution28:585-616. Goertz,G., andP. F. Diehl. 1992. Territorialchanges and internationalconflict.London:RoutledgeKegan Paul. Goldstein,J. S., andJ. R. Freeman.1990. Three-waystreet:Strategicreciprocityin worldpolitics. Chicago: Universityof Chicago Press. Gottman,J. 1973. The significance of territory.Charlottesville:UniversityPress of Virginia. Gurr,T. R. 1993.Minoritiesat risk:A global view ofethnopoliticalconflicts.Washington,DC: U.S. Institute of Peace Press. Hachey,T. E. 1972. Theproblemof partition: Peril to worldpeace. Chicago:RandMcNally. Hensel, P.2000. Territory:Theoryandevidenceon geographyandconflict. In Whatdo we knowaboutwar? edited by J. Vasquez.New York:Rowman& Littlefield. Heraclides,A. 1997. The ending of unendingconflicts: Separatistwars.Millennium26:679-707. Holsti, K. J. 1991. Peace and war: Armedconflicts and internationalorder 1648-1989. New York:Cambridge UniversityPress. Horowitz,D. L. 1985. Ethnicgroups in conflict. Berkeley:Universityof CaliforniaPress. Huth,P. 1996. Standingyour ground:Territorialdisputesand internationalconflict.Ann Arbor:University of MichiganPress. Huth, P., and T. Allee. 2002. Thedemocraticpeace and territorialconflict in the twentiethcentury.Cambridge, UK: CambridgeUniversityPress. Jaggers,K., andT. R. Gurr.1995. Trackingdemocracy'sthirdwave with the polity IIIdata.Journalof Peace Research32:469-81. Kaldor,M. 1996. Balkancarve-up.New Statesman& Society 9:24-5. Kaufmann,C. 1996.Possibleandimpossiblesolutionsto ethniccivil wars.InternationalSecurity20:136-75. 1998. When all else fails: Ethnicpopulationtransfersandpartitionsin the twentiethcentury.Inter---. national Security23:120-56. Kirby,A. M., and M. D. Ward.1987. The spatialanalysis of peace and war. ComparativePolitical Studies 20:293-313. Kocs, S. 1995. Territorialdisputesand interstatewar 1945-1987. Journalof Politics 57:159-75. Kumar,R. 1997. The troubledhistory of partition.ForeignAffairs76:22-34. Levy, J. L. 2000. Loss aversion,framingeffects, and internationalconflict:Perspectivesfromprospecttheory.InHandbookof warstudiesII, editedby M. I. Midlarski.AnnArbor:Universityof MichiganPress. Liberman,P. 1996. Does conquestpay? The exploitationof occupied industrialsocieties. Princeton,NJ: PrincetonUniversityPress. Long, J. S. 1997. Regressionmodelsfor categorical and limiteddependentvariables.ThousandOaks, CA: Sage. Lustick, I. S. 1993. Unsettledstates, disputedlands. Ithaca,NY: Cornell UniversityPress. Mackinder,H. 1919. Democratic ideals and reality.New York:Holt. Maoz, Z. 1989. Joining the club of nations:Political developmentand internationalconflict, 1816-1976. InternationalStudies Quarterly33:199-231. . 2000. Alliances: The streetgangs of world politics. In Whatdo we knowabout war? edited by J. Vasquez.New York:Rowman& Littlefield. Maoz, Z., andB. M. Russett. 1993. Normativeandstructuralcauses of democraticpeace 1946-1986.American Political Science Review87:624-38. McGarry,J., andB. O'Leary,eds. 1993. Thepolitics of ethnicconflictregulation:Case studiesofprotracted ethnic conflicts. New York:Routledge KeganPaul. Mearsheimer,J. J. 1993. ShrinkBosnia to save it. New YorkTimes,March31. ---. 1998. A peace agreementthat'sbound to fail. New YorkTimes,October 19. Mearsheimer,J. J., and S. VanEvera. 1995. The partitionthatdarenot speak its name:When peace means war. TheNew Republic,December 5. 1996. Partitionis the inevitablesolutionfor Bosnia.InternationalHeraldTribune,September25. ---. 1999. Redrawthe map, stop the killing. New YorkTimes,April 19. ----. Morgenthau,H. J. 1948. Politics among nations. New York:Knopf.

Tir/ KEEPINGTHEPEACEAFTERSECESSION

741

Most, B., andH. Starr.1980. Diffusion, reinforcement,geo-politics, andthe spreadof war.AmericanJournal of Political Science 74:932-46. Newman, D. 1999. Real spaces, symbolic spaces: Interrelatednotions of territoryin the Arab-Israeliconflict. In A road map to war: Territorialdimensionsof internationalconflict, edited by Paul F Diehl. Nashville, TN: VanderbiltUniversityPress. Organski,A. F. K., and J. Kugler. 1980. The war ledger.Chicago:Universityof Chicago Press. Posen, B. R. 1993. The securitydilemmaand ethnic conflict. In Ethnicconflict and internationalsecurity, edited by Michael E. Brown. Princeton,NJ: PrincetonUniversityPress. Rajmaira,S., and M. D. Ward.1990. Evolving foreign policy norms:Reciprocityin the superpowertriad. InternationalStudies Quarterly34:457-75. Russett,B., andJ. Oneal.2001. Triangulatingpeace: Democracy,interdependence,and internationalorganizations.New York:Norton. Sack, R. 1986. Territoriality:Its theoryand history.Cambridge,UK: CambridgeUniversityPress. Sambanis,N. 2000. Partitionas a solutionto ethnic war:An empiricalcritiqueof the theoreticalliterature. WorldPolitics 52:437-83. Schaeffer,R. K. 1990. Warpaths:Thepolitics of partition.New York:Hill & Wang. Senese, P.,andJ. Vasquez.2003. A unifiedexplanationof territorialconflict:Testingthe impactof sampling bias 1919-1992. InternationalStudies Quarterly47:275-98. Singer,J. D. 1995. Alliances, 1816-1984. Ann Arbor:Universityof Michigan,Correlatesof WarProject. Small, M., andJ. D. Singer. 1982. Resortto arms:Internationaland civil wars, 1816-1980. 2nd ed. Beverly Hills, CA: Sage. Spykman,N. 1944. Thegeographyof peace. New York:HarcourtBrace. Stinnett,D., J. Tir,P. Shafer,P.Diehl, andC. Gochman.2002. The correlatesof warprojectcontiguitydata. ConflictManagementand Peace Science 19:59-68. Tir,J. 2001. Never-endingconflicts?Transfers,partitions,andunificationsas potentialsolutionsfor territorial disputes.Ph.D. diss., Universityof Illinois, Urbana-Champaign. 2002. Letting secessionists have their way: Can partitionshelp end and preventethnic conflicts? ---. InternationalInteractions28:261-92. ----. 2003. Avertingarmedinternationalconflicts throughstate-to-stateterritorialtransfers.Journalof Politics 65:1235-57. . 2005. Dividingcountriesto promotepeace:Prospectsforlong-termsuccess of partitions.Journalof Peace Research45:545-62. Tir,J., P. Schafer,P. F Diehl, and G. Goertz. 1998. Territorialchanges, 1816-1996: Proceduresand data. ConflictManagementand Peace Science 16:89-97. Touval,S. 1972. Theboundarypolitics ofindependentAfrica.Cambridge,MA: HarvardUniversityPress. Tullberg,J., andB. S. Tullberg.1997. Separationor unity?A model for solving ethnicconflicts. Politics and the Life Sciences 16:237-48. Vanzo,J. 1999. Borderconfigurationand conflict: Geographicalcompactnessas a territorialambitionof states.InA roadmapto war: Territorialdimensionsof internationalconflict,editedby PaulDiehl. Nashville, TN: VanderbiltUniversityPress. Vasquez,J. 1983. The tangibilityof issues andglobalconflict:A test of Rosenau'sissue areatypology.Journal of Peace Research20:179-92. 1993. The war puzzle. Cambridge,UK: CambridgeUniversityPress. ---. ---. 1995. Why do neighborsfight?Proximity,interaction,andterritoriality.Journalof Peace Research 32:277-93. Waltz,K. 1979. Theoryof internationalpolitics. Reading, PA:Addison-Wesley. Werner,S. 1999.The precariousnatureof peace:Resolving the issues, enforcingthe settlement,andrenegotiatingthe terms.AmericanJournal of Political Science 43:912-34. Wood, J. 1981. Separatist:A comparativeanalyticalperspective. CanadianJournal of Political Science 14:107-34. Young,R. A. 1997. How do peaceful secessions happen?In Warsin the midstof peace: The international politics of ethnic conflict, edited by D. Carmentand P. James.Pittsburgh,PA: Universityof Pittsburgh Press.