Ethnic Politics and Armed Conflict: AConfigurational Analysis of a New GlobalData SetAndreas WimmerLars-Erik CedermanUCLAETH ZurichBrian MinUCLAQuantitative scholarship on civil wars has long debated whether ethnic diversity breedsarmed conflict. We go beyond this debate and show that highly diverse societies are notmore conflict prone. Rather, states characterized by certain ethnopolitical configurationsof power are more likely to experience violent conflict. First, armed rebellions are morelikely to challenge states that exclude large portions of the population on the basis ofethnic background. Second, when a large number of competing elites share power in asegmented state, the risk of violent infighting increases. Third, incohesive states with ashort history of direct rule are more likely to experience secessionist conflicts. We testthese hypotheses for all independent states since 1945 using the new Ethnic PowerRelations (EPR) data set. Cross-national analysis demonstrates that ethnic politics is aspowerful and robust in predicting civil wars as is a country’s level of economicto rebellion,:development. Using multinomial logitDeliveredregression,byweIngentashow thatinfighting, andUCLA Librarysecession result from high degrees ofTue,exclusion,segmentation,andincohesion,07 Apr 2009 18:50:37respectively. More diverse states, on the other hand, are not more likely to suffer fromviolent conflict.Karl Marx predicted that revolutionary classstruggles would transform the world during the twentieth century. Instead, it turned outto be the age of ethnonationalist conflicts. Warsfought in the name of national liberation or ethnic autonomy comprise only one fifth of thewars between the Congress of Vienna (1814)and the Treaty of Versailles (1919). FromVersailles to 2001, however, the share of ethnonationalist wars rose to 45 percent, and sincethe Cold War ended it has reached 75 percent.1Ethnic demands and grievances play a promi-Direct correspondence to Andreas Wimmer(email@example.com). The authors wish to thankthe many individuals who helped assemble the dataset on which this article relies. While we cannot listthe dozens of country and regional experts whogenerously shared their knowledge, we should liketo at least mention Dennis Avilés, Yuval Feinstein,Dmitry Gorenburg, Wesley Hiers, Lutz Krebs,Patrick Kuhn, Anoop Sarbahi, James Scarritt,Manuel Vogt, Judith Vorrath, Jürg Weder, andChristoph Zürcher. Luc Girardin implemented thesoftware for the online expert survey. The data project relied on f inancial support from UCLA’sInternational Institute and the Swiss NationalScience Foundation through the project“Democratizing Divided Societies in BadNeighborhoods.” For encouraging comments andcriticisms, we are grateful to Michael Ross as wellas audiences at the department of sociology of theUniversity of Arizona, the Conference onDisaggregating the Study of Civil War andTransnational Violence held at the University ofEssex, the Program of Order, Conflict, and Violenceat Yale, and the Mannheim Center for EuropeanSocial Research.1 These figures are based on the data set assembledby Wimmer and Min (2006) and concern wars withmore than 1,000 battle deaths.AMERICAN SOCIOLOGICAL REVIEW, 2009, VOL. 74 (April:316–337)
ETHNIC POLITICS AND ARMED CONFLICT—–317nent role in most conflicts reported in the dailythan high degrees of diversity, it is ethnic exclunews—from Iraq to Darfur, Kenya to Tibet,sion from state power and competition over theIsrael and Palestine to Burma. What can thespoils of government that breed ethnic conflict.social sciences offer to an understanding ofWe propose a configurational model thatthese conflicts? When do lines of conflict folidentifies three constellations in which thislow ethnic divides and what are the causal mechstruggle over the state is most likely to escalateanisms linking ethnicity to conflict?into armed conflict. First, armed rebellions areThere is no satisfactory answer to these quesmore likely when the state excludes large sections in the burgeoning quantitative literature ontions of the population from central state powercivil wars that has emerged over the past decade.on the basis of their ethnic background. Second,The most influential school of thought disthe likelihood of infighting increases when amisses ethnicity as an explanatory factor altolarge number of ethnic elites shares governgether, arguing that ethnic grievances are tooment power and engages in competitive rivalwidespread to explain the rare event of civilry. Third, both rebellion and infighting will bewar. In this view, rebels fight wherever govmore likely and take on secessionist forms whenernments are militarily weak or lootablesegments of the population have a short andresources can feed an insurgent organizationtroubled history of direct rule by the center. We(the greed-and-opportunity perspective). Otherexamine these hypotheses with quantitativescholars maintain that ethnicity does matter,analysis of all states since World War II usingand that more ethnically diverse states are morea new data set on Ethnic Power Relations (EPR).likely to see conflict (the diversity-breeds-conThis data set records all politically relevant ethflict tradition). Yet a third group examines thenic groups, minorities and majorities, and theirconditions under which discriminated ethnicdegree of access to executive-level stateminorities will rebel (the minority-mobilizapower—from total control of the government totion school). We argue that all threeDeliveredtraditionsby Ingentaovert politicaldiscrimination and exclusion.to :tend to misconceive the relationship betweenThe EPR data set overcomes the limitations ofUCLA Libraryethnicity and conflict.existingdata sets, especially the widely usedTue, 07 Apr 200918:50:37To get this relationship right, we first need toMinorities at Risk data set, which focuses exclurecognize that the modern state is not an ethnisively on disadvantaged minorities and is thuscally neutral actor or a mere arena for politicalunable to capture the dynamics of ethnic policompetition, but a central object of and partictics at the power center. The EPR data set is alsoipant in ethnopolitical power struggles. Why isan improvement over conventional demographicthis the case? Our answer takes an institutionindices of diversity that are only tangentiallyalist point of departure. Contrary to empires,related to the ethnopolitical struggle over thenation-states are governed in the name of “theirstate.peoples,” which provides incentives to alignEthnic politics, our findings reveal, helps topolitical loyalties along ethnic divides. To gainexplain the dynamics of war and peace, contrarylegitimacy, political elites in control of executo what the greed-and-opportunity school maintive-level state power will favor co-ethnics whentains. Second, our results demonstrate that moredeciding with whom to ally and to whom todiverse states are not more war-prone, in condistribute public goods. Politics will then centrast to the expectations of the diversity-breedster on the question of which ethnic group conconflict school. Third, disaggregated analysistrols which share of executive government, andusing multinomial logit regressions shows thatthe struggle over state power will pit ethnicaldifferent kinds of ethnic conflicts result from difly defined actors against each other. In thisferent causal processes: rebellions are moreview, ethnic politics is not exclusively a struglikely the higher the share of the excluded popgle to rectify the grievances of minority groups,ulation; the chance of infighting increases as theas the minority-mobilization school assumes,number of power sharing elites augments; andbut it is more generally and fundamentally aboutsecessions are more frequent in incohesive statesthe distribution of state power along ethnic lines.that lack a long history of direct rule by theThe diversity-breeds-conflict school relies oncenter. We thus follow in the footsteps of otherdemographic indices of heterogeneity that overscholars in the quantitative literature who arguelook how ethnicity relates to the state. Ratherthat different types of wars have different caus-
318—–AMERICAN SOCIOLOGICAL REVIEWes (Buhaug 2006; Sambanis 2001), and we support the recent trend of closely investigatingthe various mechanisms that lead to armed conflicts (Kalyvas 2007).conflict. Some argue that high degrees of ethnic diversity contradict the assumption of cultural homogeneity on which modernnation-states are based, thus triggering waves ofseparatist wars and ethnic cleansings (Gellner1991; Nairn 1993). Vanhanen (1999), the mostETHNICITY AND CONFLICT:ardent proponent of the diversity-breeds-conflictGETTING THE RELATIONSHIPargument, relies on van den Berghe’s sociobiRIGHTological theory of ethnic nepotism, according toTwo major shortcomings characterize the quanwhich humans tend to favor kin and quasi-kin,titative literature on ethnicity and violence. First,such as co-ethnics, over others. As a result,the mechanisms linking ethnicity to conflictmore ethnically heterogeneous states will haveare specified in theoretically problematic andmore conflict. Finally, Sambanis (2001) drawsempirically unsatisfactory ways. Second, quanon organizational economy models to arguetitative approaches tend to overaggregate thethat more ethnically divided societies face highdependent variable and treat ethnic conflicts aser risks of ethnic war because shared ethnicitythough they have uniform causes. We first disdecreases the collective action costs associatedcuss the problem of specifying relevant mechwith organizing a rebel force. Since the likelianisms, focusing on three prominent schools ofhood of ethnic rebellion does not depend onquantitative research on the outbreak of civilgroup size, he expects “the relationship betweenwars: greed and opportunity, ethnic diversityethnic war and ethnic divisions [to be] linear andbreeds conflict, and minority mobilization.positive” (Sambanis 2001:266; see also EasterlyThe most influential articles argue that ethand Levine 1997).nicity plays no role in predicting the onset ofThese two positions—the greed-and-opporcivil wars. According to authors in thistradition,tunityDelivered by Ingenta schoolto : and the diversity-breeds-conflictthe increase in ethnic conflicts during the twentradition—rely on the same type of demographicUCLA Librarytieth century does not capture any Tue,meaningfuldiversityindicators to test their core assumption07 Apr 200918:50:37trend, but is due to the unfortunate tendency ofregarding ethnicity and conflict. Many use aboth scholarly observers and rebels themselveslinguistic fractionalization index, calculated asto attribute conflict to primordial ethnic identithe likelihood that two randomly drawn indities—a collective delusion of sorts (Laitinviduals would speak a different language. This2007:20–27). More important than ethnic idenis a poor indicator for capturing the politicaltity or political exclusion along ethnic lines aredynamics associated with ethnic conflict. First,the material and organizational incentives tonot all ethnic groups matter for politics (Chandrastage a rebellion against government. Accordingand Wilkinson 2008; Posner 2004). Second,to Fearon and Laitin’s (2003) well-known insurethnic conflicts are not the outcome of everygency model, wars erupt when governmentsday encounters between individuals; they are theare weak and rebels have ample opportunitiesresult of interactions between the state and ethto hide from troops while recruiting unemployednopolitical movements that challenge stateyoung men for whatever cause: national liberauthority (Cederman and Girardin 2007).ation, revolutionary progress, the spread of trueGiven these conceptual and measurementreligion, or rich bounty. Similarly, Collier andproblems, it is not surprising that empiricalHoeffler (2004) maintain that civil wars occurstudies produce conflicting results when usingwhere rebellions are most feasible, rather thanfractionalization indices. Some find that ethnicwhere actors are motivated by ethnic inequalifractionalization does not explain high-intensity or social marginalization. More specifically,ty conflicts (defined as more than 1,000 battlethey argue that lootable economic resourcesdeaths per year) (Collier and Hoeffler 2004;make organizing and sustaining a rebel organFearon and Laitin 2003). Others show that ethization easier (see also Collier, Hoeffler, andnic fractionalization is very important if theRohner 2006).dependent variable includes low-intensity warsA second group of scholars insists that eth(Hegre and Sambanis 2006) or if one focuses onnicity does matter. They suggest various reasonsethnic wars (Sambanis 2001) or secessionistwhy ethnically diverse states experience moreconflicts only (Buhaug 2006). Some find a par-
ETHNIC POLITICS AND ARMED CONFLICT—–319abolic relationship between ethnic fractionalas ethnically neutral, making it impossible toization and the prevalence of civil war (Elbadawigrasp the dynamics of ethnic politics in theand Sambanis 2000). Still others maintain thatpower center. Second, the MAR coding schemepolarization between two equally sized ethnicdoes not fit countries with ruling minorities orgroups, rather than fractionalization, bestcomplex coalitions of ethnically defined elitesexplains conflict (Montalvo and Reynal-Querol(e.g., Nigeria, India, and Chad).3 In such coun2tries, ethnic conflict will be pursued in the name2005).of excluded majorities (rather than minorities)We move beyond these demographic indicaor ethnic groups that share power (and are thustors of ethnic diversity in the analyses that folnot at risk). Roughly half the observations in ourlow by introducing a new data set that recordsdata set conform to such ethnopolitical conpolitically relevant groups and their access tostellations and thus escape the logic of the MARexecutive state power. This allows for a directapproach. By reducing its focus to the politicaltest of how ethnic politics affects war and peace,mobilization of discriminated minorities, therather than relying on demographic proxies farminority-mobilization model overspecifies theremoved from how ethnicity works in politicalconditions under which ethnicity leads to conpractice. Once we account for the politicalflict.dynamics of ethnic exclusion and competition,All major schools in the quantitative literadiversity in and of itself has no effect on the liketure fail to specify convincing mechanisms linklihood of civil conflict.ing ethnicity and conflict. They either rely on aThe third major approach is the minorityversion of the ethnic diversity argument that ismobilization school. These scholars analyze theunrelated to the logic of ethnic politics, or theyrelationship between ethnicity and conflict at thedefine ethnic conflicts too narrowly as a matgroup level, rather than the state level. Comingter of minority mobilization. A second problemfrom a political mobilization perspective, Gurrin the existing literature is that it conceives eth(1993a) and others explore the conditions underDelivered by Ingenta to :nic conflict as a unitary phenomenon caused bywhich ethnic minorities protest or rebel. UCLAThey Libraryuniformfactors.4 Qualitative comparative workfind various factors that account forTue,the 07politiApr 200918:50:37shows the importance of taking different ethcal behavior of ethnic groups, including, as willnopolitical constellations into account and ofbe familiar to students of social movements,acknowledging the causal heterogeneity of thethe strength of communal grievances and theprocesses that lead to ethnic conflict. The folpolitical opportunity structure provided by diflowing four vignettes of well-known ethnic conferent political regimes. Gurr and colleaguesflicts illustrate this point.have also assembled a large, worldwide dataIn Ireland, when segments of the educatedset on these “Minorities at Risk” (MAR). TheCatholic middle class, inspired by the U.S. civilMAR data set has produced a quantum leap inrights movement, mobilized against their longthe study of ethnic politics and has provided anstanding exclusion from power, the state appainvaluable service to researchers in politicalscience (Elkins and Sides 2007; Saideman andAyres 2000; Toft 2003; Walter 2006) and sociology (Chai 2005; Olzak 2006).3 The MAR data set tries to address these limitaThe minority-mobilization perspective comestions by including five “advantaged” minorities whomuch closer than the other schools to the empirbenefit from political discrimination and control aically observable mechanisms linking ethnicistate apparatus. MAR also includes a series of “comty to conflict. We thus incorporate some of theirmunal contenders” (i.e., groups that share powerinsights into the model of ethnic politics develwith others while at the same time mobilizing inprotest or rebellion); these are mostly in Africa (Gurroped below. Their perspective, however, is lim1993b). Ethnically defined elites that do not mobiited by its focus on minority groups only. Thislize their constituencies in protest are omitted.has two consequences. First, the state appears2 Ellingsen (2000) finds support for both a linearrelationship to fractionalization and a U-shaped relationship to polarization.4 The MAR data set comes closest to a more disaggregated perspective by coding different types ofethnic groups. Gurr’s (1993b) analysis, however,mostly focuses on the difference between peacefulprotest and violent rebellion, irrespective of thesegroup differences.
320—–AMERICAN SOCIOLOGICAL REVIEWratus, controlled entirely by Protestants elitesgovernment and its dependence on U.S. militarywho ruled Northern Ireland as an outpost ofpower. The Sadr Army harnessed the support ofthe British state, reacted with repression andmarginalized urban youth to oppose power sharintimidation. The ensuing escalation reinvigoing with Sunni and Kurdish political parties,rated the Irish nationalist underground army,advocating instead a strong, central state underwhich fought to unite Northern Ireland withShiite command (Bengio 2004; Cole 2003;the rest of the country. This in turn led to theWimmer 2003).emergence of Protestant militias and terroristThe factors affecting these four conflicts andgroups opposed to the nationalist projectthe mechanisms at play are quite different.(Bardon 2001).While Irish Catholics and indigenousIn Bosnia shortly before independence, theChiapanecos represent excluded groups thatleadership of the Serbian territories withdrewmobilized against the state, representatives offrom the provincial government they had sharedBosnian Serbs and Shiite Arabs were partnerswith Croatian and Bosniak politicians.in coalitional governments. Serbian BosniakMobilization for war proceeded quickly on bothelites and Iraqi ethnoreligious factions faced asides. Serbian militias, supported by the armydisorganized and ethnically fragmented state,of neighboring Yugoslavia, soon attackedwhile Catholics in Northern Ireland and theCroatian and Bosniak villages that they intendZapatistas in Mexico opposed an entrencheded to incorporate into the territory of a futurestate apparatus. The IRA and the Bosnian SerbSerbian state (Burg and Shoup 1999).nationalists developed separatist agendas aimedIn January 1994, the now iconic commanat joining established neighboring states, whiledante Marcos led a group of masked men andthe Zapatistas and Iraqi groups focused onwomen to the main square of San Cristobal dechanging ethnic power relations within existingla Casas and announced that the indigenousstates. It seems doubtful that any single indicapeoples of Chiapas and Mexico wouldDeliveredno longerby Ingentator cantoaccuratelygrasp these different eth:accept their fate as second-class citizens.He Librarynopolitical dynamics. The power configurationsUCLAdemanded profound constitutional,Tue,economic,are18:50:37different, as are the mechanisms and
the distribution of state power along ethnic lines. The diversity-breeds-conflict school relies on demographic indices of heterogeneity that over-look how ethnicity relates to the state. Rather than high degrees of diversity, it is ethnic exclu-sion from state power and competition over the spoils of government that breed ethnic conflict.