Segregation and the Onset of Civil War

Abstract

The ethnic characteristics of a country have long been considered a factor in the onset of civil war. This has been extensively studied using indices of ethnic fractionalization (how many different groups there are) and polarization (the number and relative size of groups) but with inconclusive results. An aspect that has been previously ignored has been the physical distribution of groups within a country. This article takes an index of segregation and tests it against civil war onset in three well-established datasets, two with a high threshold and one with a low threshold of deaths. It finds that an index of segregation is significant in predicting the onset of civil war. While segregation must depend on fractionalization and is probably conditioned by polarization, this analysis finds that fractionalization and polarization are not significant in predicting civil conflict in the high threshold datasets once segregation is controlled for. The paper argues that segregation has effects due to the physical separation from other ethnic groups, the physical within-group proximity and homogeneity within a territory, with each of these three factors having effects on both the motivation and the feasibility of rebellion. The significance of segregation shows that the impact of ethnic characteristics on conflict is complex and the concept of segregation adds a useful dimension to the specification of ethnicity.