Working Paper 2013-294
Ethnic heterogeneity can potentially be related to the occurrence of conflicts with longlasting economic effects. Two main measures of ethnic heterogeneity are employed in the econometric literature on ethnic diversity and conflict: the Gini heterogeneity or fractionalization index and the discrete polarization index. However, still no broad consensus is reached on which distributional aspect of ethnic diversity is associated with the outbreak of conflict. In this paper we argue that the relative importance of each
pattern of ethnic diversity depends on the trade-off between the groups’ power and its interaction with other groups. Following the Esteban and Ray [On the measurement of polarization, Econometrica, 62(4), 1994] approach to social antagonism, we axiomatically derive a parametric class of indices of conflict potential that ombines the groups’ effective power and the between-groups interaction. We use a discrete metric to define the distances between groups and we do not treat each group as a unitary actor. Moreover, we assume that the effective power of a group depends not only on its own relative size but also on the relative size of all the other groups in the population. We show that for certain parameter values the obtained indices reduce to the existing indices of ethnic diversity, while in general the derived indices combine in a non-linear way three different aspects of ethnic diversity, namely the fractionalization, the polarization and the ethnic dominance.
The power component of the extreme element of the class of indices is given by the relative Penrose-Banzhaf index of voting power. The results from our empirical exercise show that the derived extreme index outperforms the existing indices of ethnic diversity in the explanation of ethnic conflict onset.
Authors: Matija Kovacic, Claudio Zoli.