We hypothesise that, given the typically uneven distribution of ethnic groups within a country, ethnic diversity leads to greater local polarization and more frequent, but smaller, conflicts that involve only some ethnic groups. These conflicts can be overlooked if the number of fatalities is small. Our empirical work exploits data on the proportion of a country affected by a conflict, and we control for country size, poverty, geography and natural resource endowments. We show that, consistent with the hypothesis, at the margin ethnic diversity makes conflict more probable, but also makes it more likely to be localized. This finding is robust to persistence in the incidence and extent of conflict. This potentially explains the apparent lack of correlation between ethnic fractionalization and the incidence of conflict found in previous research that uses a higher threshold number of fatalities to define a conflict.
