This paper ia an attempt to clarify the relationship between fractionalization,
polarization and conflict. The literature on the measurement of ethnic diversity
has taken as given that the proper measure for heterogeneity can be calculated
by using the fractionalization index. This index is widely used in industrial
economics and, for empirical purposes, the ethnolinguistic fragmentation is
ready available for regression exercises. Nevertheless the adequacy of a
synthetic index of hetergeneity depends ...
This paper ia an attempt to clarify the relationship between fractionalization,
polarization and conflict. The literature on the measurement of ethnic diversity
has taken as given that the proper measure for heterogeneity can be calculated
by using the fractionalization index. This index is widely used in industrial
economics and, for empirical purposes, the ethnolinguistic fragmentation is
ready available for regression exercises. Nevertheless the adequacy of a
synthetic index of hetergeneity depends on the intrinsic characteristics
of the heterogeneous dimension to be measured. In the case of ethnic
diversity there is a very strong conflictive dimension. For this reason
we argue that the measure of heterogeneity should be one of the class of
polarization measures. In fact the intuition of the relationship between
conflict and fractionalization do not hold for more than two groups. In
contrast with the usual problem of polarization indices, which are of
difficult empirical implementation without making some arbitrary choice
of parameters, we show that the RQ index, proposed by Reynal-Querol (2002),
is the only discrete polarization measure that satisfies the basic properties
of polarization. Additionally we present a derivation of the RQ index from
a simple rent seeking model. In the empirical section we show that while
ethnic polarization has a positive effect on civil wars and, indirectly on
growth, this effect is not present when we use ethnic fractionalization.
+
