Ethnic, Linguistic, and Religious Heterogeneity and Preferences for Public Goods and Redistribution in Latin America

Abstract

One of the most prominent political-economic arguments is that heterogeneity undermines support for public goods and redistribution. Past research, however, has been mostly cross-sectional, used weak measures of heterogeneity, under-studied Latin America, and did not examine the multiple bases of heterogeneity. We assess negative (fractionalization and between-group inequality [BGI]) and positive (compensation) hypotheses with time-varying measures of ethnic, linguistic, and religious heterogeneity. We analyze four different preferences using up to six survey waves with over 200,000 Americas Barometer respondents across 24 Latin American and Caribbean countries. We estimate both fixed-effects models focusing on within-country variation and multilevel models focusing on between-country variation. Regardless of estimation technique, the prevailing pattern is statistical insignificance for both heterogeneity and BGI coefficients. The results largely contradict the fractionalization hypothesis, as only four of the 108 relevant heterogeneity coefficients are significantly negative. There is slightly more support for the BGI hypothesis, and especially ethnic BGI. Still, most BGI coefficients are insignificant and linguistic BGI is significantly positive in most models. The compensation hypothesis receives more support, as almost half of the heterogeneity coefficients are significantly positive. We conclude by cautioning against universal claims that heterogeneity undermines support for public goods and redistribution.