Which Diversity & what Inequality? How Methodological Choices affect the Results

• In Event: Poster Session: Political Methodology

Abstract

How does diversity or ethnic inequality relate to public goods provision? Many studies engage with the increasingly debated idea that ethnic heterogeneity by itself—or conditional on pronounced between-group inequalities—leads to inefficient public outcomes. This and related hypotheses, however, are often tested using available region-specific or proxy measures, which may compromise both the internal and external validity of results. We contribute to this debate by exploiting the wide variation in racial demography and public goods provision in Brazilian municipalities adopting a large number of competing indicators. Overall, our rich data shows the striking differences between various ethnoracial measures—based on their construction and content—that are often deemed equivalent in the literature. In contrast to earlier studies, we find no robust relationship between public outcomes and racial heterogeneity or between-group inequality indices. To address endogeneity concerns, we also use the location of Brazilian slave settlements (quilombos) and historical census (1872) estimates as original instruments for ethnoracial demography, which corroborate our results. We then demonstrate that these null findings stem from using misspecified indices which obscure the possible interaction effects between racial proportions and average group standing. We conclude by exploring the practical implications and suggesting the use of disaggregated indicators and their interaction terms.

Authors

• Alexander Kustov, Princeton University

• Giuliana Pardelli, Princeton University