What is the relationship between ethnoracial demography and public goods provision? A large literature builds on the hypotheses that diversity or between-group inequality lead to poor public goods provision. However, explorations of its causal link remain scarce. To address the potential endogeneity bias, this paper exploits the historical location of Brazilian slave settlements (quilombos) as an original instrument for ethnoracial demography. Furthermore, we distinguish social spending from the quality of public sector and investigate whether the relationship changes across different types of public goods. Contrary to previous arguments, we find no consistent effect of racial fractionalization or horizontal inequality in our detailed microdata on Brazilian municipalities. Instead, communities with a larger share of racial minorities are found to be systematically less efficient in providing public goods. This finding is consistent with ``representational gap'' hypothesis. In conclusion, we discuss the theoretical implications of our findings including alternative mechanisms driving this relationship.
