An uneven landscape of public services for people of color: identifying endogeneity in the relationship between local race composition and public expenditure

Previous economic literature finds that race diversity is associated with increased tax collection and expenditure per capita, but relative under-provision of important public goods and services. Using a panel data set covering Census “Places” of over 25k population 1980–2010, I verify this conclusion in a cross section for the 2010 US Census data. I then show that the measure of racial heterogeneity commonly used throughout the literature is endogenous over time due to omitted variables, and potential reverse causality or simultaneity. As a result, cross section and pooled results are likely to be biased. Using predicted race composition from national trends as an instrumental variable, I show that the relationship between changes in race heterogeneity and both tax and spending per capita is negative. This result suggests a potential reversal of the findings following of the influential paper by Alesina et al. (Q J Econ 114:1243–1284, 1999).