Public goods, and nested subnational units: diversity, segregation, or hierarchy?

We develop a general multi-scale diversity framework to account for spatial segregation of hierarchically-ordered ethnic groups residing in politically and administratively nested geographic aggregations. We explicate how ethnic diversity, ethnic segregation, and ethnic hierarchy interact with the "public goods catchment area" to cast doubt on extant hypotheses linking diversity and public goods provisioning. We not only show how the celebrated "diversity debit" relationship is incomplete at best but also call into question the more recent literature that posits a positive association between ethnic segregation and public goods. We test our framework using a large national census dataset containing ethnicity information (aggregate caste categories) for 830 million rural residents in India. Our nested-geography models use data from villages (n 600; 000) and sub-districts containing these villages (n 6; 000) for twenty-five different public goods. We show how not accounting for the spatial structure of diversity, segregation, and hierarchy result in biased empirical models of diversity and public goods. To the best of our knowledge, the empirical evidence in this paper comes from the largest dataset used in the politics of public goods literature.