Segregation and the Spatial Externalities of Inequality: A Theory of Collateral Cooperation for Public Goods in Cities

Why are some cities much more well-endowed in public goods than others? I argue that the degree of socioeconomic segregation largely explains the formation of preferences for public goods. Socioeconomically integrated (non-segregated) cities have a higher incidence of the negative spatial externalities of inequality (e.g., crime, contamination) that spill over from impoverished localities. Such neighborhood effects, in turn, increase the neighboring middle-class' preferences for certain public goods that address them, while decreasing the perceived relative efficacy of private solutions (e.g., private guards). Thus, socioeconomic integration - through the spatial externalities mechanism - enables preference convergence on the public provision of services in place of private options. In contrast, segregation polarizes the urban electorate along class lines. To test the argument, I propose a novel instrument for segregation that interacts rural-to-urban predicted migration (shift-share instrument) of the poor with the destination locality's "urban form." I combine this quasi-experimental strategy with census area-level measures of class- and race-based segregation and an original face-to-face survey with over 4,000 households across 408 of the 456 census areas in the megacity of Sao Paulo, Brazil. Using embedded mechanism experiments, I show that the spatial externalities mechanism is empirically distinct from alternative mechanisms - racial prejudice, social affinity, contact hypothesis - proposed in the literature.