Segregation and the Spatial Externalities of Inequality: A Theory of Interdependence and Public Goods in Cities

Abstract

Many cities across the developing world have running sewers, public patrolling, and streetlights evenly distributed throughout the city. Others consistently neglect parts of the city when it comes to the coverage of these services. What explains this difference? Conventional wisdom claims that racial diversity undermines public goods provision. I show that class differences, instead, can produce incentives for cooperation for public goods. Class-based segregation reduces the spatial externalities of inequality (e.g., sewage pollution, waterborne diseases, organized crime) that spill over from impoverished areas to the middle class. Conversely, I argue that in integrated (de-segregated) cities, the scale of such externalities undermines the efficacy of private services (e.g., private security), thereby inducing middle-class preferences for externalities-correcting public goods. I illustrate how the argument applies to the location of streetlights, the extension of sewer lines, the stationing of policing units, and investments in public schooling. Thus, while segregation encourages exit into private service provision, integration --through the spatial externalities mechanism-- aligns the urban middle class with the poor in supporting public goods in place of private alternatives. Patterns of integration and segregation shape not only voter preferences, but also when and how voters mobilize for them. I argue that integrated (de-segregated) cities have higher levels of middle- and upper-class civic engagement. In integrated Brazilian cities, the public health risks of uncollected sewerage and the incidence of violent crime in neighboring favelas (slums) induce collective mobilization among the middle class. In other words, preferences for the public (as opposed to private) provision of services, in turn, wills more civic participation. Conversely, the sufficiency of private alternatives that individual households can self-supply in segregated localities undermines these selective incentives for public demand-making. Drawing on qualitative focus groups and survey data, I document the ways in which urban geographies integrated along class lines induce civic participation in both non-political venues (e.g., neighborhood associations, conselhos (participatory councils), orçamento participativo (participatory budgeting)) and political channels of turnout at the polls. The contribution is to show how patterns of urban geographies (i.e., segregation), by shaping opportunities for "exit" into private service provision, has major distributive and political implications for city politics. I test the theory using a mixed-methods approach and original data gathered over the course of 17 months of field research in Brazil. Chapter 2 first lays out the theoretical argument, using ethnographic observations and qualitative evidence from focus groups with middle-class and favela (slum) neighborhood associations. In Chapter 3, I test the argument more systematically. To address the sorting or selection effect that plagues studies of political geography, I develop a shift-share instrumental variable of predicted migration of the rural poor to cities in Brazil, using a series of rural "push factors,'' such as the share of cash crops, the adoption of genetically modified soy, incidence of drought, and the colonial legacies of land tenure patterns. I combine this quasi-experimental strategy with measures of class-based segregation calculated using the decadal censuses. I estimate the effects of segregation on voters' perceptions and preferences, measured using an original face-to-face survey experiment conducted with 4,208 households across 420 of the total 456 neighborhoods in the megacity of São Paulo, Brazil. The analysis introduces self-interest in reducing intergroup externalities as a new mechanism for cooperation for public goods even in diverse societies. Using embedded mechanism vignettes, I distinguish the mechanism from the conventional affective attitudes --e.g., racial tolerance/prejudice, social affinity-- of intergroup contact. In Chapter 7, I consider redistributive social spending --reducing inequality directly (i.e., cash transfers)-- as an alternative to public goods as a solution to the cross-class externalities problem. I re-conceptualize segregation as "spatial distance,'' a foil to "social distance'' (i.e., the distance between classes on the income distribution) featured in the redistributive politics literature. Theories of ``social distance'' posits that social proximity should generate empathy or even altruism towards the poor. On the contrary, I find that "spatial distance,'' actual physical proximity to the poor, generates aversion. Class-based integration (de-segregation, or more local inequality) causally decreases support for redistribution among the rich across cities in Brazil. I show that increasing proximity to poverty actually undermines the rich's view that government efforts to reduce inequality are effective. In Chapter 4, I illustrate how patterns of segregation (or integration) not only shape preferences for public goods but also how voters engage in civic participation for them. Drawing again on qualitative evidence, I show that integration induces civic participation in both non-political venues (e.g., neighborhood associations, conselhos (participatory councils), orçamento participativo (participatory budgeting)) and political channels of turnout at the polls. I argue that preferences for policing and security services map onto support for far-right candidates at the state level in gubernatorial elections, while preferences for education and sanitation services drive the vote for more leftist pro-poor mayors at the municipal level. Varying levels of decentralization across service types segments vote choice across different levels of government, ensuring that voters do not vote for "mano dura'' ("tough on crime'') policies and politicians from the political Right in mayoral elections. Instead, by increasing the value of also other public "externality goods'' (i.e., public schooling and sewer lines), integration induces the middle class to swing left and vote for pro-poor politicians who campaign on the public provision of services. I show integration causally reduces polarization in mayoral elections. Segregated cities, instead, are more politically polarized along class lines. Then, in Chapter 5, I use an analysis of five city cases to further illustrate the theory that the spatial externalities of inequality can forge political coalitions that cut across class lines. In Chapter 8, I consider how politicians faced with different sociopolitical geographies aggregate voters' preferences. The chapter illustrates the conditions under which the demand-side spatial externalities effect is electorally salient. Finally, in Chapter 6, I consider the ways in which racial segregation can have additive or interactive effects with class in shaping intergroup relations and voters' preferences. The chapter uses a set of conjoint survey experiments also administered face-to-face across 420 neighborhoods to systematically adjudicate between competing alternative mechanisms that can explain the main results.