The Footprints of Urban Inequality in Africa

• In Event: Remote Sensing and the Political Economy of Development

Abstract

The developing world is rapidly urbanizing, especially Sub-Saharan Africa. African cities often feature stark inequality and neighborhood-level heterogeneity in wealth, with the urban poor and an increasingly vibrant urban middle class living in close proximity. This intra-urban inequality has implications for multiple areas of research in Political Science, including explorations of how contact with other socio-economic classes affects collective action, violence, and support for redistributive spending and examinations of the economies of scale and audience costs of both clientelism and programmatic politics. But it is often difficult for scholars to investigate these topics in African cities because only a tiny handful of African countries release census or other demographic data sufficiently disaggregated to systematically measure localized intra-urban variation in wealth.

Our paper attempts to overcome this data constraint by leveraging remote sensing data on built environments to produce estimates of wealth at resolutions as fine as individual city blocks. We deploy Google’s Open Buildings dataset to produce a wealth index based on the features of all of a city’s built structures. We validate our approach using localized census data for three African countries with disparate urban morphologies for which such “ground truth” data is available -- Ghana, Uganda, and South Africa -- and then produce neighborhood-level wealth estimates for hundreds of other cities across the continent where similar census data is absent.

We show that our new measure outperforms wealth estimates from existing poverty maps produced with other remote sensing data, which are typically calculated at much higher spatial resolutions. Our approach is also more scalable and replicable than the more bespoke, case-specific efforts in existing literature that also use satellite imagery to remotely identify slum neighborhoods. With our new estimates in hand, we document descriptive variation across Sub-Saharan African cities in micro-level spatial segregation by wealth and explore correlations between this segregation and data on downstream political outcomes. Ultimately, we hope our dataset becomes a resource enabling a wide range of future studies on urban politics and inequality in African cities.

Authors

• Noah Nathan, Massachusetts Institute of Technology

• Shuning Ge, University of Pennsylvania