Spatial Distributive Politics Using Geospatial and Machine Learning Techniques

Session Submission Type: Full Paper Panel

Session Description

Innovations in computational techniques and the availability of geospatial indicators and text-as-data are opening new opportunities for research, in particular, for the study of social and economic phenomena in data-scarce contexts. This panel uses recent applications of geospatial data and techniques and machine learning methods within the field of political economy to study distributive politics in different regions across the Global South. The panel brings together an international group of scholars who focus on inequality and development.

To explain inequalities in school attendance rates in the state-building era of Turkey, Bozcaga finds that the percentage of landlessness significantly and negatively impacts compulsory primary school enrollment. The stud relies on a spatial first-difference design and an original dataset that combines geocoded archival data on landless farmers, local landlords, and rural primary school attendance rates.

Kumar investigates when citizens’ formal complaints increase bureaucrats’ responsiveness. She finds that bureaucrats are less likely to respond to complaints that require reallocation of resources. She supports her argument using data on the universe of digital complaints in Mumbai's water sector from 2016-2018, supervised learning for text classification, a differences-in-differences design, and qualitative interviews with bureaucrats.

Rettl studies how citizens respond to economic decline caused by globalization when they rely on non-state institutions to obtain insurance and compensation. To do that, she combines a shift-share instrument that exploits an arguably exogenous decline in exports from Brazil with electoral returns and geo-coded survey data. She finds that when citizens rely on Evangelical churches rather than the state to obtain insurance and compensation, bishops acting as brokers have higher leverage in persuading their congregants to vote for their preferred candidate.

Xu examines the conditions under which pro-poor politicians support the extension of infrastructural public goods, despite their gentrifying effects. Using a geocoded database of different types of infrastructural public goods across cities in Brazil, she argues that such public services vary in their gentrifying effect. Local politicians account for these gentrifying effects when strategically deciding where to place different types of public goods in the city.

Sub Unit

• Division 6: Political Economy

Individual Presentations

• Teachers against Landlords: Landlessness and Compliance with Compulsory Education - Tugba Bozcaga, King's College London; Asli Cansunar, University of Washington

• Turning Away from the State: Trade Shocks and Informal Insurance in Brazil - Paula Rettl, Harvard Business School

• The Politics of Gentrification and Slum Evictions - Alice Z Xu, University of Pennsylvania

Discussant

• Giuliana Pardelli New York University Abu Dhabi