Dual Equilibria in Natural Disaster Response:
The Effect of Natural Disasters on Inequality and Conflict in Kenya

[With Alice Xu]. The vast literature on ethnic favoritism has explored a variety of different patterns in the spatial distribution of public goods and services in African countries (Miguel and Gugerty 2005; Ejdemyr et al. 2016; Nathan Forthcoming). In this project, we use fine-grained spatial weather shocks and climate vulnerability data to examine the ways in which the spatial location of natural disasters determine post-disaster patterns in the provision of public goods and services in Kenya and Ethiopia. Using quasi-random variation in flood shocks, we employ a geographic regression discontinuity design paired with a differences-in-differences method.

We argue that spatial patterns in natural disaster climate shocks prompt social welfare policy responses that could either widen or reduce the inequality gap, producing two equilibria. More specifically, when disasters occur in spatial clusters of voters who support the incumbent, it is used by the ethnically-aligned incumbent as a justification for improving infrastructural public goods and services, widening the existing gap in goods- and services-based welfare provision. Alternatively, when disaster shocks occur in spatial enclaves of the opposition voter base, it provides an accountability shock that brings the region’s underprovision of public goods to the forefront of local politics. Under such a context, the incumbent is obliged to provide disaster relief and to increase provision of public goods to such opposition enclaves (and thus reducing inequality), or else risk violence in the form of land conflict, protests, and property destruction. We illustrate this in the case of Kenya.

 
20-year mean precipitation levels in April, mapped at the constituency level

20-year mean precipitation levels in April, mapped at the constituency level

20-year standard deviation of precipitation levels in April, mapped at the constituency level

20-year standard deviation of precipitation levels in April, mapped at the constituency level