Association for Public Policy Analysis & Management

Background: National governments have the responsibility to respond to citizens' needs, especially in developing countries where vulnerable population relies on the state for public goods’ provision. If allocated efficiently governmental resources improve economic development. Instead, if governmental actors are driven by re-election incentives, resources can be distorted. Existence of these distortions is documented in several public programs, but evidence is lacking during natural disasters. Natural disasters provide a key opportunity for governments to signal how well they can perform, and for citizens to scrutinize political actions and ensure accountability. Learning which are the incentives of politicians to respond to emergencies, and whether disasters change political support is fundamental in light of politicians' incentives to act in future crisis. The 2014 West Africa Ebola (Ebola virus disease, EVD) outbreak represents a unique setting to study both how resources were allocated and how the government’s responsiveness shaped citizens' voting behavior.

Methods: I use public available data on the response and survey data from 2,265 respondents in the 15 counties of Liberia, collected in 2015-2016, to empirically test how the government responded to the epidemic. Using an instrumental variable approach, I explore whether the government had re-election incentives in providing relief comparing more or less politically competitive places affected by EVD. I use electoral data in 2011 and 2014 and confidential data on all EVD cases from the Ministry of Health, to test, in a difference-in-difference framework, whether the experience with EVD affected voter turnout and vote share for the incumbent party. I also study the effects of EVD on individual political perceptions such as levels of trust in and corruption toward several institutions. Finally, I explore the dynamic effects of the estimates using the timing when each of the 10,000 villages in Liberia was hit by the epidemic.

Results: I find that the government provided more relief effort in places affected by EVD compared to those who never experienced EVD. Relief effort was also higher in places affected for longer time and for months consequently. However, in the first part of the epidemic citizens reported government failures in most of the services. Moreover, among EVD affected places, the government was more likely to provide relief in competitive areas especially at later stages of the epidemic. Overall, individuals affected by EVD were more likely to turn out to vote. However, while citizens hit at early stages of the epidemic (March-September 2014) were less likely to vote for the incumbent party, those who experienced EVD after the main inflow of foreign aid (October-December 2014) were much more likely to vote for it.

Discussion: This is one of the first studies that investigate a “slow-onset” disaster, and explore the possibility that the government behave strategically in providing relief during the time of a disaster. The slow evolution of EVD and the postponement of Senatorial elections allowed the government to distort resources to politically active places. The research also shows how an economic shock could both generate and lose votes for an incumbent party.