Rewarding Performance in Disaster Response: Evidence from Local Governments in Latin America

Databases were used for the "Rewarding Performance in Disaster Response: Evidence from Local Governments in Latin America" research, which seeks to determine the electoral impact of disasters in Peru and Chile. For both countries, we created a similarly structured dataset containing municipal-level information on elections and participation, disasters, budgets and spending, socioeconomic outcomes, and geography. Each observation in the dataset corresponds to a municipality per year. In the case of Peru, we included the municipal elections of 2010 and 2014, and for Chile, we analyzed the local elections of 2008, 2012, and 2016. Additionally, the datasets contain socioeconomic information, such as poverty and education. Our municipal-level treatment variable is defined as having experienced at least one disaster during the mayor's term in office. In both countries, we utilized all available information on disasters, ensuring the inclusion of only large-scale public emergencies. For Peru, we included all disasters that occurred between 2007 and 2014, such as earthquakes, floods, landslides, and frosts. To ensure that only large-scale disasters were included, we identified a public emergency using two criteria: that it was officially declared a disaster area by the central government and that it received additional funds or emergency management. Our binary measure of the occurrence of a disaster combines these two indicators: whether the disaster occurred and whether it was accompanied by an additional budget transfer from the central level to address the crisis and mitigate its effects. Using these criteria, we identified 854 disasters across 532 municipalities. For Chile, our dataset includes 243 disasters, such as earthquakes, floods, and volcanic eruptions, occurring in 220 municipalities between 2008 and 2016. We included all floods and volcanic eruptions that were officially declared emergencies by the National Disaster Prevention and Response Service (SENAPRED). For earthquakes, we used the Modified Mercalli Intensity Scale to identify and include only the most affected municipalities. Since Chile is a country prone to earthquakes, there are many small-scale seismic movements that do not cause significant impacts. To ensure that only municipalities where the earthquake was strong enough to cause significant damage requiring major mayoral intervention were included, we coded as a disaster only those municipalities where the Mercalli scale intensity exceeded 7 out of a maximum of 10. In the case of earthquakes in Peru, we had already identified the most significant events because we included only those that received additional disaster-related funding from the national government. Our performance measures focus on mayors' spending. For both Chile and Peru, we used total per capita municipal spending at the local level. Rather than only tracking funds earmarked specifically for disasters, we prefer to capture total spending to account for the use of various budget lines at mayors' disposal. We used inflation-adjusted values to compare results across years. We standardized the municipal expenditure variable (mean of zero and standard deviation of 1) to compare results across the two countries and included socioeconomic variables as controls.