Tropical cyclones can result in thousands of deaths when the exposed population is unprepared or ill-equipped to cope with the hazard. Evaluating the importance of institutions and socioeconomic conditions for these deaths is challenging due to the extreme variability in hazard exposure. Studies of socioeconomic risk factors that do not account for exposure will be imprecise and possibly biased, as a storm’s path and intensity are important determinants of mortality and may be correlated with socioeconomic conditions. I therefore model and then control for hazard exposure by spatially interacting meteorological and socioeconomic data, allowing me to develop novel evidence of socioeconomic risk factors.

In essay 1, I construct a global dataset of over one thousand tropical cyclone events occurring between 1979 and 2016. Controlling for population exposure to strong winds and rainfall, I find that higher levels of national government effectiveness are associated with lower tropical cyclone mortality. Further, deaths are higher when exposure is concentrated over a subset of the population that is already less well off. In essay 2, I investigate whether local government capacity and poverty alleviation can reduce tropical cyclone deaths, using panel data from 78 provinces and 1,426 municipalities in the Philippines. Tropical cyclone exposure is concentrated in wealthier regions of the Philippines, but once wind exposure and rainfall are controlled for I find robust evidence of a link between local poverty rates and cyclone deaths. In essay 3, I investigate the potential for leveraging policy experiments for causal inference about the effects of development interventions on disaster mortality using an existing randomized control trial in the Philippines. This empirical example illustrates how randomization overcomes issues of multicollinearity and omitted variable bias; however, the presence of outliers in exposure and vulnerability to natural hazards interact to make average treatment effect estimates highly imprecise.

Strong evidence of an association between government effectiveness and cyclone deaths suggests that capacity constraints need to be addressed in tandem with risk-specific strategies and financial transfers. Further, evidence that local poverty rates and socioeconomic conditions matter highlights the need for equitable and inclusive approaches to mitigating the risk from tropical cyclones.