Natural hazards are epistemically random events that can cause massive losses on human lives. However, how natural disasters affect human lives is a matter of human resilience. Do institutions provide mechanisms for people's vulnerability in natural disasters, and are some institutions more impairing than others? The growing literature on politically motivated disaster relief has found that measures to ensure against and prepare for disastrous events to a large extent are used as 'porks' to buy constituent support. However, there is as yet no global large-n study that sufficiently captures how this affects politically excluded groups. This thesis develops a novel approach, and examines these matters by studying all rapid onset natural disasters in the world from 1980-2008. The contributions of this project are twofold. I disaggregate institutions and study how specific institutional subcomponents affect natural disaster vulnerability. Secondly, by using geography information systems (GIS) this thesis is able to make local level inferences about which political groups that are more vulnerable, and under which conditions natural disasters are more deadly. The empirical evidence of this project reveals core features of the political dimension behind disaster vulnerability, which has not been empirically evident until now. Estimates suggest that areas that are populated by politically excluded groups on average experience twice as many casualties as areas that are not populated by excluded groups. This finding yields support to a core rational choice argument: political representation in central government affects the allocation of goods, and incumbents allocate goods favoring those constituents that ensure them in power. I also find evidence that presidential institutions on average experience 50 casualties more than parliamentary institutions. This finding lends support to the literature arguing that allocation of global goods in presidential regimes are characterized by local distribution, short-term policy, and large transaction costs, which all hampers global goods provisions.