The local economic impacts of natural disasters: A view from outer space

Abstract

In this dissertation, I study the local economic impacts of natural disasters. I use night light emissions captured by satellites as a proxy for local economic activity to bring the analysis of economic impacts of natural extremes to a finer spatial scale. In doing so, this thesis bridges a gap between the international economic literature on natural disasters and their economic impacts, and the more case study oriented or geographically confined studies on natural extremes in the fields of natural hazards and remote sensing. Using a variety of approaches and methods, I discuss how light emissions can help us understand how natural extremes affect the economies they impact. I illustrate the method with a case study of Hurricane Katrina, for which I show how the effects of the destruction of this event can be observed from space. Combined with a growing literature that makes use of night light data to proxy economic development, these findings suggest that night light data can provide a way forward for studying local economic dynamics in response to natural hazards. Moreover, night light data serve as an independently and objectively measured proxy for economic statistics that are not available at the local level in most parts of the world, hence opening up research possibilities in areas for which no data is available otherwise. In the second main chapter of this thesis, I expand this approach and turn to a global analysis in which I show worldwide adverse local impacts of natural hazards and signs of relocation of economic activity across space. I do so for a global panel of 0.5° grid cells, spanning roughly 55 by 55 kilometers at the equator. I show that these dynamics take place within countries, further strengthening the case for studying disaster impacts at this spatial scale, and opening avenues for research into spatial spillover effects between regions within countries. In the second half of this thesis, I discuss a -- at the time of writing -- relatively new suite of satellite data, with which I produce results that place findings in the literature in a different perspective. In a global panel of earthquakes between 2012 and 2016, I find no consistent relationship between changes in night light intensity and the occurrence of heavy earthquakes. This finding holds at the spatial level of the analysis of weather extremes in Chapter 2, and at the resolution of night light data – roughly 500 by 500 meters at the equator. To dive deeper, in my final chapter I conduct a detailed comparative case study on a major earthquake in Japan. Findings of this chapter indicate that night light intensity does not consistently reflect adverse impacts, even in municipalities for which adverse economic impacts and widespread destruction were recorded. Concluding, light emissions do not always reflect adverse impacts from earthquakes. Researchers should therefore be cautious when contributing to this rapidly growing literature. The work in this thesis contributes to this ongoing debate and suggests ways forward.