Citation

Czajkowski J, Kennedy E. Popul. Environ. 2010; 31(1-3): 121-149.

Copyright

(Copyright © 2010, Holtzbrinck Springer Nature Publishing Group)

DOI

10.1007/s11111-009-0097-x

PMID

unavailable

Abstract

Given the significant population increases in United States hurricane-prone coastal areas since 1970, we empirically model U.S. coastal hurricane fatalities from 1970 to 2007. The findings suggest that an apparent decline in hurricane fatality risk since 1970 may be overstated, especially considering the impact of evacuation levels achieved, and that various factors believed to affect the rate of hurricane fatalities have divergent outcomes. The research also illustrates the necessity of understanding the impact of a seminal storm such as Hurricane Katrina. In the analyses, we account for the coastal versus inland county location in the raw fatality data, as well as for the role of evacuation using zero-inflated specifications to handle the excessive number of zero fatality county observations. Initial results, with and without the fatalities from Hurricane Katrina, show that the expected count of fatalities increases by a factor of 1.7-3.4 for a one category level increase in storm strength, and begin to provide an understanding of the bias in the fatality data by varying the level of county evacuation achieved. Finally, we find a higher expected count of fatalities for more frequently struck counties, especially those in the Gulf of Mexico; a lower expected count of fatalities for counties with higher percentages of the population over 65, less than 18, and in poverty. We have also found that forecasting technology improvements have had a relatively minor effect on the expected count of fatalities.

Language: en