Predicting Catastrophic Events: Flood Losses as Heavy Tailed Distributions

Abstract

Recent weather events such as the 2019 flooding of the Missouri River, Hurricane Florence (2018) in the Carolinas, Hurricane Harvey (2017) in Texas, and Hurricane Sandy (2012) in New Jersey have reignited a discussion on flooding risks. Still fresh in people’s minds, Hurricane Katrina (2005), first refocused flooding risk as a major concern for Americans on a national level. Most climate science suggests that the intensity of flooding events is increasing, and studies show the financial impact is likewise escalating even measured in real terms. FEMA’s response to these events has been insufficient to rebuild the pre-existing capital in affected regions, and inadequate to bring back the displaced population. In addition, an analysis of FEMA’s national map of flood risk shows FEMA is underestimating the national exposure to flood risk. Despite this hole in the public model of coverage, a private sector model has not emerged to provide coverage to consumers. As a nation, there is little evidence to suggest we have taken steps to properly categorize, mitigate, and distribute flood risk effectively to entities that can manage it. This thesis uses heavy-tailed distributions to analyze the potential financial impacts of catastrophic flood losses on a private company in New Jersey administering flood insurance.