Abstract: This paper considers the interaction of two key flood policy instruments commonly used in the US, levee infrastructure and flood insurance, and measures how much flood insurance take-up changes in response to levee provision. Levees are critical infrastructure that reduce expected flood damage in a protected area. When a levee is constructed, and later accredited by the Federal Emergency Management Agency (FEMA), it alters inherent flood risk, flood insurance prices, and mandatory insurance purchase requirements. Using a novel panel dataset drawing from the National Levee Database, manually collected levee accreditation documentation, and FEMA flood insurance data, we leverage variation in levee construction and accreditation timing within a difference-in-differences design. Construction timing allows us to examine insurance take-up as a result of decreased flood risk, while take-up responses to accreditation reflect changes in insurance prices and mandatory purchase requirements. Our paper has three main findings: first, we find that levee construction decreases flood insurance by 20 percent. Second, we find that levee accreditation does not further change flood insurance take-up. Third, we find that decreases in flood insurance take-up due to levee construction decreases aggregate household insurance spending by $1.2 million per levee-mile, accounting for both extensive and intensive margin changes, and $5.7 million in averted expected damages per levee-mile, which, when compared to recent estimates of levee construction costs, corresponds to a break-even time horizon of 10 to 50 years.
Hurricanes cause tremendous economic damage to the United States. Although economists have studied macro-level indicators of economic progress following these devastating disasters, there is very little evidence of establishment-level responses to hurricanes, and heterogeneity in outcomes depending on establishment characteristics. This paper combines detailed spatial data on hurricane trajectories with county-level characteristics on establishment entry and exit to provide the first empirical micro-level estimates of establishment behavior following a natural disaster. I plan to perform a heterogeneity analysis to determine how entry and exit varies depending on establishment characteristics. My initial results indicate that establishments actually increase following a hurricane event, and this increase persists for many years following the disaster. I find no evidence of firm size heterogeneity in post-hurricane establishment counts.