Perilous Flood Risk Assessments

Flooding is among the costliest natural disasters in the United States. Although the federal government provides floodplain boundary maps and subsidizes insurance, demand remains extremely low. This paper assembles the most comprehensive set of files ever compiled on flood risk and insurance to investigate the impacts of information provision on residential demand for flood insurance. I find that map updates over the past fifteen years removed more than one million properties from the high-risk floodplains on aggregate. This nation-wide decline in nominal flood risk is inconsistent with the best available flood science. Leveraging the staggered timing of flood map updates, I estimate that households respond to changes in official flood risk information, even when this information appears erroneous. Map updates caused a net reduction in demand for flood insurance, in stark contrast with the stated policy goal of increasing insurance coverage. The largest declines were concentrated in neighborhoods with a higher share of African Americans, a pattern largely driven by rezoning these households outside of the high-risk floodplains. On the other hand, digitizing previously-available flood maps does not impact insurance take-up, suggesting that the private benefits of improving access to flood risk information are small. Employing independent flood risk estimates in a structural model of insurance demand reveals that official map updates decreased welfare by at least 300 million dollars annually. Correcting flood maps would yield annual welfare gains exceeding 20 billion dollars, primarily in wealthy neighborhoods