AbstractAs natural disasters grow in frequency and intensity with climate change, limiting the populations and properties in harm’s way will be key to adaptation. This study evaluates one approach to discouraging development in risky areas—eliminating public incentives for development, such as infrastructure investments, disaster assistance and federal flood insurance. Using machine learning and matching techniques, we examine the Coastal Barrier Resources System (CBRS), a set of lands where these federal incentives have been removed. We find that the policy leads to lower development densities inside designated areas, increases development in neighbouring areas, reduces flood damages and alters local demographics. Our results suggest that the CBRS generates substantial savings for the federal government by reducing flood claims in the National Flood Insurance Program, while increasing the property tax base in coastal counties.