Research

Estimating the Individual Effects of Flood Damage: Evidence from Hurricane Harvey

The combination of climate change and economic growth in coastal regions increasingly exposes the United States population to extreme flooding, with the number of inflation-adjusted billion dollar flood events tripling during the 2010s. Despite the aggregate cost of natural disasters and subsequent spending in recovery and mitigation measures, our knowledge of the individual impacts of these events is limited. Residential mobility is of particular importance as disaster events may push households out of their preferred neighborhoods or force them to stay in an undesirable environment. These decisions impact housing turnover and neighborhood composition, both of which affect public finance through changes in the tax base. I construct a measure of property-level damage using widely available appraisal-district data, which I apply to idiosyncratic flooding caused by Hurricane Harvey that damaged thousands of homes outside of the 100- and 500-year floodplains on the western edge of Houston, Texas. I leverage each property’s ground elevation in a regression kink design and find a distinct change in flood damage around the peak water level. In particular, the average flood damage increases by $2,200 for each inch a property lies below the peak water level, while there is no relationship between elevation and damage for homes lying above that threshold. Using the elevation variation, I find that $10,000 of damages induced homeowners to list their home for sale by 1 percentage point (or 20 percent) within four months after the storm. This 1 percentage point gap persisted through at least 2020, and a similar pattern occurred for housing sales. Flood damage (and unexpected wealth shocks, more broadly) pushed homeowners out of their current residence and increased housing turnover for several years after the disaster event.