Do people respond to low probability risks? Evidence from tornado risk and manufactured homes

Whether people perceive and respond to low-probability natural hazards is a research question of considerable policy relevance. We obtain evidence by considering the response of housing choice to tornado risk for manufactured homes. The vulnerability of manufactured housing, combined with its growing share of the U.S. housing market, has led to proposed mandates for community shelters in mobile home parks. Expected utility theory, however, predicts that households should account for tornado risk in their housing choice. We test for an effect of tornado risk on manufactured housing demand using cross-sectional state data, as well as counties in three tornado prone states. We find that people do respond to tornado risk; our estimates indicate that each expected annual state tornado death per million residents reduces demand for manufactured homes by about 3%. The estimated quantity effect is consistent with the market studies of the price elasticity of manufactured homes.