Association for Public Policy Analysis & Management

Flooding remains one of the costliest natural disasters in the United States with direct losses averaging $8 billion (2014 USD) annually over the past 30 years. A large literature exists to recover marginal willingness to pay (MWTP) for avoiding this peril. However, with these risks spread heterogeneously across space, key policy questions arise regarding how MWTP to avoid flood risk varies. In particular, are there systematic differences in the extent to which individuals of different sociodemographic backgrounds sort over flood risk? If so, are those with lower MWTP to avoid such risks also more likely to belong to populations that are considered vulnerable? These questions are central to public disaster programs and agencies including the National Flood Insurance Program (NFIP) and FEMA. 

In this paper, we estimate household preferences to avoid flood risk in the presence of heterogeneous sorting. Using data on house sales across Florida from 1998 to 2012, we start by estimating MWTP from a hedonic property value model that identifies preferences to avoid such risks using a boundary discontinuity design (BDD) first employed by Black (1999). Utilizing data on the race and income of homebuyers, we examine whether sorting induces correlations between neighborhood flood risk and individual sociodemographics. We then use a discrete choice, residential sorting model that embeds the BDD to estimate heterogeneous preferences for neighborhood attributes (e.g. flood risk). Structural preference parameters recovered from the sorting model allow us to predict changes in the distribution of household types across flood risk, had a (counterfactual) change in flood risk (e.g. from re-zoning or revisions to NFIP flood boundary maps) taken place.

We offer three contributions to the literature. First, we generate new estimates of the MWTP to avoid flood risk. To our knowledge, we are the first to apply a BDD to the hedonic and sorting literature on flood risk. Compared with previous work, this strategy allows us to better control for endogenous neighborhood and household characteristics that may correlate with flood risk and potentially confound MWTP estimates. Second, our sorting model explores heterogeneity in MWTP to avoid flood hazards, as the distribution of preferences for flood risks is key to assessing the extent to which vulnerable sub-populations systematically sort into higher risk areas, including across racial and poverty lines. Lastly, we contribute to the discourse on distributional equity regarding natural hazards policies. Using our results, we are able to assess the impact of salient flood risk boundaries, delineated by the National Flood Insurance Program, on the distribution of household types across flood risks, and its potential role in exacerbating inequitable flood risk exposure in the United States.