Association for Public Policy Analysis & Management

A theme recurring in the economic literature of natural disasters is that individuals tend to underinvest in self-protection when they face limited financial liability for losses. Such behavior is commonly dubbed as the crowding-out (Kousky et al., 2013) or moral/charity hazard (Raschky et al., 2013). This paper extends extant literature and empirically examines how different types of disaster assistance programs affect household purchases of flood insurance across all National Flood Insurance Program (NFIP) counties. Our research is the first to consider the effect of the Public Assistance (PA) Program, which accounts for the largest portion of Federal Emergency Management Agency (FEMA)’s Disaster Relief Fund. The PA program is intended to help communities absorb the costs of emergency measures such as removing debris and repairing or replacing public buildings, roads, bridges, and public utilities. While the extant research mainly focuses on the Individual Housing Assistance (IHA) program which provides direct relief to affected households to cover uninsured losses, we argue that the PA program may also signal of a federal bailout when a disaster strikes, thereby reducing private incentives to invest in risk transfer and mitigation measures. The rising federal spending and large scale of PA program expenditure merits a deeper assessment of its potential unintended consequences.

Our empirical analysis is based on an unbalanced panel of all NFIP participating counties between 1983 and 2009. We employ instrumental variable model and along other explanatory variables, use political motivation of federal disaster aid as well as historical rainfall anomalies to instrument for endogenous federal spending variables. Our results indicate increasing spending in programs targeting at rebuilding public goods and recovery as well as individual housing assistance reduce insurance take up rates as well as total dollar coverage, however no effects were detected for the average coverage per policy. This suggests that the federal assistance may affect the extensive margin rather than intensive margin. We also find that flood insurance take-up rates and coverage are higher in wealthier counties, and highly responsive to current and prior year’s flooding shocks.

Our research has important implications for U.S. disaster policy. It suggests the different policy instruments may provide conflicted incentives in private efforts of mitigating and transferring hazard risks. While the NFIP program has grown significantly since its inception in response to various reforms and amendments, low take up rates still remain a concern across communities exposed to frequent floods and inundation risk. If flood insurance take-up decisions are crowded out by federal disaster assistance, this suggests we should re-evaluate the real costs of federal assistance by accounting for the implicit losses due to flood insurance dropout. From this perspective, our research also sheds light on the increased federal financial exposure to natural hazards (i.e. floods) in the face of climate change.