Introduction

The increasing volatility of global weather patterns has underscored the essential role of natural disaster insurance. In the United States, property damage from natural disasters totalled approximately $1.4 trillion between 2015 and 2024. Although insurance is a vital tool for risk management, coverage remains incomplete, with estimates suggesting that 70% of flood damage is uninsured ex ante. Consequently, the federal government spends significant amounts on ex post assistance, primarily through Federal Emergency Management Agency (FEMA) assistance programmes. Total ex post federal disaster relief obligations reached $52 billion in fiscal year 2024. This insurance gap is severe for flooding; nationwide coverage is only 5%, and even in high-risk areas, penetration rates are often below 25%.

Background

Flood insurance in the U.S. is largely provided by the National Flood Insurance Program (NFIP). For decades, the NFIP offered premiums more than 50% lower than expected costs, which promoted over-development and under-mitigation in flood-prone regions. In 2021, the “Risk Rating 2.0” (RR2.0) reform was introduced to move towards actuarially fair premiums using geospatial data and catastrophe models. While intended to correct moral hazard, this research argues that RR2.0 creates unintended consequences: a fiscal spillover onto FEMA aid as coverage declines, and household exposure to uninsurable reclassification risk from uncertain climate projections.

A model of optimal insurance subsidies

The authors develop a model for the optimal insurance subsidy in the presence of moral hazard, ex post assistance spillovers, and evolving climate risk. Households face static uncertainty regarding flood occurrences and dynamic uncertainty regarding future risk reclassification. In this framework, a planner chooses a premium subsidy level to trade off the benefits of risk protection and reduced FEMA spillovers against the moral hazard costs of distorted coverage, mitigation, and location choices. This model defines the optimal subsidy as a function of statistics including fiscal spillovers, demand elasticity, reclassification risk loss, and locational and mitigation elasticities.

Spillovers to ex post disaster aid

Using a county-year dataset, the research identifies a sizeable offset of ex post government spending due to ex ante insurance. It estimates that each additional 1% of insured houses reduces ex post government spending by $173 to $216 per house during a flood event. The majority of this saving, approximately $183, comes from FEMA’s Individuals and Households Program (IHP). Because the average NFIP claim payment is an order of magnitude larger than a typical IHP grant, increased insurance coverage translates into substantial federal savings.

Demand estimation

The study estimates the semi-elasticity of demand for NFIP insurance to be -0.32. This indicates that households are responsive to premium changes on the extensive margin but show a negligible response on the intensive margin of coverage limits. Demand is found to be more elastic for the first dollar of subsidy and lower in high-risk zones where purchase mandates exist, despite imperfect enforcement.

Reclassification risk

Risk-based pricing exposes households to premium volatility driven by climate uncertainty. To quantify this, the researchers use state-of-the-art climate projections for hurricanes and inland flooding. The idiosyncratic component of this reclassification risk has a standard deviation of $1,842 per household. A 50% premium subsidy provides approximately $207 per house in insurance value. This protection justifies a subsidy of approximately 15%, even without fiscal spillovers.

Moral hazard: Mitigation

The research evaluates moral hazard by examining property-level elevation decisions in Florida. It finds that properties facing higher premium increases were 7% to 11% more likely to elevate. However, the implied national mitigation elasticity is only about 22% of the Florida estimate. These benefits are smaller than the costs resulting from decreased insurance take-up and increased reclassification risk.

Moral hazard: Location

To measure locational moral hazard, the study tracks population movements from tracts with high premium increases to those with lower increases. It observes a net movement of 0.77%, yielding a migration elasticity of 0.20. While households respond to price signals when choosing locations, the welfare gains from more efficient locational decisions do not outweigh the benefits of premium subsidies.

Optimal policy

Integrating these findings, the authors calculate that the optimal subsidy for flood insurance is 52%. This is comparable to the subsidisation level provided before the recent reform. Notably, a premium subsidy of up to 8% has an infinite marginal value of public funds (MVPF), essentially paying for itself through FEMA savings. The optimal subsidy rate remains robust across various settings and household characteristics.

Conclusion

The transition to risk-based pricing through RR2.0 ignores the second-best nature of the flood insurance market. The benefits of reducing moral hazard are outweighed by the costs of increased ex post recovery spending and exposure to reclassification risk. Policymakers should consider that substantial subsidies are rationalised by the federal government’s role as a de facto insurer through ex post assistance.