This dissertation consists of three separate papers all pertaining to flood risk in the United States. The first chapter documents the changes in the frequency, severity and geographic concentration of flooding in the United States between 1978 and 2010. It also includes a review of relevant literature in economics and policy that will be referenced in two consequent chapters, paying special attention to papers that focus on the National Flood Insurance Program. I document evidence of an increase in frequency and severity of flood events over the last 3 decades nationwide. In the literature, I find a focus on catastrophic events and an opportunity to increase our understanding of the challenges faced by the National Flood Insurance Program by exploring the effect of an increased risk of flooding in a national context.

The second chapter explores the optimal choices of housing and consumption inside and outside a flood plain in a general equilibrium context. It includes a two period model of housing, consumption and location decisions with endogenous prices and uncertain flooding. I explore how prices and allocations in each period adjust as I change the magnitude and probability of flooding, consistent with data that suggests that floods are becoming more frequent and severe nationwide. I conduct several experiments to analyze the welfare implications of government flood policy in this context. My results suggest that government intervention in the form of lump sum tax and state contingent transfer (similar to disaster assistance) can be welfare improving for a range of probabilities of flooding, but not for all. Furthermore, this policy reduces precautionary savings and encourages more of the population to live in the flood zone than socially optimal.

The third chapter examines the relationship between flood risk, participation in the Community Rating System (CRS) Program, and insured flood losses. The stated goal of the CRS Program, which operates within the National Flood Insurance Program, is to encourage flood-prone communities to undertake a series of hazard mitigation efforts in order to lessen the impact of a flood. Does participation in the program lessen the impact of a flood in participating communities? Does the effectiveness of the program depend on the risk a community faces? Furthermore, does the program have a different effect when considering the repetitive loss information provided to CRS communities in 2006? The data shows that communities that participate in the program experience higher insured losses, as do communities with a higher flood risk as measured either by recent insured losses or property damage. However, I find that communities that undertake mitigation efforts are predicted to experience fewer insured losses as a function of flood risk and that both expected losses and the effect of mitigation vary by level of participation in the program. I also find that, after the repeat loss information was provided, CRS communities experienced fewer losses than was expected prior to the policy change.