Website CV Paper
Fields
Industrial Organization, Applied Micro, Theory
Job market paper
Belief Stabilization and the Flood Insurance Gap: Evidence from the National Flood Insurance Program
I study how local flood experience affects a household's decision to keep flood insurance, using a panel of National Flood Insurance Program policies that I link across fifteen years of OpenFEMA records and match to every flood insurance claim. New policyholders whose census block group has a flood claim during the policy term cancel 3.2 percentage points less often than those whose block group does not, or 1.6 points when comparing households within the same block group. This claim gap falls with tenure and is close to zero after about seven years. I estimate a model in which households update their beliefs about flood risk after a neighbor's claim, with a weight that falls with tenure. At the estimates the weight falls to about a fifth of its first-year value over ten years of coverage, and changes in which households stay insured explain little of the decline in the claim gap. In the estimated model, the households still insured at long tenure mostly overestimate their risk, and removing the decline in the weight barely changes coverage. Beliefs at entry matter much more. Correct entry beliefs, or information about flood risk delivered in the first two years of coverage, raise coverage by about 30 insured years per hundred entrants over ten years, while the same information delivered at long tenure slightly lowers it. Efforts to close the flood insurance gap should reach households in their first years of coverage.