Andrew Plantinga, UC Santa Barbara

As climate risk escalates, property insurance has a critical role to play in helping households
and firms reduce risk exposure, and to recover from natural disasters when they strike. To per-
form these functions efficiently, insurers need to access high quality information about disaster
risk and set prices that accurately reflect the costs to insure it. We use proprietary data on
parcel-level wildfire risk, together with insurance premiums derived from insurers’ regulatory
filings, to investigate how insurance is priced and provisioned in a large market for homeowners’
insurance with high levels of natural disaster risk exposure. We document striking variation
in insurers’ risk pricing strategies. Firms that rely on coarser measures of wildfire risk charge
relatively high prices in high-risk market segments – or choose not to serve these areas at all.
Empirical results show evidence of a large winner’s curse, where firms with less granular pricing
strategies have higher expected losses and lower net revenue per parcel. A theoretical model of
a market for natural hazard insurance that incorporates both price regulation and asymmetric
information across insurers helps rationalize the empirical patterns we document. Our results
highlight the underappreciated importance of the winner’s curse as a driver of high prices and
limited participation in insurance markets for large, hard-to-model risks.