Asymmetric Information and Learning: Evidence from the Automobile Insurance Market

This paper tests the predictions of adverse-selection models using data from the automobile insurance market. I find that, in contrast to what recent research suggests, the evidence is consistent with the presence of informational asymmetries in this market: new customers choosing higher insurance coverage are associated with more accidents. Consistent with the possibility of policyholders' learning about their risk type, such a coverage-accidents correlation exists only for policyholders with enough years of driving experience. The informational advantage that new customers with driving experience have over the insurer appears to arise in part from customers' underreporting their past claim history: policyholders switching to new insurers are disproportionately ones with a poor claims history, and new customers tend to underreport their past claims history when joining a new insurer.