Repeated Insurance Contracts and Learning

This article considers a repeated insurance model with incomplete information in which the insurer and the consumer both learn over time about the unknown risk category of the consumer. Care choices by young consumers affect the informational value of the accident history. Under an optimal scheme of long-term insurance contracts, young consumers will always be "subsidized" by old consumers, and consumers who are reassessed as low-risk types are always subsidizers (i.e., reverse experience ratings). We compare this optimal scheme with the set of contracts that would emerge in a competitive market if only short-term contracting were possible. These sequentially competitive contracts can lead to over-investment in care by both young and old consumers, relative to the optimum, in contrast to the underinvestment problem associated with moral hazard in insurance.