Overconfidence, Insurance and Paternalism

This paper introduces a simple model of insurance market that comprises adverse-selection, overconfidence, and testing in the attempt of reconciliating the predictions of the economic theory with the empirical evidence against moral hazard, and with stylized facts such as the Uninsured Motorist puzzle and the dependence of insurance terms on driving record. Our model delivers a rich equilibrium behavior: while drivers with a good driving record are screened through insurance contracts, overconfident bad-record drivers choose to be uninsured. Moreover, equilibrium insurance coverage is not monotonic in overconfidence. Our comparative statics analysis shows that, while bad-record drivers' equilibrium insurance coverage is non-decreasing in test results, good-record drivers' coverage is strictly decreasing both in test results and in scoring precision. The policy prescriptions that follows from our model are as follows: First, we show that the introduction of scoring is Pareto improving as it helps insurance companies sorting out overconfident from high-ability drivers, and hence it allows a more effective range of insurance contracts. Second, we show that the introduction of compulsory insurance for drivers with a bad record is not Pareto improving because it penalizes high-ability drivers. Hence, compulsory insurance cannot be considered a substitute to a compulsory driving safety education programs aimed both at making drivers aware of their own driving skills and limitations.