Insurance as a lemons market: coverage denials and pooling (Q2211475): Difference between revisions

A consumer has initial wealth \(w >0\) and faces a potential loss \(\ell \in (0,w)\). The probability of the loss is \(\theta \in [0,1]\). The consumer makes decisions according to a strictly concave von Neumann-Morgenstern utility function \(u\). The consumer knows \(\theta\), the insurer does not. But the insurer knows the distribution \(F\) of \(\theta\). \(F\) is absolutely continuous \(F(\vartheta) = \int_0^\vartheta f(s) \;d \mu(s)\). \(\mu\) is either the Lebesgue measure or has finite support. The company offers contracts of the form \((x,t)\), where \(x\) is the payment in the case of a loss and \(t\) is the premium. The optimisation problem of the consumer is thus to maximise \(U(x,t,\theta) = \theta u(w-\ell+x-t) + (1-\theta) u(w-t)\). Then for each \(\theta\), the consumer chooses the contract \((x(\theta), t(\theta))\) In addition, the insurer faces (provision) expected costs \(C(x,\theta)\) for a contract. The problem of the insurer is now to construct contracts maximising the expected profit \[ \max \int_0^1 [t(\theta) - C(x(\theta),\theta)] \;d F(\theta)\] subject to \[ U(x(\theta),t(\theta),\theta) \ge U(x(\theta'),t(\theta'),\theta) \text{ for all \(\theta,\theta'\),}\] and \[ U(x(\theta),t(\theta),\theta) \ge U(0,0,\theta)\text{ for all \(\theta\).}\] The first condition just says that the consumer optimises its expected utility, the second condition just gives the opportunity not to buy insurance. The problem is solved in the paper. In particular, the problem that no insurance is offered or that only one type \((x,t)\) is offered by the company is investigated.