Efficiency and adverse selection (Q1173869)

A model is studied, similar to the lemons model as introduced by \textit{G. Akerlof} [Q. J. Econ. 84, 488-500 (1970)]. There are two agents, a buyer and a seller, each owning an initial endowment of money and the seller also owns a certain object that might be ``good'' or ``bad''. \(\theta\) is the probability that it is good, with some distribution \(f(\theta)\). The seller can observe \(\theta\) before any transaction takes place, but the buyer only knows \(f(\theta)\). The buyer attaches a higher value to the good than the seller, both when it good as when it is bad, so there are always possible gains from trade which, because of adverse selection, due to asymmetric information, cannot be fully realized. The author studies the (non-empty) set of feasible, incentive compatible and individually rational allocations, an allocation consisting of amounts of money of the two agents and a probability \(x_ j(\theta)\) that agent \(i\) obtains the object, as a function of \(\theta\). Incentive compatibility ensures that the seller truthfully reveals \(\theta\), so that the allocation is based on the true \(\theta\). Two types of constrained efficiency are introduced (before and after the seller learns \(\theta\)) and it appears that in efficient allocations \(x_ i(\theta)\) is a step function. Finally the (constrained) efficiency of competitive allocations (with or without a system of taxes) and of allocations where either the seller or the buyer sets a price, are considered.