摘要：The paper aims at extending the egalitarian principle to environments withincomplete information. The approach is primarily axiomatic, focusing on thecharacteristic property of monotonicity: no member of the society should be worseo. when more collective decisions are available. I start by showing the incompat-ibility of this property with incentive e.ciency, even in quasi-linear environments.This serious impossibility result does not follow from the mere presence of incentiveconstraints, but instead from the fact that information is incomplete (asymmetricinformation at the time of making a decision). I then weaken the monotonicityproperty so as to require it only when starting from incentive compatible mecha-nisms at which interim utilities are transferable (in a weak sense). Adding otheraxioms in the spirit of Kalai's (Econometrica, 1977, Theorem 1) classical character-ization of the egalitarian principle under complete information, I obtain a partialcharacterization of a natural extension of the lex-min solution to problems withincomplete information. Next, I prove that, in each social choice problem, there isa unique way of rescaling the participants' interim utilities so as to make this solu-tion compatible with the ex-ante utilitarian principle. These two criteria coincidesin the rescaled utilities exactly at the incentive e.cient mechanisms that maxi-mize Harsanyi and Selten's (Management Science, 1972) weighted Nash product.These concepts are illustrated on classical examples of profit-sharing, public goodproduction and bilateral trade. The richness of the topic of social choice under in-complete information is illustrated by considering two alternative extensions of theegalitarian principle – one based on an idea of equity from the point of view of theindividuals themselves (given their private information) instead of an uninformedthird party (social planner or arbitrator), and another notion based on the idea ofprocedural justice
