From behind the veil: Evaluating allocation rules by ex-ante properties

Abstract      

We study rules for allocating objects.  Departing from standard analysis, we evaluate rules according to their performance at an ex-ante stage, before individuals learn their preferences.  Introducing an appropriate notion of ex-ante efficiency, we search for rules that are both efficient and provide incentive for individuals to truthfully report their eventual preferences.  Our main results characterize the priority (or “serial dictatorship”) rules by ex-ante efficiency and either strategy-proofness or Bayesian incentive compatibility on natural preference domains.  Allowing indifferences identifies the extended priority rules.  For domains on which utilities correspond to ordinal preference rank, the implications of our incentive requirements diverge:  ex-ante efficiency and strategy-proofness continue to characterize the priority rules, but many additional rules, including rules which maximize utilitarian welfare, are Bayesian incentive compatible.

            When truly behind the veil, agents and objects are indistinguishable, which we model as symmetric problems.  Remarkably, all rules in a large family achieve the same utilitarian welfare.  Moreover, rules adapting the top trading cycles algorithm are Lorenz maximal and priority rules are Lorenz minimal within this family.  Allowing the size of the economy to grow, we find that average welfare under each rule approaches that of a utilitarian rule.  To further compare rules, we introduce solidarity properties and consider an interim participation constraint.  These considerations distinguish methods of randomizing over families of rules.