This paper studies the outcome of fully insured random selections among multiple competitive equilibria. This defines an iterative procedure of reallocation which is Pareto improving at each step. The process converges to a unique Pareto optimal allocation in finitely many steps. The key requirement is that random selections be continuous, which is a generic condition for smooth exchange economies with strictly concave utility functions.
Published In: Essays in economic theory, growth, and labor markets: A festschrift in honor of E. Drandakis (2002, pp. 33-54)