Anton and Yao (1989) show that in split-award procurement auctions bidders coordinate their bids to sustain high buyer price. We relax their assumption that the buyer has full information about the suppliers’ production costs and restore the coordination outcome.

We show that a non-Bayesian learning procedure leads to very permissive implementation results concerning the efficient allocation of resources in a dynamic environment where impatient, privately informed agents arrive over time, and where the designer gradually learns about the distribution of...