A Laboratory Investigation of Rank Feedback in Procurement Auctions

A popular procurement auction format is one in which bidders compete during a live auction event but observe only the rank of their own bid and not the price bids of their competitors. We investigate the performance of auctions with rank feedback in a simple setting for which analytical benchmarks are readily available. We test these benchmarks in the laboratory by comparing the performance of auctions with rank-based feedback to auctions with full-price feedback as well as to auctions with no price feedback (sealed-bid auctions). When bidders are risk-neutral expected-profit maximizers, the buyer's expected costs should be the same under rank and full-price feedback; moreover, expected buyer costs should be the same as in a sealed-bid auction. However, when we test this theoretical equality in a controlled laboratory setting we find that, consistent with practitioners' beliefs but contrary to our model, rank feedback results in <i>lower</i> average prices than full-price feedback. We identify two behavioral reasons for the difference. The first explanation is based on the similarity of the bidders' problem in a sealed-bid first-price auction and an open-bid auction with rank feedback. The second explanation incorporates the use of jump bids motivated by bidder impatience.