An Experimental Examination of the Assignment Problem

The problem of optimally assigning individuals to heterogeneous objects so that each individual is allocated at most one object (the assignment problem) has a long history. Algorithms• •based •on •ordinal preferences have been developed and several auctions using monetary transfers have been proposed. The performance of two auction mechanisms to solve the assignment problem is examined in an experimental setting. One of the auctions is a sealed-bid variant of the Vickrey auction for homogeneous objects and the other auction is an extension of the English auction. The auctions are tested in two diverse competitive environments (high and low contention). The experimental results show that the English auction generates higher revenues and efficiencies than its sealed-bid counterpart especially if there is a high level of contention. However, the efficiency gains of the English auction are at the expense of consumers' surplus. Indeed, a random assignment creates greater consumers' surplus relative to either auction outcomes in the high contention environment.