To discriminate or not to discriminate: how to enforce unverifiable quality in repeated procurement

Abstract We analyse a model of repeated procurement whereby a buyer may elicit unverifiable quality by relying on two types of competitive procedures. The first type is non-discriminatory, namely a low-price auction with a public reserve price, whereas the second type is a scoring auction that includes a non-financial, discriminatory dimension based on past performance. We first provide sufficient conditions for the existence of relational procurement contracts under which the buyer can elicit the desired level of quality. We then assess which mechanism is preferable in terms of (i) the buyer’s preferences and (ii) the equilibrium existence conditions. As for (i), we establish the conditions whereby the two procedures yield the buyer the same utility as well as those under which a non-discriminatory procedure ensures a lower cost of the project, although this comes with a lower quality and a positive probability of the project not being delivered altogether. As for (ii), no clear-cut results can be established. Indeed, the range of values of the project net-of-quality utility for which an equilibrium exists under the non-discriminatory procedure is always larger than under the discriminatory one. Conversely, the two procedures have a different ranking in terms of stringency of equilibrium existence requirements for the discount factor and the net social value of quality.