Multi-Attribute Procurement Auctions in the Presence of Satisfaction Risk

Abstract

Procurement auctions are widely used to solicit bids for custom items suchas services, construction projects, public infrastructure, and more. Such auctions involve significant risk because the idiosyncratic nature of the purchases means that delivered quality is uncertain. We examine the case in which buyer satisfaction is affected by both bid quality and the winning supplier’s unobservable effort. Our setting is a multi-attribute auction in which the buyer first awards a contract to the bidder with the highest score. The winning bidder then exerts effort which affects the difference between promised and delivered quality and, thus, the buyer’s satisfaction. A performance-contingent reward or penalty occurs upon delivery. Depending upon buyer preference over quality and effort, bid quality and execution effort can be complements or substitutes. Bidders must jointly optimize these prior to choosing their bid price. In contrast to traditional multi-attribute auctions, we find that bidders gain less information rent, and that the optimal scoring mechanism rule distorts bid quality downward. We also show that neither reserve quality nor price alone are sufficient to exclude undesirable bidders. Compared with the socially efficient mechanism, effort under the buyer’s optimal mechanism is distorted upward (downward) when it substitutes (complements) bid quality. We further investigate the impact of uncertainty on bidder behavior and mechanism design under both additive and multiplicative relationships between bidders’ behaviors and randomness and find that bid quality and effort can either increase or decrease. Accordingly, the design of optimal scoring rules must take the nature of uncertainty into account.