Procurement with Cost and Non-Cost Attributes : Cost-Sharing Mechanisms

A buyer faces a two-dimensional mechanism design problem for awarding a project to one among a set of contractors, each of whom is privately informed about his cost and his estimate of an a priori random non-cost attribute. The winning contractor realizes his non-cost attribute upon the project's completion and may “manipulate” it in a costless manner (if such a manipulation is beneficial to him). The non-cost attribute inflicts a disutility cost on the buyer. This procurement problem arises in situations such as highway construction projects, where completion times are a major concern. We establish the significance of incorporating the possibility of manipulation in two ways: (1) Using an optimal mechanism obtained by ignoring the possibility of manipulation can generate perverse incentives for the winning contractor to engage in manipulation. (2) The privacy of the non-cost estimates can generate information rent only due to the possibility of contractors' manipulation. We further study the family of cost-sharing mechanisms as a nonmanipulable, easy-to-implement and near-optimal solution to the buyer's procurement problem. In a cost-sharing mechanism, the winning contractor is selected via a second-price auction and needs to reimburse a pre-specified fraction – referred to as the cost-sharing fraction – of the buyer's disutility cost upon completion of the project. We show that the cost-sharing fraction plays an unequivocal role in capturing the essential tradeoff between allocative inefficiency and information rent. We also characterize the optimal cost-sharing fraction and offer prescriptive guidelines on the choice of this fraction based on the second-moment information of the buyer's belief distribution. Finally, we establish the theoretical performance guarantees for the optimal cost-sharing mechanism