Auctions, Negotiations and Hybrid Mechanisms in Procurement
Final Report Abstract
The research projects were set up to shed light on two important topics in procurement. First, they should help to better understand the popularity of sequential procurement mechanisms. How are requests for quotes used in procurement practice and how can this be rationalized. Second, the projects examined how procurement regulation must be designed to prevent discrimination, which is important to fight favoritism and corruption. Our evidence from the field suggests that the information about the number of suppliers has strong influence on the buyer’s procurement mechanism choice. Buyers typically conduct first-price auctions if the number of bidders is small and English auctions if many suppliers compete. Practitioners argue that in this case bidders overestimate competition when they participate in first-price auctions which makes the strategy profitable for the buyer. This finding, however, cannot easily be rationalized, because bidders should correctly anticipate the lack of competition in first-price auctions and bid accordingly. To gain further insights we replicated this procurement setting in controlled laboratory experiments. The experiments show that even in a simplified procurement setting the majority of individuals who participated in the role of bidders did not anticipate the strategic incentives of the buyer and hence overestimated competition in first-price auctions. This finding has direct consequences for the optimal design of procurement auctions, because it implies that the auction designer can exploit suppliers’ uncertainty about the number of competitors. Our survey with procurement practitioners shows that request for quotes do not only provide information about the actual number of bidders to the buyer, but also reveal information about bidders’ costs. To investigate this, we devise a simple two stage model in which bidders first submit an initial offer and then the auctioneer makes a counteroffer. The initial offers can be thought of as a ceiling for the subsequent negotiations and therefore bidders should, in theory, not reveal any information about their costs in the first stage. We then take this setting to the laboratory and find that on one hand, bidders still give away information about their cost structure in the first stage. On the other hand, the auctioneers understand this and are able to exploit the bidder’s behavior. Procurement is known to be a hot spot for corruption. An important step to better fight corruption is to understand how procurement regulation needs to be designed in order to rule out discrimination. In our research, we first analyzed in how far the current procurement regulation which requires equal treatment of suppliers hinders a procurement designer to discriminate between different suppliers. Deb and Pai (2017) show that the requirement of equal treatment does not prevent discrimination in any meaningful way and still allows for perfect discrimination. This means, the procurement designer can conduct a procurement mechanism that treats suppliers equally but has an equilibrium in which the procurement designer’s preferred supplier is always awarded the contract and receives the highest possible price. Based on this result we develop a simple extension to the existing rules that prevents discrimination. We call this additional requirement imitation perfection and it implies that each bidder must have had the opportunity to imitate the allocation and payment of each other bidder who placed a better bid. The developed requirement is of practical importance. As the conduction of specific procurement projects is typically delegated to a large group of agents, it is important to set up rules that allow honest agents to adjust the procurement design to the specific requirements but to limit corrupt agents ability to favor certain bidders. At the same time it is impossible for the buyer to monitor each of the procurement transactions and check whether the implemented procedures were optimal. Thus, there is a need to set general procurement rules. The set of procurement regulations should have the following properties. Firstly, it should be easy to check whether these regulations have been followed. In particular, this should not require knowledge of unobservable such as subjective beliefs, or the use of complicated calculations such as equilibrium analyses. Secondly, the regulation should restrict corrupt agents in a meaningful way. Finally, honest agents should maintain enough freedom to enable them to implement the optimal procedures. Imitation perfection has all of these desirable properties. Firstly, a quick look at the rules of the particular auction is sufficient to verify if the procurement process satisfies imitation perfection. This is due to the fact that imitation perfection is a property of the payment rule. Hence, the verification does not require information on any details of the procurement project and can also be done ex-post. Secondly, imitation perfection prevents corrupt agents from implementing perfectly discriminatory outcomes and guarantees discrimination-free outcomes. Finally, imitation perfection gives honest agents the opportunity to implement the efficient auction as well as the revenue-optimal one if bidders are symmetric. In this respect, ensuring that the procurement mechanism is imitation-perfect comes at no costs if all bidders are ex-ante homogeneous.
Publications
- (2016), Procurement under public scrutiny: Auctions vs. negotiations. RAND Journal of Economics, 47(4), 914-934
Gretschko, V., and A. Wambach
(See online at https://doi.org/10.1111/1756-2171.12159) - (2016). Bid pooling in reverse multi-unit Dutch auctions: an experimental investigation. Theory and Decision, 81(4), 511-534
Gillen, P., A. Rasch, A. Wambach, and P. Werner
(See online at https://doi.org/10.1007/s11238-016-9546-z) - (2016). Collusion in dynamic buyer-determined reverse auctions. Management Science, 62(2), 518-533
Fugger, N., E. Katok, and A. Wambach
(See online at https://doi.org/10.1287/mnsc.2014.2142) - (2016). Pre-auction or post-auction qualification? Economic Theory Bulletin, 5(2), 139-150
Gillen, P., V. Gretschko, and A. Rasch
(See online at https://doi.org/10.1007/s40505-016-0108-x) - (2017). Bidding Complexities in Combinatorial Clock Auctions, in: Martin Bichler and Jacob Gorree Handbook of Spectrum Auction Design, Cambridge University Press
Gretschko, V., S. Knapek, and A. Wambach
(See online at https://doi.org/10.1017/9781316471609.034) - (2018). Split award auctions: Insights from theory and experiments. Management Science
Chaturvedi, A., D. Beil, and E. Katok
(See online at https://doi.org/10.1287/mnsc.2017.2932) - (2018). Trust in procurement interactions. Management Science
Fugger, N., E. Katok, and A. Wambach
(See online at https://doi.org/10.1287/mnsc.2018.3196)
