Competing Combinatorial Auctions: Theory and Experiments

• Applicants: Professor Dr. Thomas Kittsteiner; Dr. Marion Ott
• Subject Area: Economic Theory
• Term: from 2014 to 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 259111989

The design of auctions and in particular the design of combinatorial auctions, i.e. the design of auctions that allow sellers to sell multiple items in a single auction and bidders to submit bids on groups of items, has become an important field in theoretical and experimental economics. Its insights have led to many practical implementations such as spectrum auctions, privatisation auctions, tender auctions, internet ad auctions, procurement auctions and liquidity auctions by central banks. Our project aims at extending knowledge on the optimal design of combinatorial auctions in the presence of competition between sellers. So far, the impact of competition between auctioneers on the performance of various auction formats has only been investigated for single-item auctions, though it is also relevant for combinatorial auctions. In many of the applications sellers potentially compete for the same set of bidders: bidders from anywhere in the EU participate in tenders by EU governments; the same international banks participate in liquidity auctions; the same suppliers bid in procurement auctions by competing companies. We thus believe that research on competition between auctioneers who each offer multiple items will initiate a new and important research field.In our project we develop a model that enables us to analyse the interplay between competition and various forms of demand-side heterogeneity (such as differences in valuations for individual items or differences in the magnitude of complementarities/substitutabilities between items). We seek to understand the benefits of offering combinatorial auctions (rather than single-unit auctions) when auctioneers compete for bidders. As sellers' auction design choices can affect the degree of heterogeneity between participants in each of the auctions, the benefits from offering a specific auction design (and in particular of offering a combinatorial design) can depend on the degree of competition between auctioneers. General insights about the relationship between intensity of competition and performance of different design choices might help to explain the choice of certain designs in reality and motivate why combinatorial auctions are predominantly used by monopolistic sellers. Our project is likely to have implications for practical design questions. It might help internet auction platforms to select the best design alternatives for competing sellers and it might help companies to incorporate the effect of competitors on their optimal procurement auction design.To complement and test theoretical predictions we investigate bidding behaviour and sellers' design choices in the laboratory. A series of experiments will explore to what extent theoretical reference points can capture a) observed bidding behaviour in combinatorial auctions, b) bidders' participation decisions when facing the choice of different auctions, and c) sellers' auction design choices in the presence of competing sellers.

• DFG Programme: Research Grants
• International Connection: United Kingdom
• Participating Person: Professor Richard Steinberg, Ph.D.