Design and implementation of combinatorail auctions for multilateral markets

点击翻译按钮获取中文摘要

In the last two decades, auctions have received a lot of attention in economics, operations research and computer science literature. All reported studies in auctions design field recognized the potential benefits of the, so-called, combinatorial auctions, especially when complementarities exist between the traded items. Combinatorial auctions refer to auction mechanisms in which bids on combinations of items are permitted. A significant effort has been devoted since the mid-1990's to the elaboration of advanced ''package bidding'' auction techniques. The majority of this research considers however unilateral markets, i.e., markets including either, one seller and multiple buyers or one buyer and multiple sellers. This research program addresses the problem of designing and implementing combinatorial auctions in multilateral markets where multiple sellers and buyers trade heterogeneous items. The objective is to provide the actors of the market with optimized decision tools to help them solve some of the problems they would face during or after the auction process. The emphasis will be put on three inter-related decision problems: (1) Who are the final winners? (2) What should every participant pay or receive at the end of the auction? (3) What is the best strategy for constructing bids? The originality of the proposed approaches consists in combining mathematical modeling and optimization techniques, proper to the operations research field, and cost sharing and pricing methods, used in cooperative game theroy. Such a combination is relevant in a multilateral context where combinatorial bids are permitted. All the proposed approaches will be evaluated through generated and real-life problem tests arising in various contexts such as transportation service procurement or electricity markets. They will thus contribute to both industrial and academic environements.