Research On Model And Mechanism Of Multi-Attribute Auction Based On Game Theory

With economic development and popularization of Internet, multi-attribute auction shows broad prospects. Because multi-attribute auction has an outstanding advantage to satisfy the diverse needs of auctioneers to the last degree and give full play to bidders'competitive advantages, it can increase their mutual benefit. Especially in procurement, adoption of multi-attribute reverse auction on the net has more advantages. However, the existing models of multi-attribute auction are simple and their parameter values are fixed so that they can not suit the need of implementing multi-attribute auction on the internet. The problem will hinder the wide use of multi-attribute auction and the rapid development online government procurement and B2B transactions. The above problem urgently needs to be solved.Firstly this thesis focuses on the current urgent need to research online multi-attribute auction, takes the method of theoretical analysis, based the domestic and foreign research status of the multi-attribute auction, combined with the reality, and finds out the shortcomings of existing classic model designed by Che, Branco and David. The model of multi-attribute auction is perfected and the model applicability is improved.Then based on the mathematical auction model in this paper, mainly for the first score-sealed bid auction mechanism and multi-attribute English auction mechanism, bidders strategy is made and the structural optimal formulas of quality attributes, price attribute and the bidders' expected benefit are given. According to the bidder's bidding strategy, the auctioneer's expected benefit is calculated, optimal score function design strategy is derived, and the formula for calculating attribute weights of score function are given.Finally, through analysis with game theory, the expected revenue between first score sealed-bid auction and multi-attribute Dutch auction is equal and so is expected revenue between multi-attribute English auction and second score sealed-bid auction. The expected revenue between first score sealed-bid auction and multi-attribute English auction is proved to be equal by mathematical computation. So one conclusion is made that the auctioneer's expected revenue of the above multi-attribute auctions is equal. In addition, at the same time of analyzing the four multi-attribute auction mechanisms, the factors of influencing the auction result are analyzed in detail, such as the number of attribute, the number of bidder and the size of the minimal increment score.