The Study On Signal Game Model For The Strategy Of Bidders And Sellers In Parallel Internet Auction

For hundreds of products, there are numerous auctions running in parallel and numerous online shops offering fixed price sales at the same time. Auction sellers are often in a highly competitive environment that is substantially different from the monopoly position they are assumed to have in classical auction theory. Buyers must choose the type of market (e.g. auction). In such settings, the seller in an auction is no longer a monopolist, but faces competition by the sellers in other auctions or markets. The bidders do not only have an entry choice, but have an outside option in the alternative sales channel.The fact that the strategic inter action quickly turns overly complicated, when competition and outside options are added to the standard auction models, has limited the amount of theoretical work. Therefore, it is significant to the study on the parallel online auctions in theory and practice.Firstly, paper reviewed parallel online auction theory and data study, and then introduced a two-period parallel online auction model and pointed out the limitations of the model's assumptions and conclusions. Then we extended the model to a three-period parallel auction, and under the assumption that original seller's strategy of maintaining a fixed (i.e. without regard to the seller strategy), we analysis of bidder's bidding strategy equilibrium, prove the feasibility of expansion. Then in the extended model, the introduction the assumption that the seller will set the reserve price below their cost and take "shill" bidding strategy in the auction, mainly for the signal game between the bidder and the seller, choose the reserve price and the bidding (including the bidder and the seller) as a transmission signal of the bidder and the seller to establish a multi-stage signaling game model, then analyzes the two sides confused strategy game. Finally, we collected the related auction data from eBay, the world's largest auction site, by regression analysis of these data, we validate some of the conclusions that the extended model obtained.