Keyword Auction Mechanisms With Budget Constraints

In recent years with the rise of online advertising and the development of search engine technology a new auction mechanism known as keyword auction emergences. The global keyword auction market has reached tens of billions of dollars in size and is the main source of the major search engines’revenue. As a successful e-commerce model, keyword auction mechanism has aroused researchers’interest and now is a hot research topic of management science, economics, and computer science. In-depth study on the keyword auction mechanism can promote the development of the new e-commerce theoretical basis and key technologies such as algorithmic game theory, mechanism design theory and provide support for search engines’decision making for mechanism innovation and increasing the auction revenue and auction efficiency.With a systematic analysis of the relevant literature the thesis is mainly on the following aspects of work:(1) GSP keyword auction mechanism with budget constraint.Static model of the GSP keyword auction with budget constraint in complete information envirement is setted up based on a long-term perspective; the exisitent of a unique pure strategy Nash equilibrium of the static GSP mechanism with budget constraint is proved. Dynamic model of the GSP keyword auction with budget constraint in incomplete information envirement is setted up based on a short-term perspective, and the advertisers’ best response bidding strategies are analized, then the fact that advertisers bidding strategies in GSP keyword auction with budget constraint converge to an equilibrium point is poroved, and the bidding value in the equilibrium point will not lower than that in the equilibrium of the static model. The equilibrium of the static and dynamic model are verified through numerical examples. The equilibrium efficiency and auctioneer’s revenue of the GSP mechanism with budget constraint are studied based on the equilibrium analysis, and there is a room for Pareto improvement of the auciton outcome, and because all winners values are higher than that of the losers so the auction mechaism is near Pareto optimal, and the upper and lower bounds of the auctioneer’s revenue are caculated.(2) Pareto optimal keyword auction mechanism with budget constraint.B-Vickrey mechanism is proposed and applied to keyword auction with budget constraint, A step function is constructed as the winners payment function, it is proved that the B-Vickrey keyword auction mechanism satisfies the individual rationality constraint and market clearing constraint. The equilibrium bidding strategies are analyzed with the method of eliminating dominated strategies, and we pointed out that advertisers will not bid lower than their true budget and value. The auction outcome is proved to be Pareto optimal, and the lower bound of the auctioneer’s revenue is occurred when all advertiser bid their true budget and valuation.(3) Competing keyword auctions based on ranking rules with budget constraint.The advertisers’equilibrium bidding strategies and the auctioneer’s revenue with different ranking rules with a single auctioneer are caculated, let S denote the single bid ranking rule and W denote the effective bid ranking rule, the numerical example shows that different ranking rules can make more or less auctioneer’s revenue with different advertising quality type profiles. A three-stage game model is established for the keyword auction in multi-auctioneers competing environment. The advertisers’budget allocation strategies under different auctioneers competing stratgy profiles and the analytical formula of auctioneers’ revenue are analyzed. The equilibrium competing strategies of auctioneers are analyzed by numerical examples:the strategy profile (S, S) and (W, W) can both become the equilibium competing strategy profile, and the S strategy is a dominated strategy if used alone.