Optimal Design Of Online Auctions For Homogeneous Goods

With the great development of the e-business, the online auction becomes increasingly popular. In the practice of online auctions, the character of auction mechanisms will directly affect the efficiency of online auction items' allocation, which then makes influence on the revenues of the sellers and auction websites, and determines the customers' purchase behavior eventually. Therefore, the study on the auction mechanism is of great significance both in theory and practice.Based on the build-up of general optimized model of the online optimal auction, the dissertation integrates such online auction characteristics as the asynchronism of the time and the distribution of the space, and studies the online auction from four aspects including the design of optimal reserve price in single-stage online auction, the design of optimal reserve price in multi-stage auction, the optimal design in multi-unit single-period auction and the optimal design of dual-mechanism in which auction and posted price mechanisms are used in parallel.Firstly, by introducing the factors of online auction like the entry fee, the commission, the penalties and the stochastic arrival of bidders, the dissertation formulates online auctions' decision-make model of maximization of seller's expected revenue in open reserve price and private reserve price respectively, and studies the online auction's optimal open reserve price and optimal private reserve price, and gives the general form of optimal reserve price. On the base of all these above, the optimal reserve price is analyzed, showing that the sellers' optimal open reserve price is the same to the optimal private reserve price, and the maximum expected revenues are equal.Secondly, the study of the optimal reserve price in the multi-stage online auction is made on the basis of single-stage online auction model, giving the general form of the optimal reserve price series. The research shows that the optimal open reserve price is the same as the optimal private reserve price in corresponding stages, and the sellers' revenues are equal in two different reserve prices in corresponding stages. When carrying out the multi-stage auction, the seller's expected revenues monotone decrease with the increase of stage number, the optimal reserve price of the seller monotone decreases at each stage when the bidders' evaluations are distributed uniformly.Thirdly, by introducing the feature of bidders' stochastic arrival and holding cost, advertisement cost, and monitor cost, the dissertation reformulates the revenue function of seller in single-stage auction, and performs the optimal design of reserve price, auction quantity, auction time and unit-cost of advertisement in open reserve price and private reserve price, respectively, and makes comparative analysis in both situations. According to the research, when the number of bidder is more than the lot-size, the expected auction closing price in open reserve price is less than that in private reserve price, and the expected revenues in private reserve price is more than that in open reserve price when the seller-set reserve price is at the lowest level.Finally, by constructing the delay cost function, the customer decision model based on the threshold strategies in dual-mechanism is extended, deducing the general form of the model based on the threshold strategies. It shows that customers with higher valuation will choose threshold strategies to make decision once arriving at websites when the delay cost function of them continuous strict increases with the auction remaining time, no matter the delay cost function is in linear or exponential form. On the basic of it, taking the auction lasting time, the quantity of units and posted price as the decision variables, the dissertation establishes the optimal decision model of seller based on the threshold strategies in dual-mechanism, and brings out the optimal auction design principle and strategy by numerical analysis, and validates the designed principle and strategy through calculated examples.The end of the dissertation makes summary of work, and brings forward the further problem to be studied.