On the one hand, time-honored practices of open outcry, bazaar and other flea markets, as well as Sotheby's and Christie's, the traditional auction houses. On the other hand, the Internet technology ...quickly and cheaply connects people anywhere in the world at anytime. A true love story then began and gave rise to online auctions, which are undoubtedly one of the greatest successes of e-commerce. Thanks to the Internet, individual consumers, average people, generally used to posted prices, discover an efficient mechanism of price setting.
The weak popularity of auctions until the dawn of the third millennium (auctions were intended to B2B transactions, or to B2C transactions but in specific markets, such as the art market) certainly justifies the lack of interest of researchers in marketing for this mechanism. On the other hand, economists made auctions one of their favorite topics, as this mechanism constitutes a wonderful application of game theory, which has been playing a preponderant role in economics for twenty-five years.
Entitled "Essays on Online Auction Design and Bidding Behavior", this doctoral dissertation helps enrich the traditional economic approach of auctions by a behavioral and dynamic approach. The original part of the research is structured in three main stages, which go deeper and deeper in the study of bidders' behavior and auction dynamics. The whole research is based on online auctions organized by the French airline company Air France and by the auction site eBay.
In the first stage of the research, on the basis of hypotheses coming from the auction theory, the impact of auction rules on bidders' participation and seller's revenue is econometrically studied, using simultaneous equations models and the 2SLS method. Does a high opening bid have a negative effect on the number of bidders and a positive one on the auction revenue, as predicted by the auction theory? Do sequential auctions of similar items lead to similar auction price? These are examples of questions that are investigated. While the auction theory literature typically takes the number of bidders as exogenously given, we chose to consider it as endogenous. Even though an increasing number of researchers acknowledge the endogenous feature of auction participation, no empirical study really takes this feature into account. Hypotheses from the auction theory turn out to be confirmed, as far as the minimum bid and the number of bidders are concerned. Interesting results are found regarding the performance of sequential auctions of similar items, since auctions in a sequence turn out to yield increasing revenues. Furthermore, the impact of the starting bid turns out to be moderated by the effect of these sequential auctions.
In the second stage of this research, a disaggregated perspective (at the bidder's level) is adopted, since a typology of bidding behaviors based on bidders' choices is built. This part is thus aimed at investigating whether heterogeneity exists in bidder's behavior in online auctions, contrary to the longstanding assumption of homogenous, rational and strategic bidders made by the auction theory. A clustering analysis is conducted, based on decisions that bidders have to make during an auction, whose keywords are certainly: when? How? How much? For example: do bidders submit a small or a large number of bids? Do they react quickly when they lose auction leadership? When do they make their bids: at the beginning of the auction, at the end of it, or during the whole auction? What increment do they use? Results, based on two different samples related to Air France and eBay auctions provide a description of different bidding behaviors. The analysis focusing on Air France auctions highlights five types of bidding behavior: jump bidders, rational bidders, active bidders, bottom fishers and pioneers. Slightly different results are obtained for eBay auctions, since six types of bidding behavior are highlighted: snipers, evaluators, unmasking bidders, bottom fishers, pioneers and late pioneers. A closer look is given to the group of bidders who turned out to win an auction, in order to determine whether these bidders are characterized by specific bidding decisions, and thus, by specific bidding behaviors.
The third stage focuses on the dynamic bidding process of an English auction, by specifically studying the impact of signals - namely, promotional messages - sent during an online auction on the final auction price. It proposes to test a model of the genesis and the impact of these messages aimed at informing current bidders and potential bidders about the item or urging them to submit a bid. This impact is modeled through a disaggregated and dynamic model. It exploits the recent behavioral view that each bid submitted by a bidder in an English auction is a particular decision that may be influenced by signals sent by the auctioneer during the auction, that is, between the decisions that the bidder makes. This model simultaneously takes into account the following three factors: (i) the direct impact of marketing messages on the auction price when messages affect bidders' valuations, (ii) the indirect impact of messages on this price when messages attract a new bidder to the auction, and (iii) the possibility that the auctioneer's strategy for sending messages depends on past events in the auction and on the timing of past messages. In this model, we thus propose that messages influence final auction prices through a dual-path system. The model also reflects an important feature of auction messages, which is the real time interactivity between the auctioneer and auction participants. The results, obtained through Bayesian inference, support the proposed model.
This doctoral dissertation globally helps better understand bidders' behavior in view of the rules set by the seller, and estimate how this seller can maximize his/her revenues. The main original aspects of this research consist in considering the English auction as a dynamic process and in focusing on bidders' behavior, these two elements being ignored by economists. From a managerial point of view, this research can help online auction designers to better design their auctions in order to maximize their revenue.
