A "matching mechanism" is the term used by economists to describe the procedures and rules that are used to match applicants to openings in a variety of applications. Examples include matching applicants to schools (especially placing students in public schools under various school choice programs), placing medical residents in training programs, and placing applicants in jobs. This project will study dynamic matching mechanisms-procedures that allow the participants to revise and extend their reported preferences during the matching process. Surprisingly, nearly all centralized matching procedures in use today are static mechanisms (akin to sealed-bid auctions). However, dynamic matching mechanisms should be expected to have advantages over static matching mechanisms in many environments, for the same reasons that dynamic auctions are preferable to sealed-bid auctions. First, when auctioning items for which participants' preferences are interdependent, the feedback about rivals' bids provided by a dynamic auction may improve outcomes over a sealed-bid auction. In particular, the additional information which bidders are able to infer concerning the items' valuations helps to enhance the efficiency of allocations, and the reduction in the winner's curse provided by the dynamic process increases the willingness to participate in the process. Similar considerations apply in matching markets with interdependent preferences. Second, when auctioning many related items, some discovery of likely outcomes is needed to promote effective preference elicitation, as participants' own preferences are not known in advance, but must be discovered through costly analysis. An iterative auction process allows participants to focus their valuation analysis on the most relevant portions of the outcome space. Again, similar considerations apply in matching markets where preference discovery (e.g., job interviews) is also costly.
The project will build upon recent work in dynamic auctions and apply it to matching problems. The research will contribute both to the theory and practice of matching markets. One of the most fruitful areas of auction research has been the comparison of static auction mechanisms with dynamic auction mechanisms and the development of new dynamic auction designs that attain improved allocative efficiency or revenues. Moreover, dynamic auction procedures have gained considerable market share as compared to static auctions in recent years. In this light, it is particularly surprisingly that nearly all centralized matching procedures in use today are static mechanisms, and that the existing theoretical analysis focuses mostly on static matching mechanisms. The project will attempt to initiate a new research program, analogous to the study of dynamic auctions, for the matching literature.
Auctions and matching markets have played an increasing role in improving the allocation of scarce resources. For auctions, this is seen in many applications such as telecommunications spectrum, energy, and transportation, and for matching, this is seen in applications such as the medical resident market and school choice programs. The principal investigators' prior research has already had a major impact on the design of electricity, spectrum and environmental auctions in the US and Europe. In a current project for the Federal Aviation Administration, the principal investigators are attempting to design an efficient auction for landing and takeoff slots at New York's LaGuardia Airport. The proposed project is likely to lead to insights that will similarly improve the working of real-world matching mechanisms.
