Cumulative Offer Processes

Stanford University, Stanford CA

Investigators

Abstract

In the past decade, several procedures devised by economic theorists have been employed for high value resource allocation problems. These include auctions for radio spectrum in many countries, electrical power sales, and carbon emissions reductions in the UK, as well as matching processes for physicians and psychologists. Most of the new processes are "cumulative offer processes:" multi-round processes in which offers either remain on the table or could remain on the table without affecting the outcome of the mechanism. Recent proposals to conduct "package auctions" (also known as "combinatorial auctions") for radio spectrum in the US and UK and to use auctions for train scheduling have met with success in laboratory experiments using other cumulative offer processes. The research proposed here is a detailed exploration of the theory of cumulative offer processes. I anticipate obtaining several kinds of results. These will include (1) a complete characterization of the market situations in which cumulative offer processes can guarantee stable ("core") allocations, (2) an extended analysis of the welfare properties of the processes, answering questions about who benefits most from which design, (3) the development of new algorithms that succeed with high probability for matching markets like the market for new medical residents, in which we know that no algorithm can always guarantee a stable outcome, (4) a unification of the theories of matching markets and auction markets, and (5) a game theoretic analysis of auction and matching processes when the participants have poor information about one another's values