Application of Game Theory in Spectrum Auctions
What we learned in class about auction theory is very much just the skeleton basics, and real life auctions can be much more complicated to theorize about as one considers the nature of the items being auctioned, the mindsets of the bidders, and the objectives of the auctioneer. One example of a large-scale complex auction guided by game theory is the spectrum auction held by the Federal Communications Commission (FCC). Social scientists Robert Wilson, Paul Milgrom, and Preston McAfee all had done significant research in the field of game theory, so naturally the FCC used their research on competitive bidding and auction theory to draft their own auction process. The FCC had the problem of there being limited radio spectrum, and high and fast growing demand for it from different industries and companies, especially phone companies using it for cellular data.
When Milgrom saw the FCC’s draft, he allegedly said he could design a much better auction, and together with his old faculty advisor at Standford, Robert Wilson, developed an auction process called a simultaneous multiple round (SMR) auction, also known as a simultaneous ascending-bid auction. Preston McAfee, who was then a professor at the University of Texas, concurrently developed a similar idea. Wilson and Milgrom contributed the fundamental idea that all of the individual auctions should conclude simultaneously, while McAfee’s work addressed fundamental issues, such as dealing with bidder defaults and ensuring participation by minority bidders. The first auction for radio spectrum by the FCC using the SMR process occurred in 1994, and the process has since been the basis for many spectrum auctions across the world.
In 2000, Milgrom published a paper “Putting Auction Theory to Work: The Simultaneous Ascending Auction,” which analyzes the capabilities and limitations of that auction model from the perspective of its application in spectrum auctions. SMR is an auction for multiple items in which bidding occurs in rounds. Milgrom addresses many practical questions in his paper, and concludes that current theory offers no answer to many of those questions, such as the “bounded rationality” constraints that limit the generalized Vickrey auction. He states that auction design is much an engineering activity, guided by theory and evidence but dependent on ad hoc problem solving when there is no feasible theory.
Auctions become very complicated as you consider the nature of the items being auctioned. Are they substitutes for each other? Are they compliments? What if small lots of items are valued highly when auctioned off together to the same bidder, but individually have low value? How do you ensure an auction terminates in a reasonable amount of time? All of these questions are subjects of research, and there remains much to be learned in game theory. Milgrom’s paper gives an insightful overview of auction theory in an economic context, as well as addresses remaining uncertainties and areas of contention.
http://web.stanford.edu/~milgrom/publishedarticles/Putting%20Auction%20Theory%20to%20Work.pdf