Auctions for Multiple Goods, and Understanding Consumer Preferences

点击翻译按钮获取中文摘要

In 2007, responding to the Northern Rock bank run--Britain's first bank run since the 1800's-- Klemperer invented the "Product-Mix Auction", to improve the Bank of England's allocation of funds to banks, building societies, etc. (Klemperer, 2008, 2010). An Executive Director of the Bank of England described it as "a world first in central banking" and "potentially a major step forward in practical policies to support financial stability" (Milnes, 2010), and the Governor (Mervyn King) told The Economist (Oct. 19, 2012) that "Klemperer auctions [are] a marvellous application of theoretical economics to a practical problem of vital importance to financial markets". The auctions improve efficiency by simultaneously allocating distinct goods - in this example, funds against different collaterals. This project will further develop the Product-Mix Auction design to make it a practical tool in a broader range of contexts. For example, versions of the auction could be used for batch processing of stock trades to eliminate high-frequency trading problems. They could also be important in achieving the U.K.'s goal "to start auctioning electricity-supply contracts. ... by the 2020s,...across technologies. 'We would be the first country in the world to achieve that', boasts [the U.K. Cabinet Minister for Energy and Climate Change]" (The Economist, Dec. 1, 2012). The theoretical work required to extend the range and flexibility of the auction will also shed light on a much wider range of issues. For example, our work will develop a better characterisation and understanding of consumer preferences when goods are indivisible, yield answers to some deep problems about the existence of economic equilibrium, and have applications to matching theory. We will use recently-developed "tropical geometric", and other geometric, techniques as well as more standard economic tools to develop a better understanding of economic agents' preferences. We will develop our theory of "demand types" which provides a framework to check and compare the properties of agents' demands, and develop ways to depict preferences of different "types". The project will thus have both theoretical and practical outputs: in addition to preparing academic papers for submission to leading economic theory journals, and presenting at academic conferences, we will also discuss our work with potential users of our auction designs, and will develop software implementations of our auction procedures that will be easy to use for demonstration and experimentation purposes and provide a platform on which problem-specific applications can be developed.