AØKK08201U Mechanism design

Usually economists are interested in how individuals and organizations behave within their environment. For example, we ask how many hours an individual will work if he faces a certain tax system (the tax system and his employment possibilities are then the environment) or we ask how many units a firm will produce given the market environment (i.e. given the number of competitors, the mode of competition and the demand). Mechanism design goes one step further and asks: What kind of environment should a "designer" create if he wants to achieve a certain goal. For example, mechanism design asks: How should a government that is concerned about its citizens welfare design the tax schedule? In the second example, mechanism design could answer the question: How should a welfare maximizing planner organize a market?

Mechanism design is, therefore, an approach that can be used and has been used in many subfields of economics.

You can find a more thorough description of mechanism design on:

http://www.tseconomist.com/1/post/2013/01/-mechanism-design-theory-takuro-yamashita.html

 This course consists of three parts.

The first part (based on chapter 23 in MasColell/Whinston/Greene) introduces the students to the classic results and methods of mechanism design. After some introductory examples, dominant strategy mechanism design is treated: This covers the revelation mechanism and continues with the Gibbard-Satterthwaite theorem and the Groves-Clarke mechanism. We show that the designer can only achieve his objectives with the Groves-Clarke mechanism if he is willing to pay/receive own money to/from the players. As this is not always realistic, we turn from dominant strategy to Bayesian mechanism design to see whether we can get around this problem of "budget balance". The expected externality mechanism gives a positive answer to this question if players can be forced to participate in the mechanism.

We talk about some technical questions concerning incentive compatibility and use these new tools to establish the famous Myerson-Satterthwaite theorem which says that fully efficient trade cannot be achieved by any mechanism if players can opt not to participate. This naturally leads to the question which mechanism is most efficient. We study this question of optimal Bayesian mechanisms in several settings including bargaining, pricing, regulation and auctions.

 The second part applies and extends the concepts of the first part. The material is based mainly on published papers and small excerpts from other textbooks. We apply the standard model to market design in order to answer questions about innovation. Also the optimal design of an income tax scheme is discussed. We analyze how optimal mechanisms are affected if the setup differs from the classical mechanism design setup. Some of the following deviations from the standard setup are considered: Agents exert externalities on each other (e.g. if Pakistan sells nuclear weapons to North Korea, US security is affected), agents' information is correlated (e.g. if a government sells drilling rights either all companies will value the right highly if there is a lot of oil and not so highly if there is none), agents' information is multi-dimensional (e.g. buyers value design as well as functionality), agents can search for more information before signing a contract.

 The third part deals with recent, applied work of economists on so called matching markets. The classic Gale-Shapley algorithm is introduced. We then use this tool (and some related tools like the top cycle algorithm) to think about the design of the following markets: Matching students to schools/universities and organizing kidney donations. If there is time, we might cover another market like the market for physicians or power markets.

 Students are expected to do some assigned reading and smaller exercises between classes.