Mechanism Design Experiments

Mechanism design experiments bridge the gap between a theoretical mechanism and an actual economic process. In the domain of public goods, matching and combinatorial auctions, laboratory experiments identify features of mechanisms which lead to good performance when implemented among boundedly rational agents. These features include dynamic stability and security in public goods mechanisms, transparency in matching mechanisms, package bidding, simultaneity and iteration in combinatorial auctions. Mechanism Design Experiments Mechanism design is the art of designing institutions that align individual incentives with overall social goals. Mechanism design theory was initiated by Hurwicz (1972) and is surveyed in Groves and Ledyard (1986). To bridge the gap between a theoretical mechanism and an actual economic process that solves fundamental social problems, it is important to observe and evaluate the performance of the mechanism in the context of actual decision problems faced by real people with real incentives. These situations can be created and carefully controlled in a laboratory. A mechanism design experiment takes a theoretical mechanism, re-creates it in a simple environment in a laboratory with human subjects as economic agents, observes the behavior of human subjects under the mechanism, and assesses its performance relative to what it was created to do and relative to the theory upon which its creation rests. The laboratory serves as a wind-tunnel for new mechanisms, providing evidence which one can use to eliminate fragile ones, and to identify the characteristics of successful ones. When a mechanism is put to test in a laboratory, behavioral assumptions made in theory are most seriously challenged. Theory assumes perfectly rational agents who can compute the equilibrium strategies via introspection. When a mechanism is implemented among boundedly rational agents, however, characteristics peripheral to theoretical implementations, such as transparency, complexity and dynamic stability, become important, or even central, to the success of a mechanism in a laboratory, and we suspect, ultimately in the real world. Mechanism design experiments cover several major domains, including public goods and externalities, matching, contract theory, auctions, market design and information markets. In what follows, we will review experimental results of some of these topics.