Experiments in economics consist of three elements: an environment defining the payoffs motivating choice; an institution defining the language and rules governing communication, actions and outcomes; behavior, given the environment and institution. Examples from game theory, auctions, and market exchange are used to illustrate this tricotomy. Five research topics illustrate some questions and learning from experiments: noncooperative behavior is universal in market experiments, but much less prominent in two-person anonymous interaction games—an anomaly that can be understood in terms of reciprocity in personal exchange; institutions matter because the rules affect incentives that affect behavior; stock markets in the laboratory produce price bubbles that are extinguished by experience as subjects gradually acquire common expectations of intrinsic value; rewards and decision cost both matter and can be modeled in a way that is testable explaining, for example, why increased rewards, or experience, sometimes substantially improve performance; finally, experiments are now regularly employed to inform the design of complex electronic trading systems in the field—pollution credit rights, spectrum auctions, gas pipeline, and electric power networks. These smart computer-assisted markets apply optimization algorithms to the decentralized messages (location-specific bids to buy, sell, and transport), combining the information advantages of decentralization with the coordination advantages of central processing in a rule-governed system.
