Computerized Double-Auction Markets: Some Initial Experimental Results

Abstract

Over the last 2 decades there has evolved a considerable volume of literature describing the results of various experimental games in market decision making as well as their implications concerning economic theories of market price and quantity determination (see, e.g., Smith 1962, 1964, 1965, 1976b; Miller, Plott, and Smith 1977; Plott and Smith 1978; Williams 1979). The relevance of using controlled laboratory experiments to study resource allocation mechanisms has recently been addressed formally by V. L. Smith (1980), who has also provided fine summaries of the more important experimental results derived from contract price observations generated under various market institutions (Smith 1976b, 1980). The potential for increased efficiency obtainable through various auction mechanisms has been recognized for many years (Cassidy 1967). However, recent discussions of automating and computerizing auction markets have generally been in relation to the creation of a congressionally mandated national securities market (e.g., Wall Street Joutrnal 1978a, 1978b, 1978c). Charged with overseeing the developIn 1975 the Securities and Exchange Commission received a congressional mandate to move toward the creation of a national stock trading system. A potentially major step toward the development of such a system is the Cincinnati Stock Exchange's highly controversial pilot project which is currently testing an all electronic computerized trading mechanism. This paper reports on the design and analysis of a series of laboratory experiments which explore the behavior of computerautomated doubleauction markets where face-to-face buyer-seller interaction has been eliminated. The experiments also serve to demonstrate the use of the PLATO computer as a research tool for social scientists interested in laboratory experimental tech-