Experimental Evidence and Selection

Abstract

The introductory discussion provides an example to illustrate the possibility of coordination failure due to the presence of multiple Pareto-ranked Nash equilibria. Of course, as remarked earlier, whether coordination failures actually occur depends on the selection of an outcome from the set of Nash equilibria. If, for instance, the Pareto-dominant Nash equilibrium is a natural focal point, as suggested, for example, by Harsanyi–Selten [1988], then understanding the macroeconomic implications of coordination failures would be somewhat less interesting. This chapter first reviews some recent experimental evidence that bears directly on equilibrium selection in coordination games. The evidence both concerns results on equilibrium selection and provides some insights into the process of equilibration. The second part of this chapter describes a variety of selection theories that bear directly on coordination games. One conclusion of the experimental evidence is that coordination problems are not a pure theoretical curiosity. In particular, coordination failures are routinely observed in experimental games. EXPERIMENTAL EVIDENCE The discussion of experimental evidence is partitioned into three parts. First, evidence from simple coordination games is presented. Given the frequency of coordination failures observed in these experiments, further treatments which explore possible remedies to coordination problems, such as preplay communication, are presented. Baseline Experiments As a starting point, consider the coordination game in Figure 1.1. This game was the focus of a study of experimental coordination games by Cooper, DeJong, Forsythe and Ross [1992]. In this game, there are two pure strategy Nash equilibria, {1, 1} and {2,2}, and a mixed strategy equilibrium.