Experiments in Location Games
Location Games Primer
The first chapter is an introduction to the field of location games—what has traditionally been referred to as spatial competition or location theory. The beginning of these models has been attributed to Hotelling, but the idea of these models predates that paper. I refer to the field as location games because experimental methods are now being introduced to the field to supplement the traditional theoretical analysis. The literature review covers the development of the field of location theory that is relevant to this research. Also included are some recent papers that begin to explore location games with experimental analysis.
The second chapter presents results of experiments on simple Hotelling location models. Results are compared to pure- and mixed-strategy Nash equilibria for the simple location problem with two to six sellers. For groups of two firms, there is clear evidence of communication for the purpose of influencing the other seller as part of a multiperiod strategy. However, whether the motivation for this signaling is to coordinate on a fair outcome or to deceive is unclear. For groups of three firms, there is evidence of risk aversion as players avoid riskier central locations. For groups of four and five firms, results do not readily attain the unique pure-strategy Nash equilibrium due to a previously unexplored coordination problem. In the four-player groups, players choose a combination of pure-strategy focal points and the mixed strategy support; however, players also choose the center of the market, which cannot be explained by theory. In the six-player case, there are two equilibria in pure strategies, providing an even more complex coordination problem. The distributions of strategies vary systematically as the number of seller increases in a manner consistent with Nash predictions.
Location Experiments with Communication
In the third chapter I introduce one-way communication in order to facilitate the solution of the coordination problem. This design has the potential to yield a cleaner test of the theory by removing the confounding coordination difficulty. I run groups of two to four firms in a fixed-matching repeated game. In these experiments, one subject in each group is designated to be the communicator. Prior to each round of play, the communicator sends a suggested location set to the other players. If the communicator understands the strategic situation, he can specify locations to the other players that result in an equilibrium outcome. Previous experiments suggest that this kind of “cheap talk” can facilitate equilibrium selection in coordination games. The results show some improvement in the ability of subjects to coordinate on Nash equilibrium play. However, there is considerable heterogeneity across groups. I conclude that one-way communication enhances the ability of subjects to coordinate, but that bounded rationality and poor leadership preclude the convergence to equilibrium.