Genetic Drift In A Model With Strategic Complementarities

This paper investigates evolutionary adaptation in a coordination game with strategic uncertainty. This game is characterized by the multiplicity of Nash equilibria that can be ranked according to the payoff that players obtain. Two different equilibrium refinement concepts predict the selection of different equilibria. Evidence from the experiments with human subjects suggests that the equilibrium selection depends on the number of players that take part in the game and on the number of repetitions of the game. In the model described in the paper, Nash equilibria are neutrally stable. This implies that any of the equilibria can be invaded by strategies that do not disappear from a population and can eventually, through the impact of genetic drift, take the population to a different Nash equilibrium. The results of simulations in which players use the genetic algorithm to update their strategies show that, regardless of the number of players that participate in the game, any equilibrium can be reached. The number of players has an impact on the time spent in each of the equilibria. In particular, the time spent in those equilibria that result in the higher payoffs is negatively related to the number of players.