Using genetic algorithms to model the evolution of heterogeneous beliefs

Genetic algorithms have been used by economists to model the process by which a population of heterogeneous agents learn how to optimize a given objective. However, most general equilibrium models in use today presume that agents already know how to optimize. If agents face any uncertainty, it is typically with regard to their expectations about the future. In this paper, we show how a genetic algorithm can be used to model the process by which a population of agents with heterogeneous beliefs learns how to form rational expectation forecasts. We retain the assumption that agents optimally solve their maximization problem at each date given their beliefs at each date. Agents initially lack the ability to form rational expectations forecasts and have, instead, heterogeneous beliefs about the future. Using a genetic algorithm to model the evolution of these beliefs, we find that our population of artificial adaptive agents eventually coordinates their beliefs so as to achieve a rational expectations equilibrium of the model. We also report the results of a number of computational experiments that were performed using our genetic algorithm model.