Dynamic Voluntary Contribution to a Public Good:Learning to be a Free Rider

This paper explores the question whether boundedly rational agents learn to behave optimally when asked to voluntarily contribute to a public good. The decision process of individuals is described by an Evolutionary Algorithm. We find that the contribution level converges towards the Nash equilibrium although exact free rider-behavior is never observed. The latter result corresponds to findings from experiments on voluntary contribution to a public good. Crucial determinants of the learning process are the population size and the propensity to experiment.