Investment Behaviour under Knightian Uncertainty - An Evolutionary Approach

The paper analyzes investment behaviour under Knightian uncertainty by means of a genetic programming algorithm. This is an experimental approach which yields analytical results at a level of generality comparable to that obtained by conventional methods. When the artificial agents receive the same information about the unknown probability distributions, they develop behaviour rules as if they were expected utility maximizers with Bayesian learning rules and logarithmic utility functions. We then introduce asymmetric information, and study how it affects the agents' implicit preferences for risk and uncertainty.