Evolutionary Computation and Economic Models: Sensitivity and Unintended Consequences

The use of evolutionary models of complex adaptive systems is gaining attention. Such systems can generate surprising and interesting emergent behaviors. The sensitivity of these models, however, is often unknown and is rarely studied. The evidence reported here demonstrates that even small changes to simple models that adopt evolutionary dynamics can engender radically different emergent properties. This gives cause for concern when modeling complex systems, such as stock markets, where the emergent behavior depends on the collective allocation of resources of many purpose-driven agents.