This chapter describes an experiment with human decisionmaking behavior in simulated microeconomic environments. Participants were asked to operate a simplified production- distribution chain to minimize costs. Performance was systematically suboptimal, however, and in many cases the subjects were unable to secure the stable operation of the system. As a result, large-scale oscillations and various forms of highly nonlinear dynamic phenomena was observed.

A model of the applied ordering policy is proposed. Econometric estimates show that the model is an excellent representation of the actual decisions. With different parameters, computer simulations of the estimated order policy produce a great variety of complex dynamic behaviors. Analyses of the parameter space reveal an extremely complex structure having a fractal boundary between the stable and unstable solutions, and with fingers of periodic solutions penetrating deeply into regions representing quasiperiodic and chaotic solutions. In certain parts of the parameter space, any neighborhood of a given solution contains a qualitatively different solution. Thus, changes on the margin can produce a completely different system behavior.

Our results provide direct experimental evidence that chaos can be produced by the decisionmaking behavior of real people in simple managerial systems. The consequent implications for the ability of human subjects to cope with complex dynamical systems are explored.