Deterministic Chaos in an Experimental Economic System

An experiment with a simulated macroeconomic system demonstrates that the decision-making processes of agents can produce deterministic chaos. Subjects managed capital investment in a simple multiplier-accelerator economy. Performance, however, was systematically suboptimal. A model of the subjects' decision rule is proposed and related to prior studies of dynamic decision making. Econometric estimates show the model is an excellent representation of the actual decisions. The estimated rules are then simulated to evaluate the stability of the subjects' decision processes. While the majority of the estimated rules are stable, approximately 40% yield a variety of dynamics including limit cycles, period multiples, and chaos. Analysis of the parameter space reveals a complex bifurcation structure. Implications for models of human systems and experimental studies of economic dynamics are explored.