The Coordination Problem in a Structural Model of Peak-Period Congestion: An Experimental Study

The purpose of our paper is to generalize the model by Arnott et al. (1990a) to situations with a single origin and destination connected by two routes and to test the analytical results in an experiment with discrete time departure choices. The experimental evidence does not support the theoretical predictions: While the accumulated experience creates significant learning effects which imply a positive evolution of the empirical travel costs towards the equilibrium, the Nash hypothesis is still rejected and the commuting system shows substantial fluctuations until the end of the experiment.