Abstract:In a transportation system with multiple equilibria, coordination failure occurs when a group of travelers could achieve the efficient equilibrium but fail to because they cannot coordinate their decision-making. It is evident that coordination failure will result in an unnecessary loss of life and property. Therefore, understanding the factors that affect the coordination behaviors in multi-equilibrium systems is crucial for working out policy measures to improve the transportation system efficiency. Here, via a series of laboratory experiments, we investigate three factors that have significant impacts on the coordination behaviors in a directed transportation network with both positive and negative externalities. Subjects who participated in the experiments were required to choose one travel mode from public transportation and private transportation in each round. Public transportation has positive externalities in which the travel costs decrease with the increase in the number of subjects who have the same choice, whereas the externalities of private transportation are negative induced by congestion. Subjects played the two-route network game under either a one-stage or two-stage framework and with two sets of different parameters in public transportation. Through the experiment results, we find (i) the transparency of descriptive information facilitates not only the convergence toward the efficient equilibrium but also the stabilization of route choice behaviors; (ii) the group decisions can converge toward the efficient equilibrium in a greater possibility when more subjects prefer public transportation in the initial state; and (iii) 'higher' barrier between the two equilibria impedes and prolongs the process of equilibrium transformation from the inefficient equilibrium to the efficient equilibrium. Finally, we build a Mixed Logit Model at the group level to further demonstrate the significant impacts of the three factors on coordination behaviors and reproduce the aggregate mean route choice behaviors in simulation results.