Inefficiency of marginal-cost tolls in transportation networks with stochastic demands

This Congestion tolls provide an efficient way to mitigate traffic congestion. In deterministic networks, an optimal routing of traffic can arise as a user equilibrium by levying marginal-cost toll. In reality, traffic demands vary day in day out. The uncertainty of demand brings difficulties to design transport policies in theory and to implement the traffic coordination in practice. The expected travel budget is defined by involving the mean marginal-cost toll into the expected travel time, and a equilibrium model is established. We compare the expected social cost of the equilibrium and the system optimum to find the efficiency loss of the marginal-cost toll. It is found that the network performance degrades as demand variation goes up. Thus the control of demand variation could be an effective assistant tool to improve the efficiency of congestion pricing policies.