Responsive state-dependent or habitual state-independent congestion pricing under dynamic congestion

In the face of capacity disruptions (due, for example, to traffic incidents or poor weather), information provision and congestion pricing are alternative alleviating policies. A state-dependent toll equals the state-dependent marginal external cost (MEC), which is higher if traffic condition is in a bad state. This raises efficiency and thus welfare, but it may also be even more unpopular with the populace than stateindependent tolling. We study this using dynamic bottleneck congestion with an uncertain capacity that can have two states: high or low. We consider two congestion pricing regimes: responsive state-dependent congestion pricing and habitual stateindependent pricing, and three information provision regimes: no information, perfect information and imperfect information. We find that, without information provision, the habitual toll equals the expected MEC. With information provision, this is a weighted average of the MEC over all states; with weights depending on the capacity distribution, the price sensitivity of demand, the values of schedule delay and the quality of the information. Responsive pricing leads to higher welfare and a lower expected price than habitual pricing, but in our numerical model the differences tend to be small. When only one policy is implemented, information provision and congestion pricing both raise welfare. Information provision is preferable when uncertainty is high, as information is more valuable at this time.