The dynamics of morning rush-hour traffic congestion have been studied extensively in recent years. In most theoretical work, however, commuters are assumed to have identical travel cost functions and to face the same arrival-time constraints at work. In this paper, we allow commuters to differ in their travel time costs, their starting time at work, and the costs incurred from early and late arrival. Early in the rush hour, the departure rate exceeds road capacity, causing a queue to develop. Commuters order themselves systematically in the departure sequence to minimise their individual travel costs. The order in which different groups depart is not necessarily efficient. A time-varying congestion toll can be constructed to eliminate queuing and induce the optimal order of departure. Travel-cost savings from such congestion tolls and from road capacity investments are computed. Estimated benefits are generally biased if computed using average travel-cost parameters and average work start times rather than actual distributions. Savings tend to be overestimated if commuters differ primarily in travel costs, but can be underestimated for capacity investments when commuters differ primarily in starting times at work.
Abstract