Traffic congestion costs money and welfare. The first initiative to quantify monetary costs in the Netherlands goes back to the late 1970s. The method was simple: estimate total queue length throughout the year, and multiply this by the estimated average delay per km queue and value-of-time to obtain annual congestion cost. Since then more accurate estimates have been obtained and reported annually by the Dutch Ministry of Transport, but essentially the method has not been changed. The key assumption behind it is that observed queues allow us to accurately estimate the total disutility caused by the traffic congestion. Behavioural changes induced by congestion, like changed timing of travel, changes in mode choice, destination choice and in other decisions made by (potential) travellers, are also a part of the costs of congestion. As the time delays on roads get worse, one might expect more and more behavioural changes. In such situations, the observed delays may seriously underestimate the true congestion costs. This is relevant both for ex-ante economic evaluations, using prediction models, and ex-post evaluations, using observed congestion. The magnitude of a number of possible effects are illustrated by using simulation results from the Dutch National Model System. Ways of obtaining better ex-ante evaluations by means of relatively simple modifications of existing software are described. For the covering abstract see ITRD E126595.
Abstract