Modelling reliability benefits

Abstract

This paper compares the performance of several models forecasting travel time variability for road traffic, using before/after data from the introduction of the Stockholm congestion charges. Models are estimated on before-data, and the models’ forecasts for the after-situation are compared to actual aftermeasurements. The accuracy of the models vary substantially, but several models are able to forecast the benefits from reduced travel time variability with sufficient accuracy to make them useful for decision making.