Predicting Travel Time Variability for Cost-Benefit Analysis

The goal of this paper is to develop an econometric model that can be used to predict travel time variability for cost-benefit-analysis (CBA). The model explains travel time variability by the size of (mean) delays, as well as by other time-variant and invariant road characteristics. The data set used for the analysis is based on travel time data of 146 Dutch highway links for the year 2008. The resulting relationship between travel time variability and its explanatory variables can be used in transport-related cost-benefit-analyses. Up till now, most CBAs do not account for travel time variability, or only in very rough ways. One reason is the lack of straightforward ways on how to include travel time variability. While forecasts of mean travel times can be derived from traffic assignment models, this is usually not the case for travel time variability. The regression analysis shows that indeed a very significant relationship between travel time variability and delay can be found. It is also fairly robust against changes in the underlying dataset, for instance in respect to days of the week, months or weather conditions. Explanatory variables other than delay and flow-capacity-ratio contribute only little to the predictive power of the model.