Implications of Congestion Charging for Morning Commute Choice in Beijing, China: Discrete Choice Model of Combined Departure Time and Route

The objective of this paper was to investigate potential impacts of implementing variable congestion charging on the peak spreading of departure time choices as well as traffic distribution between toll and free routes. This study focused on the departure time and route choice process and investigated various discrete choice models for a combined departure time and route choice. Differing model structures and varied random parameter distributions were compared in order to capture correlations among alternatives and taste variations across commuters. The inter-alternative correlations hypotheses tested in this paper were that alternatives may be correlated due to paying for various congestion charges or may be correlated by the amount of delay. The taste heterogeneity hypotheses tested were that the perceived effects of congestion charging and late arrival penalty among commuters may be different. Compared with the basic Multinomial Logit model (MNL), the estimation results indicated that Nested Logit model (NL) and Random Parameter Logit model (RPL) improved the model performance, due to taking into account inter-alternative correlations and commuters’ heterogeneity respectively. The combined choice models were used to present commuters’ trade-offs among commute time, monetary cost and schedule delay. The results showed that under the situation without flexible work patterns, the disutility of late arrival and early departure were high, which resulted in less attractiveness for commuters to switch their departure time and further weaken the performance of implementing congestion charging. Therefore, in order to enhance the performance of a congestion charging policy for relieving traffic congestion in a community, elastic work patterns as well as other complimentary measures should be introduced in parallel.