Analysis of Urban Travel Times: Hazard-Based Approach to Random Parameters

The traditional approach to the analysis of urban travel includes a detailed and complex modeling system that considers activity and trip generation (including time-of-day of trip) and destination, mode, and route choices. Although the insights that one can gain from such a comprehensive approach are undeniable, a more simplistic approach that focuses on travel time alone (which by its nature implicitly includes the complex decision making related to destination, route, activity, and time-of-day choices) can also provide valuable information for policy makers on the determinants of congestion, the behavior and destination choices of travelers, and so on. However, in such an approach, the unobserved heterogeneity that is introduced by simplifying a complex decision-making process must be addressed. In this paper, the determinants of travel time to destination for an urban area are studied while unobserved heterogeneity is explicitly accounted for with hazard-based duration models with random parameters. With an extensive geocoded trip data set from Athens, Greece, time-to-destination model estimation results indicate that travel time duration is affected significantly by a number of factors, such as sociodemographic and trip characteristics, travel mode, frequency of trip, and time of day of the trip. In addition, the effect of many of these factors was found to vary across the population and thus underscore the need for a random-parameters formulation in the study of urban travel times with this approach.