Longitudinal Vehicle Transaction Model: Assessment of Lead and Lagged Effects of Longer-Term Changes and Life-Cycle Events

Abstract

This paper presents a dynamic vehicle transaction model developed with data from a retrospective survey conducted in Halifax, Canada. The study investigated four types of vehicle transaction decisions, including first-time vehicle purchase, vehicle acquisition, vehicle disposal, and vehicle trade. A panel-based latent segment logit (PLSL) model was developed to account for repeated transaction decisions and to capture unobserved heterogeneity among the sample households. The study took a life-oriented approach by examining the lead and lagged effects of longer-term changes and life-cycle events. The PLSL model was estimated for two latent segments, and the model results suggest that profound heterogeneity exists, as evident in parametric values of the two segments. The life-cycle event represented by the birth of a child might trigger vehicle acquisition in Segment 1. In contrast, that life-cycle event might deter an acquisition decision in Segment 2. The model confirmed historical deposition effects of key life events. For instance, a 2-year lagged effect on vehicle acquisition was seen with the birth of a child. Moreover, the model results revealed that first-time vehicle purchase behavior was considerably different from the decision to acquire a vehicle. For example, for the longer-term change represented by the addition of a job, significant heterogeneity across the two segments for a first-time vehicle purchase decision was seen and a 3-year lagged effect was confirmed. A positive relationship for the decision to acquire a vehicle in both segments was found for the same variable, and a 1-year lagged effect was confirmed. Finally, the study provided important behavioral insights for targeting specific groups of the population to promote sustainable travel behavior.