Joint Modeling of Vehicle Holding Duration and Traveling Distance with a Copula-based Multivariate Survival Model

Abstract

Focusing on the household behavior of vehicle holding duration and annual traveling distance, this study proposes a Copula-based Multivariate Survival (CMS) model to capture the interdependence between these two behaviors. The copulas can be applied to the multivariate models to accommodate various correlation structures. This study applies the widely-used copulas in other research fields, including Normal, Gumbel, Clayton, and Frank copulas, and explores how these copulas could fit the data in this study. An empirical analysis is carried out based on web-survey data obtained in the Chugoku region of Japan in 2006. Model estimation results confirm the effectiveness of the suggested CMS model form both model performance and applicability. Comparative analyses of the different types of copulas support the use of the Clayton copula, and the relevant CMS estimates that vehicle holding duration and vehicle use are significantly correlated with each other.