Application of a Gumbel Multivariate Distribution in a Competing Duration Hazard-based Vehicle Transaction Decision

Vehicle ownership has been the subject of many earlier studies, which have adopted a variety of econometric frameworks at both aggregate and disaggregate choice modeling level. Car manufacturers, oil companies, Metropolitan Planning Organizations (MPOs) and environmental agencies are among many organizations interested in using accurate aggregate and/or disaggregate vehicle ownership models. Among various modeling approaches, dynamic disaggregate models such as competing duration-based models, seem to provide better modeling fit with more effective forecasting capabilities relative to other vehicle ownership modeling frameworks. This is in large part due to their capability in jointly modeling transaction timing and type. Traditionally transaction timing has been overlooked in the commonly used static vehicle ownership models despite the significant role it plays in explaining vehicle ownership behavior. Transaction failure timing, like many other time related failures such as unemployment duration, is modeled by using duration risk based models. Additionally, the decision about transaction type is typically assumed to be independently made; therefore, independent timing models are developed for each transaction type. In other words, the inclusion of inter-correlation among the error terms of the vehicle transaction type decisions is typically ignored in the competing duration models due to the significant computational burden that it imposes on the model. This study aims to introduce a joint transaction type and timing model at the disaggregate level using a discrete competing proportional hazard model. The inter-correlation among multiple transaction types is modeled in this study by utilizing a Gumbel density function to approximate the multivariate probability density function among the transaction types’ error terms. Household transaction decisions are assumed to occur in discrete time intervals and are modeled as a generalization of the Han and Hausman (‎1) formulation. More specifically, the formulation is extended to three transaction decisions instead of the previously introduced binary application. Additionally, unlike their work, the proportional hazard specification is directly estimated in this study while in their work, they approximated it with an ordered probit specification.

• Supplemental Notes:
  • The DVD lists the title of this paper as: Application of Gumbel Multivariate Distribution in a Competing Duration Hazard-Based Vehicle Transaction Decision.
• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Rashidi, Taha H
  • Mohammadian, Abolfazl
• Conference:
  • Transportation Research Board 90th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2011-1-23 to 2011-1-27
• Date: 2011

Language

• English

Media Info

• Media Type: DVD
• Features: References; Tables;
• Pagination: 20p
• Monograph Title: TRB 90th Annual Meeting Compendium of Papers DVD

Subject/Index Terms

• TRT Terms: Automobile ownership; Behavior; Decision making; Disaggregate analysis; Metropolitan planning organizations; Multivariate analysis; Traffic forecasting; Transportation planning
• Uncontrolled Terms: Automobile manufacturers; Hazard-based duration models; Timing; Transaction data
• Subject Areas: Highways; Planning and Forecasting; I72: Traffic and Transport Planning;

Filing Info

• Accession Number: 01342496
• Record Type: Publication
• Report/Paper Numbers: 11-2457
• Files: TRIS, TRB
• Created Date: Jun 23 2011 9:06AM