Joint Modeling of Household Vehicle and Activity Allocation

Abstract

Traffic is a major problem, even in countries with low per capita vehicle ownership. Governments often use vehicle restriction measures to combat severe traffic, especially in developing countries. However, such restrictions can disrupt economic efficiency as well as the daily lives of citizens. The discipline of transportation modeling has emerged as a response to these problems. In particular, activity-based travel demand models show great promise. These models require an accurate consideration of the social factors involved in travel behavior; however, recent studies have only considered intrahousehold relationships separately from vehicle usage. This paper describes the development of a joint model for household activity and vehicle allocation. A paired combinatorial logit model was used to estimate the share of each household member in nonmandatory activities. A vehicle allocation model is a multinomial logit model that assigns an allocation probability to each household member. The two models were estimated by a joint modeling framework. The model was estimated with data collected from three cities in Iran. A strong relationship was found between the activity allocation pattern and gender, the role of the individual in the family, employment status, time of day, and duration of the activity. These effects were especially pronounced in households with young children. Incorporating demographics into transportation models can more robustly model vehicle usage behavior. This study focused on Iranian cities, but the results can be extended to other countries with similar characteristics.