Modeling Household Vehicle Transaction Behavior: Competing Risk Duration Approach

Household vehicle ownership analysis is an important and emerging topic in transportation research. Transportation researchers are interested in examining policy scenarios that can decrease the number of vehicles in a household fleet or encourage alternative transit. Many researchers have studied aspects of vehicle ownership and developed various aggregate and disaggregate models in different frameworks, including hazard-based duration models. Duration models can model the competing nature of vehicle transactions in a household and predict the timing of a change in vehicle ownership. Competing duration models are used to analyze the household vehicle ownership decision. Three vehicle transaction decisions (purchase, trade, and disposal of a vehicle) are considered. Dynamic variables that represent the state of the household and household location characteristics, an exogenous logsum variable representing maximum utility of vehicle class and vintage choice, and time-varying covariates (e.g., number of adults and household income) are considered in the model. Independent hazard models for each outcome and a competing hazard model are developed and compared