Modeling household fleet choice as a function of fuel price using a multiple discrete-continuous choice model

Abstract

In this Paper, a new model of fleet choice for households is presented that uses the multiple discrete-continuous extreme value (MDCEV) model as a framework. The aim of the model is to establish a model to allocate car types to activity based microscopic agent based transport simulations. What is new in the presented model, is that in addition to socioeconomic attributes of households, the choice is also influenced by fuel price. To model a range of fuel prices up to 20 USD/gallon a data base from a sophisticated stated adaption survey about mobility residential choice among approximately 400 Swiss households was used. The model had a choice set of 17 alternatives distinguishing car type and drive train. In the MDCEV model, a household chooses multiple car types and distributes an overall budget of vehicle miles traveled among the chosen alternatives. The model shows that fuel price has a much greater influence on the selection of the car type than on the use (VMT) of the car. In a certain range of fuel prices, households tend to switch from gasoline to diesel cars. The paper also contains an assessment of the residuals of the simulation that shows a reasonable performance of the model.