Energy consumption and greenhouse gas (GHG) emissions are at their highest levels in history. One of the largest sources of GHG emissions in the United States is from burning fossil fuels for transportation. In developing countries GHG emissions from private vehicles are growing rapidly with their wealth. Government agencies attempt to reduce dependency on fossil fuels by regulating the ownership/usage of private vehicles, promoting vehicles with higher engine efficiency, introducing new fuel types, and defining stricter emission standards. Hybrid and electric vehicles are gaining consumers’ interest and trust, and their sale shares are gradually increasing. Meanwhile, environmental awareness, taxes on conventional gasoline cars, and incentives for cars with new technologies, make small and alternative-fuel vehicles more attractive. The future of personal transportation is uncertain; in particular, car ownership, vehicle type preferences and usage behavior are likely to change in surprising ways. In this context, it is important to assess the influence of the vehicle market evolution on consumer’s vehicle demands and travel behaviors.

This dissertation proposes a comprehensive modeling framework that is able to analyze different dimensions of the car purchasing and usage problem. A multi-facet approach is taken for the investigation, and different model types are proposed. The investigation starts with a mixed logit model that accounts for time-series choices, heterogeneity in preferences and correlation across alternatives. This model is estimated on Stated Preference Survey data collected in Maryland and quantifies market elasticities and willingness-to-pays for improving car characteristics. Afterward, a dynamic discrete choice model is developed to predict the diffusion of hybrid and electric cars in Maryland, with consideration of household’s forward-looking behavior and stochasticity in vehicle market evolution. This model focuses on vehicle purchase time and vehicle type choice. To further consider vehicle usage decision, an integrated discrete-continuous choice model is proposed to jointly estimate household’s discrete choices on vehicle type/ownership and continuous choice on vehicle usage. The model is applied to estimate household-level vehicle emissions in Maryland, USA and Beijing, China.

The dissertation concludes with a sequential discrete-continuous choice model. The modeling framework is applied to estimate vehicle ownership and usage decisions of forward-looking agents over time in a finite time horizon. In particular, a recursive probit model is formulated to estimate a sequence of vehicle holding decisions, while a regression is used to estimate a sequence of vehicle usage decisions. The proposed model is tested and validated on simulated discrete and continuous choices in a car ownership problem setting.

The dissertation contributes to the theory of dynamic models for discrete-continuous decisions. The sequential discrete-continuous choice model is the first to measure the dynamic interdependency between discrete choice and continuous choice over time. The dissertation also contributes to the understanding of critical transportation issues, including market penetration of new vehicle technology, estimation of household-level vehicle emissions, and policy evaluation for promoting green vehicles and reducing dependency on cars and emissions.