UC Berkeley

In an era of fluctuating energy costs where we observe increasing concerns for environmental protection, energy sources and energy efficiency, it is of great importance to understand the mechanisms underlying consumer vehicle choices. This dissertation develops a theoretical model of vehicle choice by embedding the disaggregate indirect utility function model of household choice in the aggregate framework of the threshold model of adoption/diffusion. This approach incorporates multiple sources of heterogeneity including household income levels, household structure, comfort/quality levels and attitudes towards environmental awareness. Then, we examine the diffusion of vehicle adoption through dynamic processes such as learning-by-doing (Arrow 1962). In addition, the simulation model based on this theoretical model enables forecasts particularly suited for examining both household level adoption as well as overall diffusion of different vehicle technologies with relatively few calibrated parameters, especially in dynamic economies such as China or India.

First, this dissertation starts out with the threshold model of diffusion introduced by David (1969) and Sunding and Zilberman (2001), applying the theoretical foundation commonly used in analyzing modern irrigation technology adoption to household vehicle adoption. Households are assumed to be heterogeneous and utility-maximizing through a discrete vehicle choice and a continuous choices of miles traveled, and they adopt the technology that provides the highest utility given the optimal number of miles chosen for that particular technology. Then, the theoretical model is extended to include both parameters of vehicle comfort/quality and environmental awareness in the choice process. Another extension of the theoretical model includes household heterogeneity through the inclusion of household structures where each household structure may have their own preference set for vehicle comfort/quality. On the aggregate level, the aggregate flow demand of vehicles at time t is derived and determined to be a combination of these effects: the population growth rate (Population Effect), the shift of income distribution within the population (Income Distribution Effect) and the movement of the critical income levels for each type of vehicle (affected by characteristics of each vehicle technology as well as preferences of households including the Variable Cost Effect and Fixed Cost Effect).

Next, we develop a computer simulation model with the theoretical household vehicle choice threshold model as foundation. Utilizing a CES utility functional form as the starting point, we calibrate the simulation model using data from various sources, including data of income distribution, vehicle attributes and pricing, vehicle sales data in the U.S. and environmental awareness factors. The final calibrated specification yields an R^2 equal to 0.9595, indicating the simulated results explains approximately 95.95% of the variance in historical vehicle sales data. Using this simulation model, we forecast the influence of various factors on vehicle adoption patterns and optimal miles traveled by households. We find that changes in the fixed cost of vehicles (influenced through government policies regarding rebates or through learning-by-doing) and shifts in income distribution (including both income distribution shifts and shape changes) present particularly dominant effects on vehicle adoption compared to changes in energy price, environmental awareness or vehicle comfort/quality. The main conclusion here is that as parameters are changing, households are not only changing what type of vehicle they prefer to purchase, they also adjust how much they would like to travel.

The theoretical model and the simulation model culminate in Chapter 5 where three very different case studies illustrating potential scenarios in emerging markets, ageing economies and public transportation are presented. Case Study I presents the scenarios of rapidly growing economies with rapidly changing population dynamics such as China or India. The results from this case study illustrate the potential trend towards the adoption of bigger, better (more comfortable) and newer vehicles which are less fuel-efficient and more polluting as the economy experiences high income growth and increases in inequality. The extended scenarios presented in Case Study I incorporates the increasing trend of transportation infrastructure construction. Two effects are hypothesized in this extension: increasing comfort levels due to better roadways and networks of roadways, and decreasing comfort levels due to increasing congestion or pollution caused by the increasing stock of vehicles and miles traveled. The other extension in this portion proposes increasing environmental awareness in a developing economy as it is growing, possibly due to the Environmental Kuznets Curve. Both increasing overall comfort levels and increasing environmental awareness lead to the high rates of adoption of energy-efficient vehicles such as the new hybrid vehicle or the new compact vehicle.

Case Study II of Chapter 5 examines the ageing economies of Japan and Europe. The population dynamics are also shifting in these regions, although at a slower rate, towards a larger percentage of childless households and senior households. The substitution towards energy-efficient vehicles is driven by a combination of increased environmental concern, increasing energy prices, decreasing cost of new hybrid vehicle technology as well as increasing income levels. The effects caused by the larger percentage of childless households and senior households depend on the magnitude of each as well as on the income distribution shape within each household structure.

Case Study III takes the simulation model one step further by introducing public transportation as an alternative travel mode. Public transportation is unique because it entails zero private fixed cost but with the tradeoff of higher variable cost per mile traveled compared to driving a private vehicle. In addition, public transportation provides lower levels of comfort/quality/convenience as well as higher utility from environmental awareness. Unsurprisingly, we find that public transportation ridership benefits from high energy price increases and low public transit fares. We also observe that increasing public transit fares not only decreases the overall number of households who ride public transit, but also induces them to ride less. This case study is extended to incorporate the income distribution shifts in economies such as China. We find that public transportation usage declines and new SUV adoption rapidly increases as the population experiences income increases, demonstrating the dominant effect of the income distribution shift on the adoption of vehicle technologies.