THE CONSUMER RESPONSE TO GASOLINE PRICE CHANGES: EMPIRICAL EVIDENCE AND POLICY IMPLICATIONS A DISSERTATION SUBMITTED TO THE DEPARTMENT OF MANAGEMENT SCIENCE AND ENGINEERING AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Kenneth Thomas Gillingham June 2011 © 2011 by Kenneth Thomas Gillingham. All Rights Reserved. Re-distributed by Stanford University under license with the author. This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/ This dissertation is online at: http://purl.stanford.edu/wz808zn3318 ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. James Sweeney, Primary Adviser I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. Lawrence Goulder I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. Matthew Harding I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. John Weyant Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost Graduate Education This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives. iii Abstract When gasoline prices rise, people notice: the news is filled with reports of pinched household budgets and politicians feeling pressure to do something to ameliorate the burden. Yet, raising the gasoline tax to internalize externalities is widely considered by economists to be among the most economic efficiency-improving policies we could implement in the transportation sector. This dissertation brings new evidence to bear on quantifying the responsiveness to changing gasoline prices, both on the intensive margin (i.e., how much to drive) and the extensive margin (i.e., what vehicles to buy). I assemble a unique and extremely rich vehicle-level dataset that includes all new vehicle registrations in California 2001 to 2009, and all of the mandatory smog check program odometer readings for 2002 to 2009. The full dataset exceeds 49 million observations. Using this dataset, I quantify the responsiveness to gasoline price changes on both margins, as well as the heterogeneity in the responsiveness. I develop a novel structural model of vehicle choice and subsequent utilization, where consumer decisions are modeled in a dynamic setting that explicitly accounts for selection on unobserved driving preference at both the time of purchase and the time of driving. This utility-consistent model allows for the analysis of the welfare implications to consumers and government of a variety of different policies, including gasoline taxes and feebates. I find that consumers are responsive to changing gasoline prices in both vehicle choice and driving decisions, with more responsiveness than in many recent studies in the literature. I estimate a medium-run (i.e., roughly two-year) elasticity of fuel econ- omy with respect to the price of gasoline for new vehicles around 0.1 for California, a response that varies by whether the vehicle manufacturer faces a tightly binding fuel iv economy standard. I estimate a medium-run elasticity of driving with respect to the price of gasoline around -0.15 for new personal vehicles in the first six years. Older vehicles are driven much less, but tend to be more responsive, with an elasticity of roughly -0.3. I find that the vehicle-level responsiveness in driving to gasoline price changes varies by vehicle class, income, geographic, and demographic groups. I also find that not including controls for economic conditions and not accounting for selec- tion into different types of new vehicles based on unobserved driving preference tend to bias the elasticity of driving away from zero { implying a greater responsiveness than the true responsiveness. This is an important methodological point, for much of the literature estimating similar elasticities ignores these two issues. These results have significant policy implications for policies to reduce gasoline consumption and greenhouse gas emissions from transportation. The relatively inelas- tic estimated responsiveness on both margins suggests that a gasoline tax policy may not lead to dramatic reductions in carbon dioxide emissions, but is a relatively non- distortionary policy instrument to raise revenue. When the externalities of driving are considered, an increased gasoline tax may not only be relatively non-distortionary, but even economic efficiency-improving. However, I find that the welfare changes from an increased gasoline tax vary significantly across counties in California, an important consideration for the political feasibility of the policy. Finally, I find suggestive evi- dence that the \rebound effect” of a policy that works only on the extensive margin, such as a feebate or CAFE standards, may be closer to zero than the elasticity of driving with respect to the price of gasoline. This suggestive finding is particularly important for the analysis of the welfare effects of any policy that focuses entirely on the extensive margin. v Acknowledgement I have many people to thank for to making this dissertation possible. Without ques- tion, I should start with my advisors. If it were not for Jim Sweeney, I would not have even come to Stanford. Jim taught me to think carefully about the economic logic of any contention I make in my research. I cannot thank him more for his guidance and support over the years. I also came to Stanford hoping to get a chance to work with John Weyant. I feel privileged to have had this opportunity. My perspective on energy policy modeling has been profoundly shaped by John's insights. I have had the good fortune to have Larry Goulder as an advisor since I took a series of classes in environmental economics from him early in my time at Stanford. Larry has been a truly excellent advisor, always making sure that I see the bigger picture in any analysis that I do. In the past two years, I have greatly benefited from Matt Harding's guidance in econometrics. Matt is a world-class econometrician and I have him to thank for developing my skills as an empirical economist. Several other economists in the economics department also greatly contributed to this dissertation. Chapters 2 and 3 of my dissertation began as a class paper for Jon Levin and Tim Bresnahan's Industrial Organization seminar. Both Jon and Tim have been extremely generous with their time and insights. Besides acting as my oral defense chair, Jon also greatly helped shaped the development of my structural model in Chapter 4 through a series of conversations we had last summer. Tim's comments helped me more deeply understand how my work is novel. In addition, another Industrial Organization professor, Liran Einav, generously agreed to meet with me several times and has very useful suggestions for improving the research. Wes Hartmann of the Stanford Graduate School of Business taught a class in vi advanced empirical methods that turned out to be incredibly valuable in helping me develop the tools to estimate my structural model. Wes was also generous with his time in meeting with me to discuss the research. Conversations with Lee Schipper of the Precourt Energy Efficiency Center at Stanford were not only fun, but also very useful in keeping me honest about the details of the transportation sector. There are also many economists outside of Stanford who provided invaluable help in bringing this dissertation to fruition. My initial interest in economic modeling began when I was a research assistant at Dartmouth College for Karen Fisher-Vanden. Karen encouraged me to major in economics and build a technical skill set. I have her to thank for leading me down the path towards this dissertation. My interest in transportation economics began while I was a research assistant at Resources for the Future (RFF), working with Winston Harrington, Ian Parry, Elena Safirova, Billy Pizer, Jim Sanchirico, and others. While at RFF, I also had the good fortune to work on projects relating to energy efficiency policy with Richard Newell and Karen Palmer. In many respects this dissertation combines these two interests, by looking at policies to improve the energy efficiency of the transportation sector. I became particularly interested in this intersection while working at the White House Council of Economic Advisers with Richard Newell. Richard and I spent a great deal of time thinking about the costs and benefits of fuel economy standards, which even more directly led me down the path to this dissertation topic. More recently, while I was on the academic job market, I benefited greatly from discussions with a variety of economists around the country. Most notable are the excellent suggestions from Gautam Gowrisankaran at University of Arizona, Dave Rapson at UC Davis, Rob Williams at the University of Maryland, and Jeffrey Brown and Nolan Miller at the University of Illinois Urbana-Champaign. Prior to the aca- demic job market, I also have to thank Soren Anderson of Michigan State University for pointing me to the county-level Bureau of Labor Statistics data. Of course, this dissertation would also not have been possible without the financial support of several organizations. The US Environmental Protection Agency STAR Fellowship program funded me for several years of my doctoral career, and provided the funding to purchase the first piece of the dataset that I eventually assembled.