Evaluating the Consumer Response to Fuel Economy: AReview of the LiteratureGloria Helfand and Ann WolvertonWorking Paper SeriesWorking Paper # 09-04August, 2009U.S. Environmental Protection AgencyNational Center for Environmental Economics1200 Pennsylvania Avenue, NW (MC 1809)Washington, DC 20460

Evaluating the Consumer Response to Fuel Economy: A Review ofthe LiteratureGloria Helfand and Ann WolvertonNCEE Working Paper SeriesWorking Paper # 09-04August, 2009DISCLAIMERThe views expressed in this paper are those of the author(s) and do not necessarily representthose of the U.S. Environmental Protection Agency. In addition, although the research describedin this paper may have been funded entirely or in part by the U.S. Environmental ProtectionAgency, it has not been subjected to the Agency's required peer and policy review. No officialAgency endorsement should be inferred.

Evaluating the Consumer Response to Fuel Economy: A Review of the LiteratureBy Gloria Helfand and Ann Wolverton 1U.S Environmental Protection AgencyAbstractHow consumers evaluate trade-offs between the cost of buying additional fuel economyand the expected fuel savings that result is an important underlying determinant of theoverall cost of national fuel economy standards. Models of vehicle choice are a means topredict the change in consumers’ vehicle purchase patterns, as well as the effects of thesechanges on compliance costs and consumer surplus. This paper surveys the literature onvehicle choice models and finds a wide range in methods and results. A large puzzleraised is whether automakers build into their vehicles as much fuel economy asconsumers are willing to purchase. This paper examines possible reasons why there maybe a gap between the amount consumers are willing to pay for fuel economy and theamount that automakers provide, though there is insufficient evidence on the relativeroles of these various hypotheses. Further research on the role of fuel economy inconsumer vehicle purchases is needed to assist in understanding the welfare effects offuel economy regulation.Key Words: Consumer behavior, vehicle purchase decisions, fuel economy, energyparadox, vehicle choiceJEL Codes: D11, D12, D22, R411Authors can be contacted via email at [email protected] and [email protected] The viewsexpressed in this paper are those of the authors and do not necessarily represent those of the U.S.Environmental Protection Agency. This paper has not been subjected to EPA’s formal review process andtherefore does not represent official policy or views. We are very grateful for the suggestions andcontributions of Matt Massey, Chris Moore, two anonymous reviewers, and participants at our presentationat the World Congress of Environmental and Resource Economics.1

1. IntroductionOn April 1, 2010, the U.S. Environmental Protection Agency (EPA) and theDepartment of Transportation (2010a) issued regulations to reduce greenhouse gas(GHG) emissions from vehicles and increase the stringency of Corporate Average FuelEconomy (CAFE) standards. 2A key analytic question that was discussed in theevaluation of the standards was the role that fuel economy plays in consumers’ vehiclepurchases. Do consumers properly account for the fuel savings from more fuel-efficientvehicles when making purchase decisions?Will the requirement of additional fueleconomy increase or reduce consumer and producer welfare? This topic is also likely tobe highly relevant in other countries as they contemplate setting or tightening their ownfuel efficiency and/or GHG tailpipe standards. For instance, the International Council onClean Transportation (ICCT) finds that a number of countries are contemplatingincreasingly stringent mandatory GHG standards for vehicles (ICCT 2010).In analyses of the potential economic impacts of fuel economy and GHGstandards for cars and trucks, U.S. Federal agencies have commonly assumed that themarket shares of the vehicle fleet stay constant. For rules that lead to small changes invehicle attributes and prices, this may not be a bad approximation, since we would notexpect consumers to make large changes in vehicle purchase patterns in response.Proposals to increase fuel economy substantially, on the other hand, may cause producersto change both vehicle attributes and the price of vehicles enough that consumersnoticeably alter their decisions about what vehicles to purchase.2Because the primary source of GHG emissions from vehicles is burning fuel, reducing fuel consumptionwill reduce emissions. CAFE standards require the sales-weighted fuel economy of a manufacturer’s fleetto achieve minimum levels.2

In addition, whether consumers accurately evaluate trade-offs between the costsof consuming more fuel economy and the expected fuel savings is a matter of somedebate in the literature. Government intervention in markets is commonly justified by theexistence of one or more market failures, such as environmental externalities associatedwith a private market transaction or behavior. For instance, vehicle tailpipe emissionscontribute to climate change, an impact not priced into consumer purchase or drivingdecisions. If significant numbers of consumers and/or producers also routinely makemistakes when factoring fuel economy into vehicle purchase decisions, then regulationmay save consumers money in addition to reducing externalities. 3Consumers should be interested in the fuel economy of the vehicles they purchaseapart from what is induced by government regulation: An increase in fuel economyreduces the private cost of driving. Empirical studies of the effects of higher fuel pricesindicate that higher fuel costs do seem to trigger efforts on the part of consumers toreduce their gasoline consumption through changes in driving behavior and vehiclepurchase decisions. However, simple present value calculations comparing upfront coststo future fuel savings indicate that there appear to be many low cost opportunities toreduce fuel consumption that are not undertaken, even when they could pay off overrelatively short time periods. This observation is commonly termed the “EnergyParadox.” It is unclear, however, whether such a paradox by itself justifies additionalfuel economy requirements. Some vehicle purchasers may derive pure financial gainsfrom the additional fuel savings if, for instance, they made mistakes in calculating fuelsavings or relied on rule of thumb calculations at the time of purchase. Other consumers3The effect of fuel economy regulations on consumer and producer surplus would need to be combinedwith these external effects to estimate the total benefits and costs of fuel economy regulation.3

may be made worse off in their vehicle purchase decision if, for instance, they incuradditional costs for an attribute that they view as uncertain or risky or that they wereunwilling to buy.This paper reviews the large but inconclusive literature on the role of fueleconomy in consumers’ vehicle purchase decisions. A primary objective of this review isto highlight key gaps in the existing research that make accurate modeling andquantification of the welfare effects associated with raising CAFE standards difficult.Section 2 reviews the state of the art in vehicle choice modeling, with a focus on the roleof fuel economy in these models. Section 3 examines the various explanations posited inthe literature for why consumers appear to undervalue fuel economy. Section 4 discussesthe few studies that focus on the producer side of the equation: why, even if consumersactually would buy vehicles with greater fuel economy, automakers may not supplyvehicles with the mix of attributes consumers want. Section 5 concludes.2. State of the Art in Vehicle Choice ModelingThere are a variety of tools available to examine consumers’ valuation of differentvehicle attributes. Hedonic models, for instance, regress vehicle model attributes, such asfuel economy and horsepower, on the market price of the vehicle to estimate consumers’implicit willingness-to-pay for each individual feature (e.g., Court 1939, Arguea et al.1994, Espey and Nair 2005). 4 Because a simple hedonic regression is a reduced-formequation where market price and quantity are jointly determined by supply and demand,separately identifying the demand function requires additional modeling. Papers that usea two-stage approach can examine how changes in external conditions affect demand for4Freeman (2003) is a classic reference on nonmarket valuation, including hedonic models.4

a particular attribute. For instance, Ohta and Griliches (1986) examine how changes ingasoline prices affect the shadow price for particular vehicle characteristics, includingfuel economy, power, and weight.Fan and Rubin (2010) estimate consumers’willingness-to-pay for fuel economy while accounting for demographic differences suchas age and education. Agarwal and Ratchford (1980) estimate a hedonic price function asthe first step to predicting how consumers would rank order different makes and models.When evaluating the potential impacts of fuel efficiency or GHG standards, it isimportant to understand how changes in stringency could affect the number and types ofvehicles purchased based on their new (post-policy) mix of attributes. Investigating thisempirical question requires the use of models that examine consumers’ choices ofvehicles.For instance, many studies econometrically estimate vehicle choice as afunction of prices, consumer characteristics (such as income, family size, and age), andvehicle attributes (such as a vehicle’s power and fuel economy). Once estimated, thesemodels are often used to examine how consumers’ vehicle purchase decisions areaffected by marginal changes in vehicle or personal characteristics. While the focus is onthe consumer in these papers, many also model production decisions in recognition that itis the interaction between producers and consumers that leads to observed marketoutcomes.Other papers evaluate large-scale policy changes by simulating howconsumers and producers in the vehicle market would respond to policies such as anemissions tax or substantially tighter fuel economy standards. These models often do notuniquely estimate their own parameters, instead borrowing from the vehicle choiceliterature. While we discuss both types of models in this paper, we focus mainly on issuesrelated to the econometric estimation of vehicle choice models.5

Reflecting the complexity of consumers’ decisions, models of vehicle choice varywidely. This section begins with a discussion of the range of research questions exploredby the literature to set the stage for the issues that arise in the models. Next, we discussthe modeling frameworks that have been developed to examine vehicle choice in thecontext of these questions. In particular, we review the modeling approaches, datasources, and individual buyer and vehicle characteristics included in the models. Vehiclechoice is a rich area of research both because of its direct policy relevance – for instance,the importance of the auto industry in the U.S. economy, the increasing attention to thecontributions of vehicles to greenhouse gas emissions, and large fluctuations in fuelprices over time - as well as a multitude of technical challenges. The reader also will notethe large number of working papers referenced throughout the document. Reliance ononly published work for this discussion would neglect a large portion of the literature thatoffers innovative new estimation techniques and intriguing results. With or without theinclusion of these new papers, a review of the literature suggests that the models varygreatly on a number of dimensions. In particular, and of special interest for policyanalysis, they continue to produce widely varying estimates of the role of fuel economyin consumers’ purchase decisions.2a. Research QuestionsThe auto market has attracted a great deal of research interest because of its size,its market structure, the availability of data, and its role in a number of significant policyareas, including international trade and environmental quality. Vehicle choice modelshave been developed to analyze many of these research and policy questions. Thesemodels have also served to advance the state of economic modeling: the work of Berry et6

al. (1995), for instance, is often cited outside the motor vehicle context for itsincorporation of multiple new modeling issues into its framework. 5 In the public policyarena, topics have included the effects of voluntary export restraints on Japanese vehiclescompared to tariffs and quotas (e.g., Goldberg 1995), the market acceptability ofalternative-fuel vehicles (e.g., Brownstone et al. 1996; Brownstone and Train 1999;Brownstone et al. 2000; Greene 2001; Greene et al. 2004), the effects of introducing andremoving vehicles from markets (e.g., Petrin 2002; Berry et al. 2004), causes of thedecline in market shares of U.S. automakers (e.g., Train and Winston 2007), and theeffects of gasoline taxes (e.g., Austin and Dinan 2005; Bento et al. 2005, Feng et al.2005), “feebates” 6 (e.g., Greene et al. 2005b; Feng et al. 2005; Greene 2009), and fueleconomy standards (e.g., Goldberg 1998; Austin and Dinan 2005; Klier and Linn 2010a,2010b; Jacobson 2010; Whitefoot et al. 2011).The research question of interest often affects the structure of the model used toinvestigate the choices consumers make in the vehicle market. For instance, some studiesconsider only the market shares of new vehicles (e.g., Train and Winston 2007), whileothers include the choice between new vehicles and a generic outside good (e.g., Berry etal. 1995; Klier and Linn 2010a; Whitefoot et al. 2011), and some explicitly consider therelationship between the new and used vehicle markets (e.g., Bento et al. 2005, 2009;Busse et al. 2010; Allcott and Wozny 2010). Focusing on market share is sufficient, forinstance, to examine the decline in