Energy Economics 30 (2008) 2657–2672

Contents lists available at ScienceDirect

Energy Economics j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / e n e c o

The effect of standards and fuel prices on automobile fuel economy: An international analysis Sofronis Clerides a,b, Theodoros Zachariadis a,⁎ a b

University of Cyprus, Cyprus CEPR, United Kingdom

a r t i c l e

i n f o

Article history: Received 12 December 2007 Received in revised form 10 June 2008 Accepted 10 June 2008 Available online 19 June 2008 JEL classification: Q4 Q5 Keywords: CAFE Fuel tax Greenhouse gases Rebound effect Regulation

a b s t r a c t There is an intense debate over whether fuel economy standards or fuel taxation is the more efficient policy instrument to raise fuel economy and reduce CO2 emissions of cars. The aim of this paper is to analyze the impact of standards and fuel prices on new-car fuel economy with the aid of cross-section time series analysis of data from 18 countries. We employ a dynamic specification of new-car fuel consumption as a function of fuel prices, standards and per capita income. It turns out that standards have induced considerable fuel savings throughout the world, although their welfare impact is not examined here. If standards are not further tightened then retail fuel prices would have to remain at high levels for more than a decade in order to attain similar fuel savings. Finally, without higher fuel prices or tighter standards, one should not expect any marked improvements in fuel economy under ‘business as usual’ conditions. © 2008 Elsevier B.V. All rights reserved.

1. Introduction The share of transportation in total energy consumption and greenhouse gas (GHG) emissions is increasing, particularly in OECD countries, because of continuous growth in total vehicle kilometers traveled and stagnancy in automobile fuel economy (IEA, 2006). This comes in sharp contrast to GHG mitigation achievements in other sectors like power generation and industrial processes. In the European Union (EU), for example, the transport sector almost completely cancels out other progress towards meeting the 8% GHG reduction target under the Kyoto protocol (EEA, 2005). With the exception of biofuels, which are regarded as CO2-neutral and whose production is gradually increasing and encouraged by legislation in some world regions, other fuel/engine combinations are still not mature for mass production and even commercially

⁎ Corresponding author. Economics Research Center, University of Cyprus, P.O. Box 20537, 1678 Nicosia, Cyprus. Tel.: +357 22 892425; fax: +357 22 892426. E-mail addresses: [email protected] (S. Clerides), [email protected] (T. Zachariadis). 0140-9883/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2008.06.001

2658

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

available hybrid powertrains are experiencing quite slow penetration rates. It therefore becomes imperative for OECD countries to implement more aggressive policy measures if they are to ensure progress in limiting fuel consumption and GHG emissions.1 The most widely discussed policy instruments are fuel economy standards and fuel taxes, one form of which is the carbon tax. A cap-and-trade system might be another alternative, which under specific circumstances would be equivalent to a carbon tax; however, it is not yet considered widely as a policy tool for road transportation. A fuel economy (FE) standard is usually expressed as the maximum sales-weighted average fuel consumption for the new-car fleet entering the market in a given year. It can be either mandatory or the result of a ‘voluntary’ commitment of the automotive industry. Mandatory Corporate Average Fuel Economy (CAFE) standards have been in force in the United States since 1978 but, with a minor exception for light duty trucks, they had not been tightened since 1990. This changed in 2007, when the US President announced a plan to further tighten these standards for cars from model year 2010 onwards, and the US Congress passed the “Energy Independence and Security Act of 2007”.2 Other countries followed the US example later in the 1980s, and currently Australia, Canada, China, the EU, Japan, Switzerland, South Korea and Taiwan implement some type of FE or CO2 standard.3 Specifically in the EU, the automobile industry made a voluntary commitment that by 2008/2009 new passenger cars will emit 25% less CO2 per kilometer on average compared to 1995. As the deadline approaches and this target will not be met, early in 2007 the European Commission, the EU's executive body, announced a proposal for the adoption of a more stringent and mandatory industry-wide emission standard by 2012 (EC, 2007). As of June 2008, this proposal was still under discussion between EU leaders, the European Commission and the European Parliament. There is a lively debate about whether standards or fuel taxation is the most effective way of reducing fuel consumption. In the next section we discuss the arguments and evidence in favor of each policy. Our goal is to add to the evidence by drawing from the world's collective experience over the last thirty years in order to reach some lessons about the relative effectiveness of each policy tool. We have compiled what we believe is the most comprehensive international dataset on new-car fuel economy, covering 18 countries (US, Canada, Australia, Japan, Switzerland and 13 EU countries) and spanning a period between 1975 and 2003. Our empirical approach is to analyze the evolution of FE in the various countries and identify the impact of standards and fuel prices on that process. We analyze each series individually and also all the series pooled together in a panel using a reduced form dynamic model that allows exploring short-run and long-run impacts. We also set up a difference-in-differences specification which exploits the fact that different countries implemented standards at different times. In addition to looking at the significance and magnitude of these effects, we ask whether the enactment of a mandatory or voluntary FE target has brought about a structural change in FE evolution. Our analysis shows that FE targets have been the major driver of the FE improvements that have been observed in Europe and Japan since the mid-1990s. Fuel prices do play a role in reducing fuel consumption, but the price elasticity is quite small even in the long run, so that it becomes difficult to improve FE considerably by relying on price increases alone. Aside from cost-effectiveness issues which we do not address here, we conclude that – at least in Europe and Japan – some kind of standard is necessary to induce FE improvements in the future unless fuel prices rise considerably compared to today's levels. On the other hand, in the US we find mixed evidence for the effectiveness of standards and at the same time we find a reasonably large impact from fuel prices on FE levels4. 1 The terms fuel economy (expressed in miles per gallon) and fuel consumption (expressed in litres per 100 kilometres) are linked by the following relationship: fuel consumption (l/100 km) = 235.2 / fuel economy (mpg). We will use the two terms interchangeably with the understanding that one is essentially the inverse of the other. 2 In response to this Act, the US National Highway Traffic Safety Administration proposed new FE standards for cars and light trucks in April 2008. As of this writing (June 2008), the administrative process was still under way. See http://www.nhtsa.dot.gov/ portal/site/nhtsa/menuitem.43ac99aefa80569eea57529cdba046a0 [last accessed June 2, 2008]. 3 A CO2 standard is essentially a fuel economy standard because emissions of CO2 for each unit of gasoline burnt are constant. Diesel fuel contains a higher amount of carbon than gasoline: a liter of diesel when burnt releases about 15% more CO2 in the atmo sphere compared to a liter of gasoline. 4 Although still early to build robust evidence, it seems that soaring oil prices recently have induced some changes in consumer preferences towards smaller and/or more efficient cars; see e.g. “As Gas Costs Soar, Buyers Flock to Small Cars”, The New York Times, May 2, 2008, available at http://www.nytimes.com/2008/05/02/business/02auto.html?_r=1&partner=rssuserland&emc=rss&pagewanted= all&oref=slogin.

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

2659

The international analysis presented here relies on reduced form time series relationships as it cannot employ micro-level data on the producer's side. As such, the analysis is complementary to studies that utilize micro-level data to analyze consumer and producer responses to policy changes. Particularly in Europe, it is not possible to analyze this issue on the basis of simulations of a firm's behavior because current legislation does not require that individual automobile manufacturers disclose firm-level fuel economy data. Nonetheless, the wide international and temporal coverage of the sample yields interesting and policy-relevant results. 2. Theoretical arguments and empirical evidence It is generally acknowledged that the adoption of standards has induced fuel economy improvements, or at least it has ensured that the fuel economy of new cars will not deteriorate despite stronger consumer preferences for energy-consuming amenities and safety features that usually add weight to a car and increase fuel consumption. However, there are several critical appraisals of the current CAFE system, some of which focus on the regulatory design rather than the adoption of standards as such. For example, NRC (2002) and Portney et al. (2003) suggest that, if the CAFE system is to be retained, a number of improvements should be introduced so as to avoid economic inefficiency and negative side-effects. Furthermore, the use of standards has been criticized as a second-best solution. Environmental pollution is a classic externality problem and the economic textbook remedy for it is Pigovian taxation. Many prominent economists advocate taxation over standards on the grounds that “CAFE rules are heavy-handed government regulations replete with unintended consequences”5. Taxes also raise revenue and thereby make it possible to reduce other distortionary taxes such as labor or capital income taxes. Whether standards are more heavy-handed than taxation may be a matter of debate but it is clear that each policy tool should be evaluated on its merits, both theoretical and practical. A useful way to distinguish between them is to think of overall fuel consumption – the reduction of which is the ultimate policy objective – as the product of the number of miles traveled times the amount of fuel burnt per mile. Fuel taxes raise the cost of driving a vehicle and hence should have an immediate impact on the number of miles traveled. They will also induce consumers to purchase more fuel-efficient vehicles which will, in the longer run, lead to more of them being produced. Hence taxes push both quantities in the right direction. Standards, on the other hand, target fuel per mile directly and exclusively and therefore provide no incentive for reduced driving. In fact, opponents of regulations argue that standards could lead to an increase in miles traveled as consumers switch to more fuel-efficient cars which are cheaper to drive — although the size of such an increase is currently estimated to be low in the US (Small and Van Dender, 2007).6 Proponents of standards counter that taxation may not be as effective as economic theory suggests because consumers behave myopically. Such behavior could be explained by some sort of hyperbolic discounting, whereby consumers over-discount the future cost savings of fuel-efficient cars, or by a bounded rationality argument, which would claim that consumers do not analyze their fuel costs in a systematic way and make large errors estimating gasoline costs and savings over time (Turrentine and Kurani, 2007). The latter argument implicitly assumes that consumers always under-estimate fuel costs and it is not at all clear why this should be the case. In support of the myopia argument, Glazer and Lave (1994) cite studies showing myopic behavior of consumers in their purchase decisions of automobiles and other durable goods. They further argue that, despite higher fuel prices, both consumers and manufacturers may prefer to wait until uncertainty about technology or gasoline prices is resolved before making purchase decisions or undertaking costly research on more efficient cars respectively. Hence, even if an increase in the price of gasoline has powerful effects, those effects may be delayed and regulation may have a more immediate impact. In support of the myopia argument, Greene et al. (2005) cite several studies from public

5 Greg Mankiw in “Raise the Gas Tax,” Wall Street Journal, October 20, 2006. A list of other economists who support the carbon tax appears on Mankiw's blog, http://gregmankiw.blogspot.com/2006/09/rogoff-joins-pigou-club.html [accessed July 25, 2007]. 6 Another criticism of standards is that they led to the rise in SUVs, which must meet a less stringent standard because they are considered light trucks. This, however, seems like a problem with the policy design rather than with the policy tool. Moreover, the share of SUVs had been rising already before the introduction of CAFE standards, and the less stringent safety standards for trucks may have played a role in this development too.

2660

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

authorities and auto manufacturers indicating that consumers may take into account only the first 3 years of fuel savings when considering the value of higher fuel economy. But again this behavior does not seem so unreasonable if we consider the fact that most new car buyers will keep their vehicle for much less than the 14 years which is the average vehicle lifetime. A detailed, micro-level study by Goldberg (1998) did not find any evidence of consumer myopia. Perhaps the strongest argument in favor of standards is the fact that they are politically more feasible than taxes. In the US polls show broad popular support for standards but much less support for taxation.7 In the UK the public is suspicious of green taxes because they think it is an excuse for government to raise revenue.8 Most studies (including ours) find that the level of taxation required to reduce fuel consumption markedly is quite high. A recent paper by Hughes et al. (2006) finds that gasoline demand in the US has become much less price elastic over time, meaning that the tax increase required to achieve the desired impact is even greater than previously thought. It seems unlikely that voter-consumers will accept such high levels of additional gasoline taxes. The impact of standards, and how they compare to fuel taxes, has been examined by researchers in energy economics, industrial organization and the environment. These studies have employed estimations with micro data, discrete choice modeling, cross-section time series estimations, and simulations with partial or general equilibrium models. The rest of this section summarizes most of the important contributions to this debate. Goldberg (1998) estimated models of automobile demand, vehicle utilization and automobile supply with the aid of US micro data. She then simulated the effects of the CAFE program on fuel consumption and auto manufacturer profits and found that CAFE provided incentives for manufacturers to develop more fuel-efficient vehicles without putting a large burden on consumers. She found also that a high gasoline tax (of the order of 80 US cents per gallon) would be required to yield the same fuel saving benefit as an increase in CAFE standards. From that point of view, one can conclude that there is no strong evidence for or against standards. Thorpe (1997) developed and applied a general equilibrium model of automobile choice, distinguishing seven vehicle segments and three types of auto manufacturers (American, European and Asian). By making assumptions on the price elasticities within and between firms he performed simulations and found that CAFE actually reduces fuel economy – at least in the short term – as it shifts automobile sales towards less efficient vehicles. Kleit (2004) developed a partial equilibrium model of the US automobile market including eleven vehicle categories and four different automobile firms. Using parameters from a previously existing model and accounting for externalities of automobile use, he simulated the long-run effect of raising the current CAFE standards by 3 mpg and concluded that increasing gasoline tax by 11 US cents per gallon would yield the same energy conservation effect with raising CAFE standards by 3 mpg, at significantly lower welfare costs. Austin and Dinan (2005), using a similar simulation model, reached a similar conclusion but estimated a smaller difference in welfare effects than Kleit (2004); they state, however, that Kleit's welfare impact assessment suffers from inconsistent assumptions on elasticities and costs of different policies. Parry et al. (2005) followed a similar methodological path in their welfare analysis of standards versus fuel price increases, but also included a more detailed estimation of externalities in their simulations. They found that tightening the standards would raise welfare only under severe consumer myopia, i.e. if consumers greatly undervalue fuel savings. All studies mentioned in this paragraph have also countered the argument of market failures (which is mentioned by supporters of standards), noting that consumers are very well informed about automobile fuel costs because of ample information on both new-car fuel economy and fuel prices, so that it is unlikely for this market to be inefficient.9

7 See “Many Americans Are Trimming Travel, But Few Car Pool to Cut Fuel Use,” Wall Street Journal, July 7, 2007. Available at http:// online.wsj.com/article/SB118304470725951593.html. 8 See “The Case Against Further Green Taxes — Report and Poll”. Available at http://tpa.typepad.com/research/2007/09/the-caseagains.html [last accessed September 14, 2007]. 9 Greene (1998; p. 598) rejects these arguments, stating that it is difficult for consumers to assess their own fuel economy because they do not know their own share of city and highway driving in order to calculate their average fuel economy correctly, and because the discrepancy between officially reported and real-world fuel economy is unknown as it also depends on their own driving style.

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

2661

Greene and Liu (1988) applied discrete choice modeling to estimate the change in consumer surplus between 1978 and 1985 because of changes in automobile characteristics to achieve greater fuel economy. Τhey found this change to be close to zero, which implies that standards cannot be clearly accepted or rejected as a policy measure from a welfare point of view — a finding similar to that of Goldberg (1998) mentioned above. From a methodological standpoint, our study comes closest to those of Espey (1996), Gately (1992), Greene (1990), Johansson and Schipper (1997), and Small and Van Dender (2007), all of which involved cross-section time series analyses. Espey (1996) addressed fleet-wide fuel consumption using data from 8 OECD countries from 1975 to 1990, thus not utilizing data from low-price periods. Only the US had FE standards during that period and even these standards correlated closely with the time trend of fuel consumption because standards had been rising fairly steadily until 1990. Therefore the impact of the FE standard in that work could not be separated from the overall time trend that may represent technical progress or other change. Johansson and Schipper (1997) conducted a similar analysis with cross-section time series models including 12 OECD countries for the period 1973–1992, but using again fleet-average FE as the dependent variable and without addressing standards explicitly. Using country-specific time trends, the authors found that FE improvements had been much faster in the US than in any other country since 1978, but they were reluctant to attribute all the improvement to the CAFE program. Storchmann (2005) also employed a pooled model to estimate fleet-average fuel consumption using several explanatory variables such as private income, population density, urbanization rates, fuel prices and automobile costs. He focused on the effect of income distribution on worldwide gasoline demand and did not address the issue of FE standards. Greene (1990) tackled the same question using a different methodology. He modeled the automobile manufacturers' decision-making process and concluded that CAFE standards played a greater role than fuel prices in improving new-car FE levels in the US. Data came from the US only and, as in Espey (1996), involved a period of monotonically rising standards. Gately (1992) tested equations of fleet-average FE as a function of prices and CAFE standards, allowing for potentially asymmetric price elasticities in US gasoline consumption. He found that the effect of standards was not significantly different from zero and that gasoline prices alone (with lags of up to 10 years) could sufficiently explain the evolution of FE over the years. The estimated effect of standards, however, is diluted when fleet-average FE is the dependent variable because of slow vehicle retirement rates. In order to account properly for the effect of standards, not only the current value of the standard but a sufficient number of lags of the standard variable has to be included. If that analysis had longer time series available in order to include the lagged effect of standards in his model, it might have found a statistically significant effect of standards as well. Small and Van Dender (2007) include a CAFE variable in their analysis, which turns out to be significant for determining fuel economy. However, as their main concern is the extent of the rebound effect, i.e. how much vehicle travel will increase if fuel economy decreases, they only use the CAFE variable (defined as the difference between regulated and ‘desired’ fuel economy levels) in order to derive more stable estimates of the rebound effect. We believe that this paper goes beyond previous studies and significantly adds to the literature in several ways. First, we base our analysis on data from many countries all over the world which have implemented a FE standard or voluntary target for quite some time; at the time the previous studies were performed, only the US and Canada had standards in force, which did not allow a study the impact of standards systematically in the whole sample. Second, US data in this paper cover the period from 1975 to 2004, which adds variation to our sample because it contains periods of rising as well as falling oil prices and rising as well as stagnant CAFE standards. Third, data from Europe and Japan extend over a period without standards (before the mid-1990s) and a period after 1995 when some form of standard had been enacted in these countries; this allows the exploration of the existence or not of a ‘change in regime’ associated with the adoption of standards. None of the previous studies were able to test such a hypothesis because of lack of data by the time those studies were conducted. Finally, our analysis addresses new-car (instead of fleet-averaged) fuel economy, which is the variable that is relevant for policy makers, and is easier to follow because it is directly monitored in developed countries and is not confounded by assumptions on vehicle turnover rates.

2662

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

3. Data The object of our analysis is the evolution over time of fuel economy or fuel consumption and the role of standards and fuel prices in shaping that process. To that end we collected information from several countries that introduced FE standards at some point. As mentioned in the introductory section, the US decided in 1975 to enforce CAFE standards starting in 1978 for cars and 1982 for light trucks. Canada adopted the same standards, albeit on a voluntary basis, from 1980 onwards (1990 for light trucks). Australia enacted the first voluntary code of practice for FE improvements in 1978, setting targets for years 1983 and 1987, and continued with further voluntary agreements for tighter FE targets in the 1990s. The Japanese government introduced FE standards in the mid-1990s, aiming to reduce new-car fuel consumption by 19% in year 2010 compared to 1995. In the EU, discussions between governments and the auto industry started in the mid-1990s, and a specific voluntary commitment of the industry was finalized in 1998 for European automakers and in 1999 for Japanese and Korean automakers. We collected data for all these regions. One of the contributions of our study is that it analyzes the FE of new vehicles, as opposed to the entire fleet. As mentioned in the previous section, using fleet-average FE as the dependent variable complicates the analysis because this is a derived quantity influenced both by new-car fuel economy and by the rate at which new cars enter the market. Fleet-average FE changes very slowly, hence it becomes difficult to discern the potential impact of a standard or a new technology; this was also the result of estimations of Espey (1996) and Johansson and Schipper (1997). On the other hand, the impact of fuel prices on fleetaverage FE will be immediate as both new and existing vehicles will be affected. Hence, unless one is able to capture the impact of FE standards using a large number of lagged variables (which will be difficult in practice), using fleet-average FE will tend to bias estimation in favor of taxation as the most effective tool. The data were obtained from diverse official sources. This was not a trivial task because new-car fuel consumption is not routinely recorded in many countries, and in most cases it was only after the implementation of some FE standards that this variable started being systematically measured in countries with regulations in place. There are, however, OECD countries where such information has been gathered since the late 1970s, primarily through initiatives of the International Energy Agency (IEA) with the aid of data collected from automobile manufacturers. We were able to construct consistent time series for 18 countries. Thirteen of those are EU countries and they face an EU-wide target, and it would have been inappropriate to treat them as separate observations. We grouped together all EU countries into a single ‘EU’ observation by taking the average of all variables weighted by annual new car registrations in each country. It was possible to derive this weighted average for seven EU countries (Austria, Belgium, France, Germany, Italy, Sweden and the UK) which had data available since 1980; hence the EU variables contain the weighted average of data from these seven countries only. These account for 75% of new car registrations in the whole EU. All seven European countries switched from a ‘no standards’ to a ‘with standards’ regime in the mid-1990s as a result of the auto industry's voluntary agreement with the European Commission.

Table 1 Overview of the sample used in the study Country Australia Canada Canada

Vehicle category

Cars Cars Light duty trucks European Cars Union Japan Cars United Cars States United Light duty States trucks

Sample period

Type of Enforcement First decision for the adoption First year of implementation standards type of standards/targets or first target year

1978–2002 1980–2003 1980–2003

FE FE FE

Voluntary Voluntary Voluntary

1978 1976 1982

1983 1980 1990

1980–2003

CO2

Voluntary

1998

2008

1980–2000 1975–2004

FE FE

Mandatory Mandatory

1995 1975

2010 1978

1975–2004

FE

Mandatory

1975

1982

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

2663

Table 2 Descriptive statistics

Cross-sections

Observations

Australia Canada — cars Canada — light trucks European Union Japan US — cars US — light trucks Whole sample

25 24 24 24 21 30 30 178

Median car model year

Per capita GDP (€'2005 PPP)

Gasoline price (€′2005 per liter)

Fuel consumption (l/100 km)

Average

Std. dev.

Average

Std. dev.

Average

Std. dev.

1990 1992 1992 1992 1990 1990 1990 1990

21096 22213 22213 19197 20273 26314 26314 21819

3109 2689 2689 2744 3007 3795 3795 3915

0.483 0.462 0.462 1.048 1.326 0.395 0.395 0.635

0.037 0.034 0.034 0.091 0.260 0.070 0.070 0.303

9.10 8.20 11.03 7.59 8.58 9.26 12.13 9.51

0.76 0.57 0.61 0.57 0.48 1.78 1.81 1.91

European Union refers to the 15 countries that were members already before 2004.

We thus built an unbalanced panel consisting of seven cross-sections: Australia, Canada (cars and light duty trucks separately), EU, Japan, and the US (cars and light duty trucks separately) — 178 observations in total. Table 1 provides a summary of this panel and descriptive statistics are shown in Table 2. The Appendix provides a list of the data sources we used along with a detailed description of the history of FE regulations in each country. Fig. 1 shows the evolution of fuel economy over time in the US and the EU. It is easy to note the close relation to the existence of standards or voluntary targets. FE improvements in the US after 1982 are particularly noteworthy because fuel prices decreased sharply after 1982, so that these improvements, even if partly justified by lead times for the introduction of fuel-efficient technologies as a response to the second oil shock, should not be primarily attributed to high fuel prices: improvements continued up to the late 1980s, as long as FE regulations were becoming increasingly stringent, and stagnated thereafter, in line with

Fig. 1. Evolution of new-car fuel consumption in the US and the EU and the corresponding CAFE standard (for the US) and voluntary CO2 target (for the EU). US data come from Hellmann and Heavenrich (2004); for compilation of EU data see Zachariadis (2006). The international oil price in real terms, taken from BP (2005), is also shown.

2664

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

Fig. 2. New-car fuel consumption in Japan and four EU countries.

stagnation of CAFE standards. Fig. 2 shows the evolution of new-car fuel consumption in four EU countries and Japan. It is again interesting that fuel consumption has decreased considerably since the mid-1990s, at a time of mostly low oil prices. 4. Econometric model and estimation The question of interest is how fuel prices and the imposition of standards impact the evolution of fuel economy. The dependent variable we seek to explain is FCt, the sales-weighted fuel consumption of new vehicles. Our basic model of the process describing the evolution of FC in the absence of standards is the following:  FCi;t ¼ mi þ λFCi;t−1 þ αTRi;t þ ∑Lj¼0 βj pi;t−j þ θINC i;t þ ei;t

ð1Þ

where indices i and t denote country and time respectively. FC is expressed in liters per 100 kilometers (l/ 100 km), λ is the autoregressive coefficient of the dependent variable, TR is a trend, p denotes the price of gasoline expressed in Euros at 1995 prices per liter, INC is per capita GDP in Euros at 1995 prices and ε is a residual term that is independently and normally distributed with zero mean and constant variance. νi is a time-invariant country fixed effect that captures national particularities (such as population density, urbanization rates, automobile taxation levels, existence of large national auto manufacturers etc.) which can partly explain differences in fuel consumption between countries. The autoregressive formulation of the dependent variable has also been applied in Espey (1996) and Johansson and Schipper (1997) (who modeled fleet-average FE and not new-car FE as we do here) as well as in many econometric analyses of total fuel demand. This dynamic specification allows for a partial adjustment of the dependent variable over the years, which is in line with observed patterns of consumer demand for durable goods such as automobiles. Pervasiveness of habits, uncertainty, imperfect information regarding alternatives and prices, adjustment and transaction costs, and expectations concerning permanent income may explain the gradual adjustment process that can be approximated with a dynamic model (Dargay and Vythoulkas, 1999). Moreover, such a model enables the identification of both short-run and long-run effects which is relevant and useful for policy simulations. The short-run effect of each variable is given by the values of the corresponding coefficients while the long-run effect is given by the corresponding coefficients divided by (1 − λ).

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

2665

By construction, FCt is determined by both supply and demand factors; it depends both on the fuel economy of available models and on the choices made by consumers. Hence FCt can decrease because all cars become more efficient or because consumers switch to more efficient cars. The latter effect is likely to be more important with taxation while the former effect will dominate if standards are used. Fuel prices should impact FCt in three ways: in the short-run, they discourage driving; in the medium run, they induce consumers to purchase smaller and/or more efficient cars among available models, which will tend to lower our dependent variable; and in the longer run, they give manufacturers an incentive to increase vehicle fuel economy by introducing more fuel saving technology in their new models. We allowed for up to five lags of fuel price in an attempt to capture both of these effects but the model was never able to estimate precisely more than one fuel price coefficient. For this reason from this point on we will include only the contemporaneous price in variants of Eq. (1). Income per capita is included for comparability with other studies.10 There are observable variables, such as vehicle mass or engine power, which reflect physical factors affecting fuel consumption; however, they should not be treated as exogenous in the model as they may be themselves a result of tighter FE standards or fuel prices rather than a cause of improved fuel economy. Therefore, it may be simpler and more appropriate to include a deterministic time trend in the model instead of these individual variables, which also exhibit an almost linear increase since the early 1980s. The time trend will capture any ‘autonomous’ fuel economy improvements as well as changes in consumer preferences (e.g. towards bigger or more powerful cars) to the extent that they are not related to income. The main hypothesis we want to test is whether the imposition of standards causes a change in the evolution of FE. We consider two ways of modeling this. The first approach is to test for a structural break in the rate of autonomous technical progress and in the responsiveness of FE to fuel prices: such a break might be associated with a change in technological improvement rates, and/or with a change in the responsiveness of consumers towards fuel prices (e.g. following a prolonged period of very low or very high fuel prices). In other words, we specify the following generalization of Eq. (1):   ~ ~ FCi;t ¼ mi þ λFCi;t−1 þ αTRi;t þ α Di;t 4TRi;t þ βpi;t þ β Di;t pi;t þ θINCi;t þ ei;t

ð2Þ

D is a dummy variable that becomes unity whenever existing FE regulations are tightened or new regulations are enacted. In the US and Canada, where specific FE standards apply every year, D is equal to unity when the current FE standard is tighter than the previous year's standard. In Australia, Europe and Japan D is unity if a country has introduced a future standard that requires gradual adjustment of annual FE. We describe in the Appendix the periods in each country that are treated as a ‘with standards’ regime so that their dummy variable becomes unity. Our second approach is to specify a model where the level of fuel consumption depends directly on the FE standard, particularly the distance between this year's standard and last year's FC:  FCi;t ¼ mi þ λFCi;t−1 þ αTRi;t þ βpi;t þ γ FCi;t−1 −STDi;t þ θINCi;t þ ei;t

ð3Þ

One would expect that if last period's consumption exceeded the current period's standards there would be downward pressure on this period's consumption, hence γ would be negative. Eq. (3) can be rearranged to give an autoregressive formulation: FCi;t ¼ mi þ ðλ þ γ ÞFCi;t−1 þ αTRi;t þ βpi;t −γSTDi;t þ θINC i;t þ ei;t

ð4Þ

Estimation of the dynamic models of Eqs. (2) and (4) has to be treated with care if a panel is to be estimated. The presence of a lagged dependent variable among the regressors means that not only the OLS estimator but also the usual ‘within’ estimator is biased and inconsistent because the lagged endogenous 10 There are several reasons why income might impact fleet-average FC in different ways, hence the magnitude and direction of this effect may not be a priori obvious. Cars that consume more fuel may be bigger and more luxurious (positive income effect) or older and not technologically advanced (negative effect). Existing studies provide conflicting evidence: Dahl (1995) and Johansson and Schipper (1997) find a negative impact of income on fuel consumption, Espey (1996) finds income to be insignificant, while Storchmann (2005) finds a small but significant and positive income elasticity. These studies, however, focus on fleet-average FC and not new-car FC; new-car FC does not depend on the age distribution of the existing car stock, hence a negative income effect in our model would not be probable.

2666

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

Table 3 Regression results for Eq. (2) Countries

λ

Time trend

Dummy⁎ trend

Price

Dummy⁎ price

Income

R2

Obs.

Australia

0.364 ⁎ [1.750] 0.639 ⁎⁎⁎ [65.240] 0.278 ⁎⁎ [2.280] 0.489 ⁎⁎⁎ [5.150] 0.761 ⁎⁎⁎ [15.950] 0.330 ⁎ [2.080] 0.738 ⁎⁎⁎ [98.940] 0.579 ⁎⁎⁎ [3.590] 0.741 ⁎⁎⁎ [4.620]

− 0.004 − [1.040] − 0.003 ⁎⁎ − [2.110] − 0.004 ⁎ − [2.080] − 0.014 ⁎⁎ − [2.820] 0.004 [0.980] − 0.006 − [1.240] − 0.002 ⁎⁎⁎ − [21.240] − 0.003 − [0.830] 0.000 − [0.040]

−0.002 −[1.550] 0.004 ⁎⁎ [9.820] 0.001 [0.120] 0.014 ⁎⁎⁎ [3.670] −0.005 ⁎⁎ −[2.280] −0.013 ⁎⁎ −[2.780] −0.005 ⁎⁎⁎ −[11.240] −0.014 −[1.600] −0.007 −[1.130]

−0.090 ⁎ −[1.880] −0.084 ⁎⁎⁎ −[3.740] −0.213 ⁎⁎ −[2.330] −0.057 −[0.640] −0.079 ⁎⁎⁎ −[3.190] −0.171 ⁎⁎ −[3.030] −0.165 ⁎⁎⁎ −[8.350] −0.163 ⁎ −[1.760] −0.210 −[1.450]

0.011 [1.470] −0.014 ⁎⁎⁎ −[3.510] 0.011 [0.690] −0.082 ⁎⁎⁎ −[3.430] 0.023 ⁎⁎ [2.170] 0.062 ⁎⁎ [2.830] 0.029 ⁎⁎⁎ [11.980] 0.055 [1.650] 0.046 [1.080]

0.077 [0.410] 0.142 [1.470] 0.113 [1.030] 0.389 ⁎⁎ [2.570] −0.175 −[1.210] 0.272 ⁎ [1.830] 0.128 ⁎⁎⁎ [8.790] 0.149 [0.690] 0.134 [0.320]

0.947

24

Canada (cars + trucks) Canada (cars) Canada (light trucks) EU Japan US (cars + trucks) US (cars) US (light trucks)

44 0.906

23

0.875

23

0.988

23

0.960

20 54

0.981

28

0.900

28

Notes: See text for explanation of coefficients. Autocorrelation and heteroskedasticity robust t-statistics are shown in parentheses. ⁎, ⁎⁎ and ⁎⁎⁎ denote significance at 10%, 5% and 1% level respectively. Models with one cross-section were estimated with OLS; models with two or more cross-sections (cars + trucks for Canada and the US) were estimated with the Arellano–Bond GMM estimator.

variable is correlated with the country effect νi. We therefore estimated all panel models using the Arellano–Bond GMM estimator. The hypothesis of no second-order autocorrelation in the residuals, which is fundamental for the appropriateness of the Arellano–Bond estimator, could not be rejected in any one of the estimations we present in the paper; this ensures that estimations are consistent and unbiased. We estimated Eqs. (2) and (4) separately for each available data series, with all variables except the time trend expressed in natural logarithms11. In the case of Canada and the US we also estimated the corresponding models with both cross-sections available for each country, cars and light trucks. The reason is that since both cars and trucks are sold in the same country, it is reasonable to expect that there will be a common response of consumers towards income and prices regardless of vehicle type. Results for Eq. (2) are presented in Table 3. The lagged dependent variable is significant in all cases and ranges from 0.33 (for Japan) to 0.76 (for the EU). This implies that between 24% and 67% of the long-term adjustment of fuel consumption due to prices, income and standards takes place in the first year. This quite high adjustment rate is expected because new-car FC is the dependent variable; in contrast, Espey (1996), using fleet-average FC, found a 6% annual adjustment. Fuel prices are significant in most cases, ranging from −0.08 (for the EU and the combined Canada sample) to −0.21 (for cars in Canada). This is the short-run elasticity. Long-run price elasticities not shown in Table 3, i.e. the short-run ones divided by (1 − λ), vary from −0.14 (for Australia) to −0.63 (for the combined US sample). Income comes out as significant only in the cases of Canada (light trucks), Japan and the US (cars + trucks); significant short-run income elasticities range from 0.13 to 0.39, long-run ones from 0.41 to 0.76. As regards the two interaction variables that try to capture an eventual structural change, at least one of them turns out to be significant in most cases, with the exception of Australia, Canada (cars) and the US (in the two individual samples of cars and light trucks). In Europe, Japan and the US (combined sample) the adoption of tighter standards leads to a faster decrease of the fuel consumption variable over time. For example, after the enactment of CO2 targets in Europe and FE standards in Japan, it seems that new-car fuel consumption has started decreasing with a time trend of 0.5% per year in Europe and 1.3% in Japan, compared to an insignificant ‘pre-standard’ trend in both regions. Similarly, fuel consumption in the US decreased at an annual rate of 0.5% faster during the 1978–1985 period when standards were continuously tightened, compared to the post-1985 period when CAFE standards were stagnant. At the same time fuel 11 The logarithmic form has been used widely in the literature we are citing in this paper, and helps derive elasticities directly from the estimation results. Preliminary tests using a linear specification did not lead to significantly different results.

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

2667

Table 4 Regression results for Eq. (4) Countries

λ

Time trend

Price

STD

Income

R2

Australia

0.411 ⁎ [2.020] 0.626 ⁎⁎⁎ [6.990] 0.363 [1.430] 0.451 ⁎⁎⁎ [3.750] 0.731 ⁎⁎⁎ [14.170] 0.318 ⁎ [2.110] 0.650 ⁎⁎⁎ [4.310] 0.855 ⁎⁎⁎ [9.310] 0.229 [0.770] 0.676 ⁎⁎⁎ [13.330]

−0.005 −[1.630] −0.002 ⁎⁎⁎ −[3.660] −0.003 −[1.580] 0.005 [0.790] 0.001 [0.170] −0.009 ⁎⁎ −[2.710] −0.003 −[1.230] −0.001 −[0.260] −0.009 −[1.560] 0.000 −[0.240]

−0.057 −[0.900] −0.160 ⁎ −[1.830] −0.064 −[1.060] −0.298 −[3.200] −0.093 ⁎⁎⁎ −[3.870] −0.166 ⁎⁎ −[2.640] −0.100 ⁎⁎⁎ −[4.020] −0.065 −[1.310] −0.144 ⁎⁎ −[2.100] −0.080 ⁎⁎⁎ −[4.080]

−0.063 −[0.470] −0.071 −[0.410] 0.044 [0.160] 1.193 [0.950] 0.210 ⁎⁎ [2.130] 1.005 ⁎ [1.970] 0.222 [1.070] −0.021 −[0.210] 0.902 ⁎⁎ [2.190] 0.166 ⁎ [1.800]

0.014 [0.110] 0.102 ⁎⁎ [2.580] 0.019 [0.170] −0.036 −[0.170] −0.090 −[0.670] 0.382 ⁎⁎ [2.640] 0.108 [0.730] 0.004 [0.020] 0.571 ⁎ [1.740] −0.004 −[0.050]

0.943

Canada (cars + trucks) Canada (cars) Canada (light trucks) EU Japan USA (cars + trucks) USA (cars) USA (light trucks) All countries

Obs. 24 44

0.873

23

0.805

23

0.986

23

0.950

20 54

0.977

28

0.926

28 162

Notes: See text for explanation of coefficients. Autocorrelation and heteroskedasticity robust t-statistics are shown in parentheses. ⁎, ⁎⁎ and ⁎⁎⁎ denote significance at 10%, 5% and 1% level respectively. Models with one cross-section were estimated with OLS; models with two cross-sections (cars + trucks for Canada and the US) and seven cross-sections (all countries) were estimated with the Arellano–Bond GMM estimator.

consumption has become less sensitive to fuel prices. An explanation for the price elasticity change may be that the auto industry gradually introduced more efficient cars to fulfill its CO2 commitment without letting its production decisions depend on oil prices as much as they did it in the past. At the same time, consumers may have opted for more energy saving cars because more of these car models were available in the market and also because they may have become more conscious of their energy consumption, thereby largely ignoring fuel prices. Finally, the interaction trend variable for Canada shows strange behavior, implying that the adoption of standards has led to an increase of otherwise decreasing fuel consumption; however, there are very few observations for years with tightening standards in Canada (particularly for light trucks), so that the validity of any conclusions is questionable in this case. We now turn to the estimation results of Eq. (4), shown in Table 4. Values of λ and the price and income elasticities are similar to those of Table 3. It is worth noting that for the US the long-term price elasticity is estimated at −0.29, compared to a value of approximately −0.19 that Austin and Dinan (2005) found on the basis of partial equilibrium simulations.12 In almost all cases the deterministic time trend was found to be statistically insignificant. In estimations of individual countries STD is significant only in Europe, Japan and US light trucks, and it is also significant (at the 10% level) for the whole sample. This provides some weak evidence that the enactment of standards has contributed to the decline of new-car fuel consumption worldwide. It is worth noting that separately identifying the effects of fuel prices and standards within a specification such as that in Eq. (4) is not easy because the two variables can be highly correlated. Adopting a FE standard for the first time or tightening an already existing standard often happens after an upsurge of oil prices, which makes policy makers and the public more energy conscious. This is true, for example, of CAFE standards in the US in the late 1970s and early 1980s and voluntary targets in Australia and Canada during the same period. On the other hand, the correlation between standards and fuel prices is limited by

12 The elasticity that Austin and Dinan (2005) find is 0.22 and is positive because it refers to the effect on fuel economy, whereas we use fuel consumption as a variable; see also footnote 1. For fuel economy figures around 20–25 mpg, if we convert FE to fuel consumption and re-calculate the elasticity, it falls close to – 0.19.

2668

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

the fact that standards are unlikely to be relaxed when fuel prices drop. The fact that in some cases both coefficients are precisely estimated suggests that the collinearity problem is not severe. In sum, we have found evidence of a structural break in the process governing fuel economy that happens around the time that standards are imposed. A possible criticism of this finding is that an omitted variable may be causing both the break and the standards and what we are estimating is just spurious correlation. Although the variation in the timing of the standards and in the fluctuation of fuel prices makes it improbable that such an omitted variable exists, one way to address this criticism is to pool the data and estimate Eq. (2) in its panel form. This is essentially a difference-in-differences (DiD) estimator which can identify the impact of standards by exploiting the fact that they were imposed at different times in different countries. As a rule, a DiD estimation involves a ‘control’ population and a ‘treatment’ population, with a specific ‘treatment’ given to the latter population at some time. An equation is specified which assesses the difference in the evolution of the dependent variable in each one of the two populations before and after the ‘treatment’. The difference between these two differences determines the net effect of the ‘treatment’. In this way the DiD estimator can control for the existence of omitted variables with a global impact. Such a method is often employed to study the effect of energy and environmental regulations (see e.g. Bratberg et al., 2005; Horowitz, 2007). In our case, such a method would involve an assessment of the difference in new-car fuel consumption between countries that have implemented a FE standard and countries that do not have any standards in force. However, this is not possible: we are not aware of any new-car FE data available for countries that have not implemented FE standards. Only countries which have enacted FE regulations or voluntary targets monitor the average annual new-car fuel consumption of their cars. Therefore, a typical ‘control’ population cannot be found. It is still possible, however, to proceed in a somewhat different way. FE standards or targets have existed since the late 1970s in the US, Canada and Australia; these countries have been under a ‘with standards’ regime for the whole period we are examining, albeit with some periods of stagnating standards. Conversely, European countries and Japan, for which data are available since the 1980s, only adopted some kind of FE standards in the mid-1990s. In other words, these countries have been under a ‘no standards’ regime from the 1980s to the mid-1990s and under a ‘with standards’ regime from the mid-1990s until today. Taking the US, Canada and Australia as the ‘control’ population and European countries and Japan as the ‘treatment’ population, one can then give Eq. (2) a DiD interpretation. Table 5 displays this result. Both interaction variables are significant, though not very strongly, which confirms that a structural break appears during the first introduction of standards or the enactment of tighter standards. After this break, fuel consumption decreases at a faster rate and becomes somewhat less sensitive to fuel prices. One has to keep in mind, however, that the pooled estimation of Eq. (2) assumes that all countries have the same income and price elasticities — which, judging from the results of Tables 3 and 4, seems to be unlikely. 5. Some policy implications We now use our findings to assess the effectiveness of standards and fuel tax increases in curbing automobile fuel consumption. Results from the estimation of Eq. (2), shown in Table 3, reveal that under a ‘with standards’ regime new-car fuel consumption in Europe and Japan has decreased at an annual rate of 0.5% and 1.3% respectively, compared to a statistically insignificant trend in the ‘no standards’ period, and

Table 5 Regression results for DiD equation Countries

λ

Time trend

Dummy⁎ trend

Price

Dummy⁎ price

Income

Obs.

All countries

0.771 ⁎⁎⁎ [30.190]

−0.002 ⁎ −[1.690]

− 0.001 ⁎⁎ − [2.140]

−0.079 ⁎⁎⁎ −[3.310]

0.004 ⁎ [1.700]

0.052 [1.020]

140

Notes: See text for explanation of coefficients. Estimation was carried out with the Arellano–Bond GMM procedure. Robust t-statistics are shown in parentheses. ⁎, ⁎⁎ and ⁎⁎⁎ denote significance at 10%, 5% and 1% level respectively.

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

2669

Fig. 3. Transportation energy consumption in European countries, 1960–2005. Source: International Energy Agency online databases (http://data.iea.org).

has become less sensitive to fuel prices: the ‘pre-standard’ price elasticities decreased in absolute terms in the ‘with standards’ period. A similar result is found for the US, but only in the sample comprising both cars and trucks. In Europe and Japan, if we vary the time of change in regime by ±1 year (i.e. 1994 or 1996 instead of 1995) results (not shown here) remain essentially the same. All this provides an indication that the enactment of CAFE in the US, fuel economy standards in Japan and the auto industry's voluntary CO2 commitment in Europe have made a difference in the evolution of automobile fuel consumption over the years. In the absence of such policies, new-car fuel consumption would not have improved at the observed rates. Furthermore, bearing in mind the worldwide increase in vehicle kilometers traveled during the last decades, total fuel consumption would also have been considerably higher in the absence of standards. This finding does not imply that there are no better alternatives to FE standards but that, ceteris paribus, fuel economy would have been lower and total fuel use would have been higher without them. Αs illustrated in Fig. 3, despite stabilization of transportation energy use in countries like France, Germany and Japan in very recent years, fuel consumption is generally still rising in most developed countries. This shows that even standards or voluntary commitments by themselves are not sufficient for reversing the trend in fuel use. A combination of multiple policies may be necessary to bring the desired effect, such as fuel taxes (where they are still low, e.g. in the US), CO2-based vehicle taxation systems and long-term interventions in public transport infrastructure, land use etc. With the aid of Table 4 one can assess how much fuel taxes have to increase (in order to achieve the same impact as currently discussed standards) if standards themselves are not to be tightened in the future. It is not possible to make such a calculation for the US as the STD coefficient is insignificant — which, as explained in the previous section, has to be attributed to collinearity between standards and fuel prices rather than the ineffectiveness of standards. We will therefore focus on Europe, where Table 4 shows that FE standards/targets enter significantly in Eq. (4), with a coefficient that is 2.3 times higher than that of fuel price. At a first glance this means that in the EU, where the currently discussed 2012 standard of 130 g CO2/ km constitutes a 19% decrease compared to about 160 g CO2/km realized in the year 2004, if standards are not to be tightened then retail fuel prices will have to rise by 44% in order to have an equivalent effect. However, to make a proper comparison between standards and taxes in terms of their long-term potential to reduce total fuel consumption, one needs to correct the calculation above in order to account for two facts. One is the so called rebound effect: a tighter standard would lower automobile running costs, thereby inducing more driving and hence saving less fuel than expected. It is reasonable to assume a rebound effect of 20%, i.e. that the total distance traveled annually by a car increases by 2% for every 10% improvement in fuel economy (Greene et al., 1999); this means that the overall effect of a 19% tighter standard on fuel

2670

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

consumption would come down to about 15% for every new car13. Over a period of 10–15 years (when most cars would have been replaced by the new fuel-efficient ones) such a standard would yield total fuel savings of approximately 15%. The second fact is that, in contrast to the rebound effect caused by tighter standards, higher fuel prices discourage driving, and this applies to all cars (both new and in-use ones). Therefore, higher fuel prices can yield additional fuel savings compared to a regulation applied to new cars only. Goodwin et al. (2004) have reviewed the related literature and report a long-term elasticity of vehicle kilometers with respect to fuel prices of −0.3. What would be the long-run effect of the two policies (tighter standards vs. higher fuel taxes) on total fuel consumption if we account for these facts? A 19% tighter standard, inducing 15% fuel savings in the long run as explained above, would be equivalent to an increase in retail fuel prices of 24%. At an EU average gasoline price of 1.05 Euros per liter in year 2005, this increase amounts to a price rise of 25 Eurocents per liter; for comparison, the EU average excise tax on gasoline was about 50 Eurocents per liter in 2005 (Eurostat, 2006). To conclude, if the 130 g CO2/km target had to be abandoned a 50% increase in fuel taxes would be necessary to induce similar fuel savings in Europe. The political feasibility of imposing such a tax rise is questionable, and it would also lead to European fuel taxes being too high compared to the total externalities they are supposed to tackle (Parry and Small, 2004).14 A final remark is in order concerning the question whether fuel economy can improve due to autonomous technological progress alone. Official US studies assume almost zero autonomous FE improvement in their ‘business as usual’ scenarios (EIA, 2006). Conversely, several European long-term models assume that European automobile fuel economy will continue to improve at fast rates even without post-2010 FE regulations, in ‘reference case’ scenarios that assume real oil prices below $50 per barrel (see e.g. Zachariadis, 2006). According to the results shown in Tables 3 and 4 for the EU and the US, the time trend is not significantly different from zero — except for the combined US sample in Table 3, where the time trend is significant but very close to zero. Note that this time trend is intended to capture the composite effect of autonomous technical progress, i.e. progress that is not induced by high energy prices nor by FE standards, and other factors that are not addressed by the explanatory variables. Bearing this in mind, a zero time trend does not mean that there has been no autonomous technical progress over the last 25 years, but that changing consumer preferences have led the automotive industry to exploit efficiencyimproving technological advances in ways that do not yield fuel savings (thus allowing improved comfort and performance without reducing fuel economy). The major policy implication of such a result is that, without stricter FE standards and at oil prices around $40–60 (at 2005 prices) per barrel, one should not expect any marked FE improvements in the future in the absence of major technological breakthroughs or an economic recession. This finding is also in line with the assumptions made in other international studies (e.g. WBCSD, 2004). Evidently, if oil prices observed in early 2008, i.e. above $100 per barrel, are to remain over the longer term, technological progress should be anticipated — but this would be price induced rather than autonomous progress. 6. Conclusions To our knowledge, this is the first study that attempts to explore econometrically the impact of automobile fuel economy regulations around the world and to compare it with the effect of fuel prices, including all countries that have implemented some type of FE standards for a substantial period of time. Using data from official sources, we built an unbalanced panel comprising 178 observations from Australia, Canada, the European Union, Japan and the United States spanning a period between 1975 and

13 Small and Van Dender (2007) estimate that the rebound effect diminishes as private income rises and find a long-run rebound effect of 10% for California. However, it is reasonable to assume a higher rebound effect for Europe because more densely populated urban areas and better alternatives to private car travel make driving in Europe more sensitive to changes in automobile running costs than in the US. 14 One could argue that at a gasoline price of 1.3 Euros per liter, which was approximately the EU average in spring 2008 because of soaring crude oil prices, the necessary price increase of 25 Eurocents per liter has already taken place compared to the 2005 price of 1.05 Euros per liter. However, it would be necessary to keep retail fuel prices at this level over the longer term in order to have comparable results with those of a permanent FE regulation.

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

2671

2003. Our empirical strategy was to specify reduced form dynamic models in order to look for structural breaks in the process occurring around the time of the introduction of standards. Average annual fuel consumption of new cars was the dependent variable, and explanatory variables were gasoline prices, the FE standard or target and per capita income. We first analyzed each time series independently and then pooled the series from all countries and employed a difference-in-difference estimator. A key difference of our implementation from standard DiD techniques is that in our data everyone gets a treatment (imposes a standard); identification comes from the fact that standards were implemented in different countries at different points in time. We found that the enactment of some regulatory measure, either mandatory or voluntary, has clearly contributed to FE improvements in the US, Europe and Japan. At the same time, new-car fuel economy became less sensitive to fuel prices after the adoption of standards. DiD estimation for the whole sample confirmed that the first introduction of standards or the adoption of tighter standards is associated with a structural break in the evolution of fuel economy. Moreover, in Europe and Japan the impact of a FE standard on new-car fuel consumption was found to be more pronounced than that of a rise in fuel prices. Based on these estimates, we addressed three important policy issues. First, there seems to be sufficient evidence that if there were no FE standards or targets in force, new-car fuel economy would not have improved at the rates that have been observed in Europe and Japan in recent years, and this would most probably have happened in the US as well; as a result, transportation energy use would have increased more rapidly. Second, in order to attain the desired FE improvements without imposing any further standards or voluntary targets in Europe, fuel taxes would have to increase by 50%. Third, without higher fuel prices and/or tighter FE standards, one should not expect any marked improvements in fuel economy under ‘business as usual’ conditions. Potential fuel savings due to autonomous technical progress in the past have been counterbalanced by changes in consumer preferences towards more comfortable and powerful cars, and there is no reason to believe why this trend should not continue in the future in the absence of impressive technological breakthroughs or an economic recession. Our analysis shows that the question “standards or prices?” cannot be answered in a definite way for all world regions. In the US tighter FE standards and higher gasoline taxes need to be carefully examined against their welfare impact, and a combination of both policy options should not be excluded in view of the many uncertainties about the effectiveness and the side-effects of each measure. Conversely, regulations seem to be a more feasible option for Europe and Japan as it is hardly possible to increase fuel taxes because of their already high levels; how these regulatory measures will be designed and implemented, however, is crucial in order to avoid welfare losses for producers or consumers. 7. Uncited references An and Sauer, 2005 EC (European Commission), 2005 European Conference of Ministers of Transport, 2003 Eurostat (European Statistical Service), 2006 IEA (International Energy Agency), 2001 IEA (International Energy Agency), 2005 Acknowledgments The second author has conducted his part of the research in the framework of a Marie Curie Fellowship that he has received from the European Community within its Sixth Framework Programme. Earlier versions of the manuscript were presented in an academic seminar at the University of Cyprus, at the 29th IAEE Conference in Berlin, the 6th BIEE Academic Conference in Oxford, the 2006 International Energy Workshop in Cape Town, the 2007 ECEEE summer study in La Colle sur Loup, France, and the 2007 CRETE conference in Naxos, Greece; we greatly appreciate all comments we received. We also thank David Greene, Lee Schipper, Kenneth Small and Karl Storchmann for their comments on earlier versions of this paper as well as two anonymous referees of this journal. Views expressed in this article and remaining errors or omissions are the sole responsibility of the authors.

2672

S. Clerides, T. Zachariadis / Energy Economics 30 (2008) 2657–2672

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j. eneco.2008.06.001. References An, F., Sauer, A., 2005. Comparison of passenger vehicle fuel economy and greenhouse gas emission standards around the world. Report prepared for the Pew Center on Global Climate Change, Arlington, VA. Austin, D., Dinan, T., 2005. Clearing the air: the costs and consequences of higher CAFE standards and increased gasoline taxes. J. Environ. Econ. Manage. 50, 562–582. BP Statistical Review of World Energy, London. June. Bratberg, E., Tjøtta, S., Øines, T., 2005. Do voluntary international environmental agreements work? J. Environ. Econ. Manage. 50, 583–597. Dahl, C.A., 1995. Demand for transportation fuels: a survey of demand elasticities and their components. J. Energy Lit. 1 (3–27). Dargay, J.M., Vythoulkas, P.C., 1999. Estimation of a dynamic car ownership model. J. Transp. Econ. Policy 33, 287–302. EC (European Commission), 2005. Monitoring of ACEA/JAMA/KAMA commitment on CO2 emission reductions from passenger cars. Working Paper COM(2005) 269 final, Brussels. Also at http://europa.eu.int/comm/environment/co2/co2_monitoring.htm. EC (European Commission), 2007. Results of the review of the community strategy to reduce CO2 emissions from passenger cars and light-commercial vehicles. Document COM(2007) 19 final, Brussels. February, Also at http://www.ec.europa.eu/environment/co2/ pdf/com_2007_19_en.pdf. ECMT (European Conference of Ministers of Transport), 2003. Monitoring of CO2 emissions from new cars. Document No. CEMT/CM (2003)10, Paris. March. EEA (European Environment Agency) The European environment — state and outlook 2005, State of Environment report No 1/2005, Copenhagen. Also at http://reports.eea.eu.int/state_of_environment_report_2005_1/en/tab_content_RLR. EIA (US Department of Energy, Energy Information Administration), 2006. Annual energy outlook 2006. DOE/EIA-0383(2006), Washington, DC. Espey, M., 1996. Watching the fuel gauge: an international model of automobile fuel economy. Energy Econ. 18, 93–106. Eurostat (European Statistical Service), 2006. Energy and Transport in Figures. Statistical pocketbook, Luxembourg. Eurostat, 2006. NewCronos database. Available on CD-ROM and on the World Wide Web, Luxembourg. Gately, D., 1992. Imperfect price-reversibility of U.S. gasoline demand: asymmetric responses to price increases and declines. Energy J. 13, 179–207. Glazer, A., Lave, C., How regulations can succeed where taxes do not: an examination of automobile fuel efficiency, mimeo, University of California, Irvine, 1994. Also at http://ideas.repec.org/p/wpa/wuwppe/9406002.html [accessed August 29, 2007]. Goldberg, P., 1998. The effects of the corporate average fuel efficiency standards in the US. J. Ind. Econ. XLVI, 1–33. Goodwin, P., Dargay, J., Hanly, M., 2004. Elasticities of road traffic and fuel consumption with respect to price and income: a review. Transp. Rev. 24, 275–292. Greene, D., 1990. CAFE or price? An analysis of the effects of federal fuel economy regulations and gasoline price on new car MPG, 1978–89. Energy J. 11, 37–57. Greene, D., 1998. Why CAFE worked. Energy Policy 26, 595–613. Greene, D.L., Liu, J.T., 1988. Automotive fuel economy improvements and consumers' surplus, Transportation Research Part A. Policy Pract. 22A, 203–218. Greene, D.L., Kahn, J.R., Gibson, R.C., 1999. Fuel economy rebound effect for U.S. household vehicles. Energy J. 20, 1–31. Greene, D.L., Patterson, P.D., Singh, M., Li, J., 2005. Feebates, rebates and gas-guzzler taxes: a study of incentives for increased fuel economy. Energy Policy 33, 757–775. Hellmann, K.H., Heavenrich, R.M., 2004. Light-duty automotive technology and fuel economy trends: 1975 through 2004. Report EPA420-R-04-001, US Environmental Protection Agency, Ann Arbor, Michigan. Also at http://www.epa.gov/otaq/fetrends.htm. Horowitz, M., 2007. Changes in electricity demand in the United States from the 1970s to 2003. Energy J. 28, 93–119. Hughes, J.E., Knittel, C.R., Sperling, D., 2006. Evidence of a shift in the short-run price elasticity of gasoline demand. NBER Working Paper 12530. National Bureau of Economic Research, Cambridge, MA. September. IEA (International Energy Agency), 2001. Saving Oil and Reducing CO2 Emissions in Transport — Options and Strategies, Paris, France. IEA (International Energy Agency), 2005. Energy Prices and Taxes, Paris, France. IEA (International Energy Agency), 2006. World Energy Outlook, Paris, France. Johansson, O., Schipper, L., 1997. Measuring the long-run fuel demand of cars. J. Transp. Econ. Policy 277–292 September. Kleit, A.N., 2004. Impacts of long-range increases in the fuel economy (CAFE) standard. Econ. Inq. 42, 279–294. NRC (US National Research Council), 2002. Effectiveness and Impact of Corporate Average Fuel economy (CAFE) Standards. National Academy of Sciences, Washington, DC. Parry, I.W.H., Small, K.A., 2004. Does Britain or the United States have the right gasoline tax? Discussion Paper 02-12 rev., Resources for the Future, Washington, DC. Parry, I.W.H., Fischer, C., Harrington, W., 2005. Do Market Failures Justify Tightening Corporate Average Fuel Economy (CAFE) Standards? Resources for the Future, Washington, DC. Portney, P.R., Parry, I.W.H., Gruenspecht, H.P., Harrington, W., 2003. The Economics of fuel economy standards. J. Econ. Perspect.17, 203–217. Small, K.A., Van Dender, K., 2007. Fuel efficiency and motor vehicle travel: the declining rebound effect. Energy J. 28, 25–51. Storchmann, K., 2005. Long-run gasoline demand for passenger cars: the role of income distribution. Energy Econ. 27, 25–58. Thorpe, S.G., 1997. Fuel economy standards, new vehicle sales, and average fuel efficiency. J. Econ. Perspect. 11, 311–326. Turrentine, T.S., Kurani, K.S., 2007. Car buyers and fuel economy? Energy Policy 35, 1213–1223. WBCSD (World Business Council for Sustainable Development), 2004. Mobility 2030: Meeting the Challenges to Sustainability, Geneva, Switzerland. Zachariadis, T., 2006. On the baseline evolution of automobile fuel economy in Europe. Energy Policy 34, 1773–1785.

The effect of standards and fuel prices on automobile ...

Available online 19 June 2008. There is an intense debate over ... Tel.: +357 22. 892425; fax: +357 22 892426. ...... Eurostat (European Statistical Service), 2006.

565KB Sizes 1 Downloads 185 Views

Recommend Documents

Automobile Prices, Gasoline Prices, and Consumer ...
We use a comprehensive set of manufacturer incentives to construct .... (2) where the term αjk is a demand parameter and the terms xkt and µjt capture ... price rule makes it clear that the equilibrium price of a vehicle depends on its characterist

The Effect of Credit Availability on House Prices
Apr 27, 2015 - the housing market come through agencies that act to increase liquidity in the mortgage .... Second, the original list price is highly predictive.

The Effect of Credit Availability on House Prices
Nov 30, 2014 - Second, the original list price is highly predictive .... In the rest of the analysis of this paper, I will focus on successfully matched ... 3 Descriptive evidence ... In contrast, there is no discernable difference between the pre-.

The Effect of Credit Availability on House Prices ...
Apr 27, 2015 - ∗I thank Patrick Bayer and Duke University for providing the data. Additional thanks to ... UCLA, Los Angeles, CA. 1 .... that the increase in CLLs in San Francisco and Los Angeles raised prices for the homes that are most .... then

Automobile Prices, Gasoline Prices, and Consumer ...
mobile manufacturers adjust their prices cyclically over vehicle model-years (e.g., Copeland, ... 22We use one-month futures contracts for reformulated regular gasoline at the ... Finally, a comparison of coefficients across columns suggests.

The Changing Effects of Energy Prices on Economic Activity and ...
The Changing Effects of Energy Prices on Economic Activity and Inflation.pdf. The Changing Effects of Energy Prices on Economic Activity and Inflation.pdf.

Re-examining the effect of oil prices on economic ...
The new exchange rate regime announced by the Russian Central Bank in. November 2014 is ... I use quarterly data on the Russian GDP, the price of Urals oil and the rouble/US .... Taking these considerations into account, this section ex-.

Alternative Transportation Fuel Standards: Welfare ...
2. The transportation sector in the US accounted for 29% of total US ... Renewable Fuel Standard (RFS) in the US established by the Energy ...... Here we describe the data sources for the transportation fuel sector, which include the demand for ...

Geometric Scale Effect of Flow Channels on Performance of Fuel Cells
This paper studies the effect of flow channel scaling on fuel cell performance. In particular, the impact of dimensional scale on the order of 100 micrometers and below has been investigated. A model based on three-dimensional computational flow dyna

Doing and Learning: The Effect of One on the Other
Technion, Israel Institute of Technology, Department of Education in Technology and Science. Introduction. During the last ... constructing, and debugging some amazing machinery, their social and communication skills improved. Another .... o A change

Doing and Learning: The Effect of One on the Other
design, to the amazement of his care-takers. Thinking and. Multi-Media. Doppelt, MA. Thesis, in press. 10 th graders, judged and classified by school system to low- level classes. Tagged as un-able to pass matriculation. Considered at school as troub

The effect of coherence and noise on the ...
LFMs, is shown to create large side lobes in the time domain. Alternative ..... free numerical simulations produce very similar focal patterns as shown in Fig.

The Effect of Crossflow on Vortex Rings
The trailing column enhances the entrainment significantly because of the high pressure gradient created by deformation of the column upon interacting with crossflow. It is shown that the crossflow reduces the stroke ratio beyond which the trailing c

The Effect of Crossflow on Vortex Rings
University of Minnesota, Minneapolis, MN, 55414, USA. DNS is performed to study passive scalar mixing in vortex rings in the presence, and ... crossflow x y z wall. Square wave excitation. Figure 1. A Schematic of the problem along with the time hist

EFFECT OF HIGH CALCIUM AND PHOSPHORUS ON THE ...
EFFECT OF HIGH CALCIUM AND PHOSPHORUS ON THE GROWTH.pdf. EFFECT OF HIGH CALCIUM AND PHOSPHORUS ON THE GROWTH.pdf. Open.

Effect Of Ecological Factors On The Growth And Chlorophyll A ...
Effect Of Ecological Factors On The Growth And Chlor ... ed Kappaphycus alvarezii In Coral Reef Ecosystem.pdf. Effect Of Ecological Factors On The Growth And ...

The effect of production system and age on ...
(P < 0.05). Aspects of the fatty-acid patterns that are of relevance to human nutrition tended to favour the .... Data analysis employed a block design within the.

Selection and the effect of smoking on mortality
death T, and smoking S, controlling for observed individual characteristics X and .... Within this large data set, Statistic Sweden has constructed a smaller panel.

Effect of voriconazole on the pharmacokinetics and ...
of voriconazole (Vfend tablet; Pfizer, New York, NY) .... SD, except for tmax data, which are given as median and range. ..... Measurements of recovery from.

The Effect of Emotion and Personality on Olfactory ...
Mar 23, 2005 - 1Psychology Department, Rice University, Houston, TX, USA and ... Chen, Psychology Department MS-25, Rice University, 6100 Main St., ..... on top of a television set 51 cm away from the subject. ..... and autonomic alteration by admini