THE CLIMATE IMPACT OF AMERICA’S AUTOMOBILES

Global Warming on the Road

Global Warming on the Road

THE CLIMATE IMPACT OF AMERICA’S AUTOMOBILES

AUTHORS

John DeCicco and Freda Fung

ENVIRONMENTAL DEFENSE

WITH THE ASSISTANCE OF

Feng An

ENERGY AND TRANSPORTATION TECHNOLOGIES, LLC

Cover image: Corbis.

Our missionEnvironmental Defense is dedicated to protecting the environmental rights of allpeople, including the right to clean air, clean water, healthy food and flourishingecosystems. Guided by science, we work to create practical solutions that win last-ing political, economic and social support because they are nonpartisan, cost-effective and fair.

©2006 Environmental Defense

100% recycled (100% post-consumer) totally chlorine free paper

The complete report is available online at www.environmentaldefense.org.

iii

Executive summary iv

Introduction 1

Methodology and assumptions 4

Carbon emissions from automobiles on the road 6

Beyond the carbon tally 13

Conclusions 21

Appendix A: Stock modeling methodology and assumptions 22

Notes 27

References 30

Contents

iv

For most Americans, the automobile isan essential part of daily life. It has shapedour culture and our landscape. The in-dustries that build and serve cars form akey part of our economy. The automobileis not without its faults, but they oftenare concealed by the styling, performanceand other features that make today’svehicles so desirable. Still, when aproduct is so widely used, its faults canadd up to massive unwanted side effects.

Consider global warming. Motorvehicles play a major part in whatscientists call the most serious environ-mental problem the world faces. Theautomobile’s main contribution comesfrom the carbon dioxide (CO2) emittedas the engine burns fuel.

The greenhouse gas pollution thatcauses global warming comes fromnumerous sources. Any single contribu-tion may seem small in proportion to theworld total, but collectively it becomes aproblem of vast scale. To address a prob-lem of such vast scale requires interna-tional agreements and national policies.But making good on such commitmentswill require changes in how we manageevery activity that contributes to theproblem. How much should we cut, andwhere? To answer this question, we needa clear picture of which sources con-tribute most to global warming.

The disproportionate impact

of U.S. cars and light trucks

Automobiles—by which we meanpersonal motor vehicles, including lighttrucks such as pickups, SUVs and vansas well as sedans and wagons—emitroughly 10% of global CO2 emissionsfrom fossil fuels, which are the mainform of greenhouse gas pollution.American automobiles have a dispro-

Executive summary

portionate impact: U.S. cars are drivenfurther each year and burn more fuelper mile than the international average.The United States has 5% of the world’spopulation and 30% of the world’s auto-mobiles, but it contributes 45% of theworld’s automotive CO2 emissions.

In 2004, U.S. cars and light trucksemitted 314 million metric tons ofcarbon-equivalent (MMTc). That equalsthe amount of carbon in a coal train50,000 miles long—enough to stretch 17times between New York and San Fran-cisco. In fact, the amount of CO2 emittedfrom oil used for transportation in theUnited States is similar to the amountfrom coal used to generate electricity.

America’s ‘rolling carbon’

This report details, for the first time,the global warming pollution emittedin the course of America’s daily driving.It represents a complete picture of thenation’s “rolling carbon,” from the latestluxury SUV cruising through an upscalesuburb to the oldest pickup truck bump-ing along a rural lane.

Many policy-related reports focus onemissions from new vehicles, since newvehicle sales are regularly reported indetail. Here, using national statistics onhow long vehicles last and how far theyare driven as they age, we examine theCO2 emissions from all personal vehicles,both new and used. This analysis allows usto answer questions such as, “How muchcarbon is emitted by all the Ford-built carsnow in use? From all the Hondas?” It alsoenables us to estimate the CO2 emissionsfrom all the sport-utility vehicles (SUVs)in use, from all the midsize cars, and soon—that is, according to vehicle class.

Table ES1 summarizes on-road car-bon emissions by vehicle class in 2004.

v

Perhaps surprisingly, small cars (com-pacts, subcompacts and two-seaters)were responsible for the most carbonemissions, amounting to 77 MMTc.Small cars once had been the dominantsegment by sales; given the longevity ofvehicles, many of them remain on theroad today. Thus, despite their higherthan average fuel economy, small carsstill accounted for the largest share ofrolling carbon emissions as of 2004.

This situation illustrates the relativedurability of automobiles: In terms ofusage, the U.S. light vehicle stock nowhas a “half-life” of roughly eight years;in other words, 50% of vehicles arereplaced within that time. It takes16 years for the fleet to be 90% replacedin terms of the carbon emitted duringdriving. In short, the choices maderegarding new vehicles influence emis-sions for many years to come.

SUVs represent the second largestportion of rolling carbon. They soonwill be the main source of automotiveCO2 emissions, having overtaken smallcars in terms of market share in 2002.Their impact will be all the greater dueto their lower than average fuel economy.As of 2004, all the SUVs on the road inthe United States emitted 67 MMTc,an amount equivalent to the CO2

spewed by 55 large coal-fired power

plants. Next were pickup trucks, whichcollectively emitted 60 MMTc in 2004.

Car companies vs. electric

companies: a comparison

Table ES2 gives the breakdown by carcompany, which shows that automakers’shares of rolling carbon follow historicalmarket shares, led by GM. These resultsare highlighted in Figure ES1, whichshows the quantity of carbon emitted byeach automaker’s products compared tothe carbon emissions of major electricpower companies. Car companies areon a par with—and some even higherthan—electric utilities in terms of thecarbon emissions associated with theuse of their products.

These estimates indicate that productsmade by each of the Big Three—GM,Ford and DaimlerChrysler—emittedmore carbon in 2004 than the country’slargest electric utility, American ElectricPower (AEP). GM’s products emitted99 MMTc, and Ford’s emitted 80MMTc, nearly twice AEP’s 41 MMTc.DaimlerChrysler’s cars and light trucksin use emitted 51 MMTc. The totalCO2 emissions from Big Three auto-mobiles in 2004 was comparable to thetotal from the top 11 electric companies.And the global warming pollution from

TABLE ES1

Rolling stock carbon emissions by vehicle class, 2004

Vehicle class Carbon emissions (MMTc) Carbon emissions share

Small cars 77 25%

SUVs 67 21%

Pickups 60 19%

Midsize cars 54 17%

Vans 29 9%

Large cars 26 8%

Cars 157 50%

Light trucks 157 50%

Overall 314 100%

vi

vehicles made by Toyota (the fourthamong automakers) edges out that fromthe Tennessee Valley Authority (thethird among electric companies).

Electric utilities and automakers mayseem like apples and oranges, but in factthey make a suitable comparison. In

both cases, their carbon emissionsinclude the impacts of long-livedproducts and equipment. The quantityof each sector’s emissions also is deter-mined by a number of actors, includingend-users such as consumers and busi-nesses. For example, AEP’s CO2 emis-

0 20 40 60 80 100 120

NissanProgress Energy

Edison InternationalDominion

CinergyAmeren

XcelHonda

Tennessee Valley AuthorityToyota

Southern CompanyAmerican Electric Power

DaimlerChryslerFord

General Motors

Carbon emissions (MMTc)

Major automakers’ productsMajor electric producers

99

80

51

41

37

27

26

17

17

17

16

15

15

15

15

FIGURE ES1

Carbon emissions of major automakers’ products vs. major electric

producers in the United States, 2004

Source: Table 1 and CERES et al. (2006).

TABLE ES2

Rolling stock carbon emissions by automaker, 2004

Manufacturer Carbon emissions (MMTc) Carbon emissions share

GM 99 31%

Ford 80 25%

DaimlerChrysler 51 16%

Toyota 27 9%

Honda 17 6%

Nissan 15 5%

Volkswagen 5 2%

Hyundai 4 1%

Mitsubishi 4 1%

BMW 3 1%

Kia 2 1%

Subaru 3 1%

Others 4 1%

Big Three 230 73%

Overall 314 100%

vii

sions are the result of the electricityconsumed in homes, offices, schoolsand other power-using facilities in theirservice area. Decisions by state and localofficials are also important in both cases,through regulation of electric utilityservices and rates on one hand, andthrough transportation infrastructure andland use planning on the other. More-over, greenhouse gas pollution from bothsectors is influenced by the carbon in-tensity of their fuel supply. While eachutility has its own mix of fuels, overallutility emissions reflect the fact that 54%of U.S. electricity comes from coal. Auto-mobile fuel essentially all comes from oil.

The three-legged stool propping

up emissions

Because automotive CO2 emissions areinfluenced by many different decisionmakers, finding opportunities to cut car-bon entails looking at the factors behindemissions. Figure ES2 illustrates the threemain factors propping up rolling carbon:

• Travel demand, or the amount of driv-ing, which is commonly measured asannual vehicle miles of travel (VMT).

Total U.S. car and light truck VMT in2004 amounted to 2.6 trillion miles.

• Fuel use rate, or the amount of fuelconsumed per mile, which is theinverse of fuel economy. The fueleconomy of the U.S. automobile stockaveraged 19.6 mpg in 2004, implyingan average fuel use rate of 51 gallonsper 1,000 miles of driving.

• Fuel carbon content, or the amount ofgreenhouse gases emitted per gallon offuel consumed. Counting only the CO2

directly released when fuel is burned,this factor amounts to 5.3 pounds ofcarbon per gallon of gasoline.

These three factors combined result inthe 314 MMTc emitted by U.S. auto-mobiles in 2004.

Among these factors, only decreasesin fuel use rate (achieved by increasingfuel economy) have served to partiallylimit CO2 emissions. Fuel carbon con-tent has not significantly changed,reflecting the tenacity of oil as a usefulresource and the barriers facing alterna-tive fuels. And many forces drive traveldemand, all of them serving to pushauto sector CO2 emissions upward.

Managing automotive rolling

carbon

Ultimately, cutting auto sector CO2 emis-sions will require attention to all threelegs of the stool. Thus, policy discussionsregarding transportation and climate aretypically framed around the question ofhow to change travel demand, fuel econ-omy or fuel type. A key challenge, how-ever, is that each factor is the product ofdecisions made by many actors. In addi-tion to consumers, automakers and othersinvolved in the auto market, total autosector CO2 emissions are influenced bythe energy industry as well as by themany interests and levels of government

Vehicle CO2 emissions

Traveldemand Fuel use rate

Fuelcarboncontent

FIGURE ES2

Three main factors behind rolling

carbon

viii

that shape land use, infrastructure andother parts of the transportation system.

The magnitude of the automotiveglobal warming problem and the com-plexities of the associated decision makingsuggest a need to recast the question.Instead of asking “What policies arerequired to address the factors that propup automobile carbon emissions?” deci-sion makers can ask “What policies canmotivate each actor to address the aspectsof emissions that they can influence?”Such a view shifts the focus from tech-nical factors and how they are deter-mined to individual actors and the partsof the problem they each can control.

For some major actors, such as auto-makers, how much they can address emis-sions is limited by other players. Theability to increase fuel economy by re-designing vehicles depends in part onthe extent to which consumers makefuel economy a priority. Nevertheless, itis implausible that automakers (or otheractors) have no ability to lower carbonemissions. Similarly, while individualconsumers cannot themselves redesigncars (or develop new fuels or reshapeland use), their scope of influence is not

zero. Better information on emissions,new evaluation tools and education areneeded to enable all actors to understandhow their decisions impact emissions.

This report seeks to aid that process.By underscoring the magnitude of auto-motive CO2 emissions, it highlights theneed to break the problem down intomanageable pieces. Consideration canthen be given to opportunities to reducecarbon by tackling each piece of thelarger problem. Such a carbon manage-ment paradigm would enable each actorto address the emissions each influences,complementing existing policy strategiesdefined around the technical factors thatcharacterize total emissions. A citygovernment, for example, can changethe emissions associated with its trans-portation operations. It might do thisby carrying out its functions with lessdriving, increasing the efficiency of itsvehicles or using low-carbon fuels.

Exactly how to define a set of carbonmanagement policies appropriate forthe auto sector is a subject for futureresearch as well as an important topicfor discussion by everyone who holds astake in the car-climate problem.

1

America’s cars are one of the world’slargest sources of global warming pollu-tion. Our personal vehicles—from theMini Cooper to the Hummer and every-thing in between—emitted 314 millionmetric tons of carbon-equivalent(MMTc) annually as of 2004.1 Thatequals the amount of carbon in a coaltrain 50,000 miles long, enough tostretch 17 times between New Yorkand San Francisco.2 In the UnitedStates, in fact, the amount of globalwarming pollution from oil used fortransportation is similar to that fromcoal used for electric power generation.3

Such comparisons can help Ameri-cans understand the extent of our car-bon problem, and they also help usthink about how to solve it. Emissionsfrom power plants depend on decisionsby power companies, but also on deci-sions made by businesses and consumerswho choose and use electrical equip-ment and appliances. Similarly, auto-mobile emissions depend not only onautomakers and fuel suppliers, but alsoon decisions by businesses and con-sumers who choose vehicles, and bylocal, state and federal officials whooversee America’s transportation systemand land use. Whether it is an issue ofprimary energy resources (fossil orotherwise), how the fuels are convertedand distributed to consumers, or howefficiently the fuels are used throughchoices of technology and infrastruc-ture, the interlocking nature and sheermagnitude of carbon emissions impliesa shared task.

This report provides a detailed snap-shot of carbon emissions from the “roll-ing stock” of all U.S. cars and lighttrucks on the road in 2004. We reportresults both by automaker and marketsegment. The statistics compiled here

Introduction

enable us to answer questions such as“What portion of CO2 emissions isassociated with pickup trucks?” or“What portion of CO2 emissions isrepresented by all Toyota vehicles onU.S. roads?” The results also underscorethe scope of the car-carbon problem andshow how long it takes for changes inthe characteristics of new car sales toinfluence the overall carbon emissionsfrom the auto sector.

The approach taken here is similarto that of our previous reports4 on auto-motive carbon burdens, referring to theannual CO2 emissions associated with agiven population of vehicles. However,while our Automakers’ Corporate CarbonBurdens reports calculated CO2 emis-sions as annual averages over the ex-pected lifetime of a series of new modelyear vehicles, this report addresses theemissions from all vehicles (new andused) on the road in a given year. Thisparticular analysis has never been donebefore. We present the results for 2004,the most recent year for which sufficientdata are available.5 The results dependnot only on vehicle characteristics (suchas fuel economy) but also on the extentof driving. Thus, these “rolling carbon”estimates reflect both the number ofvehicles in use and how many miles theyare driven each year. The carbon emis-sions from model year 2004 new autosales represent 10% of the emissionsfrom all automobiles on the road.

As the statistics assembled in thisreport show, the carbon emissionsfrom all General Motors’ vehicles onthe road as of 2004 were more thandouble those from America’s largestelectric generating company, AmericanElectric Power (AEP). In fact, theCO2 emissions of the Big Three—GM, Ford, DaimlerChrysler—each

2

exceeded those of AEP, and the emis-sions associated with most major auto-makers are comparable to or larger thanthose from the largest electric utilitiesin the country.

Perhaps even more surprising, thesmall car market segment (includingcompacts, subcompacts and two-seatsports cars) produced the most globalwarming pollution in 2004, simplybecause so many of them remain on theroad. The small-car carbon tally reflectsthe longevity of vehicles, underscoringhow decisions about the CO2 emissionsrate of new vehicles will affect theamount of global warming pollution foryears to come. SUVs had the secondlargest share of CO2 emissions. Whiletheir sales grew rapidly over the past 15years, SUVs only overtook small cars interms of new car market share in 2002.SUVs soon will become the dominantsource of global warming pollution onAmerica’s roads, and their impact will be

magnified because of their lower thanaverage fuel economy.

Figure 1 places these estimates ina global context. Figure 1(a) breaksdown global fossil fuel CO2 emissionsby major energy end-use in 2003.Worldwide, automobiles (cars andother light-duty vehicles) emitted 680MMTc in 2003, about 10% of globalCO2 emissions from fossil fuels.6 Oil,coal, natural gas and related fossilresources account for 84% of all human-caused CO2 emissions (including othermajor sources such as deforestation).CO2 is the dominant greenhouse gas(GHG, the term for any form of airpollution that contributes to globalwarming). It accounts for 72% of allhuman-caused GHG emissions. Thus,the light duty vehicle slice in Figure 1(a)represents an estimated 6% of all globalwarming pollution.7

Figure 1(b) breaks down global lightvehicle CO2 emissions by country or

1(a). Global carbon emissions by sector

(6,814 MMTc in 2003a)

1(b). Light duty vehicle carbon emissions

by regionb

(680 MMTc in 2003a)

Electricityand heat

production41%

Residential8%

Othertransportation

14%

Light dutyvehicles

10%

Industry18%

Other sectorsc

9%OECD

Europe21%

OECDPacific

9% Former Soviet Unionand Eastern Europe

6%China

2%Other Asia

2%India1%

Middle East1%

Latin America5%

Africa2%

United States45%

Canadaand Mexico

7%

FIGURE 1

Global fossil carbon emissions by economic sector

These estimates include only CO2 emissions from fossil fuel use, and so exclude emissions from biofuel use or deforestation.a Global emissions by sector are estimates for 2003 from IEA (2005). Global light-duty vehicle CO2 emissions for 2003 are projections from WBCSD (2004).b Light duty vehicle emission shares by region are estimates for 2000 from WBCSD (2004).c Other sectors include commercial, public services, agriculture and energy industries other than electricity and heat production.

3

major region as of 2000. U.S. cars andlight trucks—the subject of thisreport—are 45% of this total. The 202million automobiles on America’s roadsin 2000 represent 30% of the estimated683 million automobiles (light dutyvehicles) in use worldwide that year.8

The U.S. share of automotive CO2

emissions is disproportionately high

because our vehicles are driven morethan those in the rest of the world; the11,000 miles per year average for U.S.automobiles is about 29% greater thanthe global average of 8,500 miles peryear.9 U.S. automobiles also consumemore fuel, emitting about 15% moreCO2 per mile than the average lightduty vehicle globally.10

4

Characterizing rolling stock CO2

emissions involves data on new vehiclesales plus statistics on how long vehicleslast and how much they are driven asthey age. To perform this analysis, weuse what is termed a stock model: acomputer-based accounting frameworkthat calculates fuel consumption andCO2 emissions for various groupings ofvehicles. Given the sales volumes andfuel economy of a group of new vehiclesin a given year and previous years, themodel computes the CO2 emissions rateof all such vehicles remaining on theroad in the given year.

Figure 2 illustrates how a stock modelworks, showing what would happen ifthe CO2 emissions rate of new cars sud-denly were cut in half as of model year2000. The left graph shows the CO2

emissions rate of the new fleet; the rightgraph shows the resulting impact onthe rolling stock emissions rate. Inthis simplified example, we assume aninitially constant new fleet average CO2

emissions rate of 440 grams per mile

Methodology and assumptions

(g/mi) prior to 2000, correspondingto an average on-road fuel economy of20 mpg (similar to the actual on-roadaverage in recent years). If the CO2

emissions rate of new vehicles werehalved, down to 220 g/mi (correspond-ing to a doubling of on-road fuel econ-omy to 40 mpg) in 2000, the effect wouldnot be fully reflected in the rolling stockuntil 2025. The gradual change in thestock average CO2 emissions ratereflects the slow turnover of vehicles;lower-emitting new vehicles graduallyreplace older vehicles as they are drivenless and then retired. However, progressis more rapid at first; 90% of the lowerCO2 emissions rate is achieved within16 years, even though it takes approxi-mately another decade for the oldestvehicles to be completely replaced.

For our stock model, new vehiclecharacteristics are obtained from federalsales and fuel economy data, covering carsand light trucks of up to 8,500 poundsgross vehicle weight (GVW). Thestock turnover parameters, including

0

100

200

300

400

500

2030202020102000199019801970

Ne

w f

lee

t C

O2 e

mis

sio

ns

ra

te (

gra

ms

/mil

e)

New fleet CO2emissions ratehalved in 2000

0

100

200

300

400

500

2030202020102000199019801970

Sto

ck

CO

2 e

mis

sio

ns

ra

te (

gra

ms

/mil

e)

Stock CO2 emissionsrate drops gradually

over 25 years

FIGURE 2

How the rolling stock responds to a cut in the new fleet CO2 emissions rate

5

age-dependent vehicle survival proba-bilities and usage rates, are also basedon government statistics. We calibratedour fuel use (and therefore CO2 emis-sions) calculation to the most recentlyavailable data on total light duty vehiclefuel use, ensuring that our total rollingcarbon estimate matches nationalstatistics on the auto sector’s share ofU.S. CO2 emissions. We performedsensitivity analyses to examine theeffects of different vehicle survival andusage parameters, confirming that ourestimates of manufacturer and vehiclesegment shares are insensitive to changesin such parameters.

We report the CO2 emissions (givenon a carbon-mass equivalent basis)released directly from fuel combustionby the car—in other words, only theCO2 coming from the tailpipe. How-ever, the tailpipe is not the only sourceof greenhouse gases associated with eachgallon of motor fuel consumed. Addi-tional CO2 and other greenhouse gasesare released when crude oil is extractedand shipped, when it is refined intogasoline or diesel fuel, and when thesemotor fuels are distributed throughpipelines, barges and tanker trucks, evenbefore they go into a vehicle’s fuel tank.These upstream emissions add about31% to the global warming impact ofeach gallon of gasoline burned in theUnited States and about 23% to eachgallon of diesel fuel.11 The CO2 directlyreleased when burning gasoline amountsto 19.4 pounds per gallon.12 Counting

the roughly 6 pounds of CO2 and othergreenhouse gases emitted upstreamwould bring the total, full fuel cycle emis-sions to 25.3 pounds per gallon.13

While developing our vehicle stockmodel, we reviewed the models used byseveral government agencies and researchinstitutes, including the Department ofEnergy’s VISION and NEMS (EnergyInformation Administration) models,the Stockholm Environment Institute’sLEAP model and EPA’s MOVESmodel. VISION, LEAP and NEMSuse approaches similar to the one weadopted, estimating stock carbon emis-sions based on vehicle sales, fuel econ-omy and stock turnover rate. They aremore complex, however, in that they aredesigned to calculate future stock changesthat reflect numerous hypothetical alter-natives and couple the calculations toa broader energy modeling framework,neither of which are needed for our pur-poses. EPA’s MOVES model combinesvehicle-specific, second-by-second emis-sions calculations with travel activitydata to project stock fuel consumption,allowing users to generate results forsmaller geographic areas (e.g., at thecounty level) and in smaller temporalunits (e.g., by time of day). Although wedid not run these other models, we con-firmed the consistency of our stockmodeling results by comparing themto published results from NEMS andMOVES. Appendix A provides furtherdetail on our stock model and theassumptions behind it.

6

Our stock model tallies enable us tobreak down, by automaker and vehicleclass, the 314 MMTc emitted by all ofthe passenger cars and light trucks up to8,500 lbs gross vehicle weight (GVW)on U.S. roads in 2004.

Rolling stock impacts of major

automakers

Table 1 lists our rolling stock results byautomaker, including estimates of thenumber of vehicles in service, theiraverage on-road fuel economy, averageCO2 emissions rate, and contribution tooverall carbon emissions.14 These valuesfor the 2004 on-road stock can be com-pared to those for model year 2004 newsales to illustrate the trends as oldervehicles are replaced by newer ones. Forexample, the overall stock on-road fueleconomy of 19.6 mpg is greater thanthe 19.3 mpg overall average for new

(model year 2004) vehicles.15 In otherwords, the average older vehicle on theroad actually burns less fuel and emitsless carbon than the average newvehicle. This situation reflects the factthat new fleet fuel economy has beenfalling since 1988 due to the marketshift from cars to light trucks. A com-parison of rolling stock vs. new fleetcarbon burden shares by automaker isgiven in Figure 3. It shows that thedistribution of CO2 emissions hasbeen shifting away from the Big Threetoward other firms who are gainingmarket share.

Vehicles built by the “Big Six”—General Motors (GM), Ford,DaimlerChrysler, Toyota, Honda andNissan—made up an estimated 92%of the on-road light vehicle stock asof 2004. With their historically leadingmarket shares, the Big Three (GM,Ford and DaimlerChrysler) alone

TABLE 1

Light vehicle stock, fuel consumption, and carbon emissions by automaker, 2004

ROLLING STOCK AS OF 2004 (ALL VEHICLES NEW AND USED) MY2004 (NEW FLEET ONLY)

Vehicle On-road Fuel Carbon Vehicle Carbon On-road Carbon

population fuel economy consumption emissions population emissions fuel economy Market burden

Manufacturer (millions) (mpg) (Mbd) (MMTc) share share (mpg) share share

GM 64.4 19.2 2.68 98.6 31.6% 31.4% 18.9 27.5% 28.0%

Ford 49.8 18.6 2.17 80.0 24.4% 25.5% 17.4 18.6% 20.6%

DaimlerChrysler 30.4 18.0 1.39 51.3 14.9% 16.3% 17.8 14.4% 15.6%

Toyota 18.6 21.6 0.74 27.1 9.1% 8.6% 21.3 13.2% 11.9%

Honda 13.3 24.2 0.47 17.3 6.5% 5.5% 23.2 8.7% 7.2%

Nissan 10.0 20.8 0.40 14.6 4.9% 4.6% 19.3 6.1% 6.1%

Volkswagen 3.7 22.7 0.14 5.1 1.8% 1.6% 21.6 2.3% 2.0%

Hyundai 2.8 23.8 0.10 3.8 1.4% 1.2% 22.0 2.7% 2.4%

Mitsubishi 3.0 21.8 0.12 4.3 1.5% 1.4% 22.4 0.9% 0.8%

BMW 2.0 19.7 0.09 3.3 1.0% 1.0% 20.0 1.9% 1.8%

Kia 1.3 21.0 0.06 2.2 0.6% 0.7% 20.0 1.7% 1.6%

Subaru 2.0 22.4 0.07 2.7 1.0% 0.9% 21.6 1.0% 0.9%

Others 2.5 19.1 0.10 3.8 1.2% 1.2% 19.0 0.9% 1.0%

Big Three 144.6 18.7 6.24 229.9 71.0% 73.2% 18.1 60.5% 64.2%

Overall 203.7 19.6 8.53 314.2 100.0% 100.0% 19.3 100.0% 100.0%

Carbon emissions from automobiles on the road

7

accounted for 71% of the stock, com-pared to their 60% share of the 2004new vehicle market. Automakers’ sharesof rolling stock carbon emissions closelytrack their respective shares of the on-road vehicle population, with differencesdue to different average fuel economylevels and different mixes of older andnewer vehicles on the road.

GM, which for many years hasbeen the biggest automaker both inthe United States and globally, had anestimated stock share of 31.6%, with64 million light vehicles in 2004. Theaverage fuel economy of all GM carsand light trucks on the road was19.2 mpg, slightly lower than the over-all average of 19.6 mpg.16 As seen inTable 1, however, GM’s estimated shareof overall rolling CO2 emissions, 31.4%,is slightly lower than its share of thevehicle population.17 Thus, GM’s rollingcarbon amounted to 99 MMTc in 2004.This estimate compares well with the93 MMTc estimate for 1999 based oninformation that GM supplied for theU.S. voluntary GHG reporting initiative(EIA 2001).18

Ford has the second largest share ofcarbon emissions. Its vehicles comprise24.4% of the rolling stock and 25.5% ofthe stock carbon burden, amounting to80 MMTc in 2004. In Ford’s case, theshift to SUVs during the 1990s pulleddown their overall average fuel economymore than GM’s. Ford’s own reportingof CO2 emissions from its vehicles isgiven on a global, rather than U.S. basis,and Ford did not provide an estimate inits report to DOE’s Voluntary Report-ing of GHG Program, so direct com-parison is difficult. In its CorporateCitizenship Report 2003–04, Fordestimated that all the greenhouse gases(CO2 and other GHGs, includingrefrigerants) emitted by Ford vehicleson the road worldwide, including lightand heavy duty vehicles, amounted to109 MMTc in 2000.19

DaimlerChrysler follows in thirdplace, accounting for 14.9% of the roll-ing stock and for 16.3% of stock carbonburden. That amounts to 51 MMTc,a level greater than that from any U.S.electric company as of 2004. To ourknowledge, neither DaimlerChrysler

0% 5% 10% 15% 20% 25% 30% 35%

Kia

Subaru

BMW

Others

Hyundai

Mitsubishi

Volkswagen

Nissan

Honda

Toyota

DaimlerChrysler

Ford

General Motors

Carbon burden share

On-road stockNew fleet

FIGURE 3

Stock vs. new fleet carbon burden shares by automaker, 2004

8

nor other automakers (besides GM andFord) have publicly reported estimatesof GHG emissions from their vehiclesin use. Vehicles of the Big Three takentogether accounted for nearly threequarters (73%) of U.S. light vehiclecarbon emissions in 2004.

Toyota, Honda and Nissan havethe fourth, fifth and sixth largest sharesof the stock respectively. Since theirfuel economy levels are higher thanaverage, these three firms have rollingstock carbon shares (8.6%, 5.5% and4.6% respectively) that are lower thantheir shares of the on-road vehicle pop-ulation. Other players in the U.S.market, including Volkswagen, Mitsu-bishi, Hyundai, BMW and Subaru,each accounted for about 1%–2% stockshare. Kia, though it still has a smallmarket share, is an example of anothercompany where mix differences leadto a stock carbon share slightly greaterthan the population share of theirvehicles (see Table 1), even thoughtheir fleet has been more fuel efficientthan average.

Another interesting note: Eventhough the market share of the BigThree has been generally declining sincethe mid 1970s, and their combined

market share had dipped to 60% bymodel year 2004, the Big Three stillaccounted for more than 70% of thevehicle stock. This lag reflects thelifetime usage and slow turnover ofcars and light trucks, and also impliesa similar persistence of rolling carbonimpacts. According to Davis and Diegel(2004), the expected median lifetimefor a 1990 model year (MY) car is16.9 years, while that for a MY1990light truck is 15.5 years,20 and lifetimeshave been continuing to lengthen. Itshould be noted, however, that lightvehicle lifetimes are still shorter thanthose of many other energy-consumingitems; freight trucks and aircraft lastlonger than cars, and buildings lastmuch longer (although appliances andequipment may not).

Rolling stock carbon burdens by

market segment

Another way to examine carbonemissions is by major market segment(vehicle class): small cars, midsize cars,large cars, pickups, vans and SUVs.21

Table 2 shows the stock share, carbonemissions and average fuel efficiencyof these six vehicle segments. This

TABLE 2

Light vehicle stock, fuel consumption, and carbon emissions by vehicle class, 2004

ROLLING STOCK AS OF 2004 (ALL VEHICLES NEW AND USED) MY2004 (NEW FLEET ONLY)

Vehicle On-road Fuel Carbon Vehicle Carbon On-road Carbon

population fuel economy consumption emissions population emissions fuel economy Market burden

Vehicle class (millions) (mpg) (Mbd) (MMTc) share share (mpg) share share

Small cars 65.1 24.3 2.10 77.2 32.0% 25% 24.6 23% 18%

Midsize cars 38.5 21.4 1.46 53.7 18.9% 17% 23.2 17% 17%

Large cars 17.6 19.7 0.72 26.5 8.6% 8% 20.8 7% 8%

Pickups 32.2 16.3 1.63 59.9 15.8% 19% 15.6 16% 15%

Vans 17.3 17.9 0.80 29.5 8.5% 9% 19.1 5% 7%

SUVs 33.1 16.3 1.83 67.5 16.2% 21% 16.9 31% 34%

Cars 121.2 22.6 4.27 157.3 59.5% 50% 23.4 47% 43%

Light trucks 82.5 16.6 4.26 156.9 40.5% 50% 16.7 53% 57%

Overall 203.7 19.6 8.53 314.2 100.0% 100% 19.3 100% 100%

9

breakdown of carbon emissions share byvehicle class is also depicted in Figure 4.

Despite falling sales since 1987,small cars remained the largest segmentof the on-road stock. The 65 millionsmall cars (including small wagons) inuse accounted for 32% of the total U.S.light duty vehicle stock in 2004, incontrast to their 23% share of newvehicle sales. Small cars still accountedfor the largest share of rolling stockcarbon emissions, emitting 77 MMTcin 2004, or about one-fourth of totallight vehicle carbon emissions. Thesegment’s 25% carbon share is sub-stantially less than its 32% stock sharebecause of its higher than average fueleconomy. U.S. small cars averaged 24.3mpg on-road in 2004, 24% better thanthe overall average, for an average emis-sions rate of 4.3 tons CO2-equivalentper year.22

Midsize cars were the second largestsegment of vehicle population, com-prising 19% of the 2004 light vehiclestock. Averaging 21.4 mpg on the road,midsize cars accounted for 54 MMTc,or 17% of rolling stock carbon emis-

sions, putting them in fourth place interms of carbon share.

SUVs and pickups comprised thethird and fourth largest segments, with16.2% and 15.8% of the in-use vehiclepopulation respectively. However, SUVsand pickups were in second and thirdplace, respectively, in terms of carbonemissions in 2004. Each of these twosegments contributed about one-fifthof total rolling stock emissions. SUVcarbon emissions in 2004 amounted to67 MMTc and pickup trucks emitted60 MMTc. The disproportionate shareof SUV and pickup emissions resultsfrom their below-par fuel economy,averaging 16.3 mpg for both segments.This fuel economy level is 17% lowerthan the stock average and correspondsto an average emissions rate of about6.5 tons of CO2-equivalent per year.The 67 MMTc emitted by all the SUVson the road in 2004 is comparable tothe carbon emissions from 55 large coal-fired power plants.23

The remaining segments, vans(mainly minivans) and large cars,accounted for smaller portions of the

0% 10% 20% 30% 40%

Large cars

Vans

Midsize cars

Pickups

SUVs

Small cars

Carbon burden share

On-road stockNew fleet

FIGURE 4

Stock vs. new fleet carbon burden shares by vehicle class, 2004

10

stock and contributed 9% and 8% of2004 stock carbon emissionsrespectively. Overall, the 2004 vehiclestock composition was 59% cars and41% light trucks. But the rolling carbonshares were roughly 50% each for carsand light trucks, reflecting the higherCO2 emissions rates of trucks comparedto cars.

Tables 3 and 4 cross-tabulate stockshare and carbon share, respectively, bysegment and automaker. Groups of

vehicles that each contribute to morethan 5% of stock carbon emissionsinclude four segments of GeneralMotor’s products (small cars, midsizecars, pickups, SUVs) and two segmentsof Ford’s products (pickups and SUVs).GM’s and Ford’s pickups are the twogroups with the largest emissions over-all, accounting for 6.8% and 6.6% ofcarbon share, respectively. Since our dataexclude Class 2b trucks (8,501–10,000lb GVW), which are mostly three-

TABLE 3

Vehicle population cross-tabulation by automaker and vehicle class, 2004

Manufacturer Small cars Midsize cars Large cars Pickups Vans SUVs Overall

GM 8.2% 6.9% 4.4% 5.6% 2.1% 4.4% 31.6%

Ford 6.3% 4.0% 2.7% 5.6% 2.1% 3.8% 24.4%

DaimlerChrysler 2.9% 2.0% 1.2% 2.2% 3.2% 3.5% 14.9%

Toyota 3.2% 2.2% 0.4% 1.5% 0.4% 1.5% 9.1%

Honda 3.9% 1.5% 0.4% 0.8% 6.5%

Nissan 2.1% 1.1% 0.8% 0.2% 0.7% 4.9%

Volkswagen 1.4% 0.4% <0.1% <0.1% <0.1% <0.1% 1.8%

Hyundai 0.9% 0.3% 0.2% 1.4%

Mitsubishi 0.9% 0.3% 0.1% <0.1% 0.2% 1.5%

BMW 0.8% 0.1% <0.1% 0.1% 1.0%

Kia 0.3% 0.1% <0.1% 0.1% 0.2% 0.6%

Subaru 0.7% 0.2% <0.1% 0.1% 1.0%

Others 0.4% <0.1% <0.1% 0.1% <0.1% 0.7% 1.2%

Overall 32.0% 18.9% 8.6% 15.8% 8.5% 16.2% 100.0%

TABLE 4

Rolling carbon share cross-tabulation by automaker and vehicle class, 2004

Manufacturer Small cars Midsize cars Large cars Pickups Vans SUVs Overall

GM 6.3% 6.0% 4.0% 6.8% 2.4% 5.9% 31.4%

Ford 4.9% 3.5% 2.7% 6.6% 2.4% 5.3% 25.5%

DaimlerChrysler 2.3% 1.6% 1.2% 3.2% 3.3% 4.7% 16.3%

Toyota 2.2% 2.1% 0.4% 1.6% 0.5% 1.9% 8.6%

Honda 2.7% 1.5% 0.4% 0.9% 5.5%

Nissan 1.6% 1.1% 0.8% 0.2% 1.0% 4.6%

Volkswagen 1.2% 0.4% <0.1% <0.1% <0.1% <0.1% 1.6%

Hyundai 0.7% 0.3% 0.2% 1.2%

Mitsubishi 0.7% 0.3% <0.1% <0.1% 0.3% 1.4%

BMW 0.8% 0.1% 0.1% 0.1% 1.0%

Kia 0.3% 0.1% <0.1% 0.1% 0.2% 0.7%

Subaru 0.6% 0.2% <0.1% 0.1% 0.9%

Others 0.3% <0.1% <0.1% 0.1% <0.1% 0.8% 1.2%

Overall 24.6% 17.1% 8.4% 19.1% 9.4% 21.5% 100.0%

11

quarter- and one-ton pickups, the actualcarbon emissions share of pickups over-all is greater still.24

Comparing automobile and

electric sector carbon emissions

Accounting for roughly 20% of U.S.energy-related CO2 emissions, cars andlight trucks are a key source of thenation’s global warming emissions.Viewed in a corporate impact context,GM, Ford and DaimlerChryslervehicles are the first, second and thirdlargest CO2 emitters in the UnitedStates, all ahead of country’s largestelectricity producer, American ElectricPower (AEP). Such comparisons areillustrated in Figure 5.

The 99 MMTc from GM vehiclesin 2004 was more than double the41 MMTc from AEP. Ford’s cars andtrucks emitted 80 MMTc, essentiallydouble AEP’s emissions that year, andDaimlerChrysler’s products emitted51 MMTc. The total CO2 emissions

from the Big Three’s automobiles in2004 were comparable to the total fromthe top 11 electric companies. Nextdown the list, the CO2 emissions fromToyota vehicles were slightly higherthan those from the Tennessee ValleyAuthority, the nation’s third largestpower producer.

Comparing CO2 emissions from on-road stocks (rather than just new sales)to those from electric utilities is appro-priate in that both include the emissionsof long-lived products and equipment(automobiles and appliances or otherelectrical devices). Both are influencedby the actions of end-users, that is,consumers and businesses. For example,a power company’s CO2 emissions arethe result of the electricity consumed inhomes, offices, schools and other power-using facilities in its service area. Bothare influenced by the carbon intensity oftheir fuel supply. For example, coalaccounts for 54% of energy consumedby U.S. power generation25 and 83% ofcarbon emissions from the electric

0 20 40 60 80 100 120

NissanProgress Energy

Edison InternationalDominion

CinergyAmeren

XcelHonda

Tennessee Valley AuthorityToyota

Southern CompanyAmerican Electric Power

DaimlerChryslerFord

General Motors

Carbon emissions (MMTc)

Major automakers’ productsMajor electric producers

99

80

51

41

37

27

26

17

17

17

16

15

15

15

15

FIGURE 5

Carbon emissions of major automakers’ products vs. major electric

producers in the United States, 2004

Source: Table 1 and CERES et al. (2006).

12

sector.26 Government decisions andpolicies are also important in both cases.State and local regulators oversee elec-tric utility services and rates for thepower sector, and multiple levels ofgovernment oversee transportationinfrastructure, finance and land useplanning for automobiles. The sectorsare, of course, different in many other

ways, particularly in the nature of themarkets and the specific roles played bythe other actors who influence usage.

Another way to view corporate CO2

emissions impacts would be to tallythose associated with the motor fuelsales by major oil companies, but wehave not attempted such an analysisdue to lack of data.27

13

The rolling stock snapshot for 2004 isthe product of many forces. Key factorsinclude the amount of travel, the level offuel economy and the carbon content ofmotor fuel. Among these factors, onlyfuel economy has served to partiallycontrol CO2 emissions. Fuel carbonintensity has not significantly changed,reflecting the tenacity of oil as a sourceof motor fuel and the fact that no sys-tematic effort has been made to reduceor offset GHG emissions associatedwith petroleum fuels. Many forces drivetravel demand, and all of them have hada combined effect of pushing auto sectorCO2 emissions upward.

Factors determining auto sector

CO2 emissions

As illustrated in Figure 6, auto sectorCO2 emissions can be depicted as athree-legged stool, with the overall levelpropped up by the three major factorsnoted. These factors are generally

Beyond the carbon tally

quantifiable and can be tracked throughtime, and examining them forms acommon basis for analyzing the car-climate problem.

For the 2004 rolling carbon snapshotdiscussed in this report, total U.S. autosector CO2 emissions can be viewed asthe product of:

• Travel demand (2.6 × 1012 miles/year)• Fuel use rate (51 gallons/1000 miles)• Fuel carbon content (5.3 pounds of

carbon/gallon)

The result rounds out to the 314 MMTctotal of rolling stock carbon emissionsgiven in Tables 1 and 2.

The first factor, known as vehiclemiles traveled (VMT), represents thetotal amount of driving in cars and lighttrucks.28 The second factor, averagefuel use per mile, is based on the stockaverage fuel economy of 19.6 mpg.The third factor, fuel carbon contentof 5.3 pounds/gallon, represents thecarbon in gasoline.29

A look at what’s behind each of thethree legs of the car-carbon stool shedslight on the multifaceted nature of autosector carbon emissions and puts intoperspective the challenge of addressinga problem of this magnitude.

TRAVEL DEMAND

Demographic and economic factors,urban and regional development, changesin transportation infrastructure andevolving land-use patterns all influencethe amount of driving. The travel demandfactor, however, is often given but passingattention in U.S. climate policy, and dis-cussions frequently focus on the politicallycharged issue of raising fuel taxes.

In the United States, road building,automaking and the supply of petroleum

Vehicle CO2 emissions

Traveldemand Fuel use rate

Fuelcarboncontent

FIGURE 6

Three main factors behind rolling

carbon

14

fuels form a three-way symbiosis thathas served to sustain steady increases indriving. The public appreciates thebenefits of automobiles enough tosupport fuel taxes that help pay forroads. Yet taxes receive great scrutiny,and taxes on fuel generally have beenlimited to the minimum needed to coverbasic highway costs.

The personal value of the car is highenough that this major part of the trans-portation system is privately financedby consumers themselves. Even duringprice spikes, fuel cost is generally a smallfraction of the total cost of owning andoperating a car.30 Thus, U.S. fuel pricesprovide little restraint on VMT. Absenteffective policies to manage transporta-tion demand at the national scale, VMThas increased steadily for many years.Net VMT growth was 158% between1970 and 2004, for an average growthrate of 2.8% per year.31

A key factor driving travel demand ispopulation. The U.S. population in 2002was 288 million, representing anincrease of 41% since 1970, and anincrease of 90% since 1950.32 Com-

pounding population growth areincreases in vehicle ownership rates andaverage VMT per capita, as shown inFigure 7. Both have grown steadily overthe second half of the 20th century.VMT per capita has increased 82%since 1970—double the populationgrowth over this period—rising fromroughly 5,400 miles per year to 9,900 by2002. Car ownership (vehicles percapita) has risen essentially in parallel.Figure 7 shows ownership over theentire population, including children,the very old and others who do notdrive. The intensity of auto use ishighlighted by the fact that there arenow more cars than drivers, with thelatest statistics showing that the UnitedStates has 1.06 personal vehicles perlicensed driver.33

Surveys of how Americans use theircars are commonly reported on a house-hold basis, as shown in Figure 8. Theaverage household drove roughly21,000 miles in 2001, 17% more thanthe household average in 1990. Whilecommuting trips still had the largestshare of driving, American households

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

200019901980197019601950

Ve

hic

les

pe

r c

ap

ita

VM

T p

er c

ap

ita (m

iles

)

0

2,000

4,000

6,000

8,000

10,000

12,000

Vehicles per capitaVMT per capita

FIGURE 7

U.S. vehicle ownership and VMT per capita (1950–2002)

Source: Davis and Diegel (2004), Table 8.2

15

are using their vehicles more oftenfor shopping. The number of milesfor shopping trips steadily increasedsince 1969, and grew by 40% from1990 to 2001, as a result of increasedshopping trips and extended triplengths.34 These statistics highlight thetrend of Americans’ increasing relianceon personal vehicles, a natural conse-quence of burgeoning auto-orienteddevelopment that has been the normin most parts of the United States overthe past several decades.

The issue of how to manage traveldemand, which is intimately linked toland use and many other factors, is quiteinvolved. However, a growing sense ofthe limitations of traditional transporta-tion priorities reveals many reasonsto pursue policies that would serve todampen VMT growth (TRB 2005).As noted in the next section, catalyticconverters have succeeded in cuttingair pollution in spite of VMT growth;nevertheless, such technology-based

reductions have not been enough toachieve healthy air in many regions(Scott et al. 2005). Thus, in additionto ongoing efforts to improve tech-nology, renewed attention is beingplaced on improved land use, trans-portation alternatives and rationalizedpricing to achieve a more efficienttransportation system that is less relianton motor vehicles.

FUEL ECONOMY

The 1970s oil crises—with gasolinelines, fears of shortages and suddenlyhigher fuel prices, plus the policyresponse of Corporate Average FuelEconomy (CAFE) standards—drove aleap in new light vehicle fuel economybetween 1975 and 1981. Fuel economycontinued to slowly increase between1981 and 1988 but then began decliningdue to the rising popularity of lighttrucks and largely unchanged CAFEstandards. Nevertheless, the EPA-ratedfuel economy of the new automobile

0

5,000

10,000

15,000

20,000

25,000

200119951990a*1990198319771969

Mil

es

pe

r y

ea

r

Year of survey

� Other� Social and recreation� Misc. family and personal business� Shopping� To and from work

FIGURE 8

VMT trend by travel purpose

Source: Hu and Reuscher (2004), Table 6.

*The survey methodology changed in 1995 and so the 1990 results are shown in two forms: using original surveydefinitions (labeled "1990") and adjusted for comparison to the 1995 and 2001 surveys (labeled "1990a").

16

fleet saw a net 78% increase from1974–2004. Adjusting for increasedshortfall, average on-road fuel economyimproved by roughly 50% over thatperiod.35 This efficiency gain implies a33% reduction in CO2 emissions permile from today’s vehicles compared tothose of a generation ago.

It is instructive to compare the fuel-economy driven reductions in CO2

emissions with the reductions in theemissions of other forms of vehiclepollution. Efforts to seriously controlsmog-forming automobile emissionsalso began in the 1970s. The Clean AirAct limited tailpipe emissions ofhydrocarbons (HC), nitrogen oxides(NOx) and carbon monoxide (CO) bymodel year 1975 to levels that mostvehicles could attain only through theuse of catalytic converters (Mondt2000). Today, after an ongoingtightening of tailpipe standards and ageneration of effort, the United Stateshas achieved over a 60% reduction inautomotive “rolling smog,” that is,

health-harming pollution from cars andlight trucks.36 This drop in total emis-sions was achieved by cutting stock-average per-mile emission rates toroughly 15% of their pre-control levels.In short, catalyst-based emissionscontrol—incorporating computerizedengine management, reformulated low-sulfur gasoline and related devices—hasbeen a potent enough technological fixto achieve net pollution reductions inspite of increased driving.

The difference between the effective-ness of tailpipe emissions controls andfuel economy improvement over thisperiod is illustrated in Figure 9. Thebackdrop for both air pollution and CO2

emissions from cars and light trucks isVMT growth, which, as describedearlier, rose by more than a factor of2.5 since 1970. Yet total health-damagingair pollution from U.S. automobilesdropped by nearly 10 million tons ofNOx-equivalent,37 corresponding tothe 60% reduction noted. Reductionswill continue as even tighter tailpipe

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

20052000199519901985198019751970

Ind

ex

(1

97

0 =

1)

Carbon dioxideOther air pollutants

Down by 63% (–10 million tons)

Up by 70% (+114 million tons)

FIGURE 9

Trends in carbon dioxide emissions and other air pollution from U.S.

automobiles

Source: Authors’ analysis of EPA (2005b), FHWA (2006) and historical editions of Highway Statistics. The “other airpollutants” trend is based on a health-damage-weighted composite for tailpipe emissions of nitrogen oxides (NOx),volatile organic compounds (VOC), particulates (PM10), carbon monoxide (CO) and sulfur dioxide (SO2).

17

standards are phased in. AutomotiveCO2 emissions, on the other hand, haveincreased by 70% since 1970 and arecontinuing to rise. Although many tech-nologies are available to cut CO2 emis-sions rates, none of the options (not evenhybrid powertrains) are as effective as thecatalytic converter has been for cuttingconventional pollutants. Lacking such atechnological fix, vehicle design changeswill need to be complemented by otherstrategies, such as low-carbon fuels andreductions in travel demand, to see majorprogress in reducing rolling carbon.

LOW-CARBON FUELS

The 1970s energy crisis also promptedefforts to develop and commercializealternatives to petroleum-based trans-portation fuels. Since then, the U.S.economy has seen major shifts in thestructure of oil use; sectors that couldswitch to other fuels have largely doneso. Non-transportation oil use is nowlower in absolute terms than it was in1970, even after three decades of eco-nomic growth. Transportation, however,remains 96% petroleum dependent, andthe sector now accounts for two-thirdsof U.S. oil use, compared to one-halfin 1970.38

Throughout this time a variety offuels have been promoted by federaland state policies. Alternatives that haveseen support include synthetic fuels(synfuels) from coal and oil shales,natural gas, methanol (derived fromnatural gas or coal), various biofuels,electricity and hydrogen.39

Although nearly all alternative fuelsare advertised as “clean,” policies topromote them have had few if anystipulations regarding environmentalperformance beyond those alreadyrequired for gasoline. California’s 1990low-emission vehicle (LEV) programwas partly premised on levels of strin-gency that would favor alternative fuels.

However, by the mid-1990s it becameclear that all standards short of the zero-emission vehicle (ZEV) mandate couldbe met with improved catalytic controlsand reformulated gasoline.

Most alternative fuels are now pro-moted as offering low GHG emissions.An ongoing debate plays out in numer-ous paper studies (“full fuel cycle”analyses) of which fuels might have lowor zero GHG emissions under variousassumptions. However, requirements tocontrol GHG emissions throughout afuel’s life cycle are yet to be developed.Since little work has been done tosystematically measure and track theactual fuel-cycle impacts of commer-cially produced fuels, the real-worldgreenhouse gas implications of alter-native fuels remain uncertain. Moreover,one major U.S. policy to promote alter-native fuels, giving automakers CAFEcredits for selling flexible-fuel vehicles(FFVs), is counterproductive. Theresulting trade-off in vehicle efficiencyhas increased oil consumption andcarbon emissions over what they wouldhave been without the FFV policy(DOT 2002).

One alternative fuel that has seenconsistent backing is ethanol. Ethanolcan be readily blended with gasoline inlow concentrations and such “gasohol”has long been available in the Midwest.U.S. fuel ethanol use is now increasingrapidly toward the renewable fuel man-date of 7.5 billion gallons by 2012.40

That level would amount to about 3% ofU.S. light vehicle fuel consumption.41

However, because the energy use andemissions associated with agriculturaland ethanol production processes arenot tracked, it is difficult to ascertainthe net carbon impact of ethanol use.A recent meta-study found that replac-ing gasoline with corn ethanol has anambiguous effect on GHG emissions(Farrell et al. 2006). Like most commercial

18

and agricultural activity in the UnitedStates, ethanol production is now pur-sued without a need to manage carbon.To ensure reductions in GHG emissionsthrough fuel substitution, protocols willneed to be established for tracking fuel-cycle impacts and certifying that the netGHG emissions from ethanol or otherfuels are measurably lower than thosefrom gasoline.

Fuel supply infrastructure is long-lived, with large-scale investmentsneeded to develop new fuel productionfacilities, so shifting to alternative fuelsrequires longer lead times than modi-fying vehicles. Although the past fewyears have seen high oil prices, and theU.S. Department of Energy revised itsoil price outlook significantly upward(EIA 2005), the competitive environ-ment for petroleum alternatives remainsspeculative. This uncertainty in com-mercial outlook compounds the uncer-tainty in the carbon impacts of any givenfuel due to the lack of policies for man-aging GHG emissions associated withfuel production. Therefore, while shift-ing to low-carbon fuels is an essentialcomplement to cutting travel demandand raising fuel economy, how well newfuels will serve to reduce GHG emis-sions depends on programs and policiesthat largely remain to be developed.

Managing auto sector carbon

While the major factors that determineemissions—the three-legged stool oftravel demand, fuel economy and type offuel—provide a convenient snapshot ofthe sector, such an analysis has limitationswhen considering how to cut emissions.Multiple decision makers (“actors”) in-fluence each leg of the stool. In otherwords, the factors that technically deter-mine emissions are the product ofchoices made by many players in themarketplace, such as automakers, fuel

suppliers and consumers, as well as byregional planning bodies and othertransportation policy forums.

For example, consider fuel economy,a tangible and easy-to-track factorbehind automotive CO2 emissions.42

Consumers influence fuel economythrough their selection of vehicles aswell as their driving and maintenancehabits. Automakers influence fueleconomy through their product strategiesand design choices. But the decisionsmade by consumers and automakersare not independent; moreover, the netcarbon impact of fuel economy dependson the broader influences that shapevehicle use.

RECASTING THE DISCUSSION

The complex, market- and policy-mediated nature of decisions thatinfluence CO2 emissions is not uniqueto the auto sector, of course. But ourrolling stock tally underscores the mag-nitude of automotive CO2 emissionsand spotlights the roles of the manyactors whose decisions are behind the314 MMTc from U.S. cars and lighttrucks. Such a view suggests a need torecast the discussion of what to do aboutthe car-carbon problem from

What policies are needed to address

each factor behind automotive CO2

emissions?

to

What policies can motivate each actor

to address the aspects of CO2 emissions

under each actor’s control?

In a market-based framework, onlyindividual decision makers can deter-mine the actual, perceived costs andbenefits to themselves resulting fromthe decisions they make. Developingan actor-based carbon management

19

framework can clarify the roles of allplayers and enable cost-effective deci-sions about carbon reduction. Underthis framework, policies can be designedto motivate actors to consider andreduce carbon impacts of every decisionthey can control. Such an approachwould be a step beyond factor-basedanalysis, which rests on assumptionsabout the costs and benefits to particularactors presumed to hold the most influ-ence over a given factor.

A NEED FOR NEW TOOLS

Exactly how to define a set of auto-sector specific carbon managementpolicies is a subject for future work. Wecan, however, illustrate the value of suchan approach by considering the exampleof a city as an actor who can influencethe CO2 emissions associated with itstransportation operations. The vehiclesa city operates are, in fact, a microcosmof the larger rolling carbon situation.

To cut the carbon from a city’stransportation operations, its agenciesmight consider ways to carry out theirfunctions with less driving; they mightchange the number, type and efficiencyof the vehicles they operate; or theymight use low-carbon fuels.43 Changesin vehicle purchases would influence thedemand side of the market to whichautomakers respond. Moreover, undersuch a fleet carbon management ap-proach, decisions about vehicle choicewould no longer be made in isolationfrom looking at other (and sometimesmore accessible) ways for the city to cutcarbon from its vehicles.

Of course, a municipality’s vehiclefleet is just a small fraction of all thevehicles within a community. In fact,beyond its own fleet, city officials have ameasure of control over land-use, zoningand development decisions, parking andthe availability of non-auto modes oftransportation. A city can help lead other

regional governments as well as stateofficials to explicitly manage carbonimpacts in all aspects of transportationand land use planning. Such decisionsare the clear area in which local govern-ments can affect the “rolling carbon”impacts of their jurisdiction. Challengesand costs are likely be associated withany of the actions a city might considerfor cutting carbon from transportation,but there might also be cost savings.

A city is just one actor, but the aboveexample illustrates how any entity in-fluencing auto sector CO2 emissions canapproach the problem under a carbonmanagement paradigm. A private com-pany or institution that uses vehicles inthe course of business also has a measureof control over the associated CO2 emis-sions, and opportunities to reduce emis-sions can be found once an effort ismade to look for them. Right now,however, most cities (as well as mostother entities who influence auto sectorCO2 emissions) have little way of know-ing how their day-to-day or year-by-year transportation decisions impactCO2 emissions. Thus, a starting pointfor finding ways to cut emissions isdeveloping new evaluation tools thatmake the carbon impacts and costs ofdecision options explicit. Sector-specificcarbon impact evaluation tools willfacilitate “carbon-sensitive decisionmaking,” enabling each actor to makethe decisions that most affordablyreduce their portion of emissions. Abroader question is how to motivatecarbon-sensitive decision-making. Ulti-mately, those who reduce emissionsmight be able to realize a financial valuefrom their actions in the market for car-bon reductions that will emerge undernational and international climateprotection regimes.

In short, many different entities havesome measure of control over auto sectorCO2 emissions. Opportunities to cut

20

carbon can be found by examiningthe decisions made by any actor thatinfluences the future trajectory of autosector CO2 emissions. Such an actor-based approach breaks the otherwisedaunting rolling carbon problem intomanageable pieces, and so can com-plement the traditional policy focuson the factors that determine emissions

in a technical sense. Further work isneeded on the issue of how to definecarbon management policies for theauto sector; this is a subject we leavefor future research and it will also be animportant topic for discussion amongpolicy makers, automakers, the energyindustry, transportation planners andother stakeholders.

21

The global warming pollution from allU.S. cars and light trucks amounted to314 MMTc in 2004. This “rollingcarbon” accounts for about one-half ofCO2 emissions from all passengervehicles around the world and about 6%of global energy-related CO2 emissions.The amount of CO2 emitted a year fromthe U.S. vehicle stock is equivalent tothe amount of carbon in a coal train50,000 miles long.

At the national level, data on thehistory of vehicle sales and statistics onvehicle usage enable us to breakdownthe total rolling carbon by automakerand type of vehicle. Such an analysisshows, for example, that the CO2

emissions from each of GM’s, Ford’sand DaimlerChrysler’s vehicles exceedthose from any electric power company.However one looks at the issue, the car’scontribution to global warmingpollution is enormous.

Rolling carbon can be analyzed interms of the factors—VMT, fueleconomy and fuel carbon content-thatdetermine emissions. For developingnew policies, however, it is instructive toconsider the actors whose decisionsinfluence each factor. Auto companiesare, of course, a key actor through theirproduct strategies and product designdecisions. But other parties, including

Conclusions

individuals and businesses that purchaseand use vehicles, energy companies thatprovide auto fuel, and various levels ofgovernment that influence land-use andtransportation decisions, all play a partin influencing the total rolling carbon.The total CO2 emissions from allvehicles on the road are in fact the resultof complex and interdependentdecisions made by many actors.

Policy discussions on curbing auto-motive carbon emissions have tended tocenter on the technical factors that char-acterize the sector’s emissions. However,changing these factors to reduce autocarbon emissions will require comple-mentary decisions by multiple actorswho influence emissions. Our rollingstock tally underscores the magnitude ofautomotive CO2 emissions, highlightingthe potential roles of the many actors inthe system. It also suggests a need fornew tools to help each actor understandhow their decisions impact emissions,paving the way for new policies that canfoster carbon-sensitive decisions mak-ing. Such a carbon management para-digm would serve to make carbonemissions reduction an objective in day-to-day decision making, so that eachactor in sector can seek opportunities forreducing those aspects of total emissionsthat each can best influence.

22

The stock turnover model used in thisstudy is a spreadsheet-based accountingmodel for calculating fuel consumptionof all light duty vehicles on the road (the“rolling stock”). The resulting fuel useand CO2 emission estimates dependon fuel consumption rates of vehicles(computed from fuel economy data),new vehicle sales, as well as national-level statistics on how long vehicles lastand how much vehicles are driven asthey age. The last three factors—newvehicle sales, vehicle survival probabilityand usage rate by age—determine therate of vehicle stock turnover.

In our model, total stock fuel use in agiven year (Fy) is computed as the sumof fuel consumption for each group ofvehicles of a given age (cohort)weighted by the probability that avehicle survives to the given age (pi), asshown in Equation (1). In thiscalculation, the fuel consumption of avehicle cohort is estimated bymultiplying its sales (Sy) by its averagefuel consumption rate (cy) and theexpected annual miles of travel forvehicles of the given age (vI).

(1)

(2)

where

c = fuel consumption rate (e.g., gal/mi;reciprocal of fuel economy)Fy = total fuel consumption in year yg = on-road adjustment (fuel economyshortfall)i = vehicle age, i = 0, 1, ..., N (N=25)m = new fleet fuel economy (unadjustedCAFE values)

APPENDIX A

Stock modeling methodology and assumptions

pi = survival rate (probability ofsurviving to age i)Sy = new vehicle sales in year yvi = usage rate (miles per year pervehicle of age i)VMTy = total vehicle miles traveledin year yy = year of cohort (model year)

Total stock vehicle miles traveled in yeary is calculated as:

(3)

Average stock fuel economy is thencomputed as:

(4)

Stock carbon emissions are calculatedby multiplying the stock fuel useestimate (Fy) from Equation (1) by aCO2 emissions factor of 19.4 lb ofCO2/gal. Thus, the carbon emissionsestimates given here include onlyemissions from vehicle use, rather thanfull fuel cycle emissions. Assumptionsmade and parameter values used in thestock model are presented in Table A1.

Data sources and model

calibration

Historical data on new vehicle sales (Sy)and fuel economy (my) from 1979 to2004 are taken from federal fueleconomy databases, specifically, theNational Highway Traffic SafetyAdministration (NHTSA) annualCAFE data reports and EPA (2005b).Light vehicle survival and usage rates

N

23

were adopted from the PreliminaryRegulatory Impact Assessment forthe proposed MY2008-11 light truckCAFE standards (NHTSA 2005,Table VIII). The CO2 emissions esti-mates by automaker and vehicle seg-ment are based only on tallies of thefederal fuel economy data by automakerand segment; we did not have auto-maker- or segment-specific statisticson vehicle survival and usage rates.

The overall stock fuel consumptionestimates from Equation (1) were cali-brated to federal statistics for total lightduty vehicle fuel consumption in 2004(FHWA 2006, Table VM-1). Thisapproach ensures that our carbon num-bers add up to match the auto sector’sshare of national CO2 emissions in gov-ernment reports. Available fuel economyand sales data cover only cars and lighttrucks regulated under the fuel economystandards, i.e., cars and trucks with aGVW less than 8,500 lbs. However,the Federal Highway Administration(FHWA) light vehicle fuel use statisticscover cars and light trucks with GVWup to 10,000 lbs. Therefore, we parse

out the fraction of light vehicle fuel useby assuming that trucks with 8,501–10,000 lbs GVW (Class 2b trucks)account for about 4% of the overallunder 10,000 lb GVW light duty fueluse (EIA 2005, Table A7).

Because we calibrated to overall fuelconsumption, rather than attempting tojust add up fuel consumption from thevehicle stock, our modeled estimate ofthe number of vehicles in use (204 mil-lion for 2004) does not necessarily matchthe nationwide registration tallies fromthe Department of Transportation(DOT), which reported 228 million carsand light trucks in use as of 2004 (FHWA2006, Table VM-1). One reason is thatour estimates exclude Class 2b trucks; itis also likely that more very old vehiclesare in use than are captured by our stockmodeling parameters and the fact thatour sales data only go back to 1979.

Sensitivity analysis

As shown in Equation (1), stock fuelconsumption calculations entail assump-tions about vehicle usage and survival

TABLE A-1

Key assumptions for rolling stock carbon analysis

Assumption or parameter value Value

Fuel economy shortfall. Relative to CAFE (test) values, with same 20%

value assumed for all vehicles types.

CO2 emissions factor. Direct emissions only, assuming full 8,800 g/gal*

combustion but excluding upstream emissions in the fuel supply (19.4 lbs/gal)

chain (a 30% effect) and emissions from vehicle manufacturing

(a 12% effect).

Oil consumption factor. One barrel of crude oil demand per 1 : 1

barrel of gasoline or diesel fuel consumption, assuming that oil oil : gasoline

demand is driven by these high-value fuels, rather than the ratio

based on refinery yields.

Carbon factor. Million metric tons of carbon (MMTc) per year 36.83 MMTc/Mbd

corresponding to each million barrels per day of fuel consumption,

corresponding to direct CO2 emissions factor and oil consumption

factor (i.e., not full fuel cycle).

*An assumed nominal value for conventional gasoline; this value is between the CO2 emissions factor in the ArgonneNational Laboratory GREET model (8,750 g/gal) and the factor used by EPA (8,864 g/gal).

24

rates. The product of usage rate and sur-vival rate of a given cohort equals theexpected annual miles traveled of vehi-cles of that cohort. Thus, the higher theproduct of usage rates and survival ratesof a given cohort, the higher its fuel useand carbon emissions. However, becauseour estimate of overall stock fuel con-sumption is calibrated to FHWA statis-tics, it is independent of assumptionson survival probability and usage rates.Nevertheless, CO2 emission breakdownscalculated for subsets of the stock (e.g.,by automaker or vehicle type) can beaffected by such assumptions.

This analysis adopts the survival ratesand usage rates used by NHTSA (2005).Because NHTSA only provides thesurvival rates and usage rates of lighttrucks, we assumed the same survivaland usage functions for cars and lighttrucks. Even though light truckshistorically had lasted longer thanpassenger cars, the survival rates for newlight trucks have been dropping andconverging to those of cars (see

NHTSA 2006 for further discussion).44

Thus, the assumption of same survivaland usage rates for cars and trucksseems reasonable. In addition, thisanalysis also assumes the same survivalprobability and mileage for all vehiclesegments and for vehicles produced byall manufacturers, since public data arenot available for discriminating survivaland usage at that level of detail.

To examine how survival and usageassumptions affect the estimates of shareof carbon emissions among segmentsand automakers, we performed a sensi-tivity analysis using vehicle usagestatistics from the 2001 NationalHousehold Travel Survey (HTS) andthe 1990 survival probabilities estimatedby Oak Ridge National Laboratory(ORNL) (Davis and Diegel 2004,Tables 3.9 and 3.10). Figure A1compares the expected VMT per vehicleused in our analysis with values derivedfrom the HTS usage data and theORNL survival rates. The sensitivityresults suggest that our stock carbon

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

2625242322212019181716151413121110987654321

Ex

pe

cte

d a

nn

ua

l V

MT

pe

r v

eh

icle

(m

ile

s)

Vehicle age

Expected VMT used in our stock modelExpected VMT for cars using HTS usage data and ORNL survival functionExpected VMT for light trucks using HTS usage data and ORNL survival function

FIGURE A1

Comparison of expected VMT per vehicle from different sources

Expected VMT equals the product of survival probability and usage rate.

Source: NHTSA (2005), Table VIII; National Household Travel Survey (2001); Davis and Diegel (2004).

25

emissions estimates were about 2–3%higher for light trucks and 2–3% lowerfor cars than those derived using theHTS-ORNL assumptions. The impacton automakers’ oil use and carbonemissions estimates are similarly small,ranging from –2% to 2%. These effectstranslate into differences of less than1 percentage point in stock carbonshares by vehicle segment and less than0.2 point in stock shares by automaker.

Figure A2 plots the survival rates (pi)and annual usage rates (vi) used in thisanalysis plus the expected VMT byvehicle age derived as a product of thesurvival and usage rates. Thus, underthe assumptions we used, the “survivingmiles” of annual driving by vehicles ofa given age falls from just under 16,000mi/yr for new vehicles to roughly 4,000mi/yr for 16-year vehicles, for example.

Rolling stock fuel consumption andCO2 emission estimates are also sensi-tive to the choice of the on-road fueleconomy adjustment (shortfall) factor(g). Due to higher road speeds, increas-ing congestion and urbanization, in-

creased use of accessories (like airconditioning) and other factors, actualon-road fuel economy is less than theEPA-test fuel economy used for CAFEcompliance. The on-road adjustmentfactor reflects the gap between theactual fuel economy and test fueleconomy. An under-estimated shortfallfactor will result in under-estimatedstock fuel consumption, and vice versa.

For vehicle labeling and consumerinformation purposes, EPA has usedshortfall factors that average 15% since1985; the agency now has a rulemakingunderway to develop new estimates foron-road fuel economy. Although newestimates are not yet available, theexpectation is that the average estimatedshortfall will increase. Based on studiessuch as Mintz et al. (1993) and shortfallestimates used in NEMS, we adopt 20%shortfall as the assumption for thisanalysis. This level of shortfall alsohappens to yield a stock average fueleconomy modeled on the basis ofCAFE data (19.6 mpg) that closelymatches the light vehicle average of

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

2625242322212019181716151413121110987654321

Ve

hic

les

pe

r c

ap

ita

VM

T p

er c

ap

ita (m

iles

)

Vehicle age

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Survival rateUsage rateExpected annual VMT

FIGURE A2

Survival, annual usage and expected VMT by vehicle age

Source: NHTSA (2005), Table VIII, and authors' calculations.

26

19.8 mpg derived from FHWA (2006)and Energy Information Administration(EIA 2005).45

Other stock turnover models

Several government agencies and researchinstitutes have developed vehicle stockturnover models that are largely orientedto projecting future fuel use and carbonemissions under various scenarios. Thesesoftware programs include the DOEVISION model, Stockholm Environ-ment Institute-Boston Center (SEI-B)LEAP software system, the DOE-EIANEMS model and the EPA MOVESmodel. The VISION model, the trans-portation module of LEAP and theLight Duty Vehicle (LDV) Stockmodule of NEMS use a stock turnoverapproach similar to the one we adoptedin developing our stock model.

Designed as a tool for estimatingmobile source emissions to facilitateinfrastructure planning, EPA’s MOVESmodel provides more detailed estimatesof on-road vehicle fuel consumption andemissions for a small geographic region(such as at the county level), and allowsusers to generate estimates by smalltemporal units (month, day of week,hour), road type and technology type.Using second-to-second vehicle emis-sions data, the MOVES model esti-mates energy consumption rate forvehicles of different combinations offuel, engine technology, engine size,

weight and model year. Stock vehiclefuel consumption is computed by com-bining the estimated energy consump-tion rates of different vehicle combinationwith temporal VMT distribution (i.e.,VMT fraction by month, day and hour),allocation of operation mode (idle, oper-ating and braking) and vehicle popu-lation data (base year population,scrappage rates, sales growth rates andage distribution).

Even though we did not run thesemodels, we validated our stock modelingresults with published results fromNEMS and MOVES. Our stock fuelconsumption estimate of 8.53 millionbarrels per day (Mbd), which was cali-brated to the latest federal highwaystatistics, is 0.2% higher than theprojection from NEMS, while ourestimate of stock fuel economy is about3% lower than that from NEMS (EIA2005). Our stock model estimates ofon-road fuel economy of cars and lighttrucks for 2004 (22.6 mpg and 16.6 mpgrespectively) also compare well with themodeling results from MOVES for on-road fuel economy of cars and lighttrucks for 2002 (22.8 mpg and 16.6 mpgrespectively). In addition, the 2002 fleetfuel consumption estimates (8.58 Mbdor 131.5 billion gallons per year) fromMOVES, which include light vehiclesand some of the Class 2b trucks, is closeto our modeling results on light vehiclestock fuel consumption for 2004 (Koupaland Srivastava 2005).

27

1 Throughout this report, automobile GHGemissions are given on a carbon-mass basis,in million metric tons of carbon (C) mass-equivalent per year (MMTc), unless other-wise noted. Some climate-related studies(the Ceres 2006 report on electric sectoremissions, for example) give results in shorttons of CO2 mass-equivalent; the conversionfactor is 4.042 short tons of CO2 per metricton of C.

2 The coal train comparison assumes the fol-lowing: coal is 60% carbon by weight (Wang2001); an average coal car carries 111.4 shorttons (AAR 2005), implying 60 metric tonsof carbon per carload; a typical coal carlength is 50 feet (FreightCar America 2005);and New York-to-San Francisco distance of2,902 miles (maps.google.com).

3 According to EPA (2006), CO2 emissionsassociated with petroleum use for trans-portation reached 496 MMTc in 2004, equalto 96% of the CO2 emissions from coal usefrom power generation (517 MMTc).

4 See DeCicco et al. (2005) for the mostrecent update of Automakers’ CorporateCarbon Burdens, published by EnvironmentalDefense.

5 As noted below, some international com-parisons are given for earlier years if 2004data were not available.

6 Based on projections by WBCSD (2004),global light duty vehicle CO2 emissions were680 MMTc in 2003, which is also 10% ofthe 6,814 MMTc of global CO2 emissionsassociated with fossil fuel use in 2003.

7 Light duty vehicle estimates are fromWBCSD (2004); the estimates for globalfossil fuel CO2 emissions, all anthropogenicCO2 emissions, and all GHG emissions arefrom IEA (2005).

8 The U.S. vehicle stock estimate comes fromthe R.L. Polk Co., cited in Table 6.4 ofDavis and Diegel (2002). The 2000 globalvehicle stock estimate comes from WBCSD(2004). Vehicle stock estimates from othersources differ from the figures provided herebecause of different definitions for trucksand the uncertainties associated with un-registered vehicles.

9 Fulton and Eads (2004, 33) estimate worldaverage annual travel per vehicle per year at

Notes

13,755 km in 2000, which converts to 8,547miles. The annual average travel of U.S.automobiles come from Table 8.13 of Davisand Diegel (2004), with data generated fromthe 2001 National Household Travel Survey(HTS). Even though estimates for 2000were not available for comparison with the2000 global figure, HTS findings from 1990and 1995 suggest increasing levels of travel,implying that VMT per vehicle in 2000 isunlikely to be less than 11,000 miles.

10 The fuel efficiency comparison assumes thefollowing: World stock fuel economy is 22.5mpg in 2000 (WBCSD 2004); U.S. stockfuel economy is 19.6 mpg in 2004 based onour stock model.

11 Estimates from GREET 1.5a (Feb. 2000);see Wang (2001).

12 A common question is, “Since a gallonof gasoline weighs only a bit more than6 pounds, how do you get nearly 20 poundsof CO2 emissions?” The answer is that whengasoline is burned, the carbon in it (whichhas an atomic weight of 12) is combinedwith two atoms of oxygen (which each havean atomic weight of 16) to form CO2 (whichhas a molecular weight of 44). After adjust-ing for the hydrogen in the fuel (whichgets combusted into water), the CO2 re-leased when gasoline is burned weighs about3.14 times as much as the gasoline itself.

13 Wang (2001).14 Unless otherwise noted, results shown in

tables and figures are from the authors’calculations per the methodology andassumptions described in the text anddetailed in Appendix A.

15 The 19.3 mpg average on-road fuel economyfor MY2004 vehicles is based on 2004 CAFEdata, assuming 20% fuel economy shortfall.The 19.3 mpg on-road fuel economy valueconverts to 24.1 mpg in test fuel economy,which is lower than the preliminary estimateof 24.4 mpg new fleet fuel economy reportedin EPA (2005b).

16 All of the on-road MPG estimates givenhere assume a 20% fuel economy shortfallrelative to CAFE compliance values basedon EPA lab tests.

17 While an automaker with lower than aver-age fuel economy would usually have a

28

carbon burden share higher than its share ofthe vehicle population, the stock carbonburdens share also depends on the mix ofold and new vehicles in its fleet. Old vehi-cles on average are driven fewer miles peryear than new ones (see Figure A1 in theAppendix A). Thus, the larger the share ofold vehicles in an automaker’s on-road stock,the smaller their average annual miles drivenper vehicle. For the case in point, we esti-mate that the GM-built vehicles on the roadin 2004 were driven an average of 12,230miles per year, compared to an average of12,581 miles per year for all vehicles on theroad as of 2004. Thus, even though theirvehicle population-weighted fuel economyis lower than average, GM vehicles’ share ofthe rolling stock carbon burden is lower thantheir share of the stock itself.

18 The 93 MMTc value is a conversion of the340.6 MMT CO2-equivalent from theentire U.S. fleet of GM-built vehicles givenin page 47 of EIA 2001. Since 2002, GMdid not provide estimates of CO2 emissionsfrom its own fleet in the report submitted toDOE’s voluntary reporting program. Giventhat overall VMT grew by roughly 9%1999–2004 while GM’s share has declinedsince 1999, we consider these results to bereasonably consistent given their indepen-dent derivation.

19 Ford (2004) estimated that all the green-house gases (CO2 and refrigerants) emittedby Ford’s vehicles on the road worldwide,including light and heavy duty vehicles,amounted to 400 MMT CO2-equivalent,which converts to 109 MMTc in 2000.

20 The median lifetime represents the age towhich 50% of vehicles survive; it is greaterthan the median age of all vehicles on theroad, because the majority of vehicles arethose of the more recent model years.

21 Station wagons are included with cars ofcorresponding size.

22 The stock average CO2 emissions rate ispresented in metric tons and converted fromthe on-road fuel economy of 24.3 mpg of allsmall cars in 2004.

23 We assume coal generation emissions of7,100 metric tons of CO2-mass equivalentper megawatt (MW) and a typical largeplant size of 600 MW capacity, implying1.2 MMTc per year for a large coal plant.

24 NHTSA does not track fuel efficiency ofClass 2b trucks, so we are not able to include

estimates of carbon impacts of these largestversions of light-duty trucks.

25 Table 2.2 of Davis and Diegel (2004).26 Table 3.3 of EPA (2006).27 Sufficiently detailed and disaggregated data

do not appear to be available for estimatingemissions from the use of fuels supplied byoil company. The Carbon Disclosure Project(www.cdproject.net) compiles CO2 emis-sions reports voluntarily submitted by majorcorporations, sometimes including emissionsassociated with their products. Reports arelisted from several oil companies, butExxonMobil (the largest) did not provideestimates of CO2 emissions from theirproducts; also, the emissions figures aregiven on a global rather than U.S. or otherregional basis.

28 The 2.6 trillion miles figure is generatedfrom our stock model and is lower than the2.7 trillion VMT statistic in FHWA (2006),which include miles traveled by light trucksbetween 8,500–10,000 lbs GVW.

29 As noted earlier, these estimates account foronly the direct CO2 emissions from burninggasoline, not the larger, full fuel cycle emis-sions that includes CO2 as well as othergreenhouse gases emitted when producingand distributing gasoline. The fuel carboncontent of diesel is slightly higher than thatof gasoline (22.8 lbs of CO2 per gallon).However, since diesel accounts for onlyabout 3% of U.S. light duty vehicle fuel use(Davis and Diegel 2004, Table A5), thisstudy assumes light vehicle fuel carboncontent is the same as that of gasoline.

30 According to Ward’s (2005), costs of fuel onaverage account for 12–14% of total costs ofowning and operating a vehicle.

31 Light vehicle VMT from 1970-2003 comesfrom FHWA (2006) and earlier editions;2004 light vehicle VMT estimate is extrapo-lated using monthly traffic volume data fromFHWA (2004). The VMT data used hereinclude Class 2b vehicles (those between8,500 and 10,000 lbs GVW) because dataexcluding Class 2b vehicles are not available.But the growth in VMT excluding Class 2bshould be similar to that including Class 2bvehicles. Also, note that U.S. VMT growthhistorically has followed a linear rather thanexponential trend, so annual VMT growthrates computed from recent base years mightseem to be decreasing even though theVMT increase remains steady.

29

32 Statistical Abstract of the United States(Census Bureau 2005), Tables 1 and 2.

33 Table 2 of Hu and Reuscher (2005).34 Table 6 of Hu and Reuscher (2005).35 The 50% net increase compares our stock-

modeled 19.6 mpg estimate for 2004 withan on-road estimate of 13.1 mpg for 1974derived from the online tables of Davis andDiegel (2004) and earlier FHWA statistics.A 20% shortfall factor is assumed in thisanalysis (see Appendix A).

36 As of 2002, total health-damage weightedcar and light truck pollution has dropped63% since 1970, based on our calculations asshown in Figure 9. Pollution is continuing todecline as cleaner vehicles replace oldervehicles in the stock. This air qualityimprovement corresponds to a 85% reduc-tion in average per-mile emissions of thelight vehicle stock acting against the2.5-fold increase in VMT that occurred1970–2004. Note that medium and heavytrucks (most of which are diesel powered)have seen much less progress in pollutioncontrol. Moreover, the need to control fineparticles in addition to smog-formingpollutants means that additional efforts areneeded to reduce the transportation sector’scontribution to unhealthy air; see Scott et al.(2005).

37 The graph weights all tailpipe pollutantsaccording to their estimated health impactsrelative to NOx, using damage cost valuesfrom Delucchi (1997) and pollution datafrom EPA (2005a).

38 Davis and Diegel (2004), Table 2.2 andTable 1.12.

39 In recent years, for instance, ethanol, bio-diesel, and gasification-derived syntheticgasoline and diesel are receiving increasedresearch attention.

40 The renewable fuel standard established bythe Energy Policy Act of 2005 (H.R. 6,Section 1501).

41 The 3% estimate is relative to the EIA(2005) forecast of 18.25 quads for lightvehicle fuel use in 2012 and reflects the fact

that ethanol contains 34% less energy pergallon than gasoline.

42 While there is often concern about theaccuracy of fuel economy ratings and in-usefuel economy can vary greatly, on an aggre-gate basis average fuel economy is fairly wellknown, with its uncertainties being less thanthe level of growth that occurs over multi-year periods and far less than the magnitudeof change needed to address a problem suchas global warming.

43 An example of such a city-oriented policyis given by the Green Fleets program,www.greenfleets.org.

44 The NHTSA (2006) report provides up-dated survival rates and usage rates for carsand light trucks. The updated light trucksurvival rates are slightly higher than thatused in NHTSA (2005), while the lighttruck usage rates are mostly the same. Theupdated survival rates and usage rates of carssuggest that expected VMT of cars seems tobe lower than that of trucks, except for carsof age 8–11. If the most up-to-date survivalfunctions were used, the share of carbonemissions of light trucks would be higherand so would be the share of carbon emis-sions of automakers with higher share oftruck sales. However, the differences are notexpected to be significant, given the lowsensitivity we found when substitutingHTS-ORNL usage and survival assumptions.

45 The stock fuel economy of light duty vehi-cles is estimated assuming: fuel consumptionand annual VMT of Class 2b trucks is 4,922million gallons and 68,732 million milesrespectively (EIA 2005); fuel consumptionand stock fuel economy for all cars and lighttrucks with GVW up to 10,000 lbs is138,632 million gallons and 19.6 mpgrespectively (FHWA 2006). Because theClass 2b fuel economy and fuel consumptionwere estimates from the NEMS model, wechecked our estimates using Class 2b fuelconsumption and fuel economy estimatesfrom Davis and Truett (2002), which implya light vehicle fuel economy estimate of19.8 mpg.

30

AAR. 2005. Railroads and Coal.Background paper. Washington, DC:Association of American Railroads(www.aar.org). July.

Census Bureau. 2005. StatisticalAbstract of the United States.Washington, DC: U.S. Census Bureau.http://www.census.gov/prod/www/abs/statab.html, accessed 27 March 2006.

Ceres et. al. 2006. Benchmarking AirEmissions of the 100 Largest ElectricPower Producers in the United States—2004. Boston, MA: Coalition for Envi-ronmentally Responsible Economies(Ceres). April.

Davis, S. and L. F. Truett. 2002. Investi-gation of Class 2b Trucks (Vehicles of8,500 to 10,000 Lbs GVWR). ReportORNL/TM-2002/49. Oak Ridge, TN:Oak Ridge National Laboratory. March.

Davis S. and S. Diegel. 2004. Trans-portation Energy Data Book, Edition24. Report ORNL-6973. Oak Ridge,TN: Oak Ridge National Laboratory.December.

Davis S. and S. Diegel. 2002. Trans-portation Energy Data Book, Edition22. Report ORNL-6967. Oak Ridge,TN: Oak Ridge National Laboratory.September.

DeCicco, J., et. al. 2005. Automakers’Corporate Carbon Burdens, Update for1990-2003. Washington, DC: Environ-mental Defense. August.

Delucchi, M.A. 1997. The AnnualizedSocial Cost of Motor-Vehicle Use in theUnited States, 1990–1991: Summary ofTheory, Data, Methods, and Results.Report UCD-ITS-RR96-3(1). Davis,CA: Institute of Transportation Studies.June.

References

DOT. 2002. Report to Congress: Effectsof the Alternative Motor Fuels ActCAFE Incentives Policy. Washington,DC: U.S. Department of Transporta-tion. March.

EIA. 2005. Annual Energy Outlook2006 (Early Release). Washington DC:Department of Energy, Energy In-formation Administration. December.http://www.eia.doe.gov/oiaf/eao/index.html

EIA. 2001. Voluntary Reporting ofGreenhouse Gases 1999, Chapter 7.Washington DC: Department ofEnergy, Energy Information Admin-istration. February. http://www.eia.doe.gov/oiaf/1605/vr99data.

EPA. 2006. Inventory of U.S. Green-house Gas Emissions and Sinks: 1990–2004. Report EPA 430-R-06-002.Washington, DC: EnvironmentalProtection Agency. April.

EPA. 2005a. National EmissionsInventory (NEI) Air Pollutant Emis-sions Trends Data. Washington, DC:Environmental Protection Agency, July.http://www.epa.gov/ttn/chief/trends/index.html, accessed 16 Nov., 2005.

EPA. 2005b. Light-Duty AutomotiveTechnology and Fuel EconomyTrends, 1975 through 2005. ReportEPA420-R-05-001. Ann Arbor, MI:U.S. Environmental Protection Agency,Office of Transportation and Air Qual-ity. June.

Farrell, A., et al. 2006. Ethanol CanContribute to Energy and Environ-mental Goals. Science, 311: 506–8. 27January.

FHWA. 2006. Highway Statistics 2004.Washington, DC: Department of

31

Transportation, Federal HighwayAdministration. January.

FHWA. 2004. Traffic Volume TrendsMonthly Report. Washington, DC:Department of Transportation, FederalHighway Administration. December.http://www.fhwa.dot.gov/ohim/tvtw/04dectvt/index.htm, accessed 16 Jan.,2006.

Ford. 2004. 2003/4 Corporate Citizen-ship Report (web version). http://www.ford.com/en/company/about/corporateCitizenship/report/articlesClimateEstimatedGHG.htm, accessed 10 Jan.,2006.

Fulton L. and G. Eads. 2004. IEA/SMPModel Documentation and ReferenceCase Projection. Geneva, Switzerland:World Business Council for SustainableDevelopment (www.wbcsd.org).

FreightCar America, Inc. 2005. RailCar Product Line. Johnstown, PA:FreightCar America, Inc. http://www.johnstownamerica.com/products/fcp_coal.htm#, accessed 2 Dec. 2005.

Hu P. and T. Reuscher. 2004. Summaryof Travel Trends—2001 NationalHousehold Travel Survey. Washington,DC: Federal Highway Administration.December.

IEA. 2005. CO2 Emissions from FuelCombustion, 1971-2003. Paris, France:International Energy Agency.

Koupal, J. and S. Srivastava. 2005.MOVES2004 Validation Results. DraftReport. Report EPA420-P-05-002.Ann Arbor, MI: Environmental Pro-tection Agency. February.

Mintz, M., A.D. Vyas, and L.A. Conley.1993. Differences Between EPA-Testand In-Use Fuel Economy: Are theCorrection Factors Correct? Paper No.931104, Transportation Research Board,Washington, DC. January.

Mondt, J.R. 2000. Cleaner Cars: TheHistory and Technology of EmissionControl Since the 1960s. Warrendale, PA:Society of Automotive Engineers (SAE).

National Household Travel Survey 2001(HTS) website. http://nhts.ornl.gov/2001/index.shtml, accessed 15 February,2006.

NHTSA. 2006. Vehicle Survivabilityand Travel Mileage Schedule. NationalHighway Transportation and SafetyAdministration, Washington, DC.January.

NHTSA. 2005. Preliminary RegulatoryImpact Assessment of Corporate Aver-age Fuel Economy and CAFE Reformfor MY 2008-11 Light Trucks. NationalHighway Transportation and SafetyAdministration, Washington, DC.August.

Scott, J., et al. 2005. The Clean Air Actat 35: Preventing Death and Diseasefrom Particulate Air Pollution. NewYork: Environmental Defense.

TRB. 2005. Integrating Sustainabilityinto the Transportation Planning Process.Conference Proceedings 37. Washington,DC: Transportation Research Board.

Wang, M.Q. 2001. Development andUse of GREET 1.6 Fuel-Cycle Modelfor Transportation Fuels and VehicleTechnologies. Report ANL/ESD/TM-163. Argonne, IL: ArgonneNational Laboratory, Center forTransportation Research. June.Numerical estimates obtained fromGREET 1.5a (Feb. 2000).

Ward’s. 2005. Ward’s Motor VehicleFacts and Figures 2005. Southfield, MI:Ward’s Communications.

WBCSD. 2004. Mobility 2030—TheSustainable Mobility Project. Geneva,Switzerland: World Business Councilfor Sustainable Development.(http://www.wbcsd.org).

National headquarters

257 Park Avenue South

New York, NY 10010

212-505-2100

1875 Connecticut Avenue, NW

Washington, DC 20009

202-387-3500

5655 College Avenue

Oakland, CA 94618

510-658-8008

2334 North Broadway

Boulder, CO 80304

303-440-4901

2500 Blue Ridge Road

Raleigh, NC 27607

919-881-2601

44 East Avenue

Austin, TX 78701

512-478-5161

18 Tremont Street

Boston, MA 02108

617-723-5111

Project offices

3250 Wilshire Boulevard

Los Angeles, CA 90010

213-386-5501

1107 9th St., Suite 510

Sacramento, CA 95814

916-492-7078

East 3-501

No. 28 East Andingmen Street

Beijing 100007 China

+86 10 6409 7088