The effectiveness of gasoline taxation to manage air pollution

Transcription

1 Ecological Economics 36 (2001) ANALYSIS The effectiveness of gasoline taxation to manage air pollution Kristin N. Sipes *, Robert Menelsohn Yale School of Forestry an En ironmental Stuies, 360 Prospect Street, New Ha en, CT 06511, USA Receive 3 May 2000; receive in revise form 17 July 2000; accepte 21 July 2000 Abstract Higher gasoline taxes can be justifie because cars cause significant local, regional, an global air pollution amages. This stuy examines whether charging higher taxes woul result in significant emission reuctions. Both experimental survey ata an actual behavior in Southern California an Connecticut are evaluate to explore whether people woul change their riving behavior in response to higher gasoline prices. Both sets of results reveal that rivers are price inelastic in the short run ( 0.4 to 0.6) an long run ( 0.5 to 0.7). Imposing environmental surcharges on gasoline will result in only a small reuction in riving an thus only a small improvement in the environment. Such taxes will place a heavy an clear buren on rivers, however, making gasoline taxes extremely unpopular. Finally, the stuy fins that the income elasticity of gasoline is low ( ) so that the gas tax will fall heavily on the poor Elsevier Science B.V. All rights reserve. Keywors: Air pollution management; Gasoline taxation; Deman for gasoline; Price elasticity 1. Introuction * Corresponing author. Tel.: ; fax: aress: robert.menelsohn@yale.eu (K.N. Sipes). Dropping real gasoline prices an rising incomes have le to a 30% increase in motor vehicle travel in the Unite States in the past ecae (National Research Council, 1997). From 1998 to 1999, US gasoline consumption rose by 2.5% an vehicle miles travele increase by 1.4% (Feeral Highway Aministration, 1999). Although increases in gasoline prices over the last 6 months have taken many Americans by surprise, the inflation-ajuste price of gasoline is still cheaper toay than it was for most of the post-worl War II perio. Unfortunately, these increasing vehicle miles have cause serious local pollution problems, such as increasing emissions of particulate matter, an regional problems from the emissions of the precursors for ozone an photochemical smog. These pollutants are etrimental to human health (Wilson an Spengler, 1996), reuce visibility, harm crops, an amage vegetation (US Envi /01/$ - see front matter 2001 Elsevier Science B.V. All rights reserve. PII: S (00)

2 300 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) ronmental Protection Agency, 1997). Further, mobile sources prouce 22% of carbon ioxie emissions, contributing to global greenhouse gas problems (Michaelis et al., 1996). The US Environmental Protection Agency (US EPA) has mae great stries in controlling the pollutants from stationary sources, but the agency has mae much less progress controlling mobile sources. This article argues that mobile sources are intrinsically more ifficult to control because people value the opportunity to travel so highly. Several environmental policies have been consiere that woul ecrease vehicle miles travele in orer to control mobile source pollution. Fuel efficiency targets have been given to manufacturers. Public transit has been subsiize. Carpooling has been encourage. None of these programs have manage to prevent aggregate mileage from growing. Partly, these programs have not been stringent enough. For example, fuel efficiency on US cars has recently fallen as a result of the increase in SUVs. The fuel efficiency stanars i not stop SUVs because, officially, the SUVs are consiere trucks. However, we argue that these programs are not likely to succee primarily because Americans want to be able to rive. For example, many citizens support public transit subsiies because they hope that other rivers will leave congeste highways, not because they themselves plan to use public transit. In this paper, we examine a very efficient tool to iscourage riving, gasoline taxes. Economists argue this popular tool can be esigne to raise prices by the amount of amage cause. Gasoline taxes currently account for only the cost of roa construction an maintenance (an even this assertion is controversial). An aitional environmental surcharge on the price of gasoline woul charge rivers for the amage they cause to the environment, an may have the beneficial effect of encouraging people to rive less an purchase more fuel-efficient vehicles. Economists have estimate the marginal amages associate with motor vehicle air pollution using vehicle emissions, human exposures (population), human morbiity responses (Eyre et al., 1997), an morbiity values. Depening on local ispersion, populations, an values, economists have estimate that pollution from motor vehicles causes health amages ranging from $0.60 (Small an Kazimi, 1995) to $1.60 (Small, 1997) per gallon. Economists have recommene raising taxes on gasoline to reflect these local an regional amages. Economists have also recommene creating a carbon tax of $5 20/ton to reflect global amages (see summary by Pearce et al., 1996). These environmental taxes woul be economically efficient because they woul force rivers to consier the full cost of their ecision to rive an extra mile an to buy efficient vehicles. However, stuies of gasoline eman suggest that taxes woul have only a small impact on riving behavior. Archibal an Gillingham (1980) estimate an overall short-run price elasticity (percentage change in miles per percentage change in price) of 0.43 using househol ata from 1972 to 1973 Consumer Expeniture Survey. Walls et al. (1993) use 1990 National Transportation Survey to estimate a short-run price elasticity of Drollas (1984) estimates price elasticities of 0.35 over the short run an 0.73 over the long run. Dahl an Sterner (1991a,b) survey the gasoline eman literature an fin mean price elasticities for panel ata to be 0.26 over the short run an 0.86 over the long run. Johansson an Schipper (1997) fin similar long-run price elasticities of Our stuy uses experimental survey ata collecte in Los Angeles an Connecticut just prior to the recent increase in gasoline prices. The stuy poses hypothetical gasoline price increases to evaluate whether taxation will encourage a substantial change in riving behavior an a subsequent ecrease in air pollution. Given the results in the literature, we expect that gasoline eman will be price inelastic over both the short an long run. Because people can make more ajustments over time, long-run eman is expecte to be more elastic than short-run. If gasoline eman is highly price inelastic, this woul imply that gasoline taxation will have little influence on riving behavior an that it is not an effective policy to reuce air pollution. California has one of the lowest gasoline tax rates of any state ($0.18 per gallon), espite the extraorinary health risks pose by motor vehicle

3 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) air pollution in especially the Los Angeles basin. In contrast, Connecticut s gasoline taxes are higher than most states ($0.36 per gallon). In 1994, Californians rove about 272 billion miles or about 8693 miles per person. With the higher prices in Connecticut, people in Connecticut rove approximately 27 million miles (Feeral Highway Aministration, 1996) or about 8303 miles per person. 2. Methos To evaluate the effectiveness of gasoline taxation as an air pollution management policy, our stuy uses a survey esign similar to contingent valuation. The survey combines open-ene an iscrete choice questions to estimate short- an long-run changes in riving behavior ue to changes in price. Due to the tenency for aversion towar taxes, we ecie to use a gasoline price payment vehicle. To ensure that the survey was realistic, responents were aske questions relating to ecisions they were familiar with making in everyay life. The survey format poses hypothetical gasoline price increases intersperse with questions about how responents woul ajust. The plausibility of price increases was justifie through reference to current prices in Europe. Now that prices have increase substantially, these questions o not seem so hypothetical. At the time of the survey, however, people ha been facing steaily eclining real prices for ecaes. The survey began with a prompt to make responents feel comfortable answering questions that followe. Responents were aske to provie emographic an socioeconomic information, as well as information about their current riving behavior. They were then aske to escribe their behavioral changes in response to two sets of gasoline prices one slightly higher than current prices an one substantially higher. For each price level, responents were first aske how many miles they woul rive in response to an instantaneous price increase. They were then aske what ajustments they woul make over time an how many miles they woul eventually rive. Each survey was compose of two specific prices. Alternative surveys explore a wie range of prices from $1.70 to 2.90 for the low price an high prices ranging from $3.10 to A pilot phase was conucte to test whether the survey instrument was well behave. Post-survey interviews inicate that the responents unerstoo the questions an coul answer them. The survey instruments for Los Angeles an Connecticut are in Appenices A an B. The survey was conucte in the Los Angeles basin from November 20, 1999 to January 20, 2000, an in Connecticut from February 14 to February 27, Questionnaires were ranomly istribute to 201 people living in the Los Angeles basin an 336 people in south central Connecticut. The survey versions were ranomly arrange an each responent was hane a survey from the top of the pile. Each questionnaire was assigne a unique ientification number. Surveys with questionable valiity were marke with a question mark at the top right corner immeiately after being returne. A high fraction of the surveys istribute were complete because they were fille out in front of the interviewer. There were 200 complete responses in Los Angeles an 300 complete responses in Connecticut. Although the questionnaires were istribute in a ranom fashion, they are not a perfectly representative sample of either state. Table 1 compares sample an state statistics for age, sex, annual househol income, race, an vehicle miles travele. The California sample contains too many high-income people to be representative of the state. There are too many males in the California sample an not enough Hispanic people. The Connecticut sample is slightly below the income of the state an has too many young people, not enough elerly, an not quite enough males. One must be careful about generalizing from each sample to each state. However, there is no reason to expect that an empirical moel constructe from this ata woul be biase. Gasoline consumption was calculate from the vehicle miles per gallon ratio. Non-responses were roppe from the analysis. Race was ivie into three ummy variables white, black, an Hispanic. Long-run ajustments were categorize as separate ummy variables. Income was

4 302 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) classifie from the categorical response as follows, a=5000; b=17 500; c=37 500; =75 000; e= To account for changes in vehicle stock, gasoline consumption was calculate using the miles per gallon of the new vehicle if the responent chose to purchase a more fuel-efficient vehicle. Gasoline consumption was recore as zero if the responent chose to purchase an electric vehicle. If the responent chose to purchase both a more fuel-efficient car an an electric vehicle, gasoline consumption was recore relative to the fuel-efficient car response. Diagnostic resiual plots were use to ientify outliers in the ata. Any outliers were oublechecke in the original ataset to assess their valiity. Due to illogical or inconsistent answers, one survey was omitte from the ata analysis for Los Angeles an 36 were omitte for Connecticut. 3. Analysis The traitional eman moel is given by the following equation: ln Q i = Constant+ 1 ln P+ ( i X i )+ (1) where the log of iniviual gasoline consumption (ln Q i ) is a function of the log of price (ln P), an the sum of the socioeconomic variables (X i ). Assuming that gasoline eman followe the traitional moel in Eq. (1), the change in price postulate by the hypothetical question shoul have le to a change in gasoline consumption. Because only the price changes in the hypothetical case, Eq. (1)) suggests that the socioeconomic variables shoul have roppe out: ln Q 1 ln Q 0 =Constant+ 1 (ln P 1 ln P 0 )+ (2) That is, the i from Eq. (1) shoul equal zero given a change just in prices. However, there is a possibility that ifferent people in the sample woul respon ifferently to the propose changes in price, making 1 ifferent from zero for some variables. To test for a significant influence of socioeconomic characteristics on the change in gasoline consumption, Eq. (2) was moifie: ln Q 1 ln Q 0 =Constant+ 1 (ln P 1 ln P 0 ) + ( i X i )+ (3) If socioeconomic characteristics significantly influence the change in gasoline consumption, then this requires a moification of the traitional moel in Eq. (1) to inclue interaction terms between socioeconomic variables an price: ln Q i =Constant+ 1 ln P+ 2 (ln P ln X i ) + ( i X i )+ (4) where 2 (ln P ln X i ) involves interaction terms an ( i X i ) involves each socioeconomic variable inepenently. Table 1 Sample an state statistics Variable estimate LA sample CA state a CT sample CT state a Percent age Percent age Percent age 65 an over Percent male Meian househol income Percent white Percent black Percent hispanic Miles travele per week a Sources, Feeral Highway Aministration (1996), US Census Bureau (1999), Horner (2000).

5 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) Table 2 Change in gasoline consumption moel: Los Angeles a Inepenent variable Depenent variable: change in gasoline consumption Constant Intercept ( 4.66) 1.11 ( 3.30) ln(p 1 P 0 ) log of change in price ( 7.15) ( 7.62) ln Y log of income (5.47) (3.93) ln V log of number of househol vehicles ( 2.79) ( 2.42) Ajuste R a T-statistics in parentheses ( -level 0.05). b F(3308)=23.84; P c F(3304)=22.20; P Results Multiple regressions explaining the change in gasoline consumption were estimate for the California an Connecticut samples. A number of socioeconomic variables were explore incluing miles from work, miles from a train station, miles from a bus station, income, number of people in the househol, number of househol vehicles, sex, age, an race. Many of these variables were not significant an were subsequently roppe from the regression. Diagnostic tests reveale heterosceasticity with respect to prices. To correct for non-constant variance, we regresse the change in log of gasoline on the log change in price an use the preicte values of this regression to form a weighting variable (W i ), such that W i =1/preicte resiual, (Neter et al., 1996). Weighte least squares was then use throughout the analysis. Table 2 presents results for the change in gasoline consumption for Los Angeles. As hypothesize, the short- an long-run eman functions are ownwar sloping, with price elasticity estimates of 0.35 an 0.59, respectively. These estimates are consistent with the literature, but they are on the low sie. The responents were more price-responsive over the long run than the short run, as expecte. Income an the number of househol vehicles influence the short- an longrun change in gasoline consumption, although this effect is smaller than one might expect. Higher income groups are less responsive to an increase in gasoline price than lower income groups. Househols with more vehicles are likely to be more responsive to gasoline price increases. These responses suggest that socioeconomic variables influence price elasticity. In Table 3, we estimate a traitional eman moel using these ata. The results inicate that age, sex, an istance from work all influence the short-run eman for gasoline in California. For long-run eman, age an sex no longer matter. Long-run eman is influence by istance from work, income, an number of vehicles. Although higher income people buy more gasoline, the income elasticity is low (0.2). People with twice the income buy only 20% more gasoline. These income elasticities are lower than estimates from panel an cross-sectional ata. This coul be a flaw associate with opinion research as many contingent valuation stuies fin low-income elasticities. People with ifferent incomes may think they will respon the same way to some hypothetical change but, in fact, income will play a larger role than they anticipate. The results from Table 2 suggest that the traitional eman moel shoul be moifie to inclue interaction terms. As shown in Table 4, the results of regressions that inclue these interaction terms are subtly ifferent. Income interacts with price in the short-run moel, but sex an istance from work come in inepenently. In the long-run moel, income an number of vehicles interact with price an only istance from work comes in

6 304 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) inepenently. The ata closely fit both the moels in Tables 3 an 4. Traitionally, economists have assume that price elasticities are the same for everyone as in Table 3. However, what this analysis inicates is that price elasticities may be subtly ifferent for ifferent income an other groups. Table 5 presents the results for the change in gasoline consumption in Connecticut. As hypothesize, the short- an long-run eman functions are ownwar sloping; short an long-run price elasticity estimates are 0.39 an 0.49, respectively. Again, these estimates are consistent with the literature, but on the low sie. As expecte, eman is more price elastic over the long run than over the short run. Short- an long run changes in gasoline consumption are significantly influence by the number of people in a househol. As the number of people in a househol increases, the househol is more responsive to an increase in gasoline prices. Long-run changes in gasoline consumption are influence by income, Table 3 Traitional eman moel: Los Angeles a Inepenent variable Depenent variable: level of gasoline consumption Constant Intercept (0.066) ( 1.00) ln P log of price on survey ( 3.98) ( 4.68) ln age log of age (2.36) Sex 1=Reference person male (4.66) ln W log of miles live from work (8.81) (6.41) ln Y log of income (3.30) ln V log of number of househol vehicles ( 2.05) Ajuste R a T-statistics in parentheses ( -level 0.05). b F(4261)=29.84; P c F(4252)=17.56; P Variable roppe out of moel ue to insignificant coefficient. Table 4 Interaction eman moel: Los Angeles a Inepenent variable Depenent variable: level of gasoline consumption Constant Intercept 1.28 (6.85) 1.55 (8.15) ln P log of price on survey 2.13 ( 3.39) 2.57 ( 3.96) Sex 1=Reference person male (3.98) ln W log of miles live from work (7.41) (6.47) ln P ln Y Interaction term between log price an log income (2.71) (3.21) ln P ln V Interaction term between log price an log number of househol vehicles ( 1.37) ( 2.15) Ajuste R a T-statistics in parentheses ( -level 0.05). b F(5243)=20.07; P c F(4252)=17.47; P Variable roppe out of moel ue to insignificant coefficient.

7 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) Table 5 Change in gasoline consumption moel: Connecticut a Inepenent variable Depenent variable: change in gasoline consumption Constant Intercept (3.33) 1.23 ( 2.68) ln(p 1 P 0 ) log of change in price ( 8.30) ( 5.46) ln Hse log of size of househol ( 3.32) ( 2.18) ln Y log of income (2.65) White 1=Person white (3.70) Ajuste R a T-statistics in parentheses ( -level 0.05). b F(3414)=39.50; P c F(4346)=13.08; P Variable roppe out of moel ue to insignificant coefficient. Table 6 Traitional eman moel: Connecticut a Inepenent variable Depenent variable: level of gasoline consumption Constant Intercept 1.31 ( 2.26) 1.39 ( 2.95) ln P log of price on survey ( 3.37) ( 5.06) ln W log of miles live from work (12.3) ln Hse log of number of people in househol ( 2.08) ln Y log of income (3.97) (2.82) White 1=Reference person white (3.63) Ajuste R a T-statistics in parentheses ( -level 0.05). b F(3361)=73.79; P c F(4351)=12.69; P Variable roppe out of moel ue to insignificant coefficient. the size of the househol, an race. The coefficients suggest that white iniviuals an higher income groups are less likely to change their riving behavior as price rises. Table 6 presents the results for the traitional eman moel for Connecticut. Short-run gasoline consumption is best explaine by the istance an iniviual lives from work an income. Long run gasoline consumption is influence by the number of people in a househol, income, an whether the responent is white. The income elasticity reporte in Table 6 is 0.2 in the short run an 0.1 in the long run. The change in gasoline consumption moel in Table 5 suggests that the traitional eman moel shoul be moifie to inclue interaction terms. Table 7 presents the results for a gasoline eman moel with interaction terms in Connecticut. The significance of the interaction terms serves as an explanation for the influence of socioeconomic variables on the change in gasoline consumption. Interestingly, both the long-run moel of the change in gasoline consumption an the long-run moel of gasoline consumption with interaction terms suggest the ientical inepenent variables affect price elasticity. In both moels, the key variables are househol size, income, an race. R 2 in these regressions were low, ranging from 0.11 to This range of results is typical for a cross-sectional survey of the US population.

8 306 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) There are several sources of error that woul contribute to such a low R 2. First, many people are not even aware of the miles per gallon that they get from their current vehicle. Secon, many people have a har time imagining how they woul respon to a hypothetical situation, even when it is a familiar phenomenon such as a change in the price of a regular househol purchase. Thir, iniviual circumstances are ifferent so that many househols will react ifferently to the same stimulus. That is, there is every reason to expect a varie response across househols. 5. Conclusion Surveys in California an Connecticut suggest that househols have similar responses to gasoline taxation. The results suggest that higher gasoline prices woul encourage people to rive fewer miles an to purchase cars that are more fuel-efficient. Both of these responses are in a esirable irection an may encourage policy makers to consier environmental gas taxes. However, the results from several eman moels for Los Angeles an Connecticut consistently emonstrate that gasoline eman is price inelastic over both the short an long run. That is, people will make only relatively small changes in their behavior in response to higher prices. These survey results are further supporte by actual behavior. Stuies of panel an cross-sectional ata sets from the literature reveal low price elasticities in actual riving behavior. This stuy, however, reports price elasticity results that are slightly lower than the bulk of the empirical literature. The income elasticities in this stuy are conspicuously lower than the estimates from other stuies. This iscrepancy may be a flaw of hypothetical survey methos. People may fin it ifficult to imagine the hypothetical situation an how they woul actually respon to it. They ten to unerestimate how they will aapt. Several people verbalize this ilemma saying they coul not imagine how they coul survive if gasoline prices increase. It woul be interesting to see how these people have actually reacte to the gasoline price increases that have occurre since the survey was complete. Our results inicate that if an environmental surcharge is ae to gasoline taxes, then the aitional tax will ecrease gasoline consumption only slightly an, therefore, will have little effect on air pollution. For example, the price elasticity estimates suggest that a 33% increase in gasoline prices (a $0.50 per gallon tax) woul ecrease gasoline consumption by only 13 23%. Given the political opposition that followe from the moest $0.05 gasoline tax propose by the Clinton aministration in the early 1990s, it is not appar- Table 7 Interaction eman moel: Connecticut a Inepenent variable Depenent variable: level of gasoline consumption Constant Intercept ( 1.31) (0.376) ln P log of price on survey ( 6.73) 1.63 ( 4.10) ln Y log of income (4.30) ln P ln Hse Interaction term between price an number of people in househol ( 1.49) ( 1.64) ln P ln Y Interaction term between price an income (2.74) ln P*white Interaction term between price an white reference person (3.46) Ajuste R a T-statistics in parentheses ( -level 0.05). b F(4358)=52.34; P c F(4351)=11.82; P Variable not inclue in regression because insignificant with interaction terms inclue.

9 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) ent that it woul be worth pursuing such an unpopular environmental tax for such a small improvement in the environment. This analysis is not intene to be an inictment against using taxes to control pollution. It is well unerstoo that taxes are effective instruments to control pollution an other negative externalities. The analysis is instea revealing that people place great value on being allowe freeom of movement. Any attempt to restrict transport, for example, with gasoline taxation, is likely to fin significant resistance from citizens. Even attempts to provie public forms of transport are likely to be resiste unless they improve access (as they woul in highly congeste spaces such as Manhattan). In this circumstance, a tax on transport such as on gasoline will cause significant welfare losses that may not be politically worth the small environmental improvement they generate. In aition, our results raise questions about the equity of a gasoline tax. The results for Connecticut an California suggest that the income elasticity of gasoline consumption ranges from 0.1 to 0.2. The results suggest that people with twice the income buy only 10 20% more gasoline. Of course, governments coul use the revenues from gas taxes to aress equity issues by lowering taxes on poor people or subsiizing services for them. However, in practice, it is not clear that current subsiies for transport actually benefit poor people more than others. Even if the income elasticity estimates in this paper are low, a tax on gasoline woul most likely fall most heavily on the poor. Acknowlegements We woul like to thank the master s stuents at the Yale School of Forestry an Environmental Stuies for their assistance aministering the Connecticut survey. Appenix A. Survey instrument: Los Angeles Survey on responses to gasoline price increases. I am a grauate stuent at Yale University an this survey is a part of my program. I woul like to ask you a few questions about your riving behavior. There are no right or wrong answers; I am just intereste in your opinions. Age Sex (M/F) Annual househol income, (a) uner ; (b) ; (c) ; () ; (e) over Race. City of resience. Househol size. Number of househol vehicles. Approximate number of miles you live from a bus stop. Metro station. Approximate number of miles you live from work. Make an moel of the car you rive. Average miles per gallon of the car you rive. For all of the questions below, assume the average price of regular gasoline toay is $ Approximately, how many miles o you rive each week? 2. If gasoline prices rose to $1.70 per gallon toay, how many miles o you think you woul now rive each week? 3. If you ha time to ajust to this new price of $1.70 per gallon, which of the following ajustments o you think you woul make? Check each one that applies. None Move closer to your job Public Purchase a car that gets transportation miles per gallon Carpool Change jobs Purchase an electric vehicle Other 4. Consiering the above ajustments, how many miles o you think you woul now rive each week? 5. Now suppose that gasoline prices rise to $2.70 per gallon toay, which is close to the current price in Europe. How many miles o you think you woul now rive each week?

10 308 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) If you ha time to ajust to this new price of $2.70 per gallon, which of the following ajustments o you think you woul make? Check each one that applies. None Move closer to your job Public Purchase a car that gets transportation miles per gallon Carpool Change jobs Purchase an electric vehicle Other 7. Consiering the above ajustments, how many miles o you think you woul now rive each week? Appenix B. Survey instrument: Connecticut Survey on responses to gasoline price increases. I am a grauate stuent at Yale University an this survey is a part of my program. I woul like to ask you a few questions about your riving behavior. There are no right or wrong answers; I am just intereste in your opinions. Age. Sex (M/F). Annual househol income, (a) uner ; (b) ; (c) ; () ; (e) over Race. City of resience. Househol size. Number of househol vehicles. Approximate number of miles you live from a bus stop. Metro station. Approximate number of miles you live from work. Make an moel of the car you rive. Average miles per gallon of the car you rive. For all of the questions below, assume the average price of regular gasoline toay is $ Approximately, how many miles o you rive each week? 2. If gasoline prices rose to $1.70 per gallon toay, how many miles o you think you woul now rive each week? 3. If you ha time to ajust to this new price of $1.70 per gallon, which of the following ajustments o you think you woul make? Check each one that applies. None Public transportation Carpool Change jobs Move closer to your job Purchase a car that gets miles per gallon Purchase an electric vehicle Other 4. Consiering the above ajustments, how many miles o you think you woul now rive each week? 5. Now suppose that gasoline prices rise to $2.70 per gallon toay, which is close to the current price in Europe. How many miles o you think you woul now rive each week? 6. If you ha time to ajust to this new price of $2.70 per gallon, which of the following ajustments o you think you woul make? Check each one that applies. None Move closer to your job Public Purchase a car that gets transportation miles per gallon Carpool Change jobs Purchase an electric vehicle Other 7. Consiering the above ajustments, how many miles o you think you woul now rive each week? References Archibal, R., Gillingham, R., An analysis of the shortrun consumer eman for gasoline using househol survey ata. Rev. Econ. Stat. 62, Dahl, C., Sterner, T., 1991a. Analyzing gasoline eman elasticities: a survey. Energy Econ. 13, Dahl, C., Sterner, T., 1991b. A survey of econometric gasoline eman elasticities. Int. J. Energy Syst. 11 (2), Drollas, L.P., The eman for gasoline. Energy Econ. 6, Eyre, N.J., Ozemiroglu, E., Steele, P., Fuel an location effects on the amage costs of transport emission. J. Transport Econ. Policy 31, 5 23.

11 K.N. Sipes, R. Menelsohn / Ecological Economics 36 (2001) Feeral Highway Aministration, Annual Vehicle Miles of Travel an Relate Data. Office of Highway Information Management. FHWA-PL Feeral Highway Aministration, Monthly Motor Fuel Reporte by States: April Office of Highway Policy Information. FHWA-PL-002. Horner, E., Almanac of the 50 States. Information Publications, Palo Alto, CA, p Johansson, O., Schipper, L., Measuring long run automobile fuel eman: separate estimations of vehicle stock, mean fuel intensity, an mean annual riving istance. J. Transport Econ. Policy 31 (3), Michaelis, L., Bleviss, D., Orfeuil, J., Pischigner, R., Mitigation options in the transportation sector. In: Watson, R., Zinyowera, M., Moss, R. (Es.), Climate Change 1995: Impacts, Aaptations, an Mitigation of Climate Change: Scientific-Technical Analyses. Cambrige University Press, Cambrige, UK. National Research Council, Towar a Sustainable Future: Aressing the Long-Term Effects of Motor Vehicle Transportation on Climate an Ecology. National Acaemy Press, Washington, DC, p Neter, J., Kutner, M.H., Nachtsheim, C.J., Wasserman, W., Applie Linear Statistical Moels. Irwin, Chicago, IL, p Pearce, D.W., Cline, W.R., Anchanta, A.N., Fankhauser, S., Pachauri, R.K., Tol, R.S.J., Vellinga, P., The social costs of climate change: greenhouse amage an the benefits of control. In: Bruce, J.P., Lee, H., Haites, E.F. (Es.), Climate Change 1995: Economic an Social Dimensions of Climate Change. Cambrige University Press, Cambrige, Englan. Small, K.A., Economics an urban transportation policy in the Unite States. Regional Sci. Urban Econ. 27, Small, K.A., Kazimi, C., On the costs of air pollution from motor vehicles. J. Transport Econ. Policy 29, US Census Bureau, State Population Estimates: Annual Time Series, July 1, 1990 July 1, Population Estimates Program, Population Division, Washington, DC. ST-99-3, ST-99-10, ST US Environmental Protection Agency, Regional Approaches to Improving Air Quality. Office of Air Quality Planning an Stanars, Research Triangle Park, NC. EPA/ 451-K Walls, M., Krupnick, A., Hoo, C., Estimating eman for vehicle miles travele using househol survey ata: results from the Nationwie Personal Transportation Survey. Resources for the Future Discussion Paper ENR 93-25, Washington, DC. Wilson, R., Spengler, J.D., Emissions, ispersion, an concentration of particles. In: Wilson, R., Spengler, J.D. (Es.), Particles in Our Air: Concentrations an Health Effects. Harvar University Press, Cambrige, MA, pp