Cost savings of using a marketable permit system for regulating light-duty vehicle emissions

Transport Policy 1994 1 (4) 221-232

Cost savings of using a marketable permit system for regulating lightduty vehicle emissions Michael Q Wang Centerfor Transportation Research, Argonne National Laboratory, 9700 South Cass Avenue, Building 362lRoom H220, Argonne, IL 60439, USA

In the USA, each individual vehicle is required to meet uniform per-mile emission standards. The uniform standard system does not allow vehicle manufacturers flexibility in achieving overall emission reduction goals for motor vehicles. The system provides manufacturers with little or no incentive to control vehicle emissions beyond what is required. In this paper, an incentive-based marketable permit system is proposed to replace the uniform standard system. Under the marketable permit system, vehicle manufacturers are required to meet corporate average emission standards; they are allowed to buy or sell vehicle emission reduction credits among themselves to meet corporate average standards; and they are allowed to bank vehicle emission reduction credits that are accumulated in earlier years and to use the credits for meeting average standards in later years. It is estimated in this study that relative to the current uniform standard system, the marketable permit system can reduce vehicle emission control costs by $150 to $400 million per year in California, or 13-30% of the costs currently spent on vehicle emission control. Keywords:

motor vehicle emissions,

marketable

permit systems, motor vehicle emission control costs

Motor vehicles are one of the largest contributors to the US air pollution problem. They generate 27% of total US volatile organic compounds, 50% of carbon monoxide (CO), 29% of nitrogen oxides (NO,), and 17% of matter (US Environmental Protection particulate Agency (EPA), 1991). To help solve the problem, the 1990 US Clean Air Act Amendments adopt stringent per-mile vehicle emission standards (US EPA, 1990a). The amendments take the current uniform standard approach in designing vehicle emission standards: every individual vehicle engine family has to meet uniform per-mile emission standards. With the current uniform standard approach, while vehicle manufacturers are held fully responsible for meeting standards, they are This study was conducted when the author was with the Institute of Transportation Studies, University of California at Davis. The author is grateful to Professors Catherine Kling of Iowa State University and Daniel Sperling of University of California at Davis for their helpful discussions with the author. The author thanks two anonymous reviewers of this journal for their helpful comments. Primary funding for this study was provided by the California Institute for Energy Efficiency and the University of California Transportation Center. Additional funding for refining this study was provided by the Office of Environmental Analysis, US Department of Energy, under contract W-31.109.ENG-38. The author is solely responsible for the contents of the paper. 0967-070X/94/040221-12

0 1994 Butterworth-Heinemann

Ltd

allowed little flexibility in selecting emission levels for individual vehicle models and are provided with no incentives to control vehicle emissions beyond what is required. As vehicle emission control continues to become stringent, the inflexible, non-incentive system becomes costly. Consequently, alternative regulatory systems to replace the current vehicle emission regulation systems need to be investigated. In this paper, a marketable permit system for regulating vehicle emissions is proposed to replace the current uniform emission standard system. The proposed marketable permit system consists of: vehicle emission averaging across vehicle engine families (an engine family usually contains several vehicle models with the same powertrain components) produced by a vehicle manufacturer; trading of emission reduction credits among vehicle manufacturers; and banking of the emission reduction credits that are accumulated in earlier years for meeting average standards in later years. The proposed marketable permit system allows manufacturers flexibility in meeting standards and provides incentives for manufacturers to control emissions beyond what is required. Because of its flexibility, the proposed system is more cost-effective than the current uniform standard system. In this paper, an optimization 221

model is presented, which simulates manufacturers’ behavior in meeting emission requirements under the marketable permit system and under the current uniform standard system. With this optimization model, cost savings of the marketable permit system are estimated. Finally, implementation issues and policy implications of adopting the marketable permit system for vehicle emission control are explored.

Review of previous proposals and studies Previous proposals Marketable permit systems have long been proposed for regulating stationary source emissions in the USA. Particularly, since the early 1970s various formats of marketable permit systems have been designed for stationary source emission control. The cost and air quality impacts of these systems have been evaluated. As a result, the US EPA adopted in 1986 a stationary marketable permit system that includes provisions for emission averaging, emission trading and emission banking (US EPA, 1986). Most recently, in the 1990 US Clean Air Act Amendments, a marketable permit system was adopted, allowing electric utility companies to average, trade and bank sulfur oxides (SO,) emissions that are generated from electric power plants (US EPA, 1990a). For an extensive review of stationary marketable permit systems, see Wang (1992). Marketable permit systems for regulating motor vehicle emissions have been proposed in the USA to a lesser extent. White (1982) conceptualized a system whereby, at the beginning of each model year, regulatory agencies would allocate emission permits among vehicle manufacturers through a bidding process. A manufacturer would be required to limit total emissions of all its vehicles sold during a model year to those allowed by the permits held. Manufacturers would be allowed to sell or buy permits to meet their emission requirements. White foresaw that vehicle manufacturers would oppose the permit system, because they would have to pay for emission permits under the bidding process. Consequently, he proposed fleet average emission standards to replace the uniform emission standards then (and now) applied to every individual vehicle engine family. Under the emission averaging approach, salesweighted average emissions of a manufacturer’s new vehicles would have to be equal to or below fleet average standards. White also suggested that provisions for emission trading among manufacturers and emission banking over time be incorporated into the averaging system. However, he did not analyze the emission and cost impacts of his emission averaging system. In its proposed 1989 Clean Air Act Amendments, the Bush Administration proposed stringent vehicle tailpipe emission standards. To ease the difficulty of implementing the proposed standards, the Bush Administration introduced a marketable permit system for vehicle manufacturers and a marketable permit system for fuel suppliers (named as a fueling pooling averaging system) 222

(White House, 1989; US EPA, 1989). Under the proposed vehicle marketable permit system, vehicle manufacturers were required to meet corporate average emission standards. Manufacturers could earn transferable credits from sales of clean vehicles. These credits could be used to average down a manufacturer’s fleet emissions or could be sold to other manufacturers. Because of the opposition in the US Congress, the proposed marketable permit system was dropped from the final adopted 1990 Clean Air Act Amendments. Adopted marketable permit systemsfor \aehicle emission control EPA’s ulw-acging progrum to control partirulute mutter emissions of light-duty diesel \,ehicles. In 1983, the EPA adopted an averaging program to control particulate matter (PM) emissions of light-duty diesel vehicles (US EPA, 1983). The EPA claimed that the averaging program would give manufacturers flexibility in meeting diesel vehicle PM standards and would result in control cost savings. The PM averaging program requires twostep emission compliance: (1) compliance with engine-family emission limits within an engine family; and (2) compliance with average standards by a manufacturer. A manufacturer determines emission limits for its individual engine families. Vehicle models within an engine family must meet the determined emission limit. Sales-weighted averages of engine-family emission limits for a manufacturer must be equal to or below average standards. Since the early 1980s diesel light-duty vehicle sales in the USA have declined dramatically. Therefore, the share of PM emissions from diesel lightduty vehicles has become small. Consequently. the averaging program has not been used by manufacturers. EPA’s marketable permit system to control emissions of heavy-duty engines. In 1990, the EPA adopted a marketable permit system to control NOx and PM emissions of heavy-duty engines (HDEs) (US EPA, 1990b; 1990~). The system consists of emission averaging, emission trading and emission banking. Like the averaging program for light-duty diesel vehicles, the HDE marketable permit system requires two-step compliance: (I) compliance with emission limits within an engine family; and (2) compliance with average standards by a manufacturer. To help alleviate potential adverse emission impacts of the marketable permit system, the EPA applies restrictions to the permit system. For example, emission averaging is restricted within each of the subclasses - light HDE, medium HDE and heavy HDE - because of wide variations in lifetime mileage accumulation among these subclasses; ceilings are applied to engine-family emission limits in order to prevent possible introduction of gross-emitting heavyduty engine families; and a 20% one-time discount rate is applied to all tradable and bankable credits. The EPA adopted the HDE marketable permit system with the intention of helping reduce the total HDE emission control costs for manufacturers. Since gross vehicle weight of HDE vehicles ranges from 8500 to over 20

Cost savings oj’ using a marketable permit system: M Q Wang 000 lbs, meeting uniform standards for some heavier HDE vehicles could be prohibitively expensive. The permit system allows higher emissions from these vehicles as long as average emission standards are met. Therefore, total control cost of HDE vehicles can be reduced substantially. The estimated cost savings of the HDE permit system will be presented in a later section.

vidual engine families, substantially reducing the size of the base upon which averaging, trading and banking can be done. This substantial reduction in the size of the program’s base reduces the program’s flexibility and, therefore, reduces cost savings. CARB has not estimated the cost savings of its permit system.

Air Resources Board’s low-emission vehicle program. In 1990, the California Air Resources Board (CARB) adopted stringent vehicle emission standards for five vehicle types - conventional vehicles (CVs), transitional low-emission vehicles (TLEVs), low-emission vehicles (LEVs), ultra-low-emission vehicles IULEVS) and zero-emission vehicles (ZEVs) (CARB, 1990). These standards are listed in Table 1. To allow manufacturers flexibility in meeting the standards, CARB established the model-year-specific fleet average non-methane organic gas (NMOG) emission standards listed in Table 2, and designed a limited NMOG marmanufacturers to ketable permit system, allowing average, trade and bank NMOG emissions. Under the permit system, a manufacturer selects the low-emission vehicle type to which a given engine family belongs. The manufacturer must meet the NMOG, CO, and NOx emission standards for the engine family of the selected low-emission vehicle type. To meet fleet average SMOG emission standards, the manufacturer can choose any mix of the five vehicle types, as long as sales-weighted NMOG emissions are equal to or below the average NMOG standards. Besides emission averaging, manufacturers are allowed to trade and bank NMOG emissions in order to meet fleet average NMOG standards. With the banking provision, CARB allows a manufacturer to deposit credits for meeting future standards or to borrow future credits for meeting past or present standards. To discourage manufacturers from accumulating too many emission reduction credits, CARB applies high discount rates to the credits. To foster development of the cleanest vehicle technologies, CARB adopted, in conjunction with the NMOG permit system, a ZEV sales requirement for 1998 and later model years. Currently, only batterypowered electric vehicles qualify as ZEVs. CARB designed a sales permit system to allow manufacturers to meet the ZEV sales requirement - a manufacturer who sells more ZEVs than required earns sales credits. The sales credits can be used to offset the manufacturer’s future sales requirement, or they can be sold to other manufacturers. The CARB’s marketable permit system is limited in two aspects. First, emission averaging, trading and banking are applied only to NMOG emissions. No average standards are applied to CO and NO, emissions. Consequently, certain mixes of the five vehicle types chosen by manufacturers for meeting average NMOG standards may result in higher or lower fleet average CO and NO, emissions than CARB expects. Secondly, the CARB’s marketable permit system sets up emission standards for the five vehicle types rather than for indi-

One economic theory suggests that in order to minimize total emission control costs of various sources, marginal emission control costs - the costs of the last units of emissions controlled - must be equal across sources (see Baumol and Oates, 1988). Emission averaging helps to equalize marginal control costs among vehicle engine families produced by a manufacturer; emission trading helps to equalize marginal control costs of engine families produced by various manufacturers; and emission banking helps to equalize marginal control costs over model years. Thus, the three provisions help reduce total vehicle emission control costs.

California

Transport Policy

1994 Volume 1 Number- 4

Previous cost savings estimates

Marketable permit systems for light-duty vehicles. In the early 1980s the EPA funded a study to the TCS Management Group to estimate cost savings of a thenproposed marketable permit system for regulating light-duty vehicle emissions (TCS Management Group, Inc, 1984). TCS estimated the cost savings of emission averaging, emission trading, emission banking and emission charges (allowing manufacturers to pay fees for the emissions in excess of emission standards). Cost savings were estimated under two scenarios: (1) meeting 1981 emission standards; and (2) meeting 1981 emission certification levels. In order to have a safety margin for meeting emission standards, emission certification levels under the uniform standard system are always below emission standards. In the scenario of meeting 198 1 standards, average emission standards were set at 1981 standards without applying any safety margin. Under this scenario, actual emissions with the permit system were much higher than those with the uniform standard system. On the other hand, under the scenario of meeting 1981 emission certification levels, emissions were exactly the same under the two systems. Table 3 presents cost savings of the marketable permit system relative to the uniform standard system. Under the scenario of meeting 198 1 emission standards, the marketable permit system saved control costs by 18-54%, but actual emissions were increased by 33-78%. Under the scenario of meeting the 1981 certification levels, emission averaging resulted in a negligible amount of cost savings. Emission averaging and trading together saved control costs by 12%. Emission banking could save additional control costs, but the TCS study did not estimate cost savings of emission banking. The TCS study probably underestimated the potential cost savings of the permit system for several reasons. First, the study underestimated vehicle emission control costs (estimated below $200 per vehicle), thus resulting in lower monetary savings. Secondly, although the mar223

Cost savings of using a marketable permit system: M Q Waq Table 1 CARB emission standards

for low-emission

vehicles”

Emission standards (gpm at 50 000 miles) TLEV LEV ULEV ZEV cv NMOG co NOX Formaldehyde

0.25” 3.4 0.4 0.015’

0.125

0.015

0.075 3.4 0.4 0.015

0.04 3.4 0.2 0.008

1.7 0.2

0 0 0 0

“Passenger cars. The tive vehicle types are conventional vehicles (CVs), transitional low-emission vehicles (TLEVs), low-emission vehciles (LEVs), ultra-low-emission vehicles (ULEVs) and zeroemission vehicles (ZEVs). bEmission standard of non-methane hydrocarbon. %tandard of formaldehyde is applied only to vehicles fueled with methanol. Table 2 CARB average NMOG standards

(gpm at 50 000 miles)

Model year

Car and light-duty truck weighing < 3750 Ibs

Light-duty truck weighing > 3750 Ibs

I992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 and after

0.390 0.334 0.250 0.23 1 0.225 0.202 0.157 0.113 0.073 0.070 0.068 0.062

0.500 0.428 0.320 0.295 0.287 0.260 0.205 0.150 0.099 0.098 0.095 0.093

Table 3 Annual cost savings for a marketable light-duty vehicle emission control Marketable permit provisions

Savings ($ million)

Scenario of meeting 1981 emission standards” 353 Averaging Averaging and trading 737 I050 Averaging, trading and banking Scenario of meeting 1981 emission certification 9 Averaging 234 Averaging and trading

permit

system

for

As of total control cost (9%) I8 37 54 levelsh 0.4 12

Soul-~: TCS Management Group, Inc (1984). Cost savings are in 198 1 dollars. “Federal emission standards for 1981 model-year cars were 0.41 gpm for hydrocarbon (HC), 3.4 gpm for CO and 1.O gpm for NOx. bAverage emission certification levels for 198 I model-year cars were 0.23 gpm for HC, 2.0 gpm for CO and 0.6 gpm for NO\.

ketable permit system as proposed was on the basis of engine families, the TCS study simulated the permit system on the basis of three vehicle classes. The smaller size of the trading basis used in the TCS study as compared with the actual size of the trading basis certainly reduced estimated cost savings of the permit system. Thirdly, the study assumed that 1981 emission standards would be met. Vehicle emission standards will become much more stringent after 1994, which will cause emission control costs to increase substantially. At high emission control costs, the marketable permit system will result in large cost savings. 224

Marketable permit systems for heavy-duty vehicles. In 1986, the EPA funded a study to Sobotka and Company, Inc (SCI) to estimate cost savings of an HDE marketable permit system (SCI, 1986). The SC1 study grouped HDEs into four subclasses: (1) heavy-duty gasoline engines; (2) light heavy-duty diesel engines; (3) medium heavy-duty diesel engines; and (4) heavy heavy-duty diesel engines. Cost savings were estimated under three emission trading scenarios: (1) restricted trading (trading and averaging allowed only within the HDE subclass); (2) partially restricted trading (trading and averaging allowed among three diesel HDE subclasses but not between gasoline and diesel HDE subclasses); and (3) unrestricted trading (trading and averaging allowed among all HDE subclasses). Assuming average standards for NO, and PM to be 80% of the 1991 uniform HDE emission standards (to implicitly account for the difference between emission standards and emission certification levels), the SC1 study estimated substantial cost savings by the permit system (Table 4). Under the restricted and partially restricted trading scenarios, emission averaging mainly contributes to cost savings, while under the unrestricted trading scenario, both averaging and trading contribute to cost savings. The study estimated cost savings from emission banking to be about 2% of total control costs. The SC1 study did not include methanol-fueled HDEs, which are included in the EPA’s adopted HDE marketable permit system. Including methanol HDEs in the cost estimation would certainly increase cost savings, because of the increase in averaging and trading flexibility.

A complete marketable permit system The previously proposed or adopted motor vehicle marketable permit systems were either incomplete or limited. In this study, a complete marketable permit system for US light-duty vehicle emission control is proposed. The elements of this system are presented below. Emission averaging Under the marketable permit system, a manufacturer’s new vehicle fleet will be subject to fleet average emission standards, and individual vehicles will no longer be subject to uniform standards. Currently, compliance with uniform emission standards by individual vehicles is on the basis of engine families. Specifically, each individual engine family must be certified for complying with uniform emission standards. To accommodate the proposed marketable permit system in the existing emission certification process, fleet average emission standards will be enforced on the basis of engine families; that is, emission levels of individual engine families will be determined by manufacturers and certified by the EPA or the CARB. The certified engine-family emission levels and the vehicle sales will be used to calculate the average emissions of the new vehicles sold by that manufacturer. In contrast, the com-

Cost sa,@gs Table 4 Annual cost savings for an HEDE marketable Emission trading provision Emission averaging Restricted Partially restricted Unrestricted Emission averaging Restricted Partially rstricted IJnrestricted

permit system

Savings ($ million)

As of total cost (W

123.1 158.2 191.5

12.2 15.7 19.0

130.9 167.2 298.7

13.0 16.6 29.6

pliance with NMOG average standards in the CARB’s LEV program is determined on the basis of five vehicle types. Fleet average emission standards will be established for hydrocarbon (HC), CO and NO,, separately. Manufacturers will be required to meet an average standard for each pollutant. Emission averaging across the three pollutants will not be permitted, because such averaging could intensify the adverse health effects of one pollutant over another. Average emission standards will be established for passenger cars and light-duty trucks; emission averaging between them will be allowed because of their similar usage (i.e. they have similar annual vehicle miles traveled (Davis and Morris, 1992) ). Average emissions of the new vehicle fleet for a manufacturer will be calculated as follows:

I

“‘,=I El1X Salesj 1” I-1 ._ Salesi

where AER, = average emission rate for pollutant i (grams per mile (gpm) ) E,,,j = emission rate for pollutant i of engine family j (gpm) Sales. = sales volume of engine family j (equal to sales oflall models contained in the engine family) i = 1,2 or 3 to represent HC, CO or NO, j = engine family n = total number of engine families produced by a manufacturer. Under the current uniform standard system, manufacturers produce engine families having emission certification rates lower than emission standards. This difference exists because manufacturers need a safety margin to ensure that emission standards will be met, and because emission reductions in motor vehicles are usually discrete rather than continuous caused by the installation of individual emission control systems on vehicles. The difference between emission standards and emission certification rates is realized as actual emission reductions. However, under average emission standards, manufacturers to some extent may use the emission difference to lower their average emissions. Therefore, Ttmsport

if average emission standards were set to be the same as the current uniform standards, certification emissions would increase under the marketable permit system. In order to maintain identical certification emissions under the two systems, average emission standards have to be set lower than the current uniform standards. Emission trading

and trading

~OWW: SC1 (1986). Cost savings are in 1986 dollars.

AER =

ofusing a marketable permit system: M Q Wang

Policy 1994 Volume 1 Number 4

If the fleet average emissions for a manufacturer are lower than average emission standards, the manufacturer will earn emission reduction credits (ERCs). Total ERCs earned by a manufacturer will be calculated as the difference between average emission standards and the fleet average emissions multiplied by total vehicle sales in a model year. The formula below shows the calculation of ERCs for a manufacturer. ERC,= (AES, - AER,) X Sales where ERC, = emission reduction credit for pollutant i (gpm-vehicle) AES; = average emission standard for pollutant i (gpm) AER, = average emission rate for pollutant i for the manufacturer (calculated with the previous formula, gpm) Sales = total annual vehicle sales by the manufacturer. Manufacturers can sell or buy ERCs to or from each other to meet average emission standards. To effect trade among manufacturers, a formal emission trading market may be established for conducting ERC transactions. At the end of each model year, manufacturers will be required to submit the information on ERC transactions to the EPA or to the CARB, for determining compliance with standards. There are about 3.5 manufacturers involved in the US light-duty vehicle market. Thus, there are about 35 potential participants in the ERC trading market. Operation of such a small trading market should be simple. Emission bunking The marketable permit system will incorporate emission banking through which manufacturers can use the ERCs earned in one model year to meet average standards in other model years. Emission banking could be forward or backward. Forward banking works like a deposit in a financial bank. The ERCs earned in an earlier model year are deposited and used to meet emission standards in later model years. Forward banking encourages early deployment of emission control technologies. Backward banking works like a loan in a financial bank. Manufacturers can borrow future ERCs to meet emission standards in current or past model years. Backward banking may cause delays in deploying emission control technologies, and therefore in emission reductions. Enforcement of backward banking could be problematic when a manufacturer cannot later repay its earlier ERC withdrawal. It is proposed in this study that the mar225

Cost savings of using a marketable permit systenl: M Q War?,?

ketable permit system exclude backward banking. In contrast, the CARB’s LEV program includes both forward and backward banking for meeting NMOG average standards. Some manufacturers might accumulate a large number of ERCs over a period of time, which could delay development and deployment of emission control technologies over time and could distort future vehicle emission regulations. Also, by holding a large number of ERCs, these manufacturers might practice monopoly power in the ERC trading market to distort trading activities. To discourage possession of a large number of ERCs, it is proposed in this study that a discount rate be applied to the ERCs held by manufacturers each year. Discounting ERCs over time also represents an actual emission reduction benefit, because a manufacturer must reduce more than one unit of emissions now to offset one unit of emissions in the future.

Estimating cost savings of the proposed marketable permit system Simulation model

During this study, an optimization model was constructed to simulate manufacturers’ behavior when meeting emission requirements under the current uniform standard system and under the marketable permit system. The optimization model was designed to minimize total emission control costs subject to constraints of meeting emission requirements under each system. The difference between the uniform standard system and the marketable permit system is in the emission constraints. Under the uniform standard system, emission constraints are that each individual engine family must have emissions equal to or below the standards for light-duty vehicles. Under the marketable permit system, the emission constraints are that the average emissions of all engine families must be equal to or below average emisthe sion standards. Under the permit system, optimization model determines emission levels for individual engine families on the basis of minimizing total control costs. With these constraints, the optimization model calculates minimized total control costs for both the uniform standard system and the marketable permit system. The difference in total control costs between the two systems represents the control cost savings for the marketable permit system. The marketable permit system designed in this study includes emission averaging, trading and banking. In order to simulate the complete system, a dynamic optimization model that simulates the permit system over a period of time has to be designed, and information on costs and emissions over that period has to be collected. In this study, a static optimization model targeting a particular year was established to simulate emission averaging and trading, but not banking. A recent study by Rubin and Kling (1993) modified the data collected in this study, established a dynamic optimization model and simulated emission banking. 226

The specifications of the established tion model are presented mathematically

static optimizabelow.

Minimize TC =c c CF,,i(HCi ,, COi.,, NOxl,) Sales,,i i j

X

subject to

(1) C C Sales. = C C CSales. x CSa/py (2) c 1 HC. .“‘X Sales < C f: CHC (3) C C CO”’ X Sales.‘!< 1 C CC0 “‘X CSales’ “’

(4) c c NOL:,i X Sale;‘:; 5 1 C CNax,,,

x

C’Sal;!y,,

where i = vehicle engine family i j = manufacturer j TC = total cost of vehicle emission control CF,,,,(HCj,j,COI,,,NOxj,) = emission control cost as a function of emissions of HC, CO and NO, for engine family i and manufacturerj Sales,,, = sales of engine family i by manufacturer ,j under the marketable permit system CSalef,,; = current sales of engine family i by manufacturer j HC,,j = HC emissions of engine family i and manufacturerj under the marketable permit system CO,, = CO emissions of engine family i and manufacturer j under the marketable permit system NOX,,, = NO, emissions of engine family i and manufacturer j under the marketable permit system CHC,; = current HC emissions of engine family i and manufacturer j CCO,,, = current CO emissions of engine family i and manufacturer ,j CNOxid = current NO, emissions of engine family i and manufacturer j. As shown, the optimization model contains an objective function and four constraint functions. The objective function calculates total emission control costs for meeting emission standards by the automotive industry. The function indicates that total control costs can be affected by changing emission levels of HC, CO and NO,, or by changing vehicle sales, or both. Of the four constraint functions, the first function is a vehicle total sales constraint, which sets vehicle total sales identical under the uniform standard system and the marketable permit system. In other words, it is assumed here that vehicle total sales by the automotive industry will remain the same under both systems. Although emission control costs will be reduced by changing from the uniform standard system to the marketable permit system, and thus vehicle sales prices probably will be reduced slightly, any vehicle sales increase due to the slight vehicle price decrease probably will be small. This secondary vehicle sales effect of changing emission regulatory systems is ignored here. However, the vehicle sales mix of individual engine families may vary for the two systems; that is, by changing from the current uniform standard system to the marketable permit system, sales of some engine

Cost savings ofusing a marketable permit system: M Q Wang families may go up, while sales of other engine families may go down. The optimization model allows changes in vehicle sales mix. The remaining three constraints are emission constraints for HC, CO and NO,. To maintain the same actual emissions under the uniform standard system and the marketable permit system, averaged emission standards under the permit system are set at the certification emission levels which occurred under the uniform standard system (see the discussion about the difference between emission standards and certification emission levels under the two systems in the Section entitled ‘Emission averaging’). Emission control cost functions Vehicle emission control cost as a function of emissions of HC, CO and NO, is needed in the model’s objective function. During the study, data on emissions and emission control systems for California vehicle engine families were obtained from CARB. Vehicle emission control cost functions were established for light-duty vehicles sold in California in 1990. Thus, the estimation of the marketable permit system’s cost savings was conducted for light-duty vehicles sold in California in 1990. Compliance with fleet average emission standards under the marketable permit system proposed here is based on individual engine families. Currently, about 300 vehicle engine families are produced for California’s light-duty vehicle market. To precisely simulate the permit system, a cost function for each of the 300 engine families has to be estimated. Because of the lack of sufficient data on emissions and costs for each individual engine family, it was impossible to establish 300 individual cost functions. Instead, vehicle engine families of each manufacturer were grouped into three vehicle classes, on the basis of number of cylinders: (1) small (4-cylinder vehicles), (2) medium (5- and 6-cylinder vehicles), and (3) large (8- and 12-cylinder vehicles). Cost functions were established for the three vehicle classes of a given manufacturer. Because data was available, 12 vehicle manufacturers were included in this study. They are Audi, BMW, Chrysler, Ford, General Motors, Honda, Mazda, Mercedes-Benz, Mitsubishi, Toyota, Volvo, and Volkswagen. The three vehicle groups and the 12 manufacturers together form 29 combinations. Note that the potential combinations should be 36. Because some manufacturers do not produce certain vehicle groups, the actual combinations are smaller than the potential number. Establishment of a cost function for a group of engine families rather than for an individual engine family forces the simulation of the marketable permit system to be based on vehicle groups, rather than on individual engine families, despite the fact that emission averaging, trading and banking are designed to be based on engine families. The size of the trading basis in the simulation, therefore, decreases from the actual number of engine families (about 300) to the assumed combinations of vehicle groups and manufacturers (29). The

substantial decrease in the size of trading basis certainly causes the underestimation of potential cost savings of the marketable permit system. In the process of establishing cost functions, various functional forms were tested. It was found that log cost functions best represented control cost changes with emissions. The selected log cost functions take the following form: In(K) = BO + B 1 x ln(HCC0) x ln(mpg)

+ B2 x ln(N0,)

+ B3

where TC = cost of vehicle emission control ($/vehicle) HCCO = HC emissions (gpm) multiplied by CO emissions (gpm) NO, = NO, emissions (gpm) mpi: = fuel economy (miles per gallon) BO, B 1, B2 and B3 = coefficients As emission control efforts are increased, tailpipe emissions decrease, and emission control costs increase. Thus, coefficients Bl and B2 should be negative. In order to statistically generate negative coefficients for HC and CO emissions, a composite variable is generated by multiplying HC emissions and CO emissions. Engineering justification for this step is that uncontrolled HC and CO emissions tend to move in the same direction (e.g. if one increases, the other tends to increase) and that the same or similar control strategies are employed for HC and CO emission control. Emission control costs are determined by vehicle specification parameters as well as by the amount of emissions controlled. Vehicle specifications, such as engine size and vehicle weight, affect emission control costs through their effects on the amount of uncontrolled emissions. In general, large, heavy vehicles generate more engine-out emissions. To meet emission standards, these vehicles need more intensive emission control. Because of good correlation among vehicle fuel economy, engine size and vehicle weight, fuel economy was included in the cost function to reflect the effect of engine size and vehicle weight on emission control costs. To estimate coefficients for the cost function, emissions, fuel economy and emission control costs are needed. Tailpipe emission certification rates and fuel economy for 387 vehicle samples were collected from the Emission Certification Section of CARB. Vehicle emission control costs were estimated with an emission control part-pricing approach developed by Wang et ul. (1993). The part-pricing approach relies on manufacturer-suggested retail prices of vehicle emission parts. Estimation of emission control costs by the partpricing approach required several steps. First, emission control parts installed on individual vehicle models needed to be identified. Identification was accomplished through checking information contained in the emission certification application form for an engine family that had been submitted by the manufacturer to the CARB. The application form contained detailed information on 227

Cast savings of using a marketable permit system: M Q Wang emission parts installed, technical specifications, operating parameters for emission tests, and vehicle models contained in an engine family. Secondly, manufacturersuggested retail prices for emission parts were collected from vehicle dealers. Thirdly, retail prices of parts were discounted to manufacturing costs by subtracting profit and cost markups of dealers and manufacturers. Fourthly, manufacturer costs for replacement parts were converted to manufacturer costs for initial parts. This step was necessary because the manufacturer-suggested retail prices were for replacement parts, and replacement part prices are usually higher than prices for the same parts when supplied to vehicle assemblers. Fifthly, the cost of vehicle assembly and engine modifications needed to incorporate emission parts into a vehicle system was included in the estimated emission control costs. Finally, costs of individual emission parts installed in a vehicle model were added together to obtain the total emission control cost for the vehicle model. Estimated emission control cost functions Because of insufficient data, a cost function could not be estimated independently for each of the 29 combinations of vehicle groups and manufacturers. Instead, an aggregate cost function was estimated for each of the three vehicle groups (i.e. small, medium and large vehicles). The aggregate cost function for a given vehicle group was adjusted to a particular manufacturer by using the ratio of the average control cost for the manufacturer to that for all manufacturers for the given vehicle group. Through this ‘average approach’ (in the sense that average costs are used to disaggregate cost functions), all coefficients of the cost function are changed proportionately by the cost ratio. This approach implicitly assumes that the difference in emission control costs among vehicle manufacturers is caused by each coefficient of the cost function proportionally. In this way, 29 individual cost functions were established for the 29 combinations of vehicle groups and manufacturers. The 29 cost functions are presented in Table 5.

Results of cost savings estimation The established optimization model was written into a General Algebraic Modeling System (GAMS) program and run on a computer. In simulating the marketable permit system, certain assumptions were made regarding changes in vehicle sales mix and changes in emission standards. These assumptions are discussed in the following sections. Changes in vehicle sales mix Under the marketable permit system, manufacturers may meet average emission standards by changing emission levels of individual vehicle groups and/or by changing vehicle sales mix. Two cases are established to address the potential change in vehicle sales mix. One case assumes that vehicle sales mix remains exactly the same under the uniform standard system and the 228

marketable permit system; then, emission control cost savings of the marketable permit system come strictly from reallocation of emission control efforts among various engine families. The other case allows change in vehicle sales mix between the two systems. With this case, if changes in sales mix are allowed without any constraint, the optimization mode1 may result in elimination of sales of the engine families that have high control costs, which is unrealistic. The potential unrealistic results are due to the fact that the optimization model was constructed to minimize emission control costs rather than to maximize profits of vehicle sales. To prevent such unrealistic simulation results from happening, vehicle sales change for an individual engine family under the permit system are arbitrarily limited to 20% around the sales level of the engine family under the current uniform standard system. Tightening emission standards Vehicle emission standards will be tightened in the future. To test the effect of tightened emission standards on cost savings of the marketable permit system, two cases regarding emission standards were established. One case assumes emission standards of 1990 modelyear cars. The other case assumes a 70% reduction in the HC emission standard, 50% reduction in the CO standard, and 50% reduction in the NO, standard. The 70% reduction in HC standard represents the reduction rate of CARB’s TLEV NMOG standard; the 50% reduction in CO emissions represents the reduction rate of CARB’s ULEV CO standard; and the 50% reduction in NO, emissions represents the reduction rate of CARB’s LEV NO, standard; all are relative to CV emission standards. Estimated cost saliings Table 6 presents the cost savings due to use of the marketable permit system as estimated in this study. The cost savings are presented at the level of manufacturing cost. As shown, the marketable permit system can save $93 million to $21 1 million of emission control costs for lightduty vehicles sold in California each year, or 13-30% of the total costs spent on 1990 light-duty vehicles. The cost savings vary according to whether emission standards are tightened and whether changes in vehicle sales mix are allowed. The monetary cost savings translate into $50 to $120 per vehicle. As the table indicates, tightening emission standards and allowing changes in vehicle sales mix increase cost savings substantially. By accounting for profit and cost markups of manufacturers and dealers, cost savings were converted from the manufacturing cost level to the consumer cost level. At the consumer cost level, the marketable permit system saves $156 million to $354 million of emission control costs in California each year, translating into $90 to $200 per vehicle. The total cost savings due to the marketable permit system for the USA will be roughly 10 times as much as that for California, because national vehicle sales are about IO times higher than California vehicle sales.

Cost saplings of using a marketable permit system: M Q Wang Table 5 Coefficients

Table 6 Estimated annual cost savings for the light-duty ketable permit system in California”

of 29 individual cost functions”

Vehicle wouo

BO

Bl

B2

83

Audi small Audi medium Audi large BMW small BMW medium BMW large Chrysler small Chrysler medium Chrysler large Ford small ford medium Ford large GM small GM medium GM large Honda small Honda medium Mazda small Mazda medium Mercedes medium Mercedes lage Mitsubishi small Mitsubishi medium l‘oyota small l‘oyota medium Toyota lrge \‘olvo small \‘olvo medium \‘W small

4.4913 4.6996 4.9496 4.5647 4.6422 5.1133 3.1144 3.9886 4.260 3.9501 4.084 1 4.3593 3.8336 4.1094 4.2142 4.3113 4.4042 4.5700 4.7050 5.0237 5.1607 4.3745 4.5019 4.4092 4.5417 4.6850 4.5891 4.5505 4.4987

4.0649 a.0679 a.0715 a.0659 a.067 I -0.0739 -0.0545 -0.0576 4.0615 4l.057 I -0.0590 -0.0630 -0.554 -0.0594 -0.0609 -0.0632 -0.0636 -0.066 I -0.0680 -0.0726 a.0746 -0.0632 a.0650 -0.0637 -0.0656 -0.0677 -0.0663 -0.0657 -0.0650

4. I275 a.1334 -0.1405 -0.1295 -0.1317 -0.1451 -0.1071 -0.1132 -0.1208 -0.1121 -0.1159 -0.1237 -0.1088 -0.1166 a.1196 -0.1241 -0.1250 a.1297 a.1335 4). 1426 -0.1465 a.1242 -0.1277 4.1251 a.1289 a.1330 a.1303 JI.1291 -0.1277

a.4036 -0.4224 -0.4448 -0.4103 -0.4 I72 -0.4595 -0.3392 -0.3586 -0.3825 -0.3550 -0.3670 -0.39 18 -0.3445 -0.3693 -0.3788 -0.3929 -0.3958 a.4107 -0.4229 -0.45 15 -0.4638 -0.3932 -0.4046 -0.3963 -0.4082 -0.42 I 1 -0.4125 -0.4090 -0.4043

Vast function for each vehicle group takes this form: In(K) = BO + B I X In(HCC0) + B2 X ln(NOx! + 83 X In(mpR). TC is vehicle emission control cost in dollars per velncle, HCCO is HC emissions (gpm) multiplied by CO (gpm), NOx is in gpm and mpg is vehicle fuel economy, and BO, B I, B2 and B3 are coefficients presented in this table.

The cost savings of the marketable permit system estimated here are larger than those for light-duty vehicles estimated in the 1984 TCS study, but are comparable to those for heavy-duty vehicles estimated in the 1986 SC1 study. However, caution should be taken in comparing the results of this study with those of the previous two studies. Assumptions regarding the size of the trading basis and the targeted emission standards differ for this study from the TCS study, and while this study estimates cost savings for light-duty vehicles, the SC1 study estimated cost savings for heavy-duty vehicles. Nevertheless, the three studies indeed show that use of marketable permit systems for regulating motor vehicle emissions leads to large control cost savings.

Qualifications

and issues

Potential cost savings vs. estimated cost savings The estimated cost savings of the marketable permit system may underrepresent the potential cost savings of the system by a non-trivial amount, mainly because cost savings estimated in this study are based on the trading basis of 29 arbitrarily-defined vehicle groups, rather than on the actual trading basis of approximately 300 engine families. It was impossible to estimate actual potential cost savings of the permit system with the Transporr Policy 1994 Volume I Number 4

Trading case Emission trading with current standardsh Emission trading with tightened standardsC Emission trading and vehicle sales changes with current standardsd Emission trading and vehicle sales changes with tightened standards’

Savings ($ million)

vehicle mar-

As of total cost (%)

93

13.3

149

21.4

133

19.0

211

30.4

“Cost savings are at the manufacturing-cost level and in 1990 dollars. hUnder this case, emissions of individual vehicle groups are allowed to change, while vehilce sales of an individual group are fixed at the current level for the group. Total emissions after emission trading are set to be equal to those under the current system. CUnder this case, emissions of individual vehicle groups are allowed to change, while vehicle sales of an individual group are fixed at the current level for the group. Total emissions after emission trading are set to be equal to those when emission standard of HC is reduced by 70%, CO by 50% and NOx by 50%. dUnder this case, emissions of individual vehicle groups are allowed to change, and vehicle sales of an individual group are allowed to change as much as 20% around the current sales level for the group (while total vehicle sales of all groups are fixed at the current level). Total emissions after emission trading are set to be equal to those under the current system. Ylnder this case, emissions of individual vehicle groups are allowed to change, and vehilce sales of an individual group are allowed to change as much as 20% around the current sales level for the group (while total vehicle sales of all groups are fixed at the current level). Total emissions after emissiont rading are set to be equal to those when emission standard of HC is reduced by 70%, CO by 50% and NOx by 50%.

trading basis of 300 engine families, because of the extreme difficulty of estimating one cost function for each individual engine family. To demonstrate the cost savings impact of changes in trading basis, cost savings of the permit system were estimated for the trading basis of 15 vehicle groups (three vehicle classes, and three domestic manufacturers, European manufacturers together and Japanese manufacturers together). The estimation showed that the decrease in trading basis from 29 groups to 15 groups reduced cost savings of the permit system by $46 million to $74 million at the manufacturing-cost level. This finding implies that the increase in trading basis from the simulated 29 vehicle groups to the actual 300 engine families will increase cost savings substantially. Cost savings estimated in this study include savings from emission averaging and trading, but not from emission banking. However, including emission banking in the cost savings estimation would further increase the estimated cost savings of the permit system. Recently, Rubin and Kling (1993) estimated that emission banking could increase cost savings of the permit system by one-third to one-half. Emission control cost functions One of the major tasks, and probably the most difficult task, in estimating cost savings of the marketable permit system is to construct emission control cost functions. Because of lack of emission and cost data, it is not prac229

Cost savings of using a marketable permit system: M Q Wang tical to construct a cost function for each individual vehicle engine family. Without engine family-specific cost functions, the cost savings of a marketable permit system based on engine families cannot be estimated precisely and accurately. Thus, the permit system’s cost savings estimated in this study only provide information on the degree of cost savings, not on the exact amount. During construction of cost functions, various function forms have been tested, and log function form was selected as the final form. There was no definite basis for selecting one functional form over another. The only criterion used in this study was that the selected functional form would result in negative coefficients. Because of the limited number of data points used in constructing cost functions, a statistical analysis could not be conducted to test the significance of the selected function form. The most troubling matter in constructing cost functions in this study was the use of the so-called ‘average approach’ to develop disaggregate cost functions for the 29 combinations of vehicle groups and manufacturers. The approach implicitly assumed that the difference in emission control costs among vehicle groups is caused evenly by each of the independent variables (including the constant term). In reality, these independent variables contribute differently to the cost difference. Because the average approach was used, the constructed cost functions are similar to each other (the only difference among the cost functions is the ratio of corresponding coefficients). This similarity may cause cost savings of the marketable permit system estimated in this study to be smaller than cost savings estimated using the cost functions that have wide variations in functional forms and coefficients. In summary, because of uncertainty involved in constructing cost functions and because of the importance of cost functions in determining cost savings for the marketable permit system, the results presented in this paper should be treated qualitatively, not quantitatively. Issues in implementing and using a marketable permit system Average emission standards under the marketable permit system. When designing a marketable permit system, it is critical to determine the levels for average emission standards so that the amount of emissions reduced through the permit system is at least as much as that through the uniform standard system. As discussed earlier, manufacturers tend to use the emission difference between standards and certification rates to lower their fleet average emissions. Actual certification emissions under the marketable permit system would be higher than those under the uniform standard system, if the same standards were adopted under both systems. Two methods can be applied to maintain lower emissions for the permit system. One method is to lower average emission standards for the marketable permit system, relative to the uniform standards under the current system. This method is straightforward and easy to implement. The other method is to apply a discount rate to ERCs earned by manufacturers, which may create 230

adverse side-effects; for example, discounting ERCs has an asymmetrical effect on manufacturers in the sense that the manufacturers who earn ERCs are penalized. It may be politically easier to accept lowered average standards than to accept a discount rate on ERCs, because emission impacts from lowered standards are explicit, whereas those from discounting ERCs are implicit. It is proposed in this study that lowered average emission standards be adopted in the marketable permit system. However, a discount rate applied to ERCs is still proposed for the purpose of discouraging manufacturers from accumulating too many ERCs. Nevertheless, maintaining the same level of emissions under the two systems seems to be a political argument rather than an air quality argument, because the amount of emission reductions that is achieved under the uniform standard system is by no means the optimal level. Emission reduction goals for the marketable permit system should be determined on the basis of meeting air quality standards, not on the status quo reductions under the current system. In any case, designing proper average standards is important only during the transitional period from the current system to the permit system. After full implementation of the permit system, future average emission standards will be compared with the implemented ‘average standards’, not with the formerly abandoned ‘uniform standards’. Creation and operation oj’an emission trading market. To conduct emission trading and emission banking, an ERC trading market is needed. Such a market could be an EPA-operated accounting system in which ERCs owned by individual manufacturers are counted and ERC transactions among manufacturers are tracked. Alternatively, a private company could be set up to establish an emission trading market, to broker ERC transactions in the market, and to report transactions to the EPA, and the EPA could reserve authority to audit the market operation. Since there are only about 35 manufacturers supplying light-duty vehicles for the US market, operating such an accounting system or trading market will probably be simple. However, the accounting system or the trading market must not be oversimplified. The dollar value of ERCs from low-emission vehicles could motivate parties involved in after-market vehicle conversion to participate in emission trading. Currently, some companies convert gasoline vehicles to alternative-fuel vehicles. These companies could decide to claim ERCs from the converted vehicles. To avoid double-counting vehicles, the vehicles for which a conversion company claims ERCs will have to be deducted from vehicle sales of the original manufacturer. Also, since the marketable permit system is designed to target manufacturers, it would be improper to give ERCs to individual consumers who have their vehicles converted in mechanical shops. Environmental policy implications The current approach to motor vehicle emission regulation is the vestige of a first-generation regulatory

Cost savings of using a marketable permit system: M Q Wang

framework that is not suited to the changing circumstances of the future: rigid requirements limit how manufacturers achieve emission reductions, and the approach does not respond to changing economic and technological conditions. There are good reasons for the simplistic inflexibility of past and current approaches mostly associated with ease of implementation and strict but these approaches are control by regulators becoming increasingly inefficient and inappropriate. To solve those problems, two different types of incentive-based approaches can be pursued. One is to make existing market arrangements operate better by manipulating prices of vehicles to reflect their social costs, which include air pollution. The second is to create market-like arrangements that mimic real markets in the way they generate incentives. The emphasis of both approaches is on decentralized decision making, driven by self-interest but guided by the regulating body through its structuring of incentives or disincentives. The first approach typically involves use of taxes, fees and subsidies for consumers. One example is the DRIVE+ concept proposed by Gordon and Levenson I 1989), which uses a rebate-and-fee schedule for new car sales. Buyers receive a rebate if the cars they purchase have lower emissions and better fuel economy than average, or they pay a fee if the cars emit more pollution or use more fuels than average. The size of the fee or rebate is proportional to how far the emissions are above or below average. This fee-rebate concept provides incentives for individuals and organizations to purchase clean-burning and more fuel-efficient vehicles. The proposal tries to influence vehicle manufacturers to produce low-emission, fuel-efficient vehicles in response to consumers’ demands. In order to meet emission reduction and fuel-saving goals, fees or rebates will have to be large, which may cause strong opposition from certain groups. In any event, it is not certain that the fee-rebate system will be successful in encouraging production of clean-burning vehicles such that emission requirements currently applied to vehicle manufacturers can be eliminated. The second incentive-based approach primarily consists of using the marketable permit system proposed in this study. Instead of targeting millions of individual drivers as in the first approach, the marketable permit system targets a small number of vehicle manufacturers, making it easy to enforce and monitor the system. The permit system can be adapted to the current emission regulatory system in which vehicles are tested and certified and in which manufacturers are still responsible for meeting emission requirements. The emission certification process under the permit system is virtually the same as it is under the current system, except that average emissions need to be calculated and ERCs need to be counted

and

tracked.

The

additional

cost

burden

of

operating an emission trading market should be minimal, because the number of vehicle manufacturers participating will be limited. Within the existing environmental legislation, the EPA and the CARB have the authority to adopt a marketable permit system to replace Transport

Policy 1994 Volume 1 Number 4

the uniform standard system; thus a lengthy legislative process of authorizing regulators in order to reform the current motor vehicle emission regulations can be avoided. The adoption of EPA’s heavy-duty engine emission trading program and CARB’s low-emission vehicle program shows that vehicle manufacturers overwhelmingly favor and environmental groups generally support the marketable permit system, which eliminates some institutional barriers to adopting the system. As in the USA, motor vehicle emissions are currently regulated by the uniform standard approach in most West European countries and in Japan, where stringent vehicle emission standards are now proposed. Meeting these stringent standards at minimal costs becomes increasingly important in those countries, as well as in the USA. The marketable permit system proposed in this study can be applied to those countries as well as to the USA.

Conclusions This paper reviews previously proposed and adopted marketable permit systems for motor vehicle emission control and proposes a complete marketable permit system to replace the current uniform standard system. The proposed marketable permit system allows emissions averaging across a manufacturer’s vehicle engine families, emission trading among manufacturers, and emission banking over model years. The ultimate benefit of adopting such a permit system is its emission control cost savings, relative to the uniform standard system. Using an optimization model for simulating manufacturers’ behavior in meeting emission requirements, this study has estimated emission control cost savings for the proposed marketable permit system: $150 million to $350 million per year in California at the consumer cost level, or 13-30% of the current total control cost for vehicle emission control. Total cost savings attributable to the marketable permit system in the USA will probably be ten times as much as that estimated for California. To realize the cost savings, the marketable permit system should be considered as a replacment for the current uniform standard system.

References E Oates (1988) The Theory of University Press, New York California Air Resources Board (1990) ‘For consideration by the Air Resources Board at the public hearing on the proposed regulations for low-emission vehicles and clean fuels’ Los Angeles, CA September 27-28 Davis, Stacy and Melissa D Morris (1992) Transpor-ration Energy Data Book: Edition 12 ORNL-6710, Center for Transportation Analysis, Oak Ridge National Laboratory, Oak Ridge, TN Gordon, D and L Levenson (1989) ‘DRIVE+: a proposal for California to use consumer fees and rebates to reduce new motor vehicle emissions and fuel consumption’ Applied Science Division, Lawrence Berkeley Laboratory, Berkeley, CA Rubin, Jonathan and Catherine Kling (1993) ‘An emission saved is an emission earned: an empirical study of emission banking for lightduty vehicle manufacturers’ Journal of Envir-onmental Economics Baumol,

William

Environmental

Management

J and Wallace

Policy Cambridge

25 (3) 257-274

231

Cost savings of using a marketable permit system: h4 Q Wang Sobotka and Company, Inc (SCI) (1986) Savings for- the Applicatron of Trading and Averaging to Heavy-Duty Engine Regulation prepared for the US EPA, Ann Arbor, MI TCS Management Group, Inc (1984) Automobile Emission Averuging, Final Report prepared for Office of Policy Analysis, US EPA, Washington, DC TCS Management Group, Inc, Nashville, TN US Environmental Protection Agency (US EPA) (1983) ‘Control of air pollution from new motor vehicles and new motor vehicle engines; averaging of particulate emissions from 1985 and later motor year diesel-fueled light-duty vehicles and light-duty trucks’ Federal Register 48 (141) 33456-33465 US Environmental Protection Agency (US EPA) (1986) ‘Emission trading policy statement, genera1 principles for creation, banking, and use of emission reduction credits’ Federal Register 51 (233) 4381443860 US Environmental Protection Agency (US EPA) (1989) Cleun Air Act Amendments of 1989, Section-by-Section Analysis Washington, DC US Environmental Protection Agency (US EPA) (1990a) Clean Air Act Amendments of1990: Detailed Summary of Titles Washington, DC US Environmental Protection Agency (US EPA) (1990b) ‘Certification programs for banking and trading of oxides of nitro-

232

gen and particulate emission credits for heavy-duty engines; tinal rule’ Federal Register 5.5 (144) 30584-30629 US Environmental Protection Agency (US EPA) (1990~) ‘Summary and analysis of comments on the NPRM: certification programs for banking and trading of oxides of nitrogen and particulate matter emission credits for heavy-duty engines’ Office of Air and Radiation, Office of Mobile Sources. Washington, DC US Environmental Protection Agency (US EPA) (1991) National Ai, Pollutant Emission Estimates: 1940-1990 Office of Air Quality, Planning and Standards Technical Support Division, Research Triangle, NC Wang, Quanlu (1992) The Use of a Marketable Per-mit System Ji)r Light-Duty Vehicle Emission Conrrol Ph.D dissertation, University of California, Davis, CA Wang, Quanlu, Daniel Sperling and Catherine Kling (1993) ‘Emission control costs for light-duty vehicles’ Journal of Air and Waste Management Association 43 (11) 1461-1471 White, Lawrence J (1982) ‘U.S. mobile source emissions regulations: the problems of implementation’ Policy Study Journal 11 (I) 77-85 White House (1989) Thr Clean Air Au Amendments of 1989. Highlights Washington, DC

Transport Policy 1994 Volume I Number 4