![[logo] US EPA](https://webarchive.library.unt.edu/eot2008/20081107103417im_/http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Final Report: Tradable Fuel Economy Credits for Cars and Light Trucks
EPA Grant Number: R830836Title: Tradable Fuel Economy Credits for Cars and Light Trucks
Investigators: Rubin, Jonathan , Greene, David , Leiby, Paul
Institution: University of Maine
EPA Project Officer: Wheeler, William
Project Period: January 1, 2003 through December 30, 2006
Project Amount: $177,247
RFA: Market Mechanisms and Incentives for Environmental Management (2002)
Research Category: Economics and Decision Sciences
Description:
Objective:The fuel economy of new US light-duty vehicles (cars, minivans, SUVs, light trucks) has fallen significantly since its peak in 1989 following the implementation of corporate average fuel efficiency (CAFÉ) standards in 1978. CAFÉ regulations specify minimum fleet average standards for fuel efficiency that vehicle manufacturers must meet.
The effectiveness of CAFÉ standards in raising the light-duty vehicle fleet's fuel efficiency are discussed in a large body of literature. It is debated whether the improvements in average fuel efficiency are attained at a reasonable cost, whether the CAFÉ regulations induce undesirable changes in vehicles that could lower their safety, or have the unintended effect of encouraging the shift in market share from cars to light-duty trucks. In 2002, The National Research Council released a comprehensive review of the effectiveness and impact of CAFE standards. The NRC concluded that the CAFE program has clearly increased fuel economy since its inception, although certain aspects of the CAFE program have not functioned as intended. These include indirect consumer and safety costs, and the breakdown in the distinctions between minivans, SUVs and cars in the calculation of fuel economy standards. Moreover, the NRC concluded that technologies exist that, if applied to light-duty vehicles, would significantly reduce fuel consumption within 15 years.
The NRC concluded that raising the CAFE standard would reduce future fuel consumption, but that other policies could accomplish this same end at lower cost and greater flexibility. The NRC concluded (Finding 11): "Changing the current CAFE system to one featuring tradable fuel economy credits and a cap on the price of these credits appears to be particularly attractive…."
To test this conclusion, we estimate the potential cost savings from moving from the current CAFE system to one that allows vehicle manufacturers various levels of increased flexibility, including trading of CAFE credits across manufacturers. In addition, we examine the cost savings from allowing firms to buy and sell credits across time periods. Given that five firms account for a large portion of total vehicle sales, we examine the importance of imperfect competition in the credit market by modeling dominant firms as Cournot-Nash players with a competitive fringe.
Summary/Accomplishments (Outputs/Outcomes):We have developed the theoretical models for credit trading for fuel intensity credits for several different models.
- Perfect competition, single period (static) model
- Perfect competition, intertemporal model with banking and borrowing
- Cournot-Nash single period model (static)
- Cournot-Nash multi-period model without banking and borrowing.
We use data for MY 2003 vehicles sold in the United States, obtained from the National Highway Traffic Safety Administration (NHTSA) Manufacturer's Fuel Economy Reports. These give us vehicle manufacturers sales and fuel economy by vehicle class (8 cars and 7 trucks) and country of origin (foreign or domestic). The National Research Council's presents low and high retail equivalent price estimates for a low and high range of incremental fuel efficiency gains by individual technologies for 4 car and 6 truck classes (NRC, Tables 3.1 -3.4) . We use the National Research Council's high and low retail costs with low and high efficiency gains to generate low, average and high retail costs of fuel efficiency improvements that encompass the full range of cost and performance uncertainty.
A review of the technology cost literature indicates that two-parameter quadratic curves fit data from all studies reasonably well. This form is P(X)=bX+cX2, where P(x) is the retail price (cost) increase to the percentage decreases in gallons per 100 miles over a base level and b and c are parameters to be estimated. The intercept terms are omitted because, by construction, the curves pass through the origin (0% improvement has $0 cost). The parameter estimates are intended to be curve fits and not statistical estimations. The important point is that the fitted curves accurately reflect the rate of increase in retail price for a percent decrease in fuel intensity for the full range of fuel economy improvements being considered. The two-parameter quadratic functions fit the data very well, with adjusted R-squared values exceeding 0.98 in all instances. We then generate manufacturer-specific cost curves by weighting both of the estimated coefficients for the 10 size classes and vehicle types by the manufacture specific sales-weighed average of vehicles and fuel intensities. For example, to generate a particular vehicle manufacturers cost curve for cars, we combine the sales-weight average of the parameters for the 4 size classes produced by that manufacturer and weighted by fuel intensity of that manufacturer.
Trading Scenarios In order to explore the potential cost savings from allowing more regulatory flexibility by credit trading we examine 4 possible credit trading scenarios. These scenarios reflect increasing amounts of flexibility starting from the base case that does not allow any credit trading by manufacturers consistent with current CAFE regulations (see Table 1).
Table 1: Credit Trading Scenarios
| Scenario Name | Credit Trading Among Firms | Credit Trading Among Vehicle Classes | Scenario Description |
| Base | No | No | Firms must independently meet separate standards for cars and trucks |
| Class Averaging | No | Yes | Firms trade credits across vehicle classes (but not among firms) |
| Class Trading | Yes | No | Firms can trade credits in separate car and truck markets |
| Firm & Class Trading | Yes | Yes | Firms can trade credits in a single a single market |
Of particular policy concern is whether market power erodes potential savings from increased flexibility. We pursue this issue further by examining hypothetical changes to the sales shares of Japanese and US manufacturers. To test the sensitivity of our parameter assumptions we also examine each of these cases assuming low and high valuation of fuel economy by consumers, base and high projections of gasoline prices, and low, medium and high costs and effectiveness of the fuel economy technology.
Importantly, we also explore the implications of a fundamental re-consideration of our (net) cost curves. As is seen below, for low CAFE targets or low NRC technology-cost cases, the cost of additional fuel economy technology is less than the anticipated fuel savings, i.e., there is an estimated net consumer gain to increasing fuel economy. We consider three possible explanations for this apparent suboptimality of initial fuel economy: under-estimation of the costs of fuel economy technology by the NRC; systematic market failure in the market for fuel economy technology; or because the observed level of fuel economy technology in the market place reflects tradeoff with vehicle performance or other attributes missing from our model.
To evaluate the impact of this latter possibility, we re-parameterize the net cost curves by imposing the assumption that any decrease in fuel intensity must have a positive cost. We do this by setting the first derivative of the net benefit function equal to 0 at the point Xmv = 0 (no increase in fuel economy), e.g., 
. This re-calibrates (and lowers) the discounted lifetime value of fuel economy based on the observed current fuel intensity of each manufacturer.
Given the many result permutations, we focus on deviations from our base case: no credit trading among vehicle type or manufacturers, low valuation of fuel savings by consumers (3 years of fuel savings, not discounted), average costs of fuel economy technology and base projections of fuel prices. This scenario is closest to representing the current CAFE regulations coupled with manufacturers’perceptions of consumer-willingness to purchase fuel economy technology.1
Results
Table 2 report the average net costs - incremental retail fuel technology costs less fuel cost savings per vehicle - on average to manufacturers under a number of different technology cost and fuel price assumptions for decreases in fuel intensity of 30% by 2015. Columns 1-4 show the net costs under our base case assumption regarding consumer valuation of fuel economy. Positive numbers represent average per vehicle costs while negative numbers indicate discounted lifetime technology costs are less than discounted fuel savings.
What stands out is that in all cases the highest level of regulatory flexibility, firm and vehicle class trading, yields the greatest savings. For average technology costs and base fuel prices (column 1), the average net costs of increasing the CAFE standards 30% - 40% range from $44 to -$60 per vehicle. Given the construction of our cost curves and the market shares of the vehicle manufacturers, we find that “class averaging” (allowing vehicle manufacturers to trade fuel economy credits across their vehicle classes) provides the next greatest level of savings. This is followed by class trading among manufactures where manufacturers can sell or buy credits with other manufacturers in separate car and truck markets.
As shown in data columns 3 and 4 (high technology cost, high gasoline prices) the magnitude and percentage savings depends substantially on the particular scenario under examination. However, the same pattern of savings across the trading systems remains unchanged. What is clear is that the cost savings from fuel economy credit trading are potentially quite substantial for the industry as a whole. Even savings of $25 per vehicle, on average, given annual sales of 16 million units represents $400 million in annual savings.
That a significant portion of the total potential savings are gained from class averaging within firms is of particular importance for the possible impact of non-competitive behavior. This portion of savings will not be affected by the possible oligopolistic or oligopsonistic withholding of credit trades from the market in order to drive credit prices up or down. Note that adding the percentage savings of class averaging and class trading yields a greater level of savings than complete flexibility allowing both types of trading. The regulatory flexibility of class averaging and class trading are, to some extent, substitutes. However, the magnitude of the substitution effect does not appear great.
Table 2: Average Net Cost Per Vehicle From Increasing CAFE, and Percent Savings From Trading - Perfect Competition, Low (+30%) Fuel Economy Target
| Scenario Name | Base Case: Base Value, Fuel Economy, Average Cost of Fuel Economy Technology | Variant: Low Cost of Fuel Economy Technology | Variant: High Cost of Fuel Economy Technology | Variant: High Future Gasoline Prices | Variant: High Value Fuel Economy | Variant: Re-Benchmark Value of Fuel Economy |
| No-trading Cost | -$21 |
-378 |
$495 |
-$94 |
-$513 |
$170 |
| Between-Class Averaging | -$48 |
-$378
|
$419
|
-$118
|
-$519 |
$154
|
| Within-Class Trading | -$42
|
-$379
|
$431
|
-$112
|
-$520
|
$141
|
| Firm & Class Trading | -$60
|
-$379
|
$381
|
-$128
|
-$522
|
$135
|
| $/vehicle for CAFE increase averaged over all manufacturers, and % change in cost relative to the no-trade case. | ||||||
Columns 5 & 6 of Table 2 shows the impact of two alternatives: assuming full lifetime discounting of fuel economy and our re-parameterization that imposes that assumption that any incremental increases in fuel economy must have incremental net costs (described above). As seen in column 5, if consumers really do value the full lifetime discounted value of fuel cost savings, then we estimate that there are cost-effective fuel economy technologies offering an average net potential savings of about $500 available over a vehicle’s 15 year lifetime. There are modest savings available from increased regulatory flexibility.
The other perspective - that the market for vehicle fuel efficiency is currently in balance with consumers willing to pay for technology that pays back in 3 years - is shown in column 6. Here, by assumption, any increase in fuel economy has a net marginal cost. We estimate these costs to be on average $170 to $295 per vehicle. Savings of 9% to 21%, (roughly $20 to $40 per vehicle) are available with a well-functioning system of tradable CAFÉ credits.
Given the large proportion of vehicles produced by the 5 largest manufacturers, the effect of market power on the price and availability of fuel economy credits needs to be examined explicitly. In Table 3 we show the importance of oligopoly power in the credit market in terms of reductions in savings from credit trading. The first and fifth columns show the net compliance costs under perfect competition with the original and re-benchmarked cost curves. The second and sixth columns show the net costs when the big five each act as independent Cournot oligopolists. As is seen the average costs savings from trading decline very modestly. This is because imperfect competition does not affect the gains from class averaging (i.e., each vehicle manufacturers internally trading car and truck CAFÉ credits) which, given our data, generates the greatest percentage savings from the no trading baseline. As before, the cost savings shown are on average for all manufacturers jointly. Now especially, since we examine the impact of market power, the gains to individual manufacturers from credit trading will vary.
One possible impact of raising the CAFÉ standard suggested in the literature is induced sales mix changes that might favor foreign vehicle manufacturers. Given our model that focuses on the credit market, we are not able to endogenously estimate the magnitude of a policy-induced change in sales mix. Instead we explore the implications on the cost savings from credit trading with perfect competition and with oligopoly power if the Japanese vehicle manufacturers were to gain market share at the expense of domestic manufacturers. We do this by reallocating vehicle sales by manufacturer and type (car and truck) proportionally to their 2003 sales, but assuming that Japanese (Honda and Toyota) and US vehicle manufactures (General Motors, Ford, DaimlerChrysler) have equal market shares. Thus, in 2003, the big five had a 75% market share of cars with a domestic-Japanese split of 46% to 26%. We proportionally reallocate vehicles to give domestic and Japanese manufacturers each a 37% market share. For trucks we reallocate the 2003 joint share of 90% (75% domestic, 16% Japanese) to 45%-45% equal shares. We leave sales of fringe firms unaffected.
The results of this alternative sales mix are shown in columns three, four, seven and eight of Table 4. First off, comparing column three to column one and column seven to column five, we see that the absolute cost of attaining the CAFÉ standard is lower. This follows from our cost curves that show that it is relatively less costly for Japanese, compared to the domestic vehicle manufacturers, to decrease the fuel intensity of their vehicles. Thus, if Japanese vehicles make up a larger percentage of the vehicle fleet, the average cost of attaining higher efficiency standards will decrease. Comparing further, we see that the same basic pattern of cost savings occurs; the absolute and percentage savings are greatest for class averaging (within manufacturer credit trading). The impacts of imperfect competition on the savings from trading is, however, larger for both the original and re-benchmarked cost curves. This is seen by comparing column three with column 4 and comparing column seven with column 8.
Conclusions:Depending upon the case, the net cost of tightened fuel economy standards to the industry as whole may be quite large or small. This uncertainty reflects the large range of possible costs of fuel economy technology, uncertain future gasoline prices, and ambiguity regarding how consumers value future fuel economy savings. The results in this paper show how the net costs also depend on the level of future fuel economy standards, the flexibility of the standards, and the degree to which a tradable credit market is affected by non-competitive behavior. Resolving uncertainty over the engineering costs of increasing fuel economy at the firm and industry level and improving our understanding of consumers’ valuation of fuel economy is clearly needed.
For cost scenarios that impose significant costs on individual vehicle manufacturers, we find the savings from averaging and trading credits to be quite substantial. The greatest proportional savings exceeded 100% in some cases, reflecting the fact that, to the industry as a whole, average costs per vehicle turned to net savings when credit trading across manufacturers was allowed. The scenarios that show large percentage gains from trade generally reflected the middle range of net costs - where there were substantial imposed costs on some, but not all, manufacturers from increased fuel economy standards. In cases where the net costs were lower (low cost of fuel economy technology or high consumer valuation of fuel economy improvements), the gains from trading were small or non-existent reflecting the largely non-binding nature of increased fuel economy targets. At the other extreme, when fuel economy improvements were expensive, the percentage gains from being able to average and trade credits were considerably smaller (while the absolute gains in dollars-per-vehicle were greater). For many of the scenarios, the ability of each manufacturer to average credits between its car and truck classes provides greater savings than the ability to only trade credits between manufacturers in separate vehicle class markets. As expected, the greatest savings comes from the greatest flexibility, when manufacturers are able to average and trade fuel economy credits.
Given the high concentration of vehicle sales by the five largest firms, we explicitly examined the potential impact of market power in the credit markets. We modeled the largest firms as Cournot oligopolists facing a competitive fringe. The theoretical effect of imperfect competition on fuel economy credit price (compared to a perfect competition benchmark) is ambiguous since firms with market power are both sellers and buyers. Our numerical simulations show that there is a small increase in the price of credits when all five of the largest firms act as oligopolists, and seek a Cournot-Nash equilibrium. However, both sellers and buyers of credits have an incentive to reduce their net credit transactions in order to influence to the credit price. We find that the volume of credit sales can be up to 35% less compared to the perfectly competitive benchmark.
As expected, the existence of market power did lower the potential cost savings to the industry as a whole. However, the magnitude of the potential losses in efficiency from the market power were not large, usually substantially less than 25% of the potential savings from trade in all cases when considering the industry as a whole. Since some firms are net sellers and some net buyers, individual
firms experienced greater gains or losses from trading when taking market power into consideration than did the industry as a whole. Importantly, every firm was still better off from credit trading with imperfect competition compared with our no trading baseline. Imperfect competition in credits does not appear to eliminate all the gains from trading at the firm level and has relatively modest impacts on the industry as a whole.
References:
1. We differ from the current regulatory situation by assuming that each manufacturer complies with the CAFE regulations rather than falling short and paying the fines. Historically, only BMW, Porsche and the manufacturers of a few other specialty high-performance cars actually have paid fines rather than meet the standard.
Journal Articles:No journal articles submitted with this report: View all 7 publications for this project
Supplemental Keywords:Cost-Benefit, Modeling, Socio-Economic, Conservation, Energy Policy,
,
Economic, Social, & Behavioral Science Research Program, Scientific Discipline, RFA, Economics and Business, Market mechanisms, tradeable fuel economy credits, market-based mechanisms, trading systems, emissions trading, allowance market performance, market incentives, automotive fuel economy credit trading, environmental economics, pollution allowance trading, policy incentives, consumer behavior, energy efficiency
Progress and Final Reports:
2003 Progress Report
Original Abstract