Skip to Content

We invite you to try out our new beta eCFR site at We’ve made big changes to make the eCFR easier to use. Be sure to leave feedback using the 'Feedback' button on the bottom right of each page!


The Safer Affordable Fuel-Efficient (SAFE) Vehicles Rule for Model Years 2021-2026 Passenger Cars and Light Trucks

Document Details

Information about this document as published in the Federal Register.

Document Statistics
Document page views are updated periodically throughout the day and are cumulative counts for this document. Counts are subject to sampling, reprocessing and revision (up or down) throughout the day.
Enhanced Content

Relevant information about this document from provides additional context. This information is not part of the official Federal Register document.

Published Document

This document has been published in the Federal Register. Use the PDF linked in the document sidebar for the official electronic format.

Start Preamble Start Printed Page 24174


Environmental Protection Agency and National Highway Traffic Safety Administration.


Final rule.


EPA and NHTSA, on behalf of the Department of Transportation, are issuing final rules to amend and establish carbon dioxide and fuel economy standards. Specifically, EPA is amending carbon dioxide standards for model years 2021 and later, and NHTSA is amending fuel economy standards for model year 2021 and setting new fuel economy standards for model years 2022-2026. The standards set by this action apply to passenger cars and light trucks, and will continue our nation's progress toward energy independence and carbon dioxide reduction, while recognizing the realities of the marketplace and consumers' interest in purchasing vehicles that meet all of their diverse needs. These final rules represent the second part of the Administration's action related to the August 24, 2018 proposed Safer Affordable Fuel-Efficient (SAFE) Vehicles Rule. These final rules follow the agencies' actions, taken September 19, 2019, to ensure One National Program for automobile fuel economy and carbon dioxide emissions standards, by finalizing regulatory text related to preemption under the Energy Policy and Conservation Act and withdrawing a waiver previously provided to California under the Clean Air Act.


This final rule is effective on June 29, 2020.

Judicial Review: NHTSA and EPA undertake this joint action under their respective authorities pursuant to the Energy Policy and Conservation Act and the Clean Air Act. Pursuant to CAA section 307(b), 42 U.S.C. 7607(b), any petitions for judicial review of this action must be filed in the United States Court of Appeals for the D.C. Circuit. Given the inherent relationship between the agencies' action, any challenges to NHTSA's regulation under 49 U.S.C. 32909 should also be filed in the United States Court of Appeals for the D.C. Circuit.


EPA and NHTSA have established dockets for this action under Docket ID Nos. EPA-HQ-OAR-2018-0283 and NHTSA-2018-0067, respectively. All documents in the docket are listed in the index. Although listed in the index, some information is not publicly available, e.g., confidential business information (CBI) or other information whose disclosure is restricted by statute. Certain other material, such as copyrighted material, will be publicly available in hard copy in EPA's docket, and electronically in NHTSA's online docket. Publicly available docket materials can be found either electronically in by searching for the dockets using the Docket ID numbers above, or in hard copy at the following locations:

EPA: EPA Docket Center, EPA/DC, EPA West, Room 3334, 1301 Constitution Ave. NW, Washington, DC. The Public Reading Room is open from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding legal holidays. The telephone number for the Public Reading Room is (202) 566-1744.

NHTSA: Docket Management Facility, M-30, U.S. Department of Transportation (DOT), West Building, Ground Floor, Rm. W12-140, 1200 New Jersey Ave. SE, Washington, DC 20590. The DOT Docket Management Facility is open between 9 a.m. and 5 p.m. Eastern Time, Monday through Friday, except Federal holidays.

Start Further Info


EPA: Christopher Lieske, Office of Transportation and Air Quality, Assessment and Standards Division, Environmental Protection Agency, 2000 Traverwood Drive, Ann Arbor, MI 48105; telephone number: (734) 214-4584; fax number: (734) 214-4816; email address:, or contact the Assessment and Standards Division, email address: NHTSA: James Tamm, Office of Rulemaking, Fuel Economy Division, National Highway Traffic Safety Administration, 1200 New Jersey Avenue SE, Washington, DC 20590; telephone number: (202) 493-0515.

End Further Info End Preamble Start Supplemental Information


Does this action apply to me?

This action affects companies that manufacture or sell new light-duty vehicles, light-duty trucks, and medium-duty passenger vehicles, as defined under EPA's CAA regulations,[1] and passenger automobiles (passenger cars) and non-passenger automobiles (light trucks) as defined under NHTSA's CAFE regulations.[2] Regulated categories and entities include:

Start Printed Page 24175

This list is not intended to be exhaustive, but rather provides a guide regarding entities likely to be regulated by this action. To determine whether particular activities may be regulated by this action, you should carefully examine the regulations. You may direct questions regarding the applicability of this action to the person listed in FOR FURTHER INFORMATION CONTACT.

I. Executive Summary

II. Overview of Final Rule

III. Purpose of the Rule

IV. Purpose of Analytical Approach Considered as Part of Decision-Making

V. Regulatory Alternatives Considered

VI. Analytical Approach as Applied to Regulatory Alternatives

VII. What does the analysis show, and what does it mean?

VIII. How do the final standards fulfill the agencies' statutory obligations?

IX. Compliance and Enforcement

X. Regulatory Notices and Analyses

I. Executive Summary

NHTSA (on behalf of the Department of Transportation) and EPA are issuing final rules to adopt and modify standards regulating corporate average fuel economy and tailpipe carbon dioxide (CO2) emissions and use/leakage of other air conditioning refrigerants for passenger cars and light trucks for MYs 2021-2026.[3] These final rules follow the proposal issued in August 2018 and respond to each agency's legal obligation to set standards based on the factors Congress directed them to consider, as well as the direction of the United States Supreme Court in Massachusetts v. EPA, which stated that “there is no reason to think the two agencies cannot both administer their obligations and yet avoid inconsistency.” [4] These standards are the product of significant and ongoing work by both agencies to craft regulatory requirements for the same group of vehicles and vehicle manufacturers. This work aims to facilitate, to the extent possible within the statutory directives issued to each agency, the ability of automobile manufacturers to meet all requirements under both programs with a single national fleet under one national program of fuel economy and tailpipe CO2 emission regulation.

The CAFE and CO2 emissions standards established by these final rules will increase in stringency at 1.5 percent per year from MY 2020 levels over MYs 2021-2026. The “1.5 percent” regulatory alternative is new for the final rule and was not expressly analyzed in the NPRM, but it is a logical outgrowth of the NPRM analysis, being well within the range of alternatives then considered and consistent with discussions by both the agencies and commenters that there are benefits to having standards that increase at the same rate for all fleets. These standards apply to light-duty vehicles, which NHTSA divides for purposes of regulation into passenger cars and light trucks, and EPA divides into passenger cars, light-duty trucks, and medium-duty passenger vehicles (i.e., sport utility vehicles, cross-over utility vehicles, and light trucks). Both the CAFE and CO2 standards are vehicle-footprint-based, as are the standards currently in effect. These standards will become more stringent for each model year from 2021 to 2026, relative to the MY 2020 standards. Generally, the larger the vehicle footprint, the less numerically stringent the corresponding vehicle CO2 and miles-per-gallon (mpg) targets. As a result of the footprint-based standards, the burden of compliance is distributed across all vehicle footprints and across all manufacturers. Each manufacturer is subject to individualized standards for passenger cars and light trucks, in each model year, based on the vehicles it produces. When standards are carefully crafted, both in terms of the footprint curves and the rate of increase in stringency of those curves, manufacturers are not Start Printed Page 24176compelled to build vehicles of any particular size or type.

Knowing that many readers are accustomed to considering CAFE and CO2 emissions standards in terms of the mpg and grams-per-mile (g/mi) values that the standards are projected to eventually require, the agencies include those projections here. EPA's standards are projected to require, on an average industry fleet-wide basis, 201 grams per mile (g/mi) of CO2 in model year 2030, while NHTSA's standards are projected to require, on an average industry fleet-wide basis, 40.5 miles per gallon (mpg) in model year 2030. The agencies note that real-world CO2 is typically 25 percent higher and real-world fuel economy is typically 20 percent lower than the CO2 and CAFE compliance values discussed here, and also note that a portion of EPA's expected “CO2” improvements will in fact be made through improvements in minimizing air conditioning leakage and through use of alternative refrigerants, which will not contribute to fuel economy but will contribute toward reductions of climate-related emissions.

In these final rules, NHTSA and EPA are reaching similar conclusions on similar grounds: even though each agency has its own distinct statutory authority and factors, the relevant considerations overlap in many ways. Both agencies recognize that they are balancing the relevant considerations in somewhat different ways from how they may have been balanced previously, as in the 2012 final rule and in EPA's Initial Determination, but the current balancing is called for in light of the facts before the agencies. The balancing in these final rules is also somewhat different from how the agencies balanced their respective considerations in the proposal, in part because of updates to analytical inputs and methodologies, previewed in the NPRM and made in response to public comments, that collectively resulted in changes to the analytical outputs. For example, between the notice and final rule, the agencies updated fuel price projections to somewhat greater values, updated the analysis fleet to MY 2017, updated estimates of the efficacy and cost of fuel-saving technologies, revised procedures for calculating impacts on vehicle sales and scrappage, updated models for estimating highway safety impacts, updated estimates of highway congestion costs, and updated estimates of annual mileage accumulation, holding VMT (before applying the rebound effect) constant between regulatory alternative. Moreover, the cost-benefit analysis conducted for these final rules has even been overtaken by events in many ways over recent weeks. Based upon current events, and for additional reasons discussed in Section VI.D.1 the benefits of saving additional fuel through more stringent standards are potentially even smaller than estimated in this rulemaking analysis.

The standards finalized today fit the pattern of gradual, tough, but feasible stringency increases that take into account real world performance, shifts in fuel prices, and changes in consumer behavior toward crossovers and SUVs and away from more efficient sedans. This approach ensures that manufacturers are provided with sufficient lead time to achieve standards, considering the cost of compliance. The costs to both industry and automotive consumers would have been too high under the standards set forth in 2012, and by lowering the auto industry's costs to comply with the program, with a commensurate reduction in per-vehicle costs to consumers, the standards enhance the ability of the fleet to turn over to newer, cleaner and safer vehicles.

More stringent standards also have the potential for overly aggressive penetration rates for advanced technologies relative to the penetration rates seen in the final standards, especially in the face of an unknown degree of consumer acceptance of both the increased costs and of the technologies themselves—particularly given current projections of relatively low fuel prices during that timeframe. As a kind of insurance policy against future fuel price volatility, standards that increase at 1.5 percent per year for cars and trucks will help to keep fleet fuel economy higher than they would be otherwise when fuel prices are low, which is not improbable over the next several years.[5] At the same time, the standards help to address these issues by maintaining incentives to promote broader deployment of advanced technologies, and so provides a means of encouraging their further penetration while leaving manufacturers alternative technology choices. Steady, gradual increases in stringency ensure that the benefits of reduced GHG emissions and fuel consumption are achieved without the potential for disruption to automakers or consumers.

Standards that increase at 1.5 percent per year represent a reasonable balance of additional technology and required per-vehicle costs, consumer demand for fuel economy, fuel savings and emissions avoided given the foreseeable state of the global oil market and the minimal effect on climate between finalizing 1.5 percent standards versus more stringent standards. The final standards will also result in year-over-year improvements in fleetwide fuel economy, resulting in energy conservation that helps address environmental concerns, including criteria pollutant, air toxic pollutant, and carbon emissions.

The agencies project that under these final standards, required technology costs would be reduced by $86 to $126 billion over the lifetimes of vehicles through MY 2029. Equally important, purchase prices costs to U.S. consumers for new vehicles would be $977 to $1,083 lower, on average, than they would have been if the agencies had retained the standards set forth in the 2012 final rule and originally upheld by EPA in January 2017. While these final standards are estimated to result in 1.9 to 2.0 additional billion barrels of fuel consumed and from 867 to 923 additional million metric tons of CO2 as compared to current estimates of what the standards set forth in 2012 would require, the agencies explain at length below why the overall benefits of the final standards outweigh these additional costs.[6]

For the CAFE program, overall (fleetwide) net benefits vary from $16.1 billion at a 7 percent discount rate to −$13.1 billion at a 3 percent discount rate. For the CO2 program, overall (fleetwide) societal net benefits vary from $6.4 billion at a 7 percent discount rate to −$22.0 billion at a 3 percent discount rate. The net benefits straddle zero, and are very small relative to the scale of reduced required technology costs, which range from $86.3 billion to $126.0 billion for the CAFE and CO2 programs across 7 percent and 3 percent discount rates. Likewise, net benefits are very small relative to the scale of reduced retail fuel savings over the full life of all vehicles manufactured during the 2021 through 2029 model years, which range from $108.6 billion to $185.1 billion for the CAFE and CO2 programs across 7 percent and 3 percent discount rates. Similarly, all of the alternatives have small net benefits, ranging from $18.4 billion to −$31.1 Start Printed Page 24177billion for the CAFE and CO2 programs across 7 percent and 3 percent discount rates.[7]

NHTSA and EPA believe their analysis of the final rule represents the best available science, evidence, and methodologies for assessing the impacts of changes in CAFE and CO2 emission standards. In fact, the agencies note that today's analysis represents a marked improvement over prior rulemakings. Previously, the agencies were unable to model the impact of the standards on new vehicle sales or the retirement of older vehicles in the fleet, and, instead, were forced to assume, contrary to economic theory and empirical evidence, that the number of new vehicles sold and older vehicles scrapped remained static across regulatory alternatives. Today's analysis—as commenters to previous rulemakings and EPA's Science Advisory Board have argued is necessary [8] —quantifies the sales and scrappage impacts of the standards, including the associated safety benefits, and represents a significant step forward in agencies' ability to comprehensively analyze the impacts of CAFE and CO2 emission standards.

However, the agencies also believe it is important to be transparent about analytical limitations. For example, EPA's Science Advisory Board stressed that the agencies account for “evolving consumer preferences for performance and other vehicle attributes,” [9] yet due to limitations on the agencies' current ability to model buyers' choices among combinations of various attributes and their costs, the primary analysis does not account for the consumer benefits of other vehicle features that may be sacrificed for costly technologies that improve fuel economy. The agencies' analysis assumes that under these final standards, attributes of new cars and light trucks other than fuel economy would remain identical to those under the baseline standards, so that changes in sales prices and fuel economy would be the only sources of benefits or costs to new car and light truck buyers. In other words, the agencies' primary analysis does not consider that producers will likely respond to buyers' demands by reallocating some their savings in production costs due to lower technology costs to add or improve other attributes that consumers value more highly than the increases in fuel economy the augural standards would have required. The agencies have long debated whether and how best to model the consumer benefits of other vehicle attributes, and note that they have made considerable progress.[10] However, despite these potential analytical shortcomings, the agencies reaffirm that today's analysis represents the most complete and rigorous examination of CAFE and CO2 emission standards to date, and provide decision-makers a powerful analytical tool—especially since the limitations are known, do not bias the central analysis' results, and are afforded due consideration.

In terms of the agencies' respective statutory authorities, EPA is setting national tailpipe CO2 emissions standards for passenger cars and light trucks under section 202(a) of the Clean Air Act (CAA),[11] and taking other actions under its authority to establish metrics and measure passenger car and light truck fleet fuel economy pursuant to the Energy Policy and Conservation Act (EPCA),[12] while NHTSA is setting national corporate average fuel economy (CAFE) standards under EPCA, as amended by the Energy Independence and Security Act (EISA) of 2007.[13] As summarized above and as discussed in much greater detail below, the agencies believe that these represent appropriate levels of CO2 emissions standards and maximum feasible CAFE standards for MYs 2021-2026, pursuant to their respective statutory authorities. Sections III and VIII below contain detailed discussions of both agencies' statutory obligations and authorities.

Section 202(a) of the CAA requires EPA to establish standards for emissions of pollutants from new motor vehicles that cause or contribute to air pollution that may reasonably be anticipated to endanger public health or welfare. Standards under section 202(a) thus take effect only “after providing such period as the Administrator finds necessary to permit the development and application of the requisite technology, giving appropriate consideration to the cost of compliance within such period.” [14] In establishing such standards, EPA must consider issues of technical feasibility, cost, and available lead time, among other things.

EPCA, as amended by EISA, contains a number of provisions governing how NHTSA must set CAFE standards. EPCA requires that the Department of Transportation establish separate passenger car and light truck standards [15] at “the maximum feasible average fuel economy level that the Secretary decides the manufacturers can achieve in that model year,” [16] based on the agency's consideration of four statutory factors: technological feasibility, economic practicability, the effect of other standards of the Government on fuel economy, and the need of the United States to conserve energy.[17] EPCA does not define these terms or specify what weight to give each concern in balancing them—such considerations are left within the discretion of the Secretary of Transportation (delegated to NHTSA) based upon current information. Accordingly, NHTSA interprets these factors and determines the appropriate weighting that leads to the maximum Start Printed Page 24178feasible standards given the circumstances present at the time of promulgating each CAFE standard rulemaking. While EISA, for MYs 2011-2020, additionally required that standards increase “ratably” and be set at levels to ensure that the CAFE of the industry-wide combined fleet of new passenger cars and light trucks reach at least 35 mpg by MY 2020,[18] EISA requires that standards for MYs 2021-2030 simply be set at the maximum feasible level as determined by the Secretary (and by delegation, NHTSA).[19]

In the NPRM, the agencies sought comment on a variety of possible changes to existing compliance flexibilities that have been created over the past several years. The vast majority of the existing compliance flexibilities are not being changed, but a small number of flexibilities related to real-world fuel efficiency improvements are being finalized. In addition, EPA will continue to allow manufacturers to make improvements relating to air conditioning refrigerants and leakage and will credit those improvements toward CO2 compliance, and EPA is making no changes in the amounts of credits available. EPA is also not making any changes to the existing CH4 and N2 O standards. EPA is also extending the “0 g/mi upstream” incentive for electric vehicles beyond its current sunset of MY 2021, through MY 2026. EPA is also establishing a credit multiplier for natural gas vehicles through the 2026 model year. Otherwise, compliance flexibilities in the two programs do not change significantly for the final rule. These changes should help to streamline manufacturer use of those flexibilities in certain respects. While manufacturers and suppliers sought a number of other additional compliance flexibilities, the agencies have concluded that the aforementioned existing flexibilities are reasonable and appropriate, and that additional flexibilities are not justified.

Table I-1 and Table I-2 present the total costs, benefits, and net benefits for the 2021-2026 preferred alternative CAFE and CO2 levels, relative to the MY 2022-2025 existing/augural standards (with the MY 2025 standards repeated for MY 2026) and current MY 2021 standard. The preferred alternative exhibits a stringency rate increase of 1.5 percent per year for both passenger cars and light trucks. The values in Table I-1 and Table I-2 display (in total and annualized forms) costs for all MYs 1978-2029 vehicles, and the benefits and net benefits represent the impacts of the standards over the full lifetimes of the vehicles sold or projected to be sold during model years 1978-2029.

For this analysis, negative signs are used for changes in costs or benefits that decrease from those that would have resulted from the existing/augural standards. Any changes that would increase either costs or benefits are shown as positive changes. Thus, an alternative that decreases both costs and benefits, will show declines (i.e., a negative sign) in both categories. From Table I-1 and Table I-2, the preferred alternative (Alternative 3) is estimated to decrease costs relative to the baseline by $182 to $280 billion over the lifetime of MYs 1978-2029 passenger vehicles (range determined by discount rate across both CAFE and CO2 programs). It will also decrease benefits from $175 to $294 billion over the life of these MY fleets. The net impact will be a decrease from $22 billion to an increase of $16 billion in total net benefits to society over this roughly 52-year timeframe. Annualized, this amounts to roughly −$0.8 to 1.2 billion in net benefits per year.

Start Printed Page 24179

Table I-3 and Table I-4 lists costs, benefits, and net benefits for all seven alternatives that were examined.

Table I-5 and Table I-6 show a summary of various impacts of the preferred alternative for CAFE and CO2 standards. Impacts are presented in monetized and non-monetized values, as well as from the perspective of society and the consumer.

Start Printed Page 24180

Start Printed Page 24181

The agencies note that the NPRM drew more public comments (and, particularly, more pages of substantive comments) than any rulemaking in the history of the CAFE or CO2 tailpipe emissions programs—exceeding 750,000 comments. The agencies recognized in the NPRM that the proposal was significantly different from the final rules set forth in 2012, and explained at length the reasons for those differences—namely, that new information and considerations, along with an expanded and updated analysis, had led to different tentative conclusions. Today's final rules represent a further evolution of the work that supported the proposal, based on improved quantitative methodology and in careful consideration of the hundreds of thousands of public comments and deep reflection on the serious issues before the agencies. Simply put, the agencies have heard the comments, and today's analysis and decision reflect the agencies' grappling with the issues commenters raised, as well as all of the other information before the agencies. These programs and issues are weighty, and the agencies believe that a reasonable balance has been struck in these final rules between the many competing national needs that these regulatory programs collectively address.

II. Overview of Final Rule

A. Summary of Proposal

In the NPRM, the National Highway Traffic Safety Administration (NHTSA) and the Environmental Protection Agency (EPA) (collectively, “the Start Printed Page 24182agencies”) proposed the “Safer Affordable Fuel-Efficient (SAFE) Vehicles Rule for Model Years 2021-2026 Passenger Cars and Light Trucks” (SAFE Vehicles Rule). The proposed SAFE Vehicles Rule would set Corporate Average Fuel Economy (CAFE) and carbon dioxide (CO2) emissions standards, respectively, for passenger cars and light trucks manufactured for sale in the United States in model years (MYs) 2021 through 2026.[20]

The agencies explained that they must act to propose and finalize these standards and do not have discretion to decline to regulate. Congress requires NHTSA to set CAFE standards for each model year.[21] Congress also requires EPA to set emissions standards for light-duty vehicles if EPA has made an “endangerment finding” that the pollutant in question—in this case, CO2—“cause[s] or contribute[s] to air pollution which may reasonably be anticipated to endanger public health or welfare.” [22] NHTSA and EPA proposed the standards concurrently because tailpipe CO2 emissions standards are directly and inherently related to fuel economy standards,[23] and, if finalized, the rules would apply concurrently to the same fleet of vehicles. By working together to develop the proposals, the agencies aimed to reduce regulatory burden on industry and improve administrative efficiency.

The agencies discussed some of the history leading to the proposal, including the 2012 final rule, the expectations regarding a mid-term evaluation as required by EPA regulation, and the rapid process over 2016 and early 2017 by which EPA issued its first Final Determination that the CO2 standards set in 2012 for MYs 2022-2025 remained appropriate based on the information then before the EPA Administrator.[24] The agencies also discussed President Trump's direction in March 2017 to restore the original mid-term evaluation timeline, and EPA's subsequent information-gathering process and announcement that it would reconsider the January 2017 Determination.[25] EPA ultimately concluded that the standards set in 2012 for MYs 2022-2025 were no longer appropriate.[26] For NHTSA, in turn, the “augural” CAFE standards for MYs 2022-2025 were never final, and as explained in the 2012 final rule, NHTSA was obligated from the beginning to undertake a new rulemaking to set CAFE standards for MYs 2022-2025.

The NPRM thus began the rulemaking process for both agencies to establish new standards for MYs 2022-2025 passenger cars and light trucks. Standards were concurrently proposed for MY 2026 in order to provide regulatory stability for as many years as is legally permissible for both agencies together. The NPRM also included revised standards for MY 2021 passenger cars and light trucks, because the agencies tentatively concluded, based on the information and analysis then before them, that the CAFE standards previously set for MY 2021 were no longer maximum feasible, and the CO2 standards previously set for MY 2021 were no longer appropriate. Agencies always have authority under the Administrative Procedure Act to revisit previous decisions in light of new facts, as long as they provide notice and an opportunity for comment, and the agencies stated that it is plainly the best practice to do so when changed circumstances so warrant.[27]

The NPRM proposed to maintain the CAFE and CO2 standards applicable in MY 2020 for MYs 2021-2026, and took comment on a wide range of alternatives, including different stringencies and retaining existing CO2 standards and the augural CAFE standards.[28] Table II-1, Table II-2, and Table II-3 show the estimates, under the NPRM analysis, of what the MY 2020 CAFE and CO2 curves would translate to, in terms of miles per gallon (mpg) and grams per mile (g/mi), in MYs 2021-2026, as well as the regulatory alternatives considered in the NPRM. In addition to retaining the MY 2020 CO2 standards through MY 2026, EPA proposed and sought comment on excluding air conditioning refrigerants and leakage, and nitrous oxide and methane emissions for compliance with CO2 standards after model year 2020, in order to improve harmonization with the CAFE program. EPA also sought comment on whether to change existing methane and nitrous oxide standards that were finalized in the 2012 rule. The proposal was accompanied by a 1,600 page Preliminary Regulatory Impact Analysis (PRIA) and, for NHTSA, a 500 page Draft Environmental Impact Statement (DEIS), with more than 800 pages of appendices and the entire CAFE model, including the software source code and documentation, all of which were also subject to comment in their entirety and all of which received significant comments.

Start Printed Page 24183

Start Printed Page 24184

Start Printed Page 24185

The agencies explained in the NPRM that new information had been gathered and new analysis performed since publication of the 2012 final rule establishing CAFE and CO2 standards for MYs 2017 and beyond and since issuance of the 2016 Draft TAR and EPA's 2016 and early 2017 “mid-term evaluation” process. This new information and analysis helped lead the agencies to the tentative conclusion that holding standards constant at MY 2020 levels through MY 2026 was maximum feasible, for CAFE purposes, and appropriate, for CO2 purposes.

The agencies further explained that technologies had played out differently in the fleet from what the agencies previously assumed: That while there remain a wide variety of technologies available to improve fuel economy and reduce CO2 emissions, it had become clear that there were reasons to temper previous optimism about the costs, effectiveness, and consumer acceptance of a number of technologies. In addition, over the years between the previous analyses and the NPRM, automakers had added considerable amounts of technologies to their new vehicle fleets, meaning that the agencies were no longer free to make certain assumptions about how some of those technologies could be used going forward. For example, some technologies that could be used to improve fuel economy and reduce emissions had not been used entirely for that purpose, and some of the benefit of these technologies had gone instead toward improving other vehicle attributes. Other technologies had been tried, and had been met with significant customer acceptance issues. The agencies underscored the importance of reflecting the fleet as it stands today, with the technology it has and as that technology has been used, and considering what technology remains on the table at this point, whether and when it can realistically be available for widespread use in production, and how much it would cost to implement.

The agencies also acknowledged the math of diminishing returns: As CAFE and CO2 emissions standards increase in stringency, the benefit of continuing to increase in stringency decreases. In mpg terms, a vehicle owner who drives a light vehicle 15,000 miles per year (a typical assumption for analytical purposes) [31] and trades in a vehicle with fuel economy of 15 mpg for one with fuel economy of 20 mpg, will reduce their annual fuel consumption from 1,000 gallons to 750 gallons—saving 250 gallons annually. If, however, that owner were to trade in a vehicle with fuel economy of 30 mpg for one with fuel economy of 40 mpg, the owner's annual gasoline consumption would drop from 500 gallons/year to 375 gallons/year—only 125 gallons even though the mpg improvement is twice as large. Going from 40 to 50 mpg would save only 75 gallons/year. Yet each additional fuel economy improvement becomes much more expensive as the easiest to achieve low-cost technological improvement options are chosen. In CO2 terms, if a vehicle emits 300 g/mi CO2, Start Printed Page 24186a 20 percent improvement is 60 g/mi, so the vehicle would emit 240 g/mi; but if the vehicle emits 180 g/mi, a 20 percent improvement is only 36 g/mi, so the vehicle would get 144 g/mi. In order to continue achieving similarly large (on an absolute basis) emissions reductions, the percentage reduction must also continue to increase.

Related, average real-world fuel economy is lower than average fuel economy required under CAFE and CO2 standards. The 2012 Federal Register notice announcing augural CAFE and CO2 standards extending through MY 2025 indicated that, if met entirely through the application of fuel-saving technology, the MY 2025 CO2 standards would result in an average requirement equivalent to 54.5 mpg. However, because the CO2 standards provide credit for reducing leakage of AC refrigerants and/or switching to lower-GWP refrigerants, and these actions do not affect fuel economy, the notice explained that the corresponding fuel economy requirement (under the CAFE program) would be 49.7 mpg. These estimates were based on a market forecast grounded in the MY 2008 fleet. The notice also presented analysis using a market forecast grounded in the MY 2010 fleet, showing a 48.7 mpg average CAFE requirement.

In the real world, fuel economy is, on average, about 20% lower than as measured under regulatory test procedures. In the real world, then, these new standards were estimated to require 39.0-39.8 mpg.

Today's analysis indicates that the requirements under the baseline/augural CAFE standards would average 46.6 mpg in MY 2029. The lower value results from changes in the fleet forecast which reflects consumer preference for larger vehicles than was forecast for the 2012 rulemaking. In the real world, the requirements average about 37.1 mpg. Under the final standards issued today, the regulatory test procedure requirements average 40.5 mpg, corresponding to 33.2 mpg in the real world. Buyers of new vehicles experience real-world fuel economy, with levels varying among drivers (due to a wide range of factors). Vehicle fuel economy labels provide average real-world fuel economy information to buyers.

Vehicle owners also face fuel prices at the pump. The agencies noted in the NPRM that when fuel prices are high, the value of fuel saved may be enough to offset the cost of further fuel economy/emissions reduction improvements, but the agencies recognized that then-current projections of fuel prices by the Energy Information Administration did not indicate particularly high fuel prices in the foreseeable future. The agencies explained that fundamental structural shifts had occurred in global oil markets since the 2012 final rule, largely due to the rise of U.S. production and export of shale oil. The consequence over time of diminishing returns from more stringent fuel economy/emissions reduction standards, especially when combined with relatively low fuel prices, is greater difficulty for automakers to find a market of consumers willing to buy vehicles that meet the increasingly stringent standards. American consumers have long demonstrated that in times of relatively low fuel prices, fuel economy is not a top priority for the majority of them, even when highly fuel efficient vehicle models are available.

The NPRM analysis sought to improve how the agencies captured the effects of higher new vehicle prices on fleet composition as a whole by including an improved model for vehicle scrappage rates. As new vehicle prices increase, consumers tend to continue using older vehicles for longer, slowing fleet turnover and thus slowing improvements in fleet-wide fuel economy, reductions in CO2 emissions, reductions in criteria pollutant emissions, and advances in safety. That aspect of the analysis was also driven by the agencies' updated estimates of average per-vehicle cost increases due to Start Printed Page 24187higher standards, which were several hundred dollars higher than previously estimated. The agencies cited growing concerns about affordability and negative equity for many consumers under these circumstances, as loan amounts grow and loan terms extend.

For all of the above reasons, the agencies proposed to maintain the MY 2020 fuel economy and CO2 emissions standards for MYs 2021-2026. The agencies explained that they estimated, relative to the standards for MYs 2021-2026 put forth in 2012, that an additional 0.5 million barrels of oil would be consumed per day (about 2 to 3 percent of projected U.S. consumption) if that proposal were finalized, but that they also expected the additional fuel costs to be outweighed by the cost savings from new vehicle purchases; that more than 12,700 on-road fatalities and significantly more injuries would be prevented over the lifetimes of vehicles through MY 2029 as compared to the standards set forth in the 2012 final rule over the lifetimes of vehicles as more new and safer vehicles are purchased than the current (and augural) standards; and that environmental impacts, on net, would be relatively minor, with criteria and toxic air pollutants not changing noticeably, and with estimated atmospheric CO2 concentrations increasing by 0.65 ppm (a 0.08 percent increase), which the agencies estimated would translate to 0.003 degrees Celsius of additional temperature increase relative to the standards finalized in 2012.

Under the NPRM analysis, the agencies tentatively concluded that maintaining the MY 2020 curves for MYs 2021-2026 would save American auto consumers, the auto industry, and the public a considerable amount of money as compared to EPA retaining the previously-set CO2 standards and NHTSA finalizing the augural standards. The agencies explained that this had been identified as the preferred alternative, in part, because it appeared to maximize net benefits compared to the other alternatives analyzed, and recognizing the statutory considerations for both agencies. Relative to the standards issued in 2012, under CAFE standards, the NPRM analysis estimated that costs would decrease by $502 billion overall at a three-percent discount rate ($335 billion at a seven-percent discount rate) and benefits were estimated to decrease by $326 billion at a three-percent discount rate ($204 billion at a seven-percent discount rate). Thus, net benefits were estimated to increase by $176 billion at a three-percent discount rate and $132 billion at a seven-percent discount rate. The estimated impacts under CO2 standards were estimated to be similar, with net benefits estimated to increase by $201 billion at a three-percent discount rate and $141 billion at a seven-percent discount rate.

The NPRM also sought comment on a variety of potential changes to NHTSA's and EPA's compliance programs for CAFE and CO2 as well as related programs, including questions about automaker requests for additional flexibilities and agency interest in reducing market-distorting incentives and improving transparency; and on a proposal to withdraw California's CAA preemption waiver for its “Advanced Clean Car” regulations, with an accompanying discussion of preemption of State standards under EPCA.[32] The agencies sought comment broadly on all aspects of the proposal.

B. Public Participation Opportunities and Summary of Comments

The NPRM was published on NHTSA's and EPA's websites on August 2, 2018, and published in the Federal Register on August 24, 2018, beginning a 60-day comment period. The agencies subsequently extended the official comment period for an additional three days, and left the dockets open for more than a year after the start of the comment period, considering late comments to the extent practicable. A separate Federal Register notice also published on August 24, 2018, which announced the locations, dates, and times of three public hearings to be held on the proposal: One in Fresno, California, on September 24, 2018; one in Dearborn, Michigan, on September 25, 2018; and one in Pittsburgh, Pennsylvania, on September 26, 2018. Each hearing started at 10 a.m. local time; the Fresno hearing ended at 5:10 p.m. and resulted in a 235 page transcript; the Dearborn hearing ran until 5:26 p.m. and resulted in a 330 page transcript; and the Pittsburgh hearing ran until 5:06 p.m. and also resulted in a 330 page transcript. Each hearing also collected several hundred pages of comments from participants, in addition to the hearing transcripts.

Besides the comments submitted as part of the public hearings, NHTSA's docket received a total of 173,359 public comments in response to the proposal as of September 18, 2019, and EPA's docket a total of 618,647 public comments, for an overall total of 792,006. NHTSA also received several hundred comments on its DEIS to the separate DEIS docket. While the majority of individual comments were form letters, the agencies received over 6,000 pages of substantive comments on the proposal.

Many commenters generally supported the proposal and many commenters opposed it. Commenters supporting the proposal tended to cite concerns about the cost of new vehicles, while commenters opposing the proposal tended to cite concerns about additional fuel expenditures and the impact on climate change. Many comments addressed the modeling used for the analysis, and specifically the inclusion, operation, and results of the sales and scrappage modules that were part of the NPRM's analysis, while many addressed the NPRM's safety findings and the role that those findings played in the proposal's justification. Many other comments addressed California's standards and role in Federal decision-making; as discussed above, those comments are further summarized and responded to in the separate Federal Register notice published in September 2019. Nearly every aspect of the NPRM's analysis and discussion received some level of comment by at least one commenter. The comments received, as a whole, were both broad and deep, and the agencies appreciate the level of engagement of commenters in the public comment process and the information and opinions provided.

C. Changes in Light of Public Comments and New Information

The agencies made a number of changes to the analysis between the NPRM and the final rule in response to public comments and new information that was received in those comments or otherwise became available to the agencies. While these changes, their rationales, and their effects are discussed in detail in the sections below, the following represents a high-level list of some of the most significant changes:

  • Some regulatory alternatives were dropped from consideration, and one was added;
  • updated analysis fleet, and changes to technologies on “baseline” vehicles within the fleet to reflect better their current properties and improve modeling precision;
  • no civil penalties assumed to be paid after MY 2020 under CAFE program;
  • updates and expansions in accounting for certain over-compliance Start Printed Page 24188credits, including early credits earned in EPA's program;
  • updates and expansions to CAFE Model's technology paths;
  • updates to inputs defining the range of manufacturer-, technology-, and product-specific constraints;
  • updates to allow the model to adopt a more advanced technology if it is more cost-effective than an earlier technology on the path;
  • precision improvements to the modeling of A/C efficiency and off-cycle credits;
  • updates to model's “effective cost” metric;
  • extended explicit simulation of technology application through MY 2050;
  • expanded presentation of the results to include “calendar year” analysis;
  • quantifying different types of health impacts from changes in air pollution, rather than only accounting for such impacts in aggregate estimates of the social costs of air pollution;
  • updated costs to 2018 dollars;
  • updated fuel costs based on the AEO 2019 version of NEMS;
  • a variety of technology updates in response to comments and new information;
  • updated accounting of rebound VMT between regulatory alternatives;
  • updated estimates of the macroeconomic cost of petroleum dependence;
  • updated response of total new vehicle sales to increases in fuel efficiency and price; and
  • updated response of vehicle retirement rates to changes in new vehicle fuel efficiency and transaction price.

Sections IV and VI below discuss these updates in significant detail.

D. Final Standards—Stringency

As explained above, the agencies have chosen to set CAFE and CO2 standards that increase in stringency by 1.5 percent year over year for MYs 2021-2026. Separately, EPA has decided to retain the A/C refrigerant and leakage and CH4 and N2 O standards set forth in 2012 for MYs 2021 and beyond, and the stringency of the CO2 standards in this final rule reflect the “offset” also established in 2012 based on assumptions made at that time about anticipated HFC emissions reductions.

When the agencies state that stringency will increase at 1.5 percent per year, that means that the footprint curves which actually define the standards for CAFE and CO2 emissions will become more stringent at 1.5 percent per year. Consistent with Congress's direction in EISA to set CAFE standards based on a mathematical formula, which EPA harmonized with for the CO2 emissions standards, the standard curves are equations, which are slightly different for CAFE and CO2, and within each program, slightly different for passenger cars and light trucks. Each program has a basic equation for a fleet standard, and then values that change to cause the stringency changes are the coefficients within the equations. For passenger cars, consistent with prior rulemakings, NHTSA is defining fuel economy targets as follows:


TARGETFE is the fuel economy target (in mpg) applicable to a specific vehicle model type with a unique footprint combination,

a is a minimum fuel economy target (in mpg),

b is a maximum fuel economy target (in mpg),

c is the slope (in gallons per mile per square foot, or gpm, per square foot) of a line relating fuel consumption (the inverse of fuel economy) to footprint, and

d is an intercept (in gpm) of the same line.

Here, MIN and MAX are functions that take the minimum and maximum values, respectively, of the set of included values. For example, MIN[40,35] = 35 and MAX (40, 25) = 40, such that MIN[MAX (40, 25), 35] = 35.

For light trucks, also consistent with prior rulemakings, NHTSA is defining fuel economy targets as follows:


TARGETFE is the fuel economy target (in mpg) applicable to a specific vehicle model type with a unique footprint combination,

a, b, c, and d are as for passenger cars, but taking values specific to light trucks,

e is a second minimum fuel economy target (in mpg),

f is a second maximum fuel economy target (in mpg),

g is the slope (in gpm per square foot) of a second line relating fuel consumption (the inverse of fuel economy) to footprint, and

h is an intercept (in gpm) of the same second line.

The final CAFE standards (described in terms of their footprint-based curves) are as follows, with the values for the coefficients changing over time:

Start Printed Page 24189

These equations are presented graphically below, where the x-axis represents vehicle footprint and the y-axis represents fuel economy, showing that in the CAFE context, targets are higher (fuel economy) for smaller footprint vehicles and lower for larger footprint vehicles:

Start Printed Page 24190

Start Printed Page 24191

EPCA, as amended by EISA, requires that any manufacturer's domestically-manufactured passenger car fleet must meet the greater of either 27.5 mpg on average, or 92 percent of the average fuel economy projected by the Secretary for the combined domestic and non-domestic passenger automobile fleets manufactured for sale in the U.S. by all manufacturers in the model year, which projection shall be published in the Federal Register when the standard for that model year is promulgated in accordance with 49 U.S.C. 32902(b).[33] Any time NHTSA establishes or changes a passenger car standard for a model year, the MDPCS for that model year must also be evaluated or re-evaluated and established accordingly. Thus, this final rule establishes the applicable MDPCS for MYs 2021-2026. Table II-8 lists the minimum domestic passenger car standards.

EPA CO2 standards are as follows. Rather than expressing these standards as linear functions with accompanying minima and maxima, similar to the approach NHTSA has followed since 2005 in specifying attribute-based standards, the following tables specify flat standards that apply below and above specified footprints, and a linear function that applies between those footprints. The two approaches are mathematically identical. For passenger cars with a footprint of less than or equal to 41 square feet, the gram/mile CO2 target value is selected for the appropriate model year from Table II-9:

Start Printed Page 24192

For passenger cars with a footprint of greater than 56 square feet, the gram/mile CO2 target value is selected for the appropriate model year from Table II-10:

For passenger cars with a footprint that is greater than 41 square feet and less than or equal to 56 square feet, the gram/mile CO2 target value is calculated using the following equation and rounded to the nearest 0.1 grams/mile.

Start Printed Page 24193

Target CO2 = [a × f] + b

Where f is the vehicle footprint and a and b are selected from Table II-11 for the appropriate model year:

For light trucks with a footprint of less than or equal to 41 square feet, the gram/mile CO2 target value is selected for the appropriate model year from Table II-12:

Start Printed Page 24194

For light trucks with a footprint greater than the minimum value specified in the table below for each model year, the gram/mile CO2 target value is selected for the appropriate model year from Table II-13:

Start Printed Page 24195

For light trucks with a footprint that is greater than 41 square feet and less than or equal to the maximum footprint value specified in Table II-14 below for each model year, the gram/mile CO2 target value is calculated using the following equation and rounded to the nearest 0.1 grams/mile.

Target CO2 = (a × f) + b

Where f is the footprint and a and b are selected from Table II-14 below for the appropriate model year:

These equations are presented graphically below, where the x-axis represents vehicle footprint and the y-axis represents the CO2 target. The targets are lower for smaller footprint vehicles and higher for larger footprint vehicles:

Start Printed Page 24196

Start Printed Page 24197

Except that EPA elected to apply a slightly different slope when defining passenger car targets, CO2 targets may be expressed as direct conversion of fuel economy targets, as follows:

where 8887 g/gal relates grams of CO2 emitted to gallons of fuel consumed, and OFFSET reflects the fact that that HFC emissions from lower-GWP A/C refrigerants and less leak-prone A/C systems are counted toward average CO2 emissions, but EPCA provides no basis to count reduced HFC emissions toward CAFE levels.

For the reader's benefit, Table II-15, Table II-16, and Table II-17 show the estimates, under the final rule analysis, of what the MYs 2021-2026 CAFE and CO2 curves would translate to, in terms of miles per gallon (mpg) and grams per mile (g/mi).

Start Printed Page 24198

Start Printed Page 24199

As the following tables demonstrate, averages of manufacturers' estimated requirements are more stringent (i.e., for CAFE, higher, and for CO2, lower) under the final standards than under the proposed standards:

Start Printed Page 24200

E. Final Standards—Impacts

This section summarizes the estimated costs and benefits of the MYs 2021-2026 CAFE and CO2 emissions standards for passenger cars and light trucks, as compared to the regulatory alternatives considered. These estimates helped inform the agencies' choices among the regulatory alternatives considered and provide further confirmation that the final standards are maximum feasible, for NHTSA, and appropriate, for EPA. The costs and benefits estimated to result from the CAFE standards are presented first, followed by those estimated to result from the CO2 standards. For several reasons, the estimates for costs and benefits presented for the different programs, while consistent, are not identical. NHTSA's and EPA's standards are projected to result in slightly different fuel efficiency improvements. EPA's CO2 standard is nominally more stringent in part due to its assumptions about manufacturers' use of air conditioning leakage/refrigerant replacement credits, which are expected to result in reduced emissions of HFCs. NHTSA's final standards are based solely on assumptions about fuel economy improvements, and do not account for emissions reductions that do not relate to fuel economy. In addition, the CAFE and CO2 programs offer somewhat different program flexibilities and provisions, primarily because NHTSA is statutorily prohibited from considering some flexibilities when establishing CAFE standards, while EPA is not.[34] The analysis underlying this final rule reflects many of those additional EPA flexibilities, which contributes to differences in how the agencies estimate manufacturers could comply with the respective sets of standards, which in turn contributes to differences in estimated impacts of the standards. These differences in compliance flexibilities are discussed in more detail in Section IX below.

Table II-20 to Table II-23 present all subcategories of costs and benefits of this final rule for all seven alternatives proposed. Costs include application of fuel economy technology to new vehicles, consumer surplus, crash costs due to changes in VMT, as well as, noise and congestion. Benefits include fuel savings, consumer surplus, refueling time, and clean air.

Start Printed Page 24201

Start Printed Page 24202

Start Printed Page 24203

Start Printed Page 24204

Start Printed Page 24205

Start Printed Page 24206

Start Printed Page 24207

Start Printed Page 24208

Start Printed Page 24209

F. Other Programmatic Elements

1. Compliance and Flexibilities

Automakers seeking to comply with the CAFE and CO2 standards are generally expected to add fuel economy-improving technologies to their new vehicles to boost their overall fleet fuel economy levels. Readers will remember that improving fuel economy directly reduces CO2 emissions, because CO2 is a natural and inevitable byproduct of fossil fuel combustion to power vehicles. The CAFE and CO2 programs contain a variety of compliance provisions and flexibilities to accommodate better automakers' production cycles, to reward real-world fuel economy improvements that cannot be reflected in the 1975-developed test procedures, and to incentivize the production of certain types of vehicles. While the agencies sought comment on a broad variety of changes and potential expansions of the programs' compliance flexibilities in the NPRM, the agencies determined, after considering the comments, to make a few changes to the flexibilities proposed in the NPRM in this final rule. The most noteworthy change is the retention, in the CO2 program, of the flexibilities that allow automakers to continue to use HFC reductions toward their CO2 compliance, and that extend the “0 grams/mile” assumption for electric vehicles through MY 2026 (i.e., recognizing only the tailpipe emissions of full battery-electric vehicles and not recognizing the upstream emissions caused by the electricity usage of those vehicles). In the NPRM, EPA had proposed to remove and sought comment on removing those flexibilities from the CO2 program, but determined not to remove them in this final rule. EPA and NHTSA are also removing from the programs, starting in MY 2022, the credit/FCIV for full-size pickup trucks that are either hybrids or over-performing by a certain amount relative to their targets, and allowing technology suppliers to begin the petition process for off-cycle credits/adjustments.

Table II-24, Table II-25, Table II-26, and Table II-27 provide a summary of the various compliance provisions in the two programs; their authorities; and any changes included as part of this final rule:

Start Printed Page 24210

Start Printed Page 24211

Providing a technology neutral basis by which manufacturers meet fuel economy and CO2 emissions standards encourages an efficient and level playing field. The agencies continue to have a desire to minimize incentives that disproportionately favor one technology over another. Some of this may involve regulations established by other Federal agencies. In the near future, NHTSA and EPA intend to work with other relevant Federal agencies to pursue regulatory means by which we can further ensure technology neutrality in this field.

2. Preemption/Waiver

As discussed above, the issues of Clean Air Act waivers of preemption under Section 209 and EPCA/EISA preemption under 49 U.S.C. 32919 are not addressed in today's final rule, as Start Printed Page 24212they were the subject of a separate final rulemaking action by the agencies in September 2019. While many comments were received in response to the NPRM discussion of those issues, those comments have been addressed and responded to as part of that separate rulemaking action.

III. Purpose of the Rule

The Administrative Procedure Act (APA) requires agencies to incorporate in their final rules a “concise general statement of their basis and purpose.” [36] While the entire preamble document represents the agencies' overall explanation of the basis and purpose for this regulatory action, this section within the preamble is intended as a direct response to that APA (and related CAA) requirements. Executive Order 12866 further states that “Federal agencies should promulgate only such regulations as are required by law, are necessary to interpret the law, or are made necessary by compelling public need, such as material failures of private markets to protect or improve the health and safety of the public, the environment, or the well-being of the American people.” [37] Section III.C of the FRIA accompanying this rulemaking discusses at greater length the question of whether a market failure exists that these final rules may address.

NHTSA and EPA are legally obligated to set CAFE and GHG standards, respectively, and do not have the authority to decline to regulate.[38] The agencies are issuing these final rules to fulfill their respective statutory obligations to provide maximum feasible fuel economy standards and limit emissions of pollutants from new motor vehicles which have been found to endanger public health and welfare (in this case, specifically carbon dioxide (CO2); EPA has already set standards for methane (CH4), nitrous oxide (N2 O), and hydrofluorocarbons (HFCs) and is not revising them in this rule). Continued progress in meeting these statutory obligations is both legally necessary and good for America—greater energy security and reduced emissions protect the American public. The final standards continue that progress, albeit at a slower rate than the standards finalized in 2012.

National annual gasoline consumption and CO2 emissions currently total about 140 billion gallons and 5,300 million metric tons, respectively. The majority of this gasoline (about 130 billion gallons) is used to fuel passenger cars and light trucks, such as will be covered by the CAFE and CO2 standards issued today. Accounting for both tailpipe emissions and emissions from “upstream” processes (e.g., domestic refining) involved in producing and delivering fuel, passenger cars and light trucks account for about 1,500 million metric tons (mmt) of current annual CO2 emissions. The agencies estimate that under the standards issued in 2012, passenger car and light truck annual gasoline consumption would steadily decline, reaching about 80 billion gallons by 2050. The agencies further estimate that, because of this decrease in gasoline consumption under the standards issued in 2012, passenger car and light truck annual CO2 emissions would also steadily decline, reaching about 1,000 mmt by 2050. Under the standards issued today, the agencies estimate that, instead of declining from about 140 billion gallons annually today to about 80 billion gallons annually in 2050, passenger car and light truck gasoline consumption would decline to about 95 billion gallons. The agencies correspondingly estimate that instead of declining from about 1,500 mmt annually today to about 1,000 mmt annually in 2050, passenger car and light truck CO2 emissions would decline to about 1,100 mmt. In short, the agencies estimate that under the standards issued today, annual passenger car and light truck gasoline consumption and CO2 emissions will continue to steadily decline over the next three decades, even if not quite as rapidly as under the previously-issued standards.

The agencies also estimate that these impacts on passenger car and light truck gasoline consumption and CO2 emissions will be accompanied by a range of other energy- and climate-related impacts, such as reduced electricity consumption (because today's standards reduce the estimated rate at which the market might shift toward electric vehicles) and increased CH4 and N2 O emissions. These estimated impacts, discussed below and in the FEIS accompanying today's notice, are dwarfed by estimated impacts on gasoline consumption and CO2 emissions.

As explained above, these final rules set or amend fuel economy and carbon dioxide standards for model years 2021-2026. Many commenters argued that it was not appropriate to amend previously-established CO2 and CAFE standards, generally because those commenters believed that the administrative record established for the 2012 final rule and EPA's January 2017 Final Determination was superior to the record that informed the NPRM, and that that prior record led necessarily to the policy conclusion that the previously-established standards should remain in place.[39] Some commenters similarly argued that EPA's Revised Final Determination—which, for EPA, preceded this regulatory action—was invalid because, they allege, it did not follow the procedures established for the mid-term evaluation that EPA codified into regulation,[40] and also because the Revised Final Determination was not based on the prior record.[41]

The agencies considered a range of alternatives in the proposal, including the baseline/no action alternative of retaining the existing EPA carbon dioxide standards. As the agencies explained in the proposal, the proposal was entirely de novo, based on an entirely new analysis reflecting the best and most up-to-date information available to the agencies.[42] This rulemaking action is separate and distinct from EPA's Revised Final Determination, which itself was neither a proposed nor a final decision that the standards “must” be revised. EPA retained full discretion in this rulemaking to revise the standards or not revise them. In any event, the case law is clear that agencies are free to reconsider their prior decisions.[43] With that legal principle in mind, the agencies agree with commenters that the amended (and new) CO2 and CAFE standards must be consistent with the Start Printed Page 24213CAA and EPCA/EISA, respectively, and this preamble and the accompanying FRIA explain in detail why the agencies believe they are consistent. The section below discusses briefly the authority given to the agencies by their respective governing statutes, and the factors that Congress directed the agencies to consider as they exercise that authority in pursuit of fulfilling their statutory obligations.

A. EPA's Statutory Requirements

EPA is setting national CO2 standards for passenger cars and light trucks under Section 202(a) of the Clean Air Act (CAA).[44] Section 202(a) of the CAA requires EPA to establish standards for emissions of pollutants from new motor vehicles which cause or contribute to air pollution which may reasonably be anticipated to endanger public health or welfare.[45] In establishing such standards, EPA considers issues of technical feasibility, cost, available lead time, and other factors. Standards under section 202(a) thus take effect only “after providing such period as the Administrator finds necessary to permit the development and application of the requisite technology, giving appropriate consideration to the cost of compliance within such period.” [46] EPA's statutory requirements are further discussed in Section VIII.A.

B. NHTSA's Statutory Requirements

NHTSA is setting national Corporate Average Fuel Economy (CAFE) standards for passenger cars and light trucks for each model year as required under EPCA, as amended by EISA.[47] EPCA mandates a motor vehicle fuel economy regulatory program that balances statutory factors in setting minimum fuel economy standards to facilitate energy conservation. EPCA allocates the responsibility for implementing the program between NHTSA and EPA as follows: NHTSA sets CAFE standards for passenger cars and light trucks; EPA establishes the procedures for testing, tests vehicles, collects and analyzes manufacturers' data, and calculates the individual and average fuel economy of each manufacturer's passenger cars and light trucks; and NHTSA enforces the standards based on EPA's calculations.

The following sections enumerate specific statutory requirements for NHTSA in setting CAFE standards and NHTSA's interpretations of them, where applicable. Many comments were received on these requirements and interpretations. Because this is intended as an overview section, those comments will be addressed below in Section VIII rather than here, and the agencies refer readers to that part of the document for more information.

For each future model year, EPCA (as amended by EISA) requires that DOT (by delegation, NHTSA) establish separate passenger car and light truck standards at “the maximum feasible average fuel economy level that the Secretary decides the manufacturers can achieve in that model year,” [48] based on the agency's consideration of four statutory factors: “technological feasibility, economic practicability, the effect of other motor vehicle standards of the Government on fuel economy, and the need of the United States to conserve energy.” [49] The law also allows NHTSA to amend standards that are already in place, as long as doing so meets these requirements.[50] EPCA does not define these terms or specify what weight to give each concern in balancing them; thus, NHTSA defines them and determines the appropriate weighting that leads to the maximum feasible standards given the circumstances in each CAFE standard rulemaking.[51]

EISA added several other requirements to the setting of separate passenger car and light truck standards. Standards must be “based on 1 or more vehicle attributes related to fuel economy and express[ed] . . . in the form of a mathematical function.” [52] New standards must also be set at least 18 months before the model year in question, as would amendments to increase standards previously set.[53] NHTSA must regulations prescribing average fuel economy standards for at least 1, but not more than 5, model years at a time.[54] A number of comments addressed these requirements; for the reader's reference, those comments will be summarized and responded to in Section VIII. EISA also added the requirement that NHTSA set a minimum standard for domestically-manufactured passenger cars,[55] which will also be discussed further in Section VIII below.

For MYs 2011-2020, EISA further required that the separate standards for passenger cars and for light trucks be set at levels high enough to ensure that the achieved average fuel economy for the entire industry-wide combined fleet of new passenger cars and light trucks reach at least 35 mpg not later than MY 2020, and standards for those years were also required to “increase ratably.” [56] For model years after 2020, standards must be set at the maximum feasible level.[57]

1. Factors That Must Be Considered in Deciding What Levels of CAFE Standards are “Maximum Feasible”

(a) Technological Feasibility

“Technological feasibility” refers to whether a particular method of improving fuel economy can be available for commercial application in the model year for which a standard is being established. Thus, in determining the level of new standards, the agency is not limited to technology that is already being commercially applied at the time of the rulemaking. For this rulemaking, NHTSA has evaluated and considered all types of technologies that improve real-world fuel economy, although not every possible technology was expressly included in the analysis, as discussed in Section VI and also in Section VIII.

(b) Economic Practicability

“Economic practicability” refers to whether a standard is one “within the Start Printed Page 24214financial capability of the industry, but not so stringent as to” lead to “adverse economic consequences, such as a significant loss of jobs or the unreasonable elimination of consumer choice.” [58] The agency has explained in the past that this factor can be especially important during rulemakings in which the automobile industry is facing significantly adverse economic conditions (with corresponding risks to jobs). Economic practicability is a broad factor that includes considerations of the uncertainty surrounding future market conditions and consumer demand for fuel economy in addition to other vehicle attributes.[59] In an attempt to evaluate the economic practicability of different future levels of CAFE standards (i.e., the regulatory alternatives considered in this rulemaking), NHTSA considers a variety of factors, including the annual rate at which manufacturers can increase the percentage of their fleet(s) that employ a particular type of fuel-saving technology, the specific fleet mixes of different manufacturers, assumptions about the cost of the standards to consumers, and consumers' valuation of fuel economy, among other things, including, in part, safety.

It is important to note, however, that the law does not preclude a CAFE standard that poses considerable challenges to any individual manufacturer. The Conference Report for EPCA, as enacted in 1975, makes clear, and the case law affirms, “a determination of maximum feasible average fuel economy should not be keyed to the single manufacturer which might have the most difficulty achieving a given level of average fuel economy.” [60] Instead, NHTSA is compelled “to weigh the benefits to the nation of a higher fuel economy standard against the difficulties of individual automobile manufacturers.” [61] Accordingly, while the law permits NHTSA to set CAFE standards that exceed the projected capability of a particular manufacturer as long as the standard is economically practicable for the industry as a whole, the agency cannot simply disregard that impact on individual manufacturers.[62] That said, in setting fuel economy standards, NHTSA does not seek to maintain competitive positions among the industry players, and notes that while a particular CAFE standard may pose difficulties for one manufacturer as being too high or too low, it may also present opportunities for another. NHTSA has long held that the CAFE program is not necessarily intended to maintain the competitive positioning of each particular company. Rather, it is intended to enhance the fuel economy of the vehicle fleet on American roads, while protecting motor vehicle safety and paying close attention to the economic risks.

(c) The Effect of Other Motor Vehicle Standards of the Government on Fuel Economy

“The effect of other motor vehicle standards of the Government on fuel economy” involves an analysis of the effects of compliance with emission, safety, noise, or damageability standards on fuel economy capability and thus on average fuel economy. In many past CAFE rulemakings, NHTSA has said that it considers the adverse effects of other motor vehicle standards on fuel economy. It said so because, from the CAFE program's earliest years,[63] the effects of such compliance on fuel economy capability over the history of the program have been negative ones. For example, safety standards that have the effect of increasing vehicle weight lower vehicle fuel economy capability and thus decrease the level of average fuel economy that the agency can determine to be feasible. NHTSA has considered the additional weight that it estimates would be added in response to new safety standards during the rulemaking timeframe. NHTSA has also accounted for EPA's “Tier 3” standards for criteria pollutants in its estimates of technology effectiveness.[64]

The NPRM also discussed how EPA's CO2 standards for light-duty vehicles and California's Advanced Clean Cars program fit into NHTSA's consideration of “the effect of other motor vehicle standards of the Government on fuel economy.” The agencies note that on September 19, 2019, to ensure One National Program for automobile fuel economy and carbon dioxide emissions standards, the agencies finalized regulatory text related to preemption of State tailpipe CO2 standards and Zero Emission Vehicle (ZEV) mandates under EPCA and partial withdrawal of a waiver previously provided to California under the Clean Air Act.[65] This final rule's impact on State programs—including California's—will therefore be somewhat different from the NPRM's consideration. In the interest of brevity, this preamble will hold further discussion of that point, along with responses to comments received, until Section VIII.

(d) The Need of the United States To Conserve Energy

“The need of the United States to conserve energy” means “the consumer cost, national balance of payments, environmental, and foreign policy implications of our need for large quantities of petroleum, especially imported petroleum.” [66] Environmental implications principally include changes in emissions of carbon dioxide and criteria pollutants and air toxics. Prime examples of foreign policy implications are energy independence and security concerns.

(1) Consumer Costs and Fuel Prices

Fuel for vehicles costs money for vehicle owners and operators. All else equal (and this is an important qualification), consumers benefit from vehicles that need less fuel to perform the same amount of work. Future fuel prices are a critical input into the economic analysis of potential CAFE standards because they determine the value of fuel savings both to new vehicle buyers and to society, the amount of fuel economy that the new vehicle market is likely to demand in the absence of new standards, and they inform NHTSA about the consumer cost of the nation's need for large quantities of petroleum. In this final rule, NHTSA's analysis relies on fuel price projections estimated using the version of NEMS used for the U.S. Energy Information Administration's (EIA) Annual Energy Outlook for 2019.[67] Federal government agencies generally use EIA's price projections in their assessment of future energy-related policies.

(2) National Balance of Payments

Historically, the need of the United States to conserve energy has included consideration of the “national balance of payments” because of concerns that importing large amounts of oil created a Start Printed Page 24215significant wealth transfer to oil-exporting countries and left the U.S. economically vulnerable.[68] As recently as 2009, nearly half of the U.S. trade deficit was driven by petroleum,[69] yet this concern has largely lain fallow in more recent CAFE actions, in part because other factors besides petroleum consumption have since played a bigger role in the U.S. trade deficit.[70] Given significant recent increases in U.S. oil production and corresponding decreases in oil imports, this concern seems likely to remain fallow for the foreseeable future.[71] Increasingly, changes in the price of fuel have come to represent transfers between domestic consumers of fuel and domestic producers of petroleum rather than gains or losses to foreign entities.

As flagged in the NPRM, some commenters raised concerns about potential economic consequences for automaker and supplier operations in the U.S. due to disparities between CAFE standards at home and their counterpart fuel economy/efficiency and CO2 standards abroad. NHTSA finds these concerns more relevant to technological feasibility and economic practicability considerations than to the national balance of payments. The discussion in Section VIII below addresses this topic in more detail.

(3) Environmental Implications

Higher fleet fuel economy can reduce U.S. emissions of various pollutants by reducing the amount of oil that is produced and refined for the U.S. vehicle fleet, but can also increase emissions by reducing the cost of driving, which can result in more vehicle miles traveled (i.e., the rebound effect). Thus, the net effect of more stringent CAFE standards on emissions of each pollutant depends on the relative magnitude of both its reduced emissions in fuel refining and distribution and increases in its emissions from vehicle use. Fuel savings from CAFE standards also necessarily results in lower emissions of CO2, the main greenhouse gas emitted as a result of refining, distributing, and using transportation fuels. Reducing fuel consumption directly reduces CO2 emissions because the primary source of transportation-related CO2 emissions is fuel combustion in internal combustion engines.

NHTSA has considered environmental issues, both within the context of EPCA and the context of the National Environmental Policy Act (NEPA), in making decisions about the setting of standards since the earliest days of the CAFE program. As courts of appeal have noted in three decisions stretching over the last 20 years,[72] NHTSA defined “the need of the United States to conserve energy” in the late 1970s as including, among other things, environmental implications. In 1988, NHTSA included climate change concepts in its CAFE notices and prepared its first environmental assessment addressing that subject.[73] It cited concerns about climate change as one of its reasons for limiting the extent of its reduction of the CAFE standard for MY 1989 passenger cars.[74] Since then, NHTSA has considered the effects of reducing tailpipe emissions of CO2 in its fuel economy rulemakings pursuant to the need of the United States to conserve energy by reducing petroleum consumption.

(4) Foreign Policy Implications

U.S. consumption and imports of petroleum products can impose additional costs (i.e., externalities) on the domestic economy that are not reflected in the market price for crude petroleum or in the prices paid by consumers for petroleum products such as gasoline. NHTSA has said previously that these costs can include (1) higher prices for petroleum products resulting from the effect of U.S. oil demand on world oil prices, (2) the risk of disruptions to the U.S. economy caused by sudden increases in the global price of oil and its resulting impact on fuel prices faced by U.S. consumers, and (3) expenses for maintaining the strategic petroleum reserve (SPR) to provide a response option should a disruption in commercial oil supplies threaten the U.S. economy, to allow the U.S. to meet part of its International Energy Agency obligation to maintain emergency oil stocks, and to provide a national defense fuel reserve.[75] Higher U.S. consumption of crude oil or refined petroleum products increases the magnitude of these external economic costs, thus increasing the true economic cost of supplying transportation fuels above the resource costs of producing them. Conversely, reducing U.S. consumption of crude oil or refined petroleum products (by reducing motor fuel use) can reduce these external costs.

While these costs are considerations, the United States has significantly increased oil production capabilities in recent years, to the extent that the U.S. is currently producing enough oil to satisfy nearly all of its energy needs and is projected to continue to do so (or even become a net energy exporter in the near future).[76] This has added stable new supply to the global oil market, which ameliorates the U.S.' need to Start Printed Page 24216conserve energy from a security perspective even given that oil is a global commodity. The agencies discuss this issue in more detail in Section VIII below.

(2) Factors That NHTSA Is Prohibited From Considering

EPCA states that in determining the level at which it should set CAFE standards for a particular model year, NHTSA may not consider the ability of manufacturers to take advantage of several EPCA provisions that facilitate compliance with CAFE standards and thereby can reduce their costs of compliance.[77] As discussed further below, NHTSA cannot consider compliance credits that manufacturers earn by exceeding the CAFE standards and then use to achieve compliance in years in which their measured average fuel economy falls below the standards. NHTSA also cannot consider the use of alternative fuels by dual-fueled vehicles (such as plug-in hybrid electric vehicles) nor the availability of dedicated alternative fuel vehicles (such as battery electric or hydrogen fuel cell vehicles) in any model year. EPCA encourages the production of alternative fuel vehicles by specifying that their fuel economy is to be determined using a special calculation procedure that results in those vehicles being assigned a higher fuel economy level than they actually achieve. For non-statutory incentives that NHTSA developed by regulation, NHTSA does not consider these incentives subject to the EPCA prohibition on considering flexibilities. These topics will be addressed further in Section VIII below.

(3) Other Considerations in Determining Maximum Feasible CAFE Standards

NHTSA historically has interpreted EPCA's statutory factors as including consideration for potential adverse safety consequences in setting CAFE standards. Courts have consistently recognized that this interpretation is reasonable. As courts have recognized, “NHTSA has always examined the safety consequences of the CAFE standards in its overall consideration of relevant factors since its earliest rulemaking under the CAFE program.” [78] The courts have consistently upheld NHTSA's implementation of EPCA in this manner.[79] Thus, in evaluating what levels of stringency would result in maximum feasible standards, NHTSA assesses the potential safety impacts and considers them in balancing the statutory considerations and to determine the maximum feasible level of the standards.[80] Many commenters addressed the NPRM's analysis of safety impacts; those comments will be summarized and responded to in Section VI.D.2 and also in each agency's discussion in Section VIII.

The above sections explain what Congress thought was important enough to codify when it directed each agency to regulate, and begin to explain how the agencies have interpreted those directions over time and in this final rule. The next section looks more closely at the interplay between Congress's direction to the agencies and the aspects of the market that these regulations affect, as follows.

IV. Purpose of Analytical Approach Considered as Part of Decision-Making

A. Relationship of Analytical Approach to Governing Law

Like the NPRM, today's final rule is supported by extensive analysis of potential impacts of the regulatory alternatives under consideration. Below, Section VI reviews the analytical approach, Section VII summarizes the results of the analysis, and Section VIII explains how the final standards—informed by this analysis—fulfill the agencies' statutory obligations. Accompanying today's notice, a final Regulatory Impact Analysis (FRIA) and, for NHTSA's consideration, a final Environmental Impact Analysis (FEIS), together provide a more extensive and detailed enumeration of related methods, estimates, assumptions, and results. The agencies' analysis has been constructed specifically to reflect various aspects of governing law applicable to CAFE and CO2 standards, and has been expanded and improved in response to comments received to the NPRM and based on additional work by the agencies. The analysis aided the agencies in implementing their statutory obligations, including the weighing of competing considerations, by reasonably informing the agencies about the estimated effects of choosing different regulatory alternatives.

The agencies' analysis makes use of a range of data (i.e., observations of things that have occurred), estimates (i.e., things that may occur in the future), and models (i.e., methods for making estimates). Two examples of data include (1) records of actual odometer readings used to estimate annual mileage accumulation at different vehicle ages and (2) CAFE compliance data used as the foundation for the “analysis fleet” containing, among other things, production volumes and fuel economy levels of specific configurations of specific vehicle models produced for sale in the U.S. Two examples of estimates include (1) forecasts of future GDP growth used, with other estimates, to forecast future vehicle sales volumes and (2) the “retail price equivalent” (RPE) factor used to estimate the ultimate cost to consumers of a given fuel-saving technology, given accompanying estimates of the technology's “direct cost,” as adjusted to account for estimated “cost learning effects” (i.e., the tendency that it will cost a manufacturer less to apply a technology as the manufacturer gains more experience doing so).

The agencies' analysis makes use of several models, some of which are actually integrated systems of multiple models. As discussed in the NPRM, the agencies' analysis of CAFE and CO2 standards involves two basic elements: First, estimating ways each manufacturer could potentially respond to a given set of standards in a manner that considers potential consumer response; and second, estimating various impacts of those responses. Estimating manufacturers' potential responses involves simulating manufacturers' decision-making processes regarding the year-by-year application of fuel-saving technologies to specific vehicles. Estimating impacts involves calculating resultant changes in new vehicle costs, estimating a Start Printed Page 24217variety of costs (e.g., for fuel) and effects (e.g., CO2 emissions from fuel combustion) occurring as vehicles are driven over their lifetimes before eventually being scrapped, and estimating the monetary value of these effects. Estimating impacts also involves consideration of the response of consumers—e.g., whether consumers will purchase the vehicles and in what quantities. Both of these basic analytical elements involve the application of many analytical inputs.

The agencies' analysis uses the CAFE Model to estimate manufacturers' potential responses to new CAFE and CO2 standards and to estimate various impacts of those responses. The model may be characterized as an integrated system of models. For example, one model estimates manufacturers' responses, another estimates resultant changes in total vehicle sales, and still another estimates resultant changes in fleet turnover (i.e., scrappage). The CAFE model makes use of many inputs, values of which are developed outside of the model and not by the model. For example, the model applies fuel prices; it does not estimate fuel prices. The model does not determine the form or stringency of the standards; instead, the model applies inputs specifying the form and stringency of standards to be analyzed and produces outputs showing effects of manufacturers working to meet those standards, which become the basis for comparing between different potential stringencies.

The agencies also use EPA's MOVES model to estimate “tailpipe” (a.k.a. “vehicle” or “downstream”) emission factors for criteria pollutants,[81] and use four DOE and DOE-sponsored models to develop inputs to the CAFE model, including three developed and maintained by DOE's Argonne National Laboratory. The agencies use the DOE Energy Information Administration's (EIA's) National Energy Modeling System (NEMS) to estimate fuel prices,[82] and use Argonne's Greenhouse gases, Regulated Emissions, and Energy use in Transportation (GREET) model to estimate emissions rates from fuel production and distribution processes.[83] DOT also sponsored DOE/Argonne to use Argonne's Autonomie full-vehicle modeling and simulation system to estimate the fuel economy impacts for roughly a million combinations of technologies and vehicle types.[84 85] Section VI.B.3, below, and the accompanying final RIA document details of the agencies' use of these models. In addition, as discussed in the final EIS accompanying today's notice, DOT relied on a range of climate and photochemical models to estimate impacts on climate, air quality, and public health. The EIS discusses and documents the use of these models.

As further explained in the NPRM,[86] to prepare for analysis supporting the proposal, DOT expanded the CAFE model to address EPA statutory and regulatory requirements through a year-by-year simulation of how manufacturers could comply with EPA's CO2 standards, including:

  • Calculation of vehicle models' CO2 emission rates before and after application of fuel-saving (and, therefore, CO2-reducing) technologies;
  • Calculation of manufacturers' fleet average CO2 emission rates;
  • Calculation of manufacturers' fleet average CO2 emission rates under attribute-based CO2 standards;
  • Accounting for adjustments to average CO2 emission rates reflecting reduction of air conditioner refrigerant leakage;
  • Accounting for the treatment of alternative fuel vehicles for CO2 compliance;
  • Accounting for production “multipliers” for PHEVs, BEVs, compressed natural gas (CNG) vehicles, and fuel cell vehicles (FCVs);
  • Accounting for transfer of CO2 credits between regulated fleets; and
  • Accounting for carried-forward (a.k.a. “banked”) CO2 credits, including credits from model years earlier than modeled explicitly.

As further discussed in the NPRM, although EPA had previously developed a vehicle simulation tool (“ALPHA”) and a fleet compliance model (“OMEGA”), and had applied these in prior actions, having considered the facts before the Agency in 2018, EPA determined that, “it is reasonable and appropriate to use DOE/Argonne's model for full-vehicle simulation, and to use DOT's CAFE model for analysis of regulatory alternatives.” [87]

As discussed below and in Section VI.B.3, some commenters—some citing deliberative EPA staff communications during NPRM development, and one submitting comments by a former EPA staff member closely involved in the origination of the above-mentioned OMEGA model—took strong exception to EPA's decision to rely on DOE/Argonne and DOT-originated models as the basis for analysis informing EPA's decisions regarding CO2 standards. Some commenters argued that the EPA Administrator must consider exclusively models and analysis originating with EPA staff, and that to do otherwise would be arbitrary and capricious. As explained below (and as explained in the NPRM), it is reasonable for the Administrator to consider analysis and information produced from many sources, including, in this instance, the DOE/Argonne and DOT models. The Administrator has the discretion to determine what information reasonably and appropriately informs decisions regarding emissions standards. Some commenters conflated models with decisions, suggesting that the former mechanically determine the latter. The CAA authorizes the EPA Administrator, not a model, to make decisions about emissions standards, just as EPCA provides similar authority to the Secretary. Models produce analysis, the results of which help to inform decisions. However, in making such decisions, the Administrator may and should consider other relevant information beyond the outputs of any models—including public comment—and, in all cases, must exercise judgment in establishing appropriate standards.

Some commenters conflated models with inputs and/or with results of the modeling. All of the models mentioned above rely on inputs, including not only data (i.e., facts), but also estimates (inputs about the future are estimates, not data). Given these inputs, the models produce estimates—ultimately, the agencies' reported estimates of the potential impacts of standards under Start Printed Page 24218consideration. In other words, inputs do not define models; models use inputs. Therefore, disagreements about inputs do not logically extend to disagreements about models. Similarly, while models determine resulting outputs, they do so based on inputs. Therefore, disagreements about results do not necessarily imply disagreements about models; they may merely reflect disagreements about inputs. With respect to the Administrator's decisions regarding models underlying today's analysis, comments regarding inputs, therefore, are more appropriately addressed separately, which is done so below in Section VI.

The EPA Administrator's decision to continue relying on the DOE/Argonne Autonomie tool and DOT CAFE model rather than on the corresponding tools developed by EPA staff is informed by consideration of comments on results and on technical aspects of the models themselves. As discussed below, some commenters questioned specific aspects of the CAFE model's simulation of manufacturer's potential responses to CO2 standards. Considering these comments, the CAFE model applied in the final rule's analysis includes some revisions and updates. For example, the “effective cost” metric used to select among available opportunities to apply fuel-saving technologies now uses a “cost per credit” metric rather than the metric used for the NPRM. Also, the model's representation of sales “multipliers” EPA has included for CNG vehicles, PHEVs, BEVs, and FCVs reflects current EPA regulations or, as an input-selectable option, an alternative approach under consideration. On the other hand, some commenters questioning the CAFE model's approach to some CO2 program features appear to ignore the fact that prior analysis by EPA (using EPA's OMEGA) model likewise did not account for the same program features. For example, some stakeholders took issue with the CAFE model's approach to accounting for banked CO2 credits and, in particular, credits banked prior to the model years accounted for explicitly in the analysis. In the course of updating the basis for analysis fleet from model year 2016 to model year 2017, the agencies have since updated corresponding inputs. However, even though the ability to carry forward credits impacts outcomes, EPA's OMEGA model used in previous rulemakings never attempted to account for credit banking and, indeed, lacking a year-by-year structure, cannot account for credit banking. Therefore, at least with respect to this important CO2 program flexibility, the CAFE model provides a more complete and realistic basis for estimating actual impacts of new CO2 standards.

For its part, NHTSA remains confident that the combination of the Autonomie and CAFE models remains the best available for CAFE rulemaking analysis, and notes, as discussed below, that even the environmental group coalition stated that the CAFE model is aligned with EPCA requirements.[88] In late 2001, after Congress discontinued an extended series of budget “riders” prohibiting work on CAFE standards, NHTSA and the DOT Volpe Center began development of a modeling system appropriate for CAFE rulemaking analysis, because other available models were not designed with this purpose in mind, and lacked capabilities important for CAFE rulemakings. For example, although NEMS had procedures to account for CAFE standards, those procedures did not provide the ability to account for specific manufacturers, as is especially relevant to the statutory requirement that NHTSA consider the economic practicability of any new CAFE standards. Also, as early as the first rulemaking making use of this early CAFE model, commenters stressed the importance of product redesign schedules, leading developers to introduce procedures to account for product cadence. In the 2003 notice regarding light truck standards for MYs 2005-2007, NHTSA stated that “we also changed the methodology to recognize that capital costs require employment of technologies for several years, rather than a single year. . . . In our view, this makes the Volpe analysis more consistent with the [manually implemented] Stage analysis and better reflects actual conditions in the automotive industry.” [89] Since that time, NHTSA and the Volpe Center have significantly refined the CAFE model with each of rulemaking. For example, for the 2006 rulemaking regarding standards for MYs 2008-2011 light trucks, NHTSA introduced the ability to account for attribute-based standards, account for the social cost of CO2 emissions, estimate stringencies at which net benefits would be maximized, and perform probabilistic uncertainty analysis (i.e., Monte Carlo simulation).[90] For the 2009 rulemaking regarding standards for MY 2011 passenger cars and light trucks, we introduced the ability to account for attribute-based passenger car standards, and the ability to apply “synergy factors” to estimate how some technology pairings impact fuel consumption,[91] For the 2010 rulemaking regarding standards for MYs 2012-2016, we introduced procedures to account for FFV credits, and to account for product planning as a multiyear consideration.[92] For the 2012 rulemaking regarding standards for MYs 2017-2025, we introduced several new procedures, such as (1) accounting for electricity used to charge electric vehicles (EVs) and plug-in hybrid electric vehicles (PHEVs), (2) accounting for use of ethanol blends in flexible-fuel vehicles (FFVs), (3) accounting for costs (i.e., “stranded capital”) related to early replacement of technologies, (4) accounting for previously-applied technology when determining the extent to which a manufacturer could expand use of the technology, (5) applying technology-specific estimates of changes in consumer value, (6) simulating the extent to which manufacturers might utilize EPCA's provisions regarding generation and use of CAFE credits, (7) applying estimates of fuel economy adjustments (and accompanying costs) reflecting increases in air conditioner efficiency, (8) reporting privately-valued benefits, (9) simulating the extent to which manufacturers might voluntarily apply technology beyond levels needed for compliance with CAFE standards, and (10) estimating changes in highway fatalities attributable to any applied reductions in vehicle mass.[93] Also for the 2012 rulemaking, we began making use of Autonomie to estimate fuel consumption impacts of different combinations of technologies, using these estimates to specify inputs to the CAFE model.[94] In 2016, providing analyses for both the draft TAR regarding light-duty CAFE standards and the final rule regarding fuel consumption standards for heavy-duty pickup trucks and vans, we greatly expanded the agency's use of Autonomie-based full vehicle simulations and introduced the ability to simulate compliance with attribute-based standards for heavy-duty pickups and vans.[95] And, as discussed at length in the NPRM and below, for this rulemaking, we have, among other things, refined procedures to account for impacts on highway travel and safety, Start Printed Page 24219added procedures to simulate compliance with CO2 standards, refined procedures to account for compliance credits, and added procedures to account for impacts on sales, scrappage, and employment. We have also significantly revised the model's graphical user interface (GUI) in order to make the model easier to operate and understand. Like any model, both Autonomie and the CAFE model benefit from ongoing refinement. However, NHTSA is confident that this combination of models produces a more realistic characterization of the potential impacts of new standards than would another combination of available models. Some stakeholders, while commenting on specific aspects of the inputs, models, and/or results, commended the agencies' exclusive reliance on the DOE/Argonne Autonomie tool and DOT CAFE model. With respect to CO2 standards, these stakeholders noted not only technical reasons to use these models rather than the EPA models, but also other reasons such as efficiency, transparency, and ease with which outside parties can exercise models and replicate the agencies' analysis. These comments are discussed below and in Section VI.

Nevertheless, some comments regarding the model's handling of CAFE and/or CO2 standards, and some comments regarding the model's estimation of resultant impacts, led the agencies to make changes to specific aspects of the model. Comments on and changes to the inputs and model are discussed below and in Section VI; results are discussed in Section VII and in the accompanying RIA; and the meaning of results in the context of the applicable statutory requirements is discussed in Section VIII.

As explained, the analysis is designed to reflect a number of statutory and regulatory requirements applicable to CAFE and tailpipe CO2 standard setting. EPCA contains a number of requirements governing the scope and nature of CAFE standard setting. Among these, some have been in place since EPCA was first signed into law in 1975, and some were added in 2007, when Congress passed EISA and amended EPCA. The CAA, as discussed elsewhere, provides EPA with very broad authority under Section 202(a), and does not contain EPCA/EISA's prescriptions. In the interest of harmonization, however, EPA has adopted some of the EPCA/EISA requirements into its tailpipe CO2 regulations, and NHTSA, in turn, has created some additional flexibilities by regulation not expressly envisioned by EPCA/EISA in order to harmonize better with some of EPA's programmatic decisions. EPCA/EISA requirements regarding the technical characteristics of CAFE standards and the analysis thereof include, but are not limited to, the following, and the analysis reflects these requirements as summarized:

Corporate Average Standards: 49 U.S.C. 32902 requires standards that apply to the average fuel economy levels achieved by each corporation's fleets of vehicles produced for sale in the U.S.[96] CAA Section 202(a) does not preclude the EPA Administrator from expressing CO2 standards as de facto fleet average requirements, and EPA has adopted a similar approach in the interest of harmonization. The CAFE Model, used by the agencies to conduct the bulk of today's analysis, calculates the CAFE and CO2 levels of each manufacturer's fleets based on estimated production volumes and characteristics, including fuel economy levels, of distinct vehicle models that could be produced for sale in the U.S.

Separate Standards for Passenger Cars and Light Trucks: 49 U.S.C. 32902 requires the Secretary of Transportation to set CAFE standards separately for passenger cars and light trucks. CAA Section 202(a) does not preclude the EPA Administrator from specifying CO2 standards separately for passenger cars and light trucks, and EPA has adopted a similar approach. The CAFE Model accounts separately for passenger cars and light trucks, including differentiated standards and compliance.

Attribute-Based Standards: 49 U.S.C. 32902 requires the Secretary of Transportation to define CAFE standards as mathematical functions expressed in terms of one or more vehicle attributes related to fuel economy. This means that for a given manufacturer's fleet of vehicles produced for sale in the U.S. in a given regulatory class and model year, the applicable minimum CAFE requirement (i.e., the numerical value of the requirement) is computed based on the applicable mathematical function, and the mix and attributes of vehicles in the manufacturer's fleet. In the 2012 final rule that first established CO2 standards, EPA also adopted an attribute-based standard under its broad CAA Section 202(a) authority. The CAFE Model accounts for such functions and vehicle attributes explicitly.

Separately Defined Standards for Each Model Year: 49 U.S.C. 32902 requires the Secretary to set CAFE standards (separately for passenger cars and light trucks) at the maximum feasible levels in each model year. CAA Section 202(a) allows EPA to establish CO2 standards separately for each model year, and EPA has chosen to do so for this final rule, similar to the approach taken in the previous light-duty vehicle CO2 standard-setting rules. The CAFE Model represents each model year explicitly, and accounts for the production relationships between model years.[97]

Separate Compliance for Domestic and Imported Passenger Car Fleets: 49 U.S.C. 32904 requires the EPA Administrator to determine CAFE compliance separately for each manufacturers' fleets of domestic passenger cars and imported passenger cars, which manufacturers must consider as they decide how to improve the fuel economy of their passenger car fleets. CAA 202(a) does not preclude the EPA Administrator from determining compliance with CO2 standards separately for a manufacturer's domestic and imported car fleets, but EPA did not include such a distinction in either the 2010 or 2012 final rules, and EPA did not propose or ask for comment on taking such an approach in the proposal. The CAFE Model is able to account explicitly for this requirement when simulating manufacturers' potential responses to CAFE standards, but combines any given manufacturer's domestic and imported cars into a single fleet when simulating that manufacturer's potential response to CO2 standards.

Minimum CAFE Standards for Domestic Passenger Car Fleets: 49 U.S.C. 32902 requires that domestic passenger car fleets achieve CAFE levels no less than 92 percent of the industry-wide average level required under the applicable attribute-based CAFE standard, as projected by the Secretary at the time the standard is promulgated. CAA 202(a) does not preclude the EPA Administrator from correspondingly requiring that domestic passenger car fleets achieve CO2 levels no greater than 108.7 percent (1/0.92 = 1.087) of the projected industry-wide average CO2Start Printed Page 24220requirement under the attribute-based standard, but the GHG program that EPA designed in the 2010 and 2012 final rules did not include such a distinction, and EPA did not propose or seek comment on such an approach in the proposal. The CAFE Model is able to account explicitly for this requirement for CAFE standards, and sets this requirement aside for CO2 standards.

Civil Penalties for Noncompliance: 49 U.S.C. 32912 prescribes a rate (in dollars per tenth of a mpg) at which the Secretary is to levy civil penalties if a manufacturer fails to comply with a CAFE standard for a given fleet in a given model year, after considering available credits. Some manufacturers have historically demonstrated a willingness to treat CAFE noncompliance as an “economic” choice, electing to pay civil penalties rather than achieving full numerical compliance across all fleets. The CAFE Model calculates civil penalties for CAFE shortfalls and provides means to estimate that a manufacturer might stop adding fuel-saving technologies once continuing to do so would be effectively more “expensive” (after accounting for fuel prices and buyers' willingness to pay for fuel economy) than paying civil penalties. In contrast, the CAA does not authorize the EPA Administrator to allow manufacturers to sell noncompliant fleets and instead only pay civil penalties; manufacturers who choose to pay civil penalties for CAFE compliance tend to employ EPA's more-extensive programmatic flexibilities to meet tailpipe CO2 emissions standards. Thus, the CAFE Model does not allow civil penalty payment as an option for CO2 standards.

Dual-Fueled and Dedicated Alternative Fuel Vehicles: For purposes of calculating CAFE levels used to determine compliance, 49 U.S.C. 32905 and 32906 specify methods for calculating the fuel economy levels of vehicles operating on alternative fuels to gasoline or diesel through MY 2020. After MY 2020, methods for calculating alternative fuel vehicle (AFV) fuel economy are governed by regulation. The CAFE Model is able to account for these requirements explicitly for each vehicle model. However, 49 U.S.C. 32902 requires that maximum feasible CAFE standards be set in a manner that does not presume manufacturers can respond by producing new dedicated alternative fuel vehicle (AFV) models. The CAFE model can be run in a manner that excludes the additional application of dedicated AFV technologies in model years for which maximum feasible standards are under consideration. As allowed under NEPA for analysis appearing in EISs informing decisions regarding CAFE standards, the CAFE Model can also be run without this analytical constraint. CAA 202(a) does not preclude the EPA Administrator adopting analogous provisions, but EPA has instead opted through regulation to “count” dual- and alternative fuel vehicles on a CO2 basis (and through MY 2026, to set aside emissions from electricity generation). The CAFE model accounts for this treatment of dual- and alternative fuel vehicles when simulating manufacturers' potential responses to CO2 standards. For natural gas vehicles, both dedicated and dual-fueled, EPA is establishing a multiplier of 2.0 for model years 2022-2026.

Creation and Use of Compliance Credits: 49 U.S.C. 32903 provides that manufacturers may earn CAFE “credits” by achieving a CAFE level beyond that required of a given fleet in a given model year, and specifies how these credits may be used to offset the amount by which a different fleet falls short of its corresponding requirement. These provisions allow credits to be “carried forward” and “carried back” between model years, transferred between regulated classes (domestic passenger cars, imported passenger cars, and light trucks), and traded between manufacturers. However, these provisions also impose some specific statutory limits. For example, CAFE compliance credits can be carried forward a maximum of five model years and carried back a maximum of three model years. Also, EPCA/EISA caps the amount of credit that can be transferred between passenger car and light truck fleets, and prohibits manufacturers from applying traded or transferred credits to offset a failure to achieve the applicable minimum standard for domestic passenger cars. The CAFE Model explicitly simulates manufacturers' potential use of credits carried forward from prior model years or transferred from other fleets.[98] 49 U.S.C. 32902 prohibits consideration of manufacturers' potential application of CAFE compliance credits when setting maximum feasible CAFE standards. The CAFE Model can be operated in a manner that excludes the application of CAFE credits after a given model year. CAA 202(a) does not preclude the EPA Administrator adopting analogous provisions. EPA has opted to limit the “life” of compliance credits from most model years to 5 years, and to limit borrowing to 3 years, but has not adopted any limits on transfers (between fleets) or trades (between manufacturers) of compliance credits. The CAFE Model is able to account for the absence of limits on transfers of CO2 standards. Insofar as the CAFE model can be exercised in a manner that simulates trading of CO2 compliance credits, such simulations treat trading as unlimited.[99] EPA has considered manufacturers' ability to use credits as part of its decisions on these final standards, and the CAFE model is now able to account for that.

Statutory Basis for Stringency: 49 U.S.C. 32902 requires the Secretary to set CAFE standards at the maximum feasible levels, considering technological feasibility, economic practicability, the need of the Nation to conserve energy, and the impact of other government standards. EPCA/EISA authorizes the Secretary to interpret Start Printed Page 24221these factors, and as the Department's interpretation has evolved, NHTSA has continued to expand and refine its qualitative and quantitative analysis. For example, as discussed below in Section VI.B.3, the Autonomie simulations reflect the agencies' judgment that it would not be economically practicable for a manufacturer to “split” an engine shared among many vehicle model/configurations into a myriad of versions each optimized to a single vehicle model/configuration. Also responding to evolving interpretation of these EPCA/EISA factors, the CAFE Model has been expanded to address additional impacts in an integrated manner. For example, the CAFE Model version used for the NPRM analysis included the ability to estimate impacts on labor utilization internally, rather than as an external “off model” or “post processing” analysis. In addition, NEPA requires the Secretary to issue an EIS that documents the estimated impacts of regulatory alternatives under consideration. The EIS accompanying today's notice documents changes in emission inventories as estimated using the CAFE model, but also documents corresponding estimates—based on the application of other models documented in the EIS, of impacts on the global climate, on tropospheric air quality, and on human health. Regarding CO2 standards, CAA 202(a) provides general authority for the establishment of motor vehicle emissions standards, and the final rule's analysis, like that accompanying the agencies' proposal, addresses impacts relevant to the EPA Administrator's decision making, such as technological feasibility, air quality impacts, costs to industry and consumers, and lead time necessary for compliance.

Other Factors: Beyond these statutory requirements applicable to DOT and/or EPA are a number of specific technical characteristics of CAFE and/or CO2 regulations that are also relevant to the construction of today's analysis. These are discussed at greater length in Section II.F. For example, EPA has defined procedures for calculating average CO2 levels, and has revised procedures for calculating CAFE levels, to reflect manufacturers' application of “off-cycle” technologies that increase fuel economy (and reduce CO2 emissions) in ways not reflected by the long-standing test procedures used to measure fuel economy. Although too little information is available to account for these provisions explicitly in the same way that the agencies have accounted for other technologies, the CAFE Model does include and makes use of inputs reflecting the agencies' expectations regarding the extent to which manufacturers may earn such credits, along with estimates of corresponding costs. Similarly, the CAFE Model includes and makes use of inputs regarding credits EPA has elected to allow manufacturers to earn toward CO2 levels (not CAFE) based on the use of air conditioner refrigerants with lower global warming potential (GWP), or on the application of technologies to reduce refrigerant leakage. In addition, EPA has elected to provide that through model year 2021, manufacturers may apply “multipliers” to plug-in hybrid electric vehicles, dedicated electric vehicles, fuel cell vehicles, and hydrogen vehicles, such that when calculating a fleet's average CO2 levels (not CAFE), the manufacturer may, for example, “count” each electric vehicle twice. The CAFE Model accounts for these multipliers, based on either current regulatory provisions or on alternative approaches. Although these are examples of regulatory provisions that arise from the exercise of discretion rather than specific statutory mandate, they can materially impact outcomes. Section VI.B explains in greater detail how today's analysis addresses them.

Benefits of Analytical Approach

The agencies' analysis of CAFE and CO2 standards involves two basic elements: First, estimating ways each manufacturer could potentially respond to a given set of standards in a manner that considers potential consumer response; and second, estimating various impacts of those responses. Estimating manufacturers' potential responses involves simulating manufacturers' decision-making processes regarding the year-by-year application of fuel-saving technologies to specific vehicles. Estimating impacts involves calculating resultant changes in new vehicle costs, estimating a variety of costs (e.g., for fuel) and effects (e.g., CO2 emissions from fuel combustion) occurring as vehicles are driven over their lifetimes before eventually being scrapped, and estimating the monetary value of these effects. Estimating impacts also involves consideration of the response of consumers—e.g., whether consumers will purchase the vehicles and in what quantities. Both of these basic analytical elements involve the application of many analytical inputs.

As mentioned above, the agencies' analysis uses the CAFE model to estimate manufacturers' potential responses to new CAFE and CO2 standards and to estimate various impacts of those responses. DOT's Volpe National Transportation Systems Center (often simply referred to as the “Volpe Center”) develops, maintains, and applies the model for NHTSA. NHTSA has used the CAFE model to perform analyses supporting every CAFE rulemaking since 2001, and the 2016 rulemaking regarding heavy-duty pickup and van fuel consumption and CO2 emissions also used the CAFE model for analysis.[100]

NHTSA recently arranged for a formal peer review of the model. In general, reviewers' comments strongly supported the model's conceptual basis and implementation, and commenters provided several specific recommendations. The agency agreed with many of these recommendations and has worked to implement them wherever practicable. Implementing some of the recommendations would require considerable further research, development, and testing, and will be considered going forward. For a handful of other recommendations, the agency disagreed, often finding the recommendations involved considerations (e.g., other policies, such as those involving fuel taxation) beyond the model itself or were based on concerns with inputs rather than how the model itself functioned. A report available in the docket for this rulemaking presents peer reviewers' detailed comments and recommendations, and provides DOT's detailed responses.[101]

As also mentioned above, the agencies use EPA's MOVES model to estimate tailpipe emission factors, use DOE/EIA's NEMS to estimate fuel prices,[102] and use Argonne's GREET model to estimate downstream emissions rates.[103] DOT also sponsored DOE/Argonne to use the Autonomie full-vehicle modeling and simulation tool to estimate the fuel economy impacts for roughly a million Start Printed Page 24222combinations of technologies and vehicle types.[104 105]

EPA developed two models after 2009, referred to as the “ALPHA” and “OMEGA” models, which provide some of the same capabilities as the Autonomie and CAFE models. EPA applied the OMEGA model to conduct analysis of tailpipe CO2 emissions standards promulgated in 2010 and 2012, and the ALPHA and OMEGA models to conduct analysis discussed in the above-mentioned 2016 Draft TAR and Proposed and 2017 Initial Final Determinations regarding standards beyond 2021. In an August 2017 notice, the agencies requested comments on, among other things, whether EPA should use alternative methodologies and modeling, including DOE/Argonne's Autonomie full-vehicle modeling and simulation tool and DOT's CAFE model.[106]

Having reviewed comments on the subject and having considered the matter fully, the agencies have determined it is reasonable and appropriate to use DOE/Argonne's model for full-vehicle simulation, and to use DOT's CAFE model for analysis of regulatory alternatives. EPA interprets Section 202(a) of the CAA as giving the agency broad discretion in how it develops and sets CO2 emissions standards for light-duty vehicles. Nothing in Section 202(a) mandates that EPA use any specific model or set of models for analysis of potential CO2 standards for light-duty vehicles. EPA weighs many factors when determining appropriate levels for CO2 standards, including the cost of compliance (see Section 202(a)(2)), lead time necessary for compliance (id.), safety (see NRDC v. EPA, 655 F.2d 318, 336 n. 31 (D.C. Cir. 1981)) and other impacts on consumers,[107] and energy impacts associated with use of the technology.[108] Using the CAFE model allows consideration of a number of factors. The CAFE model explicitly evaluates the cost of compliance for each manufacturer, each fleet, and each model year; it accounts for lead time necessary for compliance by directly incorporating estimated manufacturer production cycles for every vehicle in the fleet, ensuring that the analysis does not assume vehicles can be redesigned to incorporate more technology without regard to lead time considerations; it provides information on safety effects associated with different levels of standards and information about many other impacts on consumers, and it calculates energy impacts (i.e., fuel saved or consumed) as a primary function, besides being capable of providing information about many other factors within EPA's broad CAA discretion to consider.

Because the CAFE model simulates a wide range of actual constraints and practices related to automotive engineering, planning, and production, such as common vehicle platforms, sharing of engines among different vehicle models, and timing of major vehicle redesigns, the analysis produced by the CAFE model provides a transparent and realistic basis to show pathways manufacturers could follow over time in applying new technologies, which helps better assess impacts of potential future standards. Furthermore, because the CAFE model also accounts fully for regulatory compliance provisions (now including CO2 compliance provisions), such as adjustments for reduced refrigerant leakage, production “multipliers” for some specific types of vehicles (e.g., PHEVs), and carried-forward (i.e., banked) credits, the CAFE model provides a transparent and realistic basis to estimate how such technologies might be applied over time in response to CAFE or CO2 standards.

There are sound reasons for the agencies to use the CAFE model going forward in this rulemaking. First, the CAFE and CO2 fact analyses are inextricably linked. Furthermore, the analysis produced by the CAFE model and DOE/Argonne's Autonomie addresses the agencies' analytical needs. The CAFE model provides an explicit year-by-year simulation of manufacturers' application of technology to their products in response to a year-by-year progression of CAFE standards and accounts for sharing of technologies and the implications for timing, scope, and limits on the potential to optimize powertrains for fuel economy. In the real world, standards actually are specified on a year-by-year basis, not simply some single year well into the future, and manufacturers' year-by-year plans involve some vehicles “carrying forward” technology from prior model years and some other vehicles possibly applying “extra” technology in anticipation of standards in ensuing model years, and manufacturers' planning also involves applying credits carried forward between model years. Furthermore, manufacturers cannot optimize the powertrain for fuel economy on every vehicle model configuration—for example, a given engine shared among multiple vehicle models cannot practicably be split into different versions for each configuration of each model, each with a slightly different displacement. The CAFE model is designed to account for these real-world factors.

Considering the technological heterogeneity of manufacturers' current product offerings, and the wide range of ways in which the many fuel economy-improving/CO2 emissions-reducing technologies included in the analysis can be combined, the CAFE model has been designed to use inputs that provide an estimate of the fuel economy achieved for many tens of thousands of different potential combinations of fuel-saving technologies. Across the range of technology classes encompassed by the analysis fleet, today's analysis involves more than a million such estimates. While the CAFE model requires no specific approach to developing these inputs, the National Academy of Sciences (NAS) has recommended, and stakeholders have commented, that full-vehicle simulation provides the best balance between realism and practicality. DOE/Argonne has spent several years developing, applying, and expanding means to use distributed computing to exercise its Autonomie full-vehicle modeling and simulation tool over the scale necessary for realistic analysis of CAFE or average tailpipe CO2 emissions standards. This scalability and related flexibility (in terms of expanding the set of technologies to be simulated) makes Autonomie well-suited for developing inputs to the CAFE model.

In addition, DOE/Argonne's Autonomie also has a long history of development and widespread application by a much wider range of users in government, academia, and industry. Many of these users apply Start Printed Page 24223Autonomie to inform funding and design decisions. These real-world exercises have contributed significantly to aspects of Autonomie important to producing realistic estimates of fuel economy levels and CO2 emission rates, such as estimation and consideration of performance, utility, and driveability metrics (e.g., towing capability, shift business, frequency of engine on/off transitions). This steadily increasing realism has, in turn, steadily increased confidence in the appropriateness of using Autonomie to make significant investment decisions. Notably, DOE uses Autonomie for analysis supporting budget priorities and plans for programs managed by its Vehicle Technologies Office (VTO). Considering the advantages of DOE/Argonne's Autonomie model, it is reasonable and appropriate to use Autonomie to estimate fuel economy levels and CO2 emission rates for different combinations of technologies as applied to different types of vehicles.

Commenters have also suggested that the CAFE model's graphical user interface (GUI) facilitates others' ability to use the model quickly—and without specialized knowledge or training—and to comment accordingly.[109] For the NPRM, NHTSA significantly expanded and refined this GUI, providing the ability to observe the model's real-time progress much more closely as it simulates year-by-year compliance with either CAFE or CO2 standards.[110] Although the model's ability to produce realistic results is independent of the model's GUI, the CAFE model's GUI appears to have facilitated stakeholders' meaningful review and comment during the comment period.

The question of whether EPA's actions should consider and be informed by analysis using non-EPA-staff-developed modeling tools has generated considerable debate over time. Even prior to the NPRM, certain commenters had argued that EPA could not consider, in setting tailpipe CO2 emissions standards, any information derived from non-EPA-staff-developed modeling. Many of the pre-NPRM concerns focused on inputs used by the CAFE model for prior rulemaking analyses.[111 112 113] Because inputs are exogenous to any model, they do not determine whether it would be reasonable and appropriate for EPA to use NHTSA's model for analysis. Other concerns focused on certain characteristics of the CAFE model that were developed to align the model better with EPCA and EISA. The model has been revised to accommodate both EPCA/EISA and CAA analysis, as explained further below. Some commenters also argued that use of any models other than ALPHA and OMEGA for CAA analysis would constitute an arbitrary and capricious delegation of EPA's decision-making authority to NHTSA, if NHTSA models are used for analysis instead.[114] As discussed above, the CAFE Model—as with any model—is used to provide analysis, and does not result in decisions. Decisions are made by EPA in a manner that is informed by modeling outputs, sensitivity cases, public comments, any many other pieces of information.

Comments responding to the NPRM's use of the CAFE model and Autonomie rather than also (for CO2 standards) ALPHA and OMEGA were mixed. For example, the environmental group coalition stated that the CAFE model is aligned with EPCA requirements,[115] but also argued (1) that EPA is legally prohibited from “delegat[ing] technical decision-making to NHTSA;” [116] (2) that “EPA must exercise its technical and scientific expertise” to develop CO2 standards and “Anything less is an unlawful abdication of EPA's statutory responsibilities;” [117] (3) that EPA staff is much more qualified than DOT staff to conduct analysis relating to standards and has done a great deal of work to inform development of standards; [118] (4) that “The Draft TAR and 2017 Final Determination relied extensively on use of sophisticated EPA analytic tools and methodologies,” i.e., the “peer reviewed simulation model ALPHA,” “the agency's vehicle teardown studies,” and the “peer-reviewed OMEGA model to make reasonable estimates of how manufacturers could add technologies to vehicles in order to meet a fleet-wide [CO2 emissions] standard;” [119] (5) that NHTSA had said in the MYs 2012-2016 rulemaking that the Volpe [CAFE] model was developed to support CAFE rulemaking and incorporates features “that are not appropriate for use by EPA in setting [tailpipe CO2] standards;” [120] (6) allegations that some EPA staff had disagreed with aspects of the NPRM analysis and had requested that “EPA's name and logo should be removed from the DOT-NHTSA Preliminary Regulatory Impact Analysis document” and stated that “EPA is relying upon the technical analysis performed by DOT-NHTSA for the [NPRM];” [121] (7) that EPA had developed “a range of relevant new analysis” that the proposal “failed to consider,” including “over a dozen 2017 and 2018 EPA peer reviewed SAE articles;” [122] (8) that EPA's OMEGA modeling undertaken during NPRM development “found costs half that of NHTSA's findings,” “Yet NHTSA did not correct the errors in its modeling and analysis, and the published proposal drastically overestimates the cost of complying . . . .;” [123] (9) that some EPA staff had requested that the technology “HCR2” be included in the NPRM analysis, “Yet NHTSA overruled Start Printed Page 24224EPA and omitted the technology;” [124] (10) that certain EPA staff had initially “rejected use of the CAFE model for development of the proposed [tailpipe CO2] standards;” [125] (11) that there are “many specific weaknesses of the modeling results derived in this proposal through use of the Volpe and Autonomie models” and that the CAFE model is “not designed in accordance with” Section 202(a) of the CAA because (A) EPA “is not required to demonstrate that standards are set at the maximum feasible level year-by-year,” (B) because EPCA “preclude[s NHTSA] from considering vehicles powered by fuels other than gas or diesel” and EPA is not similarly bound, and (C) because the CAFE model assumed that the value of an overcompliance credit equaled $5.50, the value of a CAFE penalty.[126] Because of all of these things, the environmental group coalition stated that the proposal was “unlawful” and that “Before proceeding with this rulemaking, EPA must consider all relevant materials including these excluded insights, perform its own analysis, and issue a reproposal to allow for public comment.” [127]

Some environmental organizations and States contracted for external technical analyses augmenting general comments such as those summarized above. EDF engaged a consultant, Richard Rykowski, for a detailed review of the agencies' analysis.[128] Among Mr. Rykowski's comments, a few specifically involve differences between these two models. Mr. Rykowski recommended NHTSA's CAFE model replace its existing “effective cost” metric (used to compare available options to add specific technologies to specific vehicles) with a “ranking factor” used for the same purpose. As discussed below in Section VI.A, the model for today's final rule adopts this recommendation. He also states that (1) “EPA has developed a better way to isolate and reject cost ineffective combinations of technologies . . . [and] includes only these 50 or so technology combinations in their OMEGA model runs;” (2) “NHTSA's arbitrary and rigid designation of leader-follower vehicles for engine, transmission and platform level technologies unrealistically slows the rollout of technology into the new vehicle fleet;” (3) “the Volpe Model is not capable of reasonably simulating manufacturers' ability to utilize CO2 credits to smooth the introduction of technology throughout their vehicle line-up;” and (4) “the Volpe Model is not designed to reflect the use of these [A/C leakage] technologies and refrigerants.” [129]

Mr. Rogers's analysis focuses primarily on the agencies' published analysis, but mentions that some engine “maps” (estimates—used as inputs to full vehicle simulation—of engine fuel consumption under a wide range of engine operating conditions) applied in Autonomie show greater fuel consumption benefits of turbocharging than those applied previously by EPA to EPA's ALPHA model, and these benefits could have caused NHTSA's CAFE model to estimate an unrealistically great tendency toward turbocharged engines (rather than high compression ratio engines).[130] Mr. Rogers also presents alternative examples of year-by-year technology application to specific vehicle models, contrasting these with year-by-year results from the agencies' NPRM analysis, concluding that “that the use of logical, unrestricted technology pathways, with incremental benefits supported by industry-accepted vehicle simulation and dynamic system optimization and calibration, together with publicly-defensible costs, allows cost-effective solutions to achieve target fuel economy levels which meet MY 2025 existing standards.” [131]

Mr. Duleep's analysis also focuses primarily on the agencies' published analysis, but does mention that (1) “the Autonomie modeling assumes no engine change when drag and rolling resistance reductions are implemented, as well as no changes to the transmission gear ratios and axle ratios, . . . [but] the EPA ALPHA model adjusts for this effect;” (2) “baseline differences in fuel economy [between two manufacturers' different products using similar technologies] are carried for all future years and this exaggerates the differences in technology adoption requirements and costs between manufacturers; (3) “assumptions [that most technology changes are best applied as part of a vehicle redesign or freshening] result in unnecessary distortion in technology paths and may bias results of costs for different manufacturers;” and (4) that for the sample results shown for the Chevrolet Equinox “the publicly available EPA lumped parameter model (which was used to support the 2016 rulemaking) and 2016 TAR cost data . . . results in an estimate of attaining 52.2 mpg for a cost of $2110, which is less than half the cost estimated in the PRIA.” [132]

Beyond these comments regarding differences between EPA's models and the Argonne and DOT models applied for the NPRM, these and other technical reviewers had many specific comments about the agencies' analysis for the NPRM, and these comments are discussed in detail below in Section VI.B.

Manufacturers, on the other hand, supported the agencies' use of Autonomie and the CAFE model rather than, in EPA's case, the ALPHA and OMEGA models. Expressly identifying the distinction between models and model inputs, Global Automakers stated that:

The agencies provided a new, updated analysis based on the most up-to-date data, using a proven and long-developed modeling tool, known as the Volpe model, and offering numerous options to best determine the right regulatory and policy path for ongoing fuel efficiency improvements in our nation. Now, all stakeholders have an opportunity to come to the table as part of the public process to provide input, data, and information to help shape the final rule.[133]

This NPRM's use of a single model to evaluate alternative scenarios for both programs provides consistency in the technical analysis, and Global Automakers supports the Volpe model's use as it has proven to be a transparent and user-friendly option in this current analysis. The use of the Volpe model has allowed for a broad range of stakeholders, with varying degrees of technical expertise, to review the data inputs to provide feedback on this proposed rule. The Volpe model's accompanying documentation has historically provided a clear explanation of all sources of input and constraints critical to a transparent modeling process. Other inputs have come from modeling that is used widely by other sources, specifically the Autonomie model, allowing for a robust validation, review and reassessment.[134]

The Alliance commented, similarly, that “at least at this time, NHTSA's modeling systems are superior to EPA's” and “as such, we support the Agencies' decision to use NHTSA's modeling tools for this rulemaking and recommend that both Agencies continue on this path. We encourage Agencies to work together to provide input to the single common set of tools.” [135]

Start Printed Page 24225

Regarding the agencies' use of Argonne's Autonomie model rather than EPA's ALPHA model, the Alliance commented that (1) “the benefits of virtually all technologies and their synergistic effects are now determined with full vehicle simulations;” (2) “vehicle categories have been increased to 10 to better recognize the range of 0-60 performance characteristics within each of the 5 previous categories, in recognition of the fact that many vehicles in the baseline fleet significantly exceeded the previously assumed 0-60 performance metrics. This provides better resolution of the baseline fleet and more accurate estimates of the benefits of technology. . . .;” (3) “new technologies (like advanced cylinder deactivation) are included, while unproven combinations (like Atkinson engines with 14:1 compression, cooled EGR, and cylinder deactivation in combination) have been removed;” (4) “Consistent with the recommendation of the National Academy of Sciences and manufacturers, gradeability has been included as a performance metric used in engine sizing. This helps prevent the inclusion of small displacement engines that are not commercially viable and that would artificially inflate fuel savings;” (5) “the Alliance believes NHTSA's tools (Autonomie/Volpe) are superior to EPA's (APLHA[sic]/LPM/OMEGA). This is not surprising since NHTSA's tools have had a significant head start in development. . . .” (6) “the Autonomie model was developed at Argonne National Lab with funding from the Department of Energy going back to the PNGV (Partnership for Next Generation Vehicles) program in the 1990s. Autonomie was developed from the start to address the complex task of combining 2 power sources in a hybrid powertrain. It is a physics-based, forward looking, vehicle simulator, fully documented with available training,” and (7) “EPA's ALPHA model is also a physics-based, forward looking, vehicle simulator. However, it has not been validated or used to simulate hybrid powertrains. The model has not been documented with any instructions making it difficult for users outside of EPA to run and interpret the model.” [136]

Regarding the use of NHTSA's CAFE model rather than EPA's OMEGA model, the Alliance stated that (1) NHTSA's model appropriately differentiate between domestic and imported automobiles; (2) in NHTSA's model, “dynamic estimates of vehicle sales and scrappage in response to price changes replace unrealistic static sales and scrappage numbers;” (3) NHTSA's model “has new capability to perform [CO2 emissions] analysis with [tailpipe CO2] program flexibilities;” (4) “the baseline fleet [used in NHTSA's model] has been appropriately updated based on both public and manufacturer data to reflect the technologies already applied, particularly tire rolling resistance;” and (5) “some technologies have been appropriately restricted. For example, low rolling resistance tires are no longer allowed on performance vehicles, and aero improvements are limited to maximum levels of 15% for trucks and 10% for minivans.” [137] The Alliance continued, noting that “NHTSA's Volpe model also predates EPA's OMEGA model. More importantly, the new Volpe model considers several factors that make its results more realistic.” [138] As factors leading the Volpe model to produce results that are more realistic than those produced by OMEGA, the Alliance commented that (1) “The Volpe model includes estimates of the redesign and refresh schedules of vehicles based on historical trends, whereas the OMEGA model uses a fixed, and too short, time interval during which all vehicles are assumed to be fully redesigned. . . .;” (2) “The Volpe model allows users to phase-in technology based on year of availability, platform technology sharing, phase-in caps, and to follow logical technology paths per vehicle. . . .;” (3) “The Volpe model produces a year-by year analysis from the baseline model year through many years in the future, whereas the OMEGA model only analyzes a fixed time interval. . . .;” (4) “The Volpe model recognizes that vehicles share platforms, engines, and transmissions, and that improvements to any one of them will likely extend to other vehicles that use them” whereas “The OMEGA model treats each vehicle as an independent entity. . . .;” (5) “The Volpe model now includes sales and scrappage effects;” and (6) “The Volpe model is now capable of analyzing for CAFE and [tailpipe CO2] compliance, each with unique program restrictions and flexibilities.” [139] The Alliance also incorporated by reference concerns it raised regarding EPA's OMEGA-based analysis supporting EPA's proposed and prior final determinations.[140]

The Alliance further stated that “For all of the above reasons and to avoid duplicate efforts, the Alliance recommends that the Agencies continue to use DOT's Volpe and Autonomie modeling system, rather than continuing to develop two separate systems. EPA has demonstrated through supporting Volpe model code revisions and by supplying engine maps for use in the Autonomie model that their expertise can be properly represented in the rulemaking process without having to develop separate or new tools.” [141]

Some individual manufacturers provided comments supporting and elaborating on the above comments by Global Automakers and the Alliance. For example, FCA commented that “the modeling performed by the agencies should illuminate the differences between the CAFE and [tailpipe CO2 emissions] programs. This cannot be accomplished when each agency is using different tools and assumptions. Since we believe NHTSA possesses the better set of tools, we support both agencies using Autonomie for vehicle modeling and Volpe (CAFE) for fleet modeling.” [142]

Honda stated that “The current version of the CAFE model is reasonably accurate in terms of technology efficiency, cost, and overall compliance considerations, and reflects a notable improvement over previous agency modeling efforts conducted over the past few years. We found the CAFE model's characterization of Honda's “baseline” fleet—critical modeling minutiae that provide a technical foundation of the agencies' analysis—to be highly accurate. We commend NHTSA and Volpe Center staff on these updates, as well as on the overall transparency of the model. The model's graphical user interface (GUI) makes it easier to run, model functionality is thoroughly documented, and the use of logical, traceable input and output files accommodates easy tracking of results.” [143] Similarly, in an earlier presentation to the agencies, Honda included the following slide comparing EPA's OMEGA model to DOT's CAFE (Volpe) model, and making recommendations regarding future improvements to the latter: [144]

Start Printed Page 24226

Toyota, in addition to arguing that the agencies' application of model inputs (e.g., an analysis fleet based on MY 2016 compliance data) produced more realistic results than in the draft TAR and in EPA's former proposed and final determinations, also stressed the importance of the CAFE model's year-by-year accounting for product redesigns, stating that this produces more realistic results than the OMEGA-based results shown previously by EPA:

The modeling now better accounts for factors that limit the rate at which new technologies enter and then diffuse through a manufacturer's fleet. Bringing new or improved vehicles and technologies to market is a several-year, capital-intensive undertaking. Once new designs are introduced, a period of stability is required so investments can be amortized. Vehicle and technology introductions are staggered over time to manage limited resources. Agency modeling now better recognizes the inherent constraints imposed by realities that dictate product cadence. We agree with the agencies' understanding that “the simulation of compliance actions that manufacturers might take is constrained by the pace at which new technologies can be applied in the new vehicle market,” and we are encouraged to learn that “agency modeling can now account for the fact that individual vehicle models undergo significant redesigns relatively infrequently.” The preamble correctly notes that manufacturers try to keep costs down by applying most major changes mainly during vehicle redesigns and more modest changes during product refresh, and that redesign cycles for vehicle models can range from six to ten years, and eight to ten-years for powertrains. This appreciation for standard business practice enables the modeling to more accurately capture the way vehicles share engines, transmissions, and platforms. There are now more realistic limits placed on the number of engines and transmissions in a powertrain portfolio which better recognizes manufacturers must manage limited engineering resources and control supplier, production, and service costs. Technology sharing and inheritance between vehicle models tends to limit the rate of improvement in a manufacturer's fleet.[145]

These comments urging EPA to use NHTSA's CAFE model echo comments provided in response to a 2018 peer review of the model. While identifying various opportunities for improvement, peer reviewers expressed strong overall support for the CAFE model's technical approach and execution. For example, one reviewer, after offering many specific technical recommendations, concluded as follows:

The model is impressive in its detail, and in the completeness of the input data that it uses. Although the model is complex, the reader is given a clear account of how variables are variously divided and combined to yield appropriate granularity and efficiency within the model. The model tracks well a simplified version of the real-world and manufacturing/design decisions. The progression of technology choices and cost benefit choices is clear and logical. In a few cases, the model simply explains a constraint, or a value assigned to a variable, without defending the choice of the value or commenting on real-world variability, but these are not substantive omissions. The model will lend itself well to future adaptation or addition of variables, technologies and pathways.[146]

Although the peer review charge focused solely on the CAFE model, another peer reviewer separately recommended that EPA “consider opportunities for EPA to use the output from the Volpe Model in place of their OMEGA Model output” [147]

More recently, in response to the NPRM, Dr. Julian Morris, an economist at George Washington University, commented extensively on the superiority of the agencies' NPRM analysis to previous analyses, offering the following overall assessment:

I have assessed the plausibility of the analyses undertaken by NHTSA and EPA in relation to the proposed SAFE rule. I found that the agencies have undertaken a thorough—one might even say exemplary—analysis, improving considerably on earlier analyses undertaken by the agencies of previous rules relating to CAFE standards and [tailpipe CO2] emission standards. Of particular note, the agencies included more realistic estimates of the rebound effect, developed a sophisticated model of the Start Printed Page 24227scrappage effect, and better accounted for various factors affecting vehicle fatality rates.[148]

The agencies carefully considered these and other comments regarding which models to apply when estimating potential impacts of each of the contemplated regulatory alternatives. For purposes of estimating the impacts of CAFE standards, even the coalition of environmental advocates observed that the CAFE model reflects EPCA's requirements. As discussed below in Section VI.A, EPCA imposes specific requirements not only on how CAFE standards are to be structured (e.g., including a minimum standard for domestic passenger cars), but also on how CAFE standards are to be evaluated (e.g., requiring that the potential to produce additional AFVs be set aside for the model years under consideration), and the CAFE model reflects these requirements, and the agencies consider the CAFE model to be the best available tool for CAFE rulemaking analysis. Regarding the use of Autonomie to construct fuel consumption (i.e., efficiency) inputs to the CAFE model, the agencies recognize that other vehicle simulation tools are available, including EPA's recently-developed ALPHA model. However, as also discussed in Section VI.B.3, Autonomie has a much longer history of development and refinement, and has been much more widely applied and validated. Moreover, Argonne experts have worked carefully for several years to develop methods for running large arrays of simulations expressly structured and calibrated for use in DOT's CAFE model. Therefore, the agencies consider Autonomie to be the best available tool for constructing such inputs to the CAFE model. While the agencies have also carefully considered potential specific model refinements, as well as the merits of potential changes to model inputs and assumptions, none of these potential refinements and input have led either agency to reconsider using the CAFE model and Autonomie for CAFE rulemaking analysis.

With respect to estimating the impacts of CO2 standards, even though Argonne and the agencies have adapted Autonomie and the CAFE model to support the analysis of CO2 standards, environmental groups, California, and other States would prefer that EPA use the models it developed during 2009-2018 for that purpose.[149] Arguments that EPA revert to its ALPHA and OMEGA models fall within three general categories: (1) Arguments that EPA's models would have selected what commenters consider better (i.e., generally more stringent) standards, (2) arguments that EPA's models are technically superior, and (3) arguments that the law requires EPA use its own models.

The first of these arguments—that EPA's models would have selected better standards—conflates the analytical tool used to inform decision-making with the action of making the decision. As explained elsewhere in this document and as made repeatedly clear over the past several rulemakings, the CAFE model (or, for that matter, any model) neither sets standards nor dictates where and how to set standards; it simply informs as to the potential effects of setting different levels of standards. In this rulemaking, EPA has made its own decisions regarding what CO2 standards would be appropriate under the CAA.

The third of these arguments—that EPA is legally required to use only models developed by its own staff—is also without merit. The CAA does not require the agency to create or use a specific model of its own creation in setting tailpipe CO2 standards. The fact that EPA's decision may be informed by non-EPA-created models does not, in any way, constitute a delegation of its statutory power to set standards or decision-making authority.[150] Arguing to the contrary would suggest, for example, that EPA's decision would be invalid because it relied on EIA's Annual Energy Outlook for fuel prices for all of its regulatory actions rather than developing its own model for estimating future trends in fuel prices. Yet, all Federal agencies that have occasion to use forecasts of future fuel prices regularly (and appropriately) defer to EIA's expertise in this area and rely on EIA's NEMS-based analysis in the AEO, even when those same agencies are using EIA's forecasts to inform their own decision-making. Similarly, this argument would mean that the agencies could not rely on work done by contractors or other outside consultants, which is contrary to regular agency practice across the entirety of the Federal Government.

The specific claim here that use of the CAFE model instead of ALPHA and OMEGA is somehow illegitimate is similarly unpersuasive. The CAFE and CO2 rules have, since Massachusetts v. EPA, all been issued as joint rulemakings, and, thus are the result of a collaboration between the two agencies. This was true when the rulemakings used separate models for the different programs and continues to be true in today's final rule, where the agencies take the next step in their collaborative approach by now using simply one model to simulate both programs. In 2007, immediately following this Supreme Court decision, the agencies worked together toward standards for model years 2011-2015, and EPA made use of the CAFE model for its work toward possible future CO2 standards. That the agencies would need to continue the unnecessary and inefficient process of using two separate combinations of models as the joint National Program continues to mature, therefore, runs against the idea that the agencies, over time, would best combine resources to create an efficient and robust regulatory program. For the reasons discussed throughout today's final rule, the agencies have jointly determined that Autonomie and the CAFE model have significant technical advantages, including important additional features, and are therefore the more appropriate models to use to support both analyses.

Further, the fact that Autonomie and CAFE models were initially developed by DOE/Argonne and NHTSA does not mean that EPA has no role in either these models or their inputs. That is, the development process for CAFE and CO2 standards inherently requires technical and policy examinations and deliberations between staff experts and decision-makers in both agencies. Such engagements are a healthy and important part of any rulemaking activity—and particularly so with joint rulemakings. The Supreme Court stated in Massachusetts v. EPA that, “The two obligations [to set CAFE standards under EPCA and to set tailpipe CO2 emissions standards under the CAA] may overlap, but there is no reason to think the two agencies cannot both administer their obligations and yet Start Printed Page 24228avoid inconsistency.” [151] When agency experts consider analytical issues and agency decision-makers decide on policy, which is informed (albeit not dictated) by the outcome of that work, they are working together as the Court appears to have intended in 2007, even if legislators' intentions have varied in the decades since EPCA and the CAA have been in place.[152] Regulatory overlap necessarily involves deliberation, which can lead to a more balanced, reasonable, and improved analyses, and better regulatory outcomes. It did here. The existence of deliberation is not per se evidence of unreasonableness, even if some commenters believe a different or preferred policy outcome would or should have resulted.[153]

Over the 44 years since EPCA established the requirement for CAFE standards, NHTSA, EPA and DOE career staff have discussed, collaborated on, and debated engineering, economic, and other aspects of CAFE regulation, through focused meetings and projects, informal exchanges, publications, conferences and workshops, and rulemakings.

Part of this expanded exchange has involved full vehicle simulation. While tools such as PSAT (the DOE-sponsored simulation tool that predated Autonomie) were in use prior to 2007, including for discrete engineering studies supporting inputs to CAFE rulemaking analyses, these tools' information and computing requirements were such that NHTSA had determined (and DOE and EPA had concurred) that it was impractical to more fully integrate full vehicle simulation into rulemaking analyses. Since that time, computing capabilities have advanced dramatically, and the agencies now agree that such integration of full vehicle simulation—such as the large-scale exercise of Autonomie to produce inputs to the CAFE Model—can make for more robust CAFE and CO2 rulemaking analysis. This is not to say, though, that experts always agree on all methods and inputs involved with full vehicle simulation. Differences in approach and inputs lead to differences in results. For example, compared to other publicly available tools that can be practicably exercised at the scale relevant to fleetwide analysis needed for CAFE and CO2 rulemaking analysis, DOE/Argonne's Autonomie model is more advanced, spans a wider range of fuel-saving technologies, and represents them in more specific detail, leaving fewer “gaps” to be filled with other models (risking inconsistencies and accompanying errors). These differences discussed in greater detail below in Section VI.B.3. Perhaps most importantly, the CAFE model considers fuel prices in determining both which technologies are applied and the total amount of technology applied, in the case where market forces demand fuel economy levels in excess of the standards. While OMEGA can apply technology in consideration of fuel prices, OMEGA will apply technology to reach the same level of fuel economy (or CO2 emissions) if fuel prices are 3, 5, or 20 dollars, which violates the SAB's requirement that the analysis “account for [. . .] future fuel prices .” [154] Furthermore, it produces a counterintuitive result. If fuel prices become exorbitantly high, we would expect consumers to place an emphasis on additional fuel efficiency as the potential for extra fuel savings is tremendous.

Moreover, DOE has for many years used Autonomie (and its precursor model, PSAT) to produce analysis supporting fuel economy-related research and development programs involving billions of dollars of public investment, and NHTSA's CAFE model with inputs from DOE/Argonne's Autonomie model has produced analysis supporting rulemaking under the CAA. In 2015, EPA proposed new tailpipe CO2 standards for MY 2021-2027 heavy-duty pickups and vans, finalizing those standards in 2016. Supporting the NPRM and final rule, EPA relied on analysis implemented by NHTSA using NHTSA's CAFE model, and NHTSA used inputs developed by DOE/Argonne using DOE/Argonne's Autonomie model. CBD questioned this history, asserting that, “EPA conducted a separate analysis using a different iteration of the CAFE model rather than rely on the version which NHTSA used, again resulting and parallel but corroborative modeling results.” [155] CBD's comment mischaracterizes EPA's actual use of the CAFE Model. As explained in the final rule, EPA's “Method B” analysis was developed as follows:

In Method B, the CAFE model from the NPRM was used to project a pathway the industry could use to comply with each regulatory alternative, along with resultant impacts on per-vehicle costs. However, the MOVES model was used to calculate corresponding changes in total fuel consumption and annual emissions for pickups and vans in Method B. Additional calculations were performed to determine corresponding monetized program costs and benefits.[156]

In other words, a version of NHTSA's CAFE Model was used to perform the challenging part of the analysis—that is, the part that involves accounting for manufacturers' fleets, accounting for available fuel-saving technologies, accounting for standards under consideration, and estimating manufacturers' potential responses to new standards—EPA's MOVES model was used to perform “downstream” calculations of fuel consumption and tailpipe emissions, and used spreadsheets to calculate even more straightforward calculations of program costs and benefits. While some stakeholders perceive these differences as evidencing a meaningfully independent approach, in fact, the EPA staff's analysis was, at its core, wholly dependent on NHTSA's CAFE Model, and on that model's use of Autonomie simulations.

Given the above, the only remaining argument for EPA to revert to its previously-developed models rather than relying on Autonomie and the CAFE model would be that the former are so technically superior to the latter that even model refinements and input changes cannot lead Autonomie and the CAFE model to produce appropriate and reasonable results for CO2 rulemaking analysis. As discussed below, having considered a wide range of technical differences, the agencies find that the Autonomie and CAFE models currently provide the best analytical combination for CAFE and tailpipe CO2 emissions rulemaking analysis. As discussed Start Printed Page 24229below in Section VI.B.3, Autonomie not only has a longer and wider history of development and application, but also DOE/Argonne's interaction with automakers, supplier and academies on continuous bases had made individual sub-models and assumptions more robust. Argonne has also been using research from DOE's Vehicle Technology Office (VTO) at the same time to make continuous improvements in Autonomie.[157] Also, while Autonomie uses engine maps as inputs, and EPA developed engine maps that could have been used for today's analysis, EPA declined to do so, and those engine maps were only used in a limited capacity for reasons discussed below in Section VI.C.1.

As also discussed below in Section VI.A.4, the CAFE model accounts for some important CO2 provisions that EPA's OMEGA model cannot account for. For example, the CAFE model estimates the potential that any given manufacturer might apply CO2 compliance credits it has carried forward from some prior model year. While one commenter, Mr. Rykowski, takes issue with how the CAFE model handles credit banking, he does not acknowledge that EPA's OMEGA model, lacking a year-by-year representation of compliance, is altogether incapable of accounting for the earning and use of banked compliance credits. Also, although Mr. Rykowski's comments regarding A/C leakage and refrigerants are partially correct insofar as the CAFE model does not account for leakage-reducing technologies explicitly, the comment is as applicable to OMEGA as it is to the CAFE model and, in any event, data regarding which vehicles have which leakage-reducing technologies was not available for the MY 2016 fleet. Nevertheless, as discussed in Section VI.A.4, NHTSA has refined the CAFE model's accounting for the cost of leakage reduction technologies.

The agencies have also considered Mr. Rykowski's comments suggesting that using OMEGA would be preferable because, rather than selecting from hundreds of thousands of potential combinations of technologies, OMEGA includes only the “50 or so” combinations that EPA has already determined to be cost-effective. The “better way” of making this determination is also effectively a model, but the separation of this model from OMEGA is, as evidenced by manufacturers' comments, obfuscatory, especially in terms of revealing how specific vehicle model/configurations initial engineering properties are aligned with specific initial technology combinations. By using a full set of technology combinations, the CAFE model makes very clear how each vehicle model/configuration is assigned to a specific initial combination and, hence, how subsequently fuel consumption and cost changes are accounted for. Moreover, EPA's separation of “thinning” process from OMEGA's main compliance simulation makes sensitivity analysis difficult to implement, much less follow. The agencies find, therefore, that the CAFE model's approach of retaining a full set of vehicle simulation results throughout the compliance simulation to be more realistic (e.g., more capable of reflecting manufacturer- and vehicle-specific factors), more responsive to changes in model inputs (e.g., changes to fuel prices, which could impact the relative attractiveness of different technologies), more transparent, and more amenable to independent corroboration the agencies' analysis.

Regarding comments by Messrs. Duleep, Rogers, and Rykowski suggesting that the CAFE model, by tying most technology application to planned vehicle redesigns and freshening, is too restrictive, the agencies disagree. As illustrated by manufacturers' comments cited above, as reinforced by both extensive product planning information provided to the agencies, and as further reinforced by extensive publicly available information, manufacturers tend to not make major changes to a specific vehicle model/configuration in one model year, and then make further major changes to the same vehicle model/configuration the next model year. There is ample evidence that manufacturers strive to avoid such discontinuity, complexity, and waste, and in the agencies' view, while it is impossible to represent every manufacturer's decision-making process precisely and with certainty, the CAFE model's approach of using estimated product design schedules provides a realistic basis for estimating what manufacturers could practicably do. Also, the relevant inputs are simply inputs to the CAFE model, and if it is actually more realistic to assume that a manufacturer can change major technology on all of its products every year, the CAFE model can easily be operated with every model year designated as a redesign year for every product, but as discussed throughout this document, the agencies consider this to be extremely unrealistic. While this means the CAFE model can be run without a year-by-year representation that carries forward technologies between model years, doing so would be patently unrealistic (as reflected in some stakeholders' comments in 2002 on the first version of the CAFE model). Conversely, the OMEGA model cannot be operated in a way that accounts for what the agencies consider to be very real product planning considerations.

However, having also considered Mr. Rykowski's comments about the CAFE model's “effective cost” metric, and having conducted side-by-side testing documented in the accompanying FRIA, the agencies are satisfied that an alternative “cost per credit” metric is also a reasonable metric to use for estimating how manufacturers might selected among available options to add specific fuel-saving technologies to specific vehicles.[158] Therefore, NHTSA has revised the CAFE model accordingly, as discussed below in Section VI.A.4.

Section VI.C.1 also addresses Mr. Rogers's comments on engine maps used as estimates to full vehicle simulation. In any event, because engine maps are inputs to full vehicle modeling and simulation, the relative merits of specific maps provide no basis to prefer one vehicle simulation modeling system over another. Similarly, Section VI.B.3 also addresses Mr. Duleep's comments preferring EPA's prior approach, using ALPHA, of effectively assuming that a manufacturer would incur no additional cost by reoptimizing every powertrain to extract the full fuel economy potential of even the smallest incremental changes to aerodynamic drag and tire rolling resistance. Mr. Duleep implies that Autonomie is flawed because the NPRM analysis did not apply Autonomie in a way that makes such assumptions. The agencies discuss powertrain sizing and calibration in Section VI.B.3, and note here that such assumptions are not inherent to Start Printed Page 24230Autonomie; like engine maps, these are inputs to full vehicle simulation. Therefore, neither of these comments by Mr. Rogers and Mr. Duleep lead the agencies to find reason not to use Autonomie.

None of this is to say that Autonomie and the CAFE model as developed and applied for the NPRM left no room for improvement. In the NPRM and RIA, the agencies discussed plans to continue work in a range of specific technical areas, and invited comment on all aspects of the analysis. As discussed below in Chapter VI, the agencies received extensive comment on the published model, inputs, and analysis, both in response to the NPRM and, for newly-introduced modeling capabilities (estimation of sales, scrappage, and employment effects), in response to additional peer review conducted in 2019. The agencies have carefully considered these comments, refined various specific technical aspects of the CAFE model (like the “effective cost” metric mentioned above), and have also updated inputs to both Autonomie and the CAFE model. Especially given these refinements and updates, as discussed throughout this rule, EPA maintains that for CO2 rulemaking analysis, Autonomie and the CAFE model have advantages that warrant relying on them rather than on EPA's ALPHA and OMEGA models. Some examples of such advantages include: A longer history of ongong development and application for rulemaking, including by EPA; documentation and model operation stakeholders have found to be comparatively clear and enabling of independent replication of agency analyses; a mechanism to explicitly reflect the fact that manufacturers' product decisions are likely to be informed by fuel prices; better integration of various model functions, enabling efficient sensitivity analysis; and an annual time step that makes it possible to conduct report results on both a calendar year and model year basis, to estimate accruing impacts on vehicle sales and scrappage, and to account for the fact that not every vehicle can be designed in every model year; and other advantages discussed throughout today's notice. Therefore, recognizing that models inform but do not make regulatory decisions, EPA has elected to rely solely on the Autonomie and CAFE models to produce today's analysis of regulatory alternatives for CO2 standards.

The following sections provide a brief technical overview of the CAFE model, including changes NHTSA made to the model since 2012, and differences between the current analysis, the analysis for the 2016 Draft TAR and for the 2017 Proposed Determination/2018 Final Determination, and the 2018 NPRM, before discussing inputs to the model and then diving more deeply into how the model works. For more information on the latter topic, see the CAFE model documentation, available in the docket for this rulemaking and on NHTSA's website.

1. What assumptions have changed since the 2012 final rule?

Any analysis of regulatory actions that will be implemented several years in the future, and whose benefits and costs accrue over decades, requires a large number of assumptions. Over such time horizons, many, if not most, of the relevant assumptions in such an analysis are inevitably uncertain.[159] The 2012 CAFE/CO2 rule considered regulatory alternatives for model years through MY 2025 (17 model years after the 2008 market information that formed the basis of the analysis) that accrued costs and benefits into the 2060s. Not only was the new vehicle market in 2025 unlikely to resemble the market in 2008, but so, too, were fuel prices. It is natural, then, that each successive CAFE/CO2 analysis should update assumptions to reflect better the current state of the world and the best current estimates of future conditions.[160] However, beyond the issue of unreliable projections about the future, a number of agency assertions have proven similarly problematic. In fact, Securing America's Future Energy (SAFE) stated in their comments on the NPRM:

Although the agencies argue “circumstances have changed” and “analytical methods and inputs have been updated,” a thorough analysis should provide a side-by-side comparison of the changing circumstances, methods, and inputs used to arrive at this determination . . . They represent a rapid, dramatic departure from the agencies' previous analyses, without time for careful review and consideration.[161]

We describe in detail (below) the changes to critical assumptions, perspectives, and modeling techniques that have created substantive differences between the current analysis and the analysis conducted in 2012 to support the final rule. To the greatest extent possible, we have calculated the impacts of these changes on the 2012 analysis.

a) The Value of Fuel Savings

The value of fuel savings associated with the preferred alternative in the 2012 final rule is primarily a consequence of two assumptions: [162] The fuel price forecast and the assumed growth in fuel economy in the baseline alternative against which savings are measured. Therefore, as the value of fuel savings accounted for nearly 80 percent of the total benefits of the 2012 rule, each of these assumptions is consequential. With a lower fuel price projection and an expectation that new vehicle buyers respond to fuel prices, the 2012 rule would have shown much smaller fuel savings attributable to the more stringent standards. Projected fuel prices are considerably lower today than in 2012, the agencies now understand new vehicle buyers to be at least somewhat responsive to fuel prices, and the agencies have therefore updated corresponding model inputs to produce an analysis the agencies consider to be more realistic.

The first of these assumptions, fuel prices, was simply an artifact of the timing of the rule. Following recent periodic spikes in the national average gasoline price and continued volatility after the great recession, the fuel price forecast then produced by EIA (as part of AEO 2011) showed a steady march toward historically high, sustained gasoline prices in the United States. However, the actual series of fuel prices has skewed much lower. As it has turned out, the observed fuel price in the years between the 2012 final rule and this rule has frequently been lower than the “Low Oil Price” sensitivity case in the 2011 AEO, even when adjusted for inflation. The following graph compares fuel prices underlying the 2012 final rule to fuel prices applied in the analysis reported in today's notice, expressing both projections in 2010 dollars. The differences are clear and significant:

Start Printed Page 24231

The discrepancy in fuel prices is important to the discussion of differences between the current rule and the 2012 final rule, because that discrepancy leads in turn to differences in analytical outputs and thus to differences in what the agencies consider in assessing what levels of standards are reasonable, appropriate, and/or maximum feasible. As an example, the agencies discuss in Sections VI.D.3 Simulating Environmental Impacts of Regulatory Alternatives and VIII.A.3 EPA's Conclusion that the Final CO2 Standards are Appropriate and Reasonable that fuel price projections from the 2012 rule were one assumption, among others, that could have led to overestimates of the health benefits that resulted from reducing criteria pollutant emissions. Yet the agencies caution readers not to interpret this discrepancy as a reflection of negligence on the part of the agencies, or on the part of EIA. Long-term predictions are challenging and the fuel price projections in the 2012 rule were within the range of conventional wisdom at the time. However, it does suggest that fuel economy and tailpipe CO2 regulations set almost two decades into the future are vulnerable to surprises, in some ways, and reinforces the value of being able to adjust course when critical assumptions are proven inaccurate. This value was codified in regulation when EPA bound itself to the mid-term evaluation process as part of the 2012 final rule.[163]

To illustrate this point clearly, substituting the current (and observed) fuel price forecast for the forecast used in the 2012 final rule creates a significant difference in the value of fuel savings. Even under identical discounting methods (see Section 2, below), and otherwise identical inputs in the 2012 version of the CAFE Model, the current (and historical) fuel price forecast reduces the value of fuel savings by $150 billion— from $525 billion to $375 billion (in 2009 dollars).

The second assumption employed in the 2012 (as well as the 2010) final rule, that new vehicle fuel economy never improves unless manufacturers are required to increase fuel economy in the new vehicle market by increasingly stringent regulations, is more problematic. Despite the extensive set of recent academic studies showing, as discussed in Section VI.D.1.a)(2), that consumers value at least some portion, and in some studies nearly all, of the potential fuel savings from higher levels of fuel economy at the time they purchase vehicles, the agencies assumed in past rulemakings that buyers of new vehicles would never purchase, and manufacturers would never supply, vehicles with higher fuel economy than those in the baseline (MY 2016 in the 2012 analysis), regardless of technology cost or prevailing fuel prices in future model years. In calendar year 2025, the 2012 final rule assumed gasoline would cost nearly $4.50/gallon in today's dollars, and continue to rise in subsequent years. Even recognizing that higher levels of fuel economy would be achieved under the augural/existing standards than without them, the assertion that fuel economy and CO2 emissions would not improve beyond 2016 levels in the presence of nearly $5/gallon gasoline is not supportable. This is highlighted by the observed increased consumer demand for higher-fuel-economy vehicles during the gas price spike of 2008, when average U.S. prices briefly broke $4/gallon. In the 2012 final rule, this assumption—that fuel economy and emissions would never improve absent regulation—created a persistent gap in fuel economy between Start Printed Page 24232the baseline and augural standards that grew to 13 mpg (at the industry average, across all vehicles) by MY 2025. In the 2016 Draft TAR, NHTSA's analysis included the assumption that manufacturers would deploy, and consumers would demand, any technology that recovered its own cost in the first year of ownership through avoided fuel costs. However, in both the Draft TAR and the Proposed and Final Determination documents, EPA's analysis assumed that the fuel economy levels achieved to reach compliance with MY 2021 standards would persist indefinitely, regardless of fuel prices or technology costs.

By substituting the conservative assumption that consumers are willing to purchase fuel economy improvements that pay for themselves with avoided fuel expenditures over the first 2.5 years [164] (identical to the assumption in this final rule's central analysis) the gap in industry average fuel economy between the baseline and augural scenarios narrows from 13 mpg in 2025 to 6 mpg in 2025. As a corollary, acknowledging that fuel economy would continue to improve in the baseline under the fuel price forecast used in the final rule erodes the value of fuel savings attributable to the preferred alternative. While each gallon is still worth as much as was assumed in 2012, fewer gallons are consumed in the baseline due to higher fuel economy levels in new vehicles. In particular, the number of gallons saved by the preferred alternative selected in 2012 drops from about 180 billion to 50 billion once we acknowledge the existence of even a moderate market for fuel economy.[165] The value of fuel savings is similarly eroded, as higher fuel prices lead to correspondingly higher demand for fuel economy even in the baseline—reducing the value of fuel savings from $525 billion to $190 billion.

The magnitude of the fuel economy improvement in the baseline is a consequence of both the fuel prices assumed in the 2012 rule (already discussed as being higher than both subsequent observed prices and current projections) and the assumed technology costs. In 2012, a number of technologies were assumed to have negative incremental costs—meaning that applying those technologies to existing vehicles would both improve their fuel economy and reduce the cost to produce them. Asserting that the baseline would experience no improvement in fuel economy without regulation is equivalent to asserting that manufacturers, despite their status as profit maximizing entities, would not apply these cost-saving technologies unless forced to do so by regulation. While this issue is discussed in greater detail in Section VI.B the combination of inexpensive (or free) technology and high fuel prices created a logically inconsistent perspective in the 2012 rule—where consumers never demanded additional fuel economy, despite high fuel costs, and manufacturers never supplied additional fuel economy, despite the availability of inexpensive (or cost saving) technology to do so.

Many commenters on earlier rules supported the assumption that fuel economy would not improve at all in the absence of standards. In fact, some commenters still support this position. For example, EDF commented to the NPRM that, “NHTSA set the Volpe model to project that, with CAFE standards remaining flat at MY 2020 levels through MY 2026, automakers would over-comply with the MY 2020 standards by 9 grams/mile of CO2 for cars and 15 g/mi of CO2 for light trucks during the 2029-2032 timeframe, plus 1%/year improvements beyond MY 2032. This assumption unreasonably obscures the impacts of the rollback and is not reflective of historical compliance performance.” [166]

EDF is mistaken in two different ways: (1) By acknowledging the existence of a well-documented market for fuel economy, rather than erroneously inflating the benefits associated with increasing standards, this assumption serves to isolate the benefits actually attributable to each regulatory alternative, and (2) it is, indeed, reflective of historical compliance performance. While the agencies rely on the academic literature (and comments from companies that build and sell automobiles) to defend the assertion that a market for fuel economy exists, the industry's historical CAFE compliance performance is a matter of public record.[167] As shown in Figure IV-3, Figure IV-4, and Figure IV-5 for more than a decade, the industry average CAFE has exceeded the standard for each regulatory class—by several mpg during periods of high fuel prices.

Start Printed Page 24233

Start Printed Page 24234

While this rulemaking has shown the impact of deviations from the 2012 rule assumptions individually, these two assumptions affect the value of fuel savings jointly. Replacing the fuel price forecast with the observed historical and current projected prices, and including any technology that pays for itself in the first 2.5 years of ownership through avoided fuel expenditures, reduces the value of fuel savings from $525 billion in the 2012 rule to $140 billion, all else equal. Interestingly, this reduction in the value of fuel savings is smaller than the result when assuming only that the desired payback period is nonzero. While it may seem counterintuitive, it is entirely consistent.

The number of gallons saved under the preferred alternative is actually higher when modifying both assumptions, compared to only modifying the payback period. Updating both assumptions leads to about 100 billion gallons saved by the preferred alternative in 2012, compared to only 50 billion from changing only the payback period, and 180 billion in the 2012 analysis. This occurs because the fuel economy in the baseline is lower when updating both the fuel price and the payback period—the gap between the augural standards and the baseline grows to 9 mpg, rather than only 6 mpg when updating only the payback period. Despite the existence of inexpensive Start Printed Page 24235technology in both cases, with lower fuel prices there are fewer opportunities to apply technology that will pay back quickly. As a consequence, the number of gallons saved by the preferred alternative in 2012 increases—but each gallon saved is worth less because the price of fuel is lower.

b) Technology Cost

While the methods used to identify cost-effective technologies to improve fuel economy in new vehicles have continuously evolved since 2012 (as discussed further in Section IV.B.1), as have the estimated cost of individual technologies, the inclusion of a market response in all scenarios (including the baseline) has changed the total technology cost associated with a given alternative. As also discussed in Section VI.B, acknowledging the existence of a market for fuel economy leads to continued application of the most cost-effective technologies in the baseline—and in other less stringent alternatives—up to the point at which there are no remaining technologies whose cost is fully offset by the value of fuel saved in the first 30 months of ownership. The application of this market-driven technology has implications for fuel economy levels under lower stringencies (as discussed earlier), but also for the incremental technology cost associated with more stringent alternatives. As lower stringency alternatives (including the 2012 baseline) accrue more technology, the incremental cost of more stringent alternatives decreases.

By including a modest market for fuel economy, and preserving all other assumptions from the 2012 final rule, the incremental cost of technology attributable to the preferred alternative decreases from about $140 billion to about $72 billion. This significant reduction in technology cost is somewhat diminished by the associated reduction in the value of fuel savings (a decrease of $385 billion) when acknowledging the existence of a market for fuel economy. Another consequence of these changes is that the incremental cost of fuel economy technology is responsive to fuel price, as it should be. Under higher prices (as were assumed in 2012), consumers demand higher fuel economy in the new vehicle market. Under lower prices (as have occurred since the 2012 rule) consumers demand less fuel economy than would have been consistent with the fuel price assumptions in 2012.[168] Including a market response in the analysis ensures that, in each case, the cost of fuel economy technology within an alternative is consistent with those assumptions. Using the same fuel price forecast that supports this rule, and the same estimate of market demand for fuel economy, the incremental cost of technology in the preferred alternative would rise back up to about $110 billion.

c) The Social Cost of Carbon (SCC) Emissions

As discussed extensively in the NPRM, the agencies' perspective regarding the social cost of carbon has narrowed in focus. While the 2012 final rule considered the net present value of global damages resulting from carbon emitted by vehicles sold in the U.S. between MY 2009 and MY 2025, the NPRM (and this final rule) consider only those damages that occur to the United States and U.S. territories. As a result of this change in perspective, the value of estimated damages per-ton of carbon is correspondingly smaller. Had the 2012 final rule utilized the same perspective on the social cost of carbon, the benefits associated with the preferred alternative would have been about $11 billion, rather than $53 billion. However, the savings associated with carbon damages are a consequence of both the assumed cost per-ton of damages and the number of gallons of fuel saved. As discussed above, the gallons saved in the 2012 final rule were likely inflated as a result of both fuel price forecasts and the assumption that no market exists for fuel economy improvements. Correcting the estimate of gallons saved from the preferred alternative in the 2012 rule and considering only the domestic social cost of carbon further reduces the savings in carbon damages to $6 billion.

d) Safety Neutrality

In the 2012 final rule, the agencies showed a “safety neutral” compliance solution; that is, a compliance solution that produced no net increase in on-road fatalities for MYs 2017-2025 vehicles as a result of technology changes associated with the preferred alternative. In practice, safety neutrality was achieved by expressly limiting the availability of mass reduction technology to only those vehicles whose usage causes fewer fatalities with decreased mass. This result was discussed as one possible solution, where manufacturers chose technology solutions that limited the amount of mass reduction applied, and concentrated the application on vehicles that improve the safety of other vehicles on the roads (primarily by reducing the mass differential in collisions). However, it implicitly assumed that each and every manufacturer would leave cost-effective technologies unused on entire market segments of vehicles in order to preserve a safety neutral outcome at the fleet level for a given model year (or set of model years) whose useful lives stretched out as far as the 2060s. Removing these restrictions tells a different story.

When mass reduction technology, determined in the model to be a cost-effective solution (particularly in later model years, when more advanced levels of mass reduction were expected to be possible), is unrestricted in its application, the 2012 version of the CAFE Model chooses to apply it to vehicles in all segments. This has a small effect on technology costs, increasing compliance costs in the earliest years of the program by a couple billion dollars, and reducing compliance costs for MYs 2022—2025 by a couple billion dollars. However, the impact on safety outcomes is more pronounced.

Also starting with the model and inputs used for the 2012 final rule (and, as an example, focusing on that rule's 2008-based market forecast), removing the restrictions on the application of mass reduction technology results in an additional 3,400 fatalities over the full lives of MYs 2009-2025 vehicles in the baseline,[169] and another 6,900 fatalities over those same vehicle lives under the preferred alternative. The result, a net increase of 3,500 fatalities under the preferred alternative relative to the baseline, also produces a net social cost of $18 billion. The agencies' current treatment of both mass reduction technology, which can greatly improve the effectiveness of certain technology packages by reducing road load, and estimated fatalities and now account for both general exposure (omitted in the 2012 final rule modeling) and fatality risk by age of the vehicle, further changes the story around mass reduction technology application for compliance and its relationship to on-road safety.

2. What methods have changed since the 2012 final rule?

Simulating how manufacturers might respond to CAFE/CO2 standards Start Printed Page 24236requires information about existing products being offered for sale, as well as information about the costs and effectiveness of technologies that could be applied to those vehicles to bring the fleets in which they reside into compliance with a given set of standards. Following extensive additional work and consideration since the 2012 analysis, both agencies now use the CAFE Model to simulate these compliance decisions. This has several practical implications which are discussed in greater detail in Section VI.A. Briefly, this change represents a shift toward including a number of real-world production constraints—such as component sharing across a manufacturer's portfolio—and product cadence, where only a subset of vehicles in a given model year are redesigned (and thus eligible to receive fuel economy technology). Furthermore, the year-by-year accounting ensures a continuous evolution of a manufacturer's product portfolio that begins with the market data of an initial model year (model year 2017, in this analysis) and continues through the last year for which compliance is simulated. Finally, the modeling approach has migrated from one that relied on the simple product of single values to estimate technology effectiveness to a model that relies on full vehicle simulation to determine the effectiveness of any combination of fuel economy technologies. The combination of these changes has greatly improved the realism of simulated vehicle fuel economy for combinations of technologies across vehicle systems and classes.

In addition to these changes to the portions of the analysis that represent the supply of fuel economy (by manufacturer, fleet, and model year) in the new vehicle market, this analysis contains changes to the representation of consumer demand for fuel economy. One such measure was discussed above—the notion that consumers will demand some amount of fuel economy improvement over time, consistent with technology costs and fuel prices. However, another deviation from the 2012 final rule analysis reflects overall demand for new vehicles. Across ten alternatives, ranging from the baseline (freezing future standards at 2016 levels) to scenarios that increased stringency by seven percent per year, from 2017 through 2025, the 2012 analysis showed no response in new vehicle sales, down to the individual model level. This implied that, regardless of changes to vehicle cost or attributes driven by stringency increases, no fewer (or possibly more) units of any single model would be sold in any year, in any alternative. Essentially, that analysis asserted that the new vehicle market does not respond, in any way, to average new vehicle prices across the alternatives—regardless of whether the incremental cost is $1,600/vehicle (as it was estimated to be under the preferred alternative) or nearly $4,000/vehicle (as it was in under the 7 percent alternative). Both the NPRM and this final rule, while not employing pricing models or full consumer choice models to address differentiated demand within brands or manufacturer portfolios, have incorporated a modeled sales response that seeks to quantify what was not quantified in previous rulemakings.

An important accounting method has also changed since the 2012 final rule was published. At the time of that rule, the agencies used an approach to discounting that combined attributes of a private perspective and a social perspective in their respective benefit cost analyses. This approach was logically inconsistent, and further reinforced some of the exaggerated estimates of fuel savings, social benefits (from reduced externalities), and technology costs described above. The old method discounted the value of all incremental quantities, whether categorized as benefits or costs, to the model year of the vehicle to which they accrued. This approach is largely acceptable for use in a private benefit cost analysis, where the costs and benefits accrue to the buyer of a new vehicle (in the case of this policy) who weighs their discounted present values at the time of purchase. However, the private perspective would not include any costs or benefits that are external to the buyer (e.g., congestion or the social cost of carbon emissions). For an analysis that compares benefits and costs from the social perspective, attempting to estimate the relative value of a policy to all of society rather than just buyers of new vehicles, this approach is more problematic.

The discounting approach in the 2012 final rule was particularly distortionary for a few reasons. The fact that benefits and costs occurred over long time periods in the 2012 rule, and the standards isolated the most aggressive stringency increases in the latter years of the program, served to allow benefits that occurred in 2025 (for example) to enter the accounting without being discounted, provided that they accrued to the affected vehicles during their first year of ownership. In a setting where numerous inputs (e.g., fuel price and social cost of carbon) increase over time, benefits were able to grow faster than the discount rate in some cases—essentially making them infinite. The interpretation of discounting for externalities was equally problematic. For example, the discounting approach in the 2012 final rule would have counted a ton of CO2 not emitted in CY 2025 in multiple ways, despite the fact that the social cost of carbon emissions was inherently tied to the calendar year in which the emissions occurred. Were the ton avoided by a MY 2020 vehicle, which would have been five years old in CY 2025, the value of that ton would have been the social cost of carbon times 0.86, but would have been undiscounted if that same ton had been saved by a MY 2025 vehicle in its initial year of usage.

This approach was initially updated in the 2016 Draft TAR to be consistent with common economic practice for benefit-cost analysis, and this analysis continues that approach. In the social perspective, all benefits and costs are discounted back to the decision year based on the calendar year in which they occur. Had the agencies utilized such an approach in the 2012 final rule, net benefits would have been reduced by about 20 percent, from $465 billion to $374 billion—not accounting for any of the other adjustments discussed above.

3. How have conditions changed since the 2012 final rule was published?

The 2012 final rule relied on market and compliance information from model year 2008 to establish standards for model years 2017-2025. However, in the intervening years, both the market and the industry's compliance positions have evolved. The industry has undergone a significant degree of change since the MY 2008 fleet on which the 2012FR was based. Entire brands (Pontiac, Oldsmobile, Saturn, Hummer, Mercury, etc.) and companies (Saab, Suzuki, Lotus) have exited the U.S. market, while others (most notably Tesla) have emerged. Several dozen nameplates have been retired and dozens of other created in that time. Overall, the industry has offered a diverse set of vehicle models that have generally higher fuel economy than the prior generation, and an ever-increasing set of alternative fuel powertrains.

As Table IV-1 shows, alternative powertrains have steadily increased under CAFE/CO2 regulations. Under the standards between 2011 and 2018, the number of electric vehicle offerings in the market has increased from 1 model to 57 models (inclusive of all plug-in vehicles that are rated for use on the highway), and hybrids (like the Toyota Prius) have increased from 20 models to Start Printed Page 2423743 models based on data from DOE's Alternative Fuels Data Center. Fuel efficient diesel vehicles have similarly been on the rise in that period, more than doubling the number of offerings. Flexible fuel vehicles (FFVs), capable of operating on both gasoline and E85 remain readily available in the market, but have been excluded from the table due to both their lower fuel economy and demonstrated consumer reluctance to operate FFVs on E85. They have historically been used to improve a manufacturer's compliance position, rather than other alternative fuel systems that reduce fuel consumption and save buyers money.

Not only have alternative powertrain options proliferated since the 2012 FR, the average fuel economy of new vehicles within each body style has increased. However, the more dramatic effect may lie in the range of fuel economies available within each body style. Figure IV-6 shows the distribution of new vehicle fuel economy (in miles per gallon equivalent) by body style for MYs 2008, 2016, and 2020 (simulated). Each box represents the 25th and 75th percentiles, where 25 and 75 percent of new models offered are less fuel efficient than that level. Not only has the median fuel economy improved (the median shows the point at which 50 percent of new models are less efficient) under the CAFE/CO2 programs, but the range of available fuel economies (determined by the length of the boxes and their whiskers) has increased as well. For example, the 25th percentile of pickup truck fuel economy in 2020 is expected to be significantly more efficient than 75 percent of the pickups offered in 2008. In MY 2008, there were only a few SUVs offered with rated fuel economies above 34MPG. By MY 2020 almost half of the SUVs offered will have higher fuel economy ratings—with almost 20 percent of offerings exceeding 40MPG.

The improvement in passenger car styles has been no less dramatic. As with the other styles, the range of available fuel economies has increased under the CAFE/CO2 programs and the distribution of available fuel economies skewed higher—with 40 percent of MY 2020 models exceeding 40MPG. The attribute-based standards are designed to encourage manufacturers to improve vehicle fuel economy across their portfolios, and they have clearly done so. Not only have the higher ends of the distributions increased, the lower ends in all body styles have improved as well, where the least fuel efficient 25 percent of vehicles offered in MY 2016 (and simulated in 2020) are more fuel efficient than the most efficient 25 percent of vehicles offered in MY 2008.

Start Printed Page 24238

Some commenters have argued that consumers will be harmed by any set of standards lower than the baseline (augural) standards because buyers of new vehicles will be forced to spend more on fuel than they would have under the augural standards. However, as Figure IV-6 shows, the range of fuel economies available in the new market is already sufficient to suit the needs of buyers who desire greater fuel economy rather than interior volume or some other attributes. Full size pickup trucks are now available with smaller turbocharged engines paired with 8 and 10-speed transmissions and some mild electrification. Buyers looking to transport a large family can choose to purchase a plug-in hybrid minivan. There were 57 electric models available in 2018, and hybrid powertrains are no longer limited to compact cars (as they once were). Buyers can choose hybrid SUVs with all-wheel and four-wheel drive. While these kinds of highly efficient options were largely absent from some body styles in MY 2008, this is no longer the case. Given that high-MPG vehicles are widely available, consumers must also value other vehicle attributes (e.g., acceleration and load-carrying capacity) that can can also be improved with the same technologies that can be used to improve fuel economy.

Manufacturers have accomplished a portfolio-wide improvement by improving the combustion efficiency of engines (through direct injection and Start Printed Page 24239turbocharging), migrating from four and five speed transmissions to 8 and 10 speed transmissions, and electrifying to varying degrees. All of this has increased both production costs and fuel efficiency during a period of economic expansion and low energy prices. While the vehicles offered for sale have increased significantly in efficiency between MY 2008 and MY 2020, the sales-weighted average fuel economy has achieved less improvement. Despite stringency increases of about five percent (year-over-year) between 2012 and 2016, the sales-weighted average fuel economy increased marginally. Figure IV-7 shows an initial increase in average new vehicle fuel economy (the heavy solid line, shown in mpg as indicated on the left y axis), followed by relative stagnation as fuel prices (the light dashed lines, shown in dollars per gallon as indicated on the right y axis) fell and remained low.[171] It is worth noting that average new vehicle fuel economy observed a brief spike during the year that the Tesla Model 3 was introduced (as a consequence of strong initial sales volumes, as pre-orders were satisfied, and fuel economy ratings that are significantly higher than the industry average), and settled around 27.5 MPG in Fall 2019. Average fuel economy receded further over the next several months to 26.6 MPG in February 2020.[172]

In their NPRM comments, manufacturers expressed concern that CAFE standards had already increased to the point where the price increases necessary to recoup manufacturers' increased costs for providing further increases in fuel economy outweigh the value of fuel savings.[173 174] The agencies do not agree that this point has already been reached by previous stringency increases, but acknowledge the reality of diminishing marginal returns to improvements in fuel economy. A driver with a 40MPG vehicle uses about 300 gallons of fuel per year. Increasing the fuel economy of that vehicle to 50MPG, a 25 percent increase, would likely be over $1000 in additional technology cost. However, that driver would only save 25 percent of their annual fuel consumption, or 75 gallons out of 300 gallons. Even at $3/gallon, higher than the current national average, that represents $225 per year in fuel savings. That means that the buyer's $1000 investment in additional fuel economy pays back in just under 4.5 years (undiscounted). The agencies' respective programs have created greater access to high MPG vehicles in all classes and encouraged the proliferation of alternative fuels and powertrains. But if the value of the fuel savings is insufficient to motivate buyers to invest in ever greater levels of fuel economy, manufacturers will face challenges in the market.

While Figure IV-3 through Figure IV-5 illustrate the trends in historical CAFE compliance for the entire industry, the figures contain another relevant fact. After several consecutive years of increasing standards, the achieved and required levels converge. When the standards began increasing again for passenger cars in 2011, the prior year had industry CAFE levels 5.6 mpg and 7.7 mpg in excess of their standards for domestic cars and imported cars, respectively. Yet, by 2016, the consecutive year-over-year increases had eroded the levels of over-compliance. Light trucks similarly exceeded their standard prior to increasing standards, which began in 2005. Yet, by 2011, after several consecutive years of stringency increases, the industry light-truck average CAFE was merely compliant with the rising standard.

This is largely due to the fact that stringency requirements have increased at a faster rate than achieved fuel Start Printed Page 24240economy levels for several years. The attribute-based standards took effect in 2011 for all regulatory classes, although light truck CAFE standards had been increasing since 2005. Since 2004, light truck stringency has increased an average of 2.7 percent per year, while light truck's compliance fuel economy has increased by an average of 1.7 percent over the same period.[175] For the passenger classes, a similar story unfolds over a shorter period of time. Year over year stringency increases have averaged 4.7 percent per year for domestic cars (though increases in the first two years were about 8 percent—with lower subsequent increases), but achieved fuel economy increases averaged only 2.2 percent per year over the same period. Imported passenger cars were similar to domestic cars, with average annual stringency increases of 4.4 percent but achieved fuel economy levels increasing an average of only 1.4 percent per year from 2011 through 2017. Given that each successive percent increase in stringency is harder to achieve than the previous one, long-term discrepancies between required and achieved year-over-year increases cannot be offset indefinitely with existing credit banks, as they have been so far.

With the fuel price increases fresh in the minds of consumers, and the great recession only recently passed, the CAFE stringency increases that began in 2011 (and subsequent CAFE/CO2 stringency increases after EPA's program was first enforced in MY 2012) had something of a head start. As Figure IV-3 through Figure IV-5 illustrate, the standards were not binding in MY 2011—even manufacturers that had historically paid civil penalties were earning credits for overcompliance. It took two years of stringency increase to catch up to the CAFE levels already present in MY 2011. However, seven consecutive years of increases for passenger cars and a decade of increases for light trucks has changed the credit situation. Figure IV-8 shows CAFE credit performance for regulated fleets—the solid line represents the number of fleets generating shortfalls and the dashed line represents the number of fleets earning credits in each model year.

Fewer than half as many fleets earned surplus credits for over-compliance in MY 2017 compared to MY 2011—and this trend is persistent. The story varies from one manufacturer to another, but it seems sufficient to state the obvious—when the agencies conducted the analysis to establish standards through MY 2025 back in 2012, most (if not all) manufacturers had healthy credit positions. That is no longer the case, and each successive increase requires many fleets to not only achieve the new level from the resulting increase, but to resolve deficits from the prior year as well. The large sums of credits, which last five years under both programs, have allowed most manufacturers to resolve shortfalls. But the light truck fleet, in particular, has a dwindling supply of credits available for purchase or trade. The CO2 program has a provision that allows credits earned during the early years of over-compliance to be applied through MY 2021. This has reduced the compliance burden in the last several years, as intended, but will not mitigate the compliance challenges some OEMs would face if the baseline standards remained in place and energy prices persisted at current levels.

Start Printed Page 24241

Table IV-2 shows the credits earned by each manufacturer over time.[176] As the table shows, when the agencies considered future standards in 2012, most manufacturers were earning credits in at least one fleet. However, the bold values show years with deficits and even some manufacturers who started out in strong positions, such as Ford's passenger car fleet, have seen growing deficits in recent years. While Start Printed Page 24242the initial banks for early-action years eases the burden of CO2 compliance for many OEMs, the year-to-year compliance story is similar to CAFE, see Table IV-3.

Credit position and shortfall rates clearly illustrate manufacturers' fleet performance relative to the standards. Recognizing that manufacturers plan compliance over several model years at any given time, sporadic shortfalls may not be evidence of undue difficulty, but sustained, widespread, growing shortfalls should probably be viewed as evidence that standards previously believed to be manageable might no longer be so. While NHTSA is prohibited by statute from considering availability of credits (and thus, size of credit banks) in determining maximum feasible standards, it does consider shortfalls as part of its determination. EPA has no such prohibition under the CAA and is free to consider both credits and shortfalls.

These increasing credit shortfalls are occurring at a time that the industry is deploying more technology than the agencies anticipated when establishing future standards in 2012. The agencies' projections of transmission technologies were mixed. While the agencies expected the deployment of 8-speed transmissions to about 25 percent of the market by MY 2018, transmissions with eight or more gears account for almost 30 percent of the market. However, the agencies projected no CVT transmissions in future model years, instead projecting high penetration of DCTs. However, CVTs currently make up more than 20 percent of new transmissions. The tradeoff between advanced engines and electrification was also underestimated. While the agencies projected penetration rates of turbocharged engines that are higher than we've observed in the market (45 percent compared to 30 percent), the estimated penetration of electric technologies was significantly lower. The agencies projected a couple percent of strong hybrids—which we've seen—but virtually no PHEVs or EVs. While the volumes of those vehicles are still only a couple percent of the new vehicle market, they are heavily credited under both programs and can significantly improve compliance positions even at smaller volumes. Even lower-level electrification technologies, like stop-start systems, are significantly more prevalent than we anticipated (stop-start systems were projected to be in about 2 percent of the market, compared to over 20 percent in the 2018 fleet). Despite technology deployment that is comparable to 2012 projections, and occasionally more aggressive, passenger car and light truck fleets have slightly lower fuel economy than projected. As fleet volumes have shifted along the footprint curve, the standards have decreased as well (relative to the expectation in 2012), but less so. While compliance deficits have been modest, they have been accompanied by record sales for several years. This has not only depleted existing credit banks, but created significant shortfalls that may be more difficult to overcome if sales recede from record levels.

V. Regulatory Alternatives Considered

Agencies typically consider regulatory alternatives in proposals as a way of evaluating the comparative effects of different potential ways of accomplishing their desired goal. NEPA Start Printed Page 24243requires agencies (in this case, NHTSA, but not EPA) to compare the potential environmental impacts of their proposed actions to those of a reasonable range of alternatives. Executive Orders 12866 and 13563 and OMB Circular A-4 also encourage agencies to evaluate regulatory alternatives in their rulemaking analyses. Alternatives analysis begins with a “no-action” alternative, typically described as what would occur in the absence of any regulatory action. This final rule, like the proposal, includes a no-action alternative, described below, as well as seven “action alternatives.” The final standards may, in places, be referred to as the “preferred alternative,” which is NEPA parlance, but NHTSA and EPA intend “final standards” and “preferred alternative” to be used interchangeably for purposes of this rulemaking.

In the proposal, NHTSA and EPA defined the different regulatory alternatives (other than the no-action alternative) in terms of percent-increases in CAFE and CO2 stringency from year to year. Percent increases in stringency referred to changes in the standards year over year—as in, standards that become 1 percent more stringent each year. Readers should recognize that those year-over-year changes in stringency are not measured in terms of mile per gallon or CO2 gram per mile differences (as in, 1 percent more stringent than 30 miles per gallon in one year equals 30.3 miles per gallon in the following year), but in terms of shifts in the footprint functions that form the basis for the actual CAFE and CO2 standards (as in, on a gallon or gram per mile basis, the CAFE and CO2 standards change by a given percentage from one model year to the next). Under some alternatives, the rate of change was the same for both passenger cars and light trucks; under others, the rate of change differed. Like the no-action alternative, all of the alternatives considered in the proposal were more stringent than the preferred alternative.

Alternatives considered in the proposal also varied in other significant ways. Alternatives 3 and 7 in the proposal involved a gradual discontinuation of CAFE and average CO2 adjustments reflecting the use of technologies that improve air conditioner efficiency or otherwise improve fuel economy under conditions not represented by long-standing fuel economy test procedures (off-cycle adjustments, described in further detail in Section IX, although the proposal itself would have retained these flexibilities. Commenters responding to the request for comment on phasing out these flexibilities generally supported maintaining the existing program, as proposed. Some commenters suggested changes to the existing program that were not discussed in the NPRM. Such changes would be beyond the scope of this rulemaking and were not considered. Section IX contains a more thorough summary of these comments and the issues they raise, as well as the agencies' responses. Consistent with the decision to retain these flexibilities in the final rule, alternatives reflecting their phase-out have not been considered in this final rule.

Additionally, in the NPRM for this rule, EPA proposed to exclude the option for manufacturers partially to comply with tailpipe CO2 standards by generating CO2-equivalent emission adjustments associated with air conditioning refrigerants and leakage after MY 2020. This approach was proposed in the interest of improved harmonization between the CAFE and tailpipe CO2 emissions programs because this optional flexibility cannot be available in the CAFE program.[177] Alternatives 1 through 8 excluded this option. EPA requested comment “on whether to proceed with [the] proposal to discontinue accounting for A/C leakage, methane emissions, and nitrous oxide emissions as part of the CO2 emissions standards to provide for better harmony with the CAFE program, or whether to continue to consider these factors toward compliance and retain that as a feature that differs between the programs.” [178] EPA stated that if EPA were to proceed with excluding A/C refrigerant credits as proposed, “EPA would consider whether it is appropriate to initiate a new rulemaking to regulate these programs independently . . . .” [179] EPA also stated that “[i]f the agency decides to retain the A/C leakage . . . provisions for CO2 compliance, it would likely re-insert the current A/C leakage offset and increase the stringency levels for CO2 compliance by the offset amounts described above (i.e., 13.8 g/mi equivalent for passenger cars and 17.2 g/mi equivalent for light trucks). EPA received comments from a wide range of stakeholders, most of whom opposed the elimination of these flexibility provisions.

Specifically, the two major trade organizations of auto manufacturers, as well as some individual automakers, supported retaining these provisions. Global Automakers commented that “[a]ir conditioning refrigerant leakage . . . should be included for compliance with the EPA standards for all MYs, even if it means a divergence from the NHTSA standards.” [180] Global provides several detailed reasons for their comments, including that the existing provisions are “. . . important to maintaining regulatory flexibility through real [CO2] emission reductions and would prevent the potential for additional bifurcated, separate programs at the state level.” [181] The Alliance similarly commented that it “supports continuation of the full air conditioning refrigerant leakage credits under the [CO2] standards.” [182] Some individual Start Printed Page 24244manufacturers, including General Motors,[183] Fiat Chrysler,[184] and BMW,[185] also commented in support of maintaining the current A/C refrigerant and leakage credits.

Auto manufacturing suppliers who addressed A/C refrigerant and leakage credits also generally supported retaining the existing provisions. MEMA commented that “It is essential for supplier investment and jobs, and continuous innovation and improvements in the technologies that the credit programs continue and expand to broaden the compliance pathways. MEMA urges the agencies to continue the current credit and incentives programs . . . . ” [186] DENSO also supported maintaining the current provisions.[187] However, BorgWarner supported the proposed removal of A/C refrigerant credits “for harmonization reasons,” while encouraging EPA to regulate A/C refrigerants and leakage separately from the CO2 standards.[188]

The two producers of a lower GWP refrigerant, Chemours and Honeywell, commented extensively in support of continuing to allow A/C refrigerant and leakage credits for CO2 compliance, making both economic and legal arguments. Both Chemours and Honeywell stated that A/C refrigerant and leakage credits were a highly cost-effective way for OEMs to comply with the CO2 standards,[189] with Chemours suggesting that OEM compliance strategies are based on the assumption that these credits will be available for CO2 compliance [190] and that any increase in stringency above the proposal effectively necessitates that the credits remain part of the program.[191] Honeywell stated that all OEMs (and a variety of other parties) supported retaining the credits for CO2 compliance,[192] and Chemours, Honeywell, and CBD et al. all noted that OEMs are already using the credits for low GWP refrigerants in more than 50 percent of the MY 2018 vehicles produced for sale in the U.S.[193] The American Chemistry Council also stated that the “auto industry widely supports the credits, and U.S. chemical manufacturers are at a loss as to why EPA would propose to eliminate such a successful flexible compliance program.” [194] In response to NPRM statements expressing concern that the A/C refrigerant and leakage credits could be market distorting, both Chemours and Honeywell disagreed,[195] arguing that the credits were simply a highly cost-effective means of complying with the CO2 standards,[196] and that removal of the credits at this point would, itself, distort the market for refrigerants. Honeywell argued that eliminating the A/C credit program from CO2 compliance would put the U.S. at a competitive disadvantage with other countries, and would risk U.S. jobs.[197]

Regarding the NPRM's statements that removing the A/C refrigerant and leakage credits from CO2 compliance would promote harmonization with the CAFE program, these commenters argued that harmonization was not a valid basis for that aspect of the proposal. Chemours, Honeywell, and CBD et al. all argued that Section 202(a) creates no obligation to harmonize the [CO2] program with the CAFE program.[198] Chemours further argued that to the extent disharmonization between the programs existed, it should be addressed via stringency changes (i.e., reducing CAFE stringency relative to CO2 stringency) rather than “dropping low-cost compliance options.” [199]

These commenters also expressed concern that the proposal constituted an EPA decision not to regulate HFC emissions from motor vehicles at all. Commenters argued that the NPRM provided no legal analysis or reasoned explanation for stopping regulation of HFCs,[200] and that Massachusetts v. EPA requires any final rule to regulate all greenhouse gases from motor vehicles and not CO2 alone,[201] suggesting that there was a high likelihood of a lapse in regulation because EPA had not yet proposed a new way of regulating HFC emissions.[202] Because the NPRM provided no specific information about how EPA might regulate non-CO2 emissions separately, commenters argued that the lack of clarity was inherently disruptive to OEMs.[203] CBD et al. argued that any lapse in regulation is “illegal on its face” and that even creating a separate standard for HFC emissions would be “illegal” because it “would increase costs to manufacturers and result in environmental detriment by removing any incentive to use the most aggressive approaches to curtail emissions of these highly potent GHGs.” [204]

Environmental organizations,[205] other NGOs, academic institutions, consumer organizations, and state governments [206] also commented in support of continuing the existing provisions.

EPA has considered its proposed approach to A/C refrigerant and leakage Start Printed Page 24245credits in light of these comments. EPA believes that maintaining this element of its program is consistent with EPA's authority under Section 202(a) to establish standards for reducing emissions from LDVs. Thus, maintaining the optional HFC credit program is appropriate. In addition, EPA recognizes the value of regulatory flexibility and compliance options, and has concluded that the advantages from retaining the existing A/C refrigerant/leakage credit program and associated offset between the CO2 and CAFE standards—in terms of providing for a more-comprehensive regulation of emissions from light-duty vehicles—outweigh the disadvantages resulting from the lack of harmonization.

Regarding the comment from BorgWarner about how having a separate A/C refrigerant and leakage regulation would allow for better harmonization between the programs, the agencies accept this to be an accurate statement, but believe the benefits of continued refrigerant regulation as an option for CO2 compliance outweigh the problems associated with lack of harmonization with the CAFE program.

For these reasons, EPA is not finalizing the proposed provisions, and is making no changes in the A/C refrigerant and leakage-related provisions of the current program. In light of this conclusion, EPA does not need to address the legal arguments made by CBD et al. and CARB about regulating refrigerant-related emissions separately, or potential lapses in regulation of refrigerant emissions while such a program could be developed.

As with A/C refrigerant and leakage credits, EPA proposed to exclude nitrous oxide and methane from average performance calculations after model year 2020, thereby removing these optional program flexibilities. Alternatives 1 through 8 excluded this option. EPA sought comment on whether to remove those aspects of the program that allow a manufacturer to use nitrous oxide and methane emissions reductions for compliance with its CO2 average fleet standards because such a flexibility is not allowed in the NHTSA CAFE program, or whether to retain the flexibilities as a feature that differs between the programs. Further, EPA sought comment on whether to change the existing methane and nitrous oxide standards. Specifically, EPA requested information from the public on whether the existing standards are appropriate, or whether they should be revised to be less stringent or more stringent based on any updated data.

The Alliance in its comments may have misunderstood EPA's proposal to mean that EPA was proposing to eliminate regulation of methane and nitrous oxide emissions altogether. The Alliance commented in support of such a proposal as they understood it, to eliminate the standards to provide better harmony between the two compliance programs.[207] The Alliance commented that “[n]ot only is emission of these two substances from vehicles a relatively minor contribution to GHG emissions, the Alliance has continuing concern regarding measurement and testing technologies for nitrous oxide.” [208] The Alliance commented further that if “EPA decides instead to continue to regulate methane and nitrous oxide, the Alliance recommends that EPA re-assess whether the levels of the standards remain appropriate and to retain the current compliance flexibilities. Furthermore, in this scenario, the Alliance also recommends that methane and nitrous oxide standards be assessed as a fleet average and as the average of FTP and HFET test cycles.” [209] Several individual manufacturers submitted similar comments, including Ford,[210] FCA,[211] Volvo,[212] and Mazda.[213] Ford also commented that it does not support the proposal to maintain the existing N2 O/CH4 standards while removing the program flexibilities.[214]

The Alliance further commented that “data from the 2016 EPA report on light-duty vehicle emissions supports the position that CH4 and N2 O have minimal impact on total GHG emissions, reporting only 0.045 percent in exceedance of the standard. This new information makes it apparent that CH4 and N2 O contribute a de minimis amount to GHG emissions. Additionally, gasoline CH4 and N2 O performance is within the current standards. Finally, the main producers of CH4 and N2 O emissions are flex fuel (E85) and diesel vehicles, and these vehicles have been declining in sales as compared to gasoline-fueled vehicles.” [215] The Alliance also commented that CH4 and N2 O have minimal opportunities to be catalytically treated, as N2 O is generated in the catalyst and CH4 has a low conversion efficiency compared to other emissions. EPA did not intend that additional hardware should be required to comply with the CH4 or N2 O standards on any vehicle.” [216]

Global Automakers commented in support of continuing inclusion of nitrous oxide and methane emissions standards for all MYs, even if it means a divergence from the NHTSA standards for these program elements in the regulations, “because they are complementary to EPA's program, and are better managed through a coordinated federal policy. They are also important to maintaining regulatory flexibility through real [CO2] emission reductions and would prevent the potential for additional bifurcated, separate programs at the state level.” [217] Global Automakers recommended that they remain in place per the existing program but continued to support that the N2 O testing is not necessary. Global Automakers commented that it “strongly recommends reducing the need for N2 O testing or eliminating these test requirements in their entirety. It should be sufficient to allow manufacturers to attest to compliance with the N2 O capped standards based upon good engineering judgment, development testing, and correlation to NOX emissions. EPA could, however, maintain the option to request testing to be performed for new technologies only, which could have unknown impacts on N2 O emissions.” [218] Hyundai [219] and Kia [220] submitted similar comments.

Others commented in support of retaining the existing program. MECA commented that it supports the existing standards for methane and nitrous oxide because catalyst technologies provided by MECA members that reduce these climate forcing gases are readily Start Printed Page 24246available and cost-effective.[221] MECA also commented that the ability to trade reductions in these pollutants in exchange for CO2 gives vehicle manufacturers the flexibilities they need to comply with the emission limits by the most cost-effective means.[222] CBD et al. commented that the alternative compliance mechanisms currently available in the program exist to provide cost-effective options for compliance, and were considered by manufacturers to be a necessary element of the program for certain types of vehicles.[223] CBD et al. further argued that “[e]liminating these flexibilities consequently imposes costs on manufacturers without discernible environmental benefits,” and suggested that harmonization with the CAFE program was not a relevant decision factor for EPA.[224] Several other parties commented generally in support of retaining the existing program for A/C leakage credits, discussed above, and N2 O and CH4 standards.[225]

After considering these comments, EPA is retaining the regulatory provisions related to the N2 O and CH4 standards with no changes, specifically including the existing flexibilities that accompany those standards. EPA is not adopting its proposal to exclude nitrous oxide and methane emissions from average performance calculations after model year 2020 or any other changes to the program. The standards continue to serve their intended purpose of capping emissions of those pollutants and providing for more-comprehensive regulation of emissions from light-duty vehicles. The standards were intended to prevent future emissions increases, and these standards were generally not expected to result in the application of new technologies or significant costs for manufacturers using current vehicle designs.[226] The program flexibilities are working as intended and all manufacturers are successfully complying with the standards. Most vehicle models are well below the standards and for those that are above the standards, manufacturers have used the flexibilities to offset exceedances with CO2 improvements to demonstrate compliance. EPA did not receive any data in response to its request for comments supporting potential alternative levels of stringency.

While the Alliance and several individual manufacturers recommended eliminating the standards altogether, EPA did not propose to eliminate the standards, but to eliminate the optional flexibilities, and solicited comment on adjusting the standards to be more or less stringent. Thus, EPA does not believe it would be appropriate to eliminate completely the standards in this final rule without providing an opportunity for comment on that idea. Furthermore, as noted above, EPA believes the standards are continuing to serve their intended purpose of capping emissions and remain appropriate. Manufacturers have been subject to the standards for several years, and the Alliance acknowledges in their comments that the exceedance of the standards, which is offset by manufacturers using compliance flexibilities, is very small and that most vehicles meet the standards. Regarding the Alliance comments that the standards should be based on a fleet average approach, EPA notes that the purpose of the standards is to cap emissions, not to achieve fleet-wide reductions.[227] The fleet average emissions for N2 O and CH4 are well below the numerical level of the cap standards and therefore the existing cap standards would not be an appropriate fleet average standard. Adopting a fleet average approach using the same numerical level as the established cap standards would not achieve the intended goal of capping emissions at current levels. If technologies lead to exceedances of the caps, automakers have the opportunity to apply appropriate flexibilities under the current program to achieve GHG emission neutrality. EPA is not aware of any manufacturer that has been prevented from bringing a technology to the marketplace because of the current cap levels or approach. EPA believes it would need to consider all options further, with an opportunity for public comment, before adopting such a significant change to the program.

As explained above, the agencies have changed the alternatives considered for the final rule, partly in response to comments. The basic form of the standards represented by the alternatives—footprint-based, defined by particular mathematical functions—remains the same and as described in the NPRM. For the EPA program, EPA has chosen in this final rule to retain the existing program for regulation of A/C refrigerant leakage, nitrous oxide, and methane emissions as part of the CO2 standard. This allows manufacturers to continue to rely on this flexibility which they describe as extremely important for compliance, although it results in continued differences between EPA's and NHTSA's programs. This approach also avoids the possibility of gaps in the regulation of HFCs, CH4, and N2 O while EPA developed a different way of regulating the non-CO2 emissions as part of or concurrent with the NPRM, and thereby allows EPA to continue to regulate GHE emissions from light-duty vehicles on a more-comprehensive basis. Thus, all alternatives considered in this final rule reflect inclusion of CH4, N2 O, and HFC in EPA's overall “CO2” (more accurately, CO2-equivalent, or CO2 e) requirements. Besides this change, the alternatives considered for the final rule differ from the NPRM in two additional ways: First, alternatives reflecting the phase-out of the A/C efficiency and off-cycle programs have been dropped in response to certain comments and in recognition of the potential real-world benefits of those programs. And second, the preferred alternative for this final rule reflects a 1.5 percent year-over-year increase for both passenger cars and light trucks. These changes will be discussed further below, following a brief discussion of the form of the standards.

A. Form of the Standards

As in the CAFE and CO2 rulemakings in 2010 and 2012, NHTSA and EPA proposed in the NPRM to set attribute-based CAFE and CO2 standards defined by a mathematical function of vehicle footprint, which has observable correlation with fuel economy and vehicle emissions. EPCA, as amended by EISA, expressly requires that CAFE standards for passenger cars and light trucks be based on one or more vehicle attributes related to fuel economy and be expressed in the form of a mathematical function.[228] While the CAA includes no specific requirements regarding CO2 regulation, EPA has chosen to adopt attribute-based CO2 standards consistent with NHTSA's EPCA/EISA requirements in the interest of harmonization and simplifying compliance. Such an approach is permissible under section 202(a) of the Start Printed Page 24247CAA, and EPA has used the attribute-based approach in issuing standards under analogous provisions of the CAA. Thus, both the proposed and final standards take the form of fuel economy and CO2 targets expressed as functions of vehicle footprint (the product of vehicle wheelbase and average track width). Section V.A.2 below discusses the agencies' continued reliance on footprint as the relevant attribute.

Under the footprint-based standards, the function defines a CO2 or fuel economy performance target for each unique footprint combination within a car or truck model type. Using the functions, each manufacturer thus will have a CAFE and CO2 average standard for each year that is almost certainly unique to each of its fleets,[229] based upon the footprints and production volumes of the vehicle models produced by that manufacturer. A manufacturer will have separate footprint-based standards for cars and for trucks. The functions are mostly sloped, so that generally, larger vehicles (i.e., vehicles with larger footprints) will be subject to lower CAFE mpg targets and higher CO2 grams/mile targets than smaller vehicles. This is because, generally speaking, smaller vehicles are more capable of achieving higher levels of fuel economy/lower levels of CO2 emissions, mostly because they tend not to have to work as hard (and therefore require as much energy) to perform their driving task. Although a manufacturer's fleet average standards could be estimated throughout the model year based on the projected production volume of its vehicle fleet (and are estimated as part of EPA's certification process), the standards to which the manufacturer must comply are determined by its final model year production figures. A manufacturer's calculation of its fleet average standards as well as its fleets' average performance at the end of the model year will thus be based on the production-weighted average target and performance of each model in its fleet.[230]

For passenger cars, consistent with prior rulemakings, NHTSA is defining fuel economy targets as follows:


TARGETFE is the fuel economy target (in mpg) applicable to a specific vehicle model type with a unique footprint combination,

a is a minimum fuel economy target (in mpg),

b is a maximum fuel economy target (in mpg),

c is the slope (in gallons per mile per square foot, or gpm, per square foot) of a line relating fuel consumption (the inverse of fuel economy) to footprint, and

d is an intercept (in gpm) of the same line.

Here, MIN and MAX are functions that take the minimum and maximum values, respectively, of the set of included values. For example, MIN[40,35] = 35 and MAX (40, 25) = 40, such that MIN[MAX (40, 25), 35] = 35.

For light trucks, also consistent with prior rulemakings, NHTSA is defining fuel economy targets as follows:


TARGETFE is the fuel economy target (in mpg) applicable to a specific vehicle model type with a unique footprint combination,

a, b, c, and d are as for passenger cars, but taking values specific to light trucks,

e is a second minimum fuel economy target (in mpg),

f is a second maximum fuel economy target (in mpg),

g is the slope (in gpm per square foot) of a second line relating fuel consumption (the inverse of fuel economy) to footprint, and

h is an intercept (in gpm) of the same second line.

Although the general model of the target function equation is the same for each vehicle category (passenger cars and light trucks) and each model year, the parameters of the function equation differ for cars and trucks. For MYs 2020-2026, the parameters are unchanged, resulting in the same stringency in each of those model years.

Mathematical functions defining the CO2 targets are expressed as functions that are similar, with coefficients a-h corresponding to those listed above.[231] For passenger cars, EPA is defining CO2 targets mathematically equivalent to the following:



TARGETCO2 is the is the CO2 target (in grams per mile, or g/mi) applicable to a specific vehicle model configuration,

a is a minimum CO2 target (in g/mi),

b is a maximum CO2 target (in g/mi),

c is the slope (in g/mi, per square foot) of a line relating CO2 emissions to footprint, and

d is an intercept (in g/mi) of the same line.

For light trucks, CO2 targets are defined as follows:

TARGETCO2 = MIN[MIN[b, MAX[a, c × FOOTPRINT + d]], MIN[f, MAX[e, g × FOOTPRINT + h]]

Start Printed Page 24248


TARGETCO2 is the is the CO2 target (in g/mi) applicable to a specific vehicle model configuration,

a, b, c, and d are as for passenger cars, but taking values specific to light trucks,

e is a second minimum CO2 target (in g/mi),

f is a second maximum CO2 target (in g/mi),

g is the slope (in g/mi per square foot) of a second line relating CO2 emissions to footprint, and

h is an intercept (in g/mi) of the same second line.

To be clear, as has been the case since the agencies began establishing attribute-based standards, no vehicle need meet the specific applicable fuel economy or CO2 targets, because compliance with either CAFE or CO2 standards is determined based on corporate average fuel economy or fleet average CO2 emission rates. In this respect, CAFE and CO2 standards are unlike, for example, safety standards and traditional vehicle emissions standards. CAFE and CO2 standards apply to the average fuel economy levels and CO2 emission rates achieved by manufacturers' entire fleets of vehicles produced for sale in the U.S. Safety standards apply on a vehicle-by-vehicle basis, such that every single vehicle produced for sale in the U.S. must, on its own, comply with minimum FMVSS. Similarly, criteria pollutant emissions standards are applied on a per-vehicle basis, such that every vehicle produced for sale in the U.S. must, on its own, comply with all applicable emissions standards. When first mandating CAFE standards in the 1970s, Congress specified a more flexible averaging-based approach that allows some vehicles to “under comply” (i.e., fall short of the overall flat standard, or fall short of their target under attribute-based standards) as long as a manufacturer's overall fleet is in compliance.

The required CAFE level applicable to a given fleet in a given model year is determined by calculating the production-weighted harmonic average of fuel economy targets applicable to specific vehicle model configurations in the fleet, as follows:


CAFErequired is the CAFE level the fleet is required to achieve,

i refers to specific vehicle model/configurations in the fleet,

PRODUCTIONi is the number of model configuration i produced for sale in the U.S., and

TARGETFE,i the fuel economy target (as defined above) for model configuration i.

Similarly, the required average CO2 level applicable to a given fleet in a given model year is determined by calculating the production-weighted average (not harmonic) of CO2 targets applicable to specific vehicle model configurations in the fleet, as follows:


CO2required is the average CO2 level the fleet is required to achieve,

i refers to specific vehicle model/configurations in the fleet,

PRODUCTIONi is the number of model configuration i produced for sale in the U.S., and

TARGETCO2,i is the CO2 target (as defined above) for model configuration i.

Section VI.A.1 describes the advantages of attribute standards, generally. Section VI.A.2 explains the agencies' specific decision to use vehicle footprint as the attribute over which to vary stringency for past and current rules. Section VI.A.3 discusses the policy considerations in selecting the specific mathematical function. Section VI.A.4 discusses the methodologies used to develop current attribute-based standards, and the agencies' current proposal to continue to do so for MYs 2021-2026. Section VI.A.5 discusses the methodologies used to reconsider the mathematical function for the proposed standards.

1. Why attribute-based standards, and what are the benefits?

Under attribute-based standards, every vehicle model has fuel economy and CO2 targets, the levels of which depend on the level of that vehicle's determining attribute (for the MYs 2021-2026 standards, footprint is the determining attribute, as discussed below). The manufacturer's fleet average CAFE performance is calculated by the harmonic production-weighted average of those targets, as defined below:

Here, i represents a given model [232] in a manufacturer's fleet, Productioni represents the U.S. production of that model, and Targeti represents the target as defined by the attribute-based standards. This means no vehicle is required to meet its target; instead, manufacturers are free to balance improvements however they deem best within (and, given credit transfers, at least partially across) their fleets.

Because CO2 is on a gram per mile basis rather a mile per gallon basis, Start Printed Page 24249harmonic averaging is not necessary when calculating required CO2 levels:

The idea is to select the shape of the mathematical function relating the standard to the fuel economy-related attribute to reflect the trade-offs manufacturers face in producing more of that attribute over fuel efficiency (due to technological limits of production and relative demand of each attribute). If the shape captures these trade-offs, every manufacturer is more likely to continue adding fuel-efficient technology across the distribution of the attribute within their fleet, instead of potentially changing the attribute—and other correlated attributes, including fuel economy—as a part of their compliance strategy. Attribute-based standards that achieve this have several advantages.

First, assuming the attribute is a measurement of vehicle size, attribute-based standards help to at least partially reduce the incentive for manufacturers to respond to CAFE and CO2 standards by reducing vehicle size in ways harmful to safety, as compared to “flat,” non-attribute based standards.[233] Larger vehicles, in terms of mass and/or crush space, generally consume more fuel and produce more carbon dioxide emissions, but are also generally better able to protect occupants in a crash.[234] Because each vehicle model has its own target (determined by a size-related attribute), properly fitted attribute-based standards reduce the incentive to build smaller vehicles simply to meet a fleet-wide average, because smaller vehicles are subject to more stringent compliance targets.

Second, attribute-based standards, if properly fitted, provide automakers with more flexibility to respond to consumer preferences than do single-valued standards. As discussed above, a single-valued standard encourages a fleet mix with a larger share of smaller vehicles by creating incentives for manufacturers to use downsizing the average vehicle in their fleet (possibly through fleet mixing) as a compliance strategy, which may result in manufacturers building vehicles for compliance reasons that consumers do not want. Under a size-related, attribute-based standard, reducing the size of the vehicle for compliance's sake is a less-viable strategy because smaller vehicles have more stringent regulatory targets. As a result, the fleet mix under such standards is more likely to reflect aggregate consumer demand for the size-related attribute used to determine vehicle targets.

Third, attribute-based standards provide a more equitable regulatory framework across heterogeneous manufacturers who may each produce different shares of vehicles along attributes correlated with fuel economy.[235] An industry-wide single-value CAFE standard imposes disproportionate cost burden and compliance challenges on manufacturers who produce more vehicles with attributes inherently correlated with lower fuel economy—i.e. manufacturers who produce, on average, larger vehicles. As discussed above, retaining flexibility for manufacturers to produce vehicles which respect heterogeneous market preferences is an important consideration. Since manufacturers may target different markets as a part of their business strategy, ensuring that these manufacturers do not incur a disproportionate share of the regulatory cost burden is an important part of conserving consumer choices within the market.

Industry commenters generally supported attribute-based standards, while other commenters questioned their benefits. IPI argued that preserving the current vehicle mix was not necessarily desirable or necessary for consumer welfare, and suggested that some vehicle downsizing in the fleet might be beneficial both for safety and for compliance.[236] IPI also argued that compliance credit trading would “help smooth out any disproportionate impacts on certain manufacturers” and “ensure that manufacturers with relatively efficient fleets still have an incentive to continue improving fuel economy (in order to generate credits)” [237] Similarly, citing Ito and Sallee, Kathryn Doolittle commented that “. . . Ito and Sallee (2018) have found ABR [“attribute-based regulations”] inefficient in cost when juxtaposed with flat standard with compliance trading.” [238]

The agencies have considered these comments. IPI incorrectly characterizes the agencies' prior statements as claims that it is important to preserve the current vehicle mix. EPA and NHTSA have never claimed, and are not today claiming that it is important to preserve the current fleet mix. The agencies have said, and are today reiterating, that it is reasonable to expect that reducing the tendency of standards to distort the market should reduce at least part of the tendency of standards to reduce consumer welfare. Or, more concisely, it is better to work with the market than against it. Single-value (aka flat) CAFE standards in place from the 1970s through 2010 were clearly distortionary. Recognizing this, the National Academy of Sciences recommended in 2002 that NHTSA adopt attribute-based CAFE standards. NHTSA did so in 2006, for light trucks produced starting MY 2008. As mentioned above, in 2007, Congress codified the requirement for attribute-based passenger car and light truck CAFE standards. Agreeing with this history, premise, and motivation, EPA has also adopted attribute-based CO2 standards. None of this is to say the agencies consider it important to hold fleet mix constant. Rather, the agencies expect that, compared to flat standards, attribute-based standards can allow the market—including fleet mix—to better Start Printed Page 24250follow its natural course, and all else equal, consumer acceptance is likely to be greater if the market does so.

The agencies also disagree with comments implying that compliance credit trading can address all of the market distortion that flat standards would entail. Evidence thus far suggests that trading is fragmented, with some manufacturers apparently willing to trade only with some other specific manufacturers. The Ito and Sallee article cited by one commenter is a highly idealized theoretical construction, with the authors noting, inter alia, that their model “assumes perfect competition.” [239] Its findings regarding comparative economic efficiency of flat- and attribute-based standards are, therefore, merely hypothetical, and the agencies find little basis in recent transactions to suggest the compliance credit trading market reflects the authors' idealized assumptions. Even if the agencies did expect credit trading markets to operate as in an idealized textbook example, basing the structure of standards on the presumption of perfect trading would not be appropriate. FCA commented that “. . . when flexibilities are considered while setting targets, they cease to be flexibilities and become simply additional technology mandates,” and the Alliance commented, similarly, that “the Agencies should keep `flexibilities' as optional ways to comply and not unduly assume that each flexibility allows additional stringency of footprint-based standards.” [240] Perhaps recognizing this reality, Congress has barred NHTSA from considering manufacturers' ability to use compliance credits (even credits earned and used by the same OEM, much less credits traded between OEMs). As discussed further in Section VIII.A.2, EPA believes that while credit trading may be a useful flexibility to reduce the overall costs of the program, it is important to set standards in a way that does not rely on credit purchasing availability as a compliance mechanism.

Considering these comments and realities, considering EPCA's requirement for attribute-based CAFE standards, and considering the benefits of regulatory harmonization, the agencies are, again, finalizing attribute-based CAFE and CO2 standards rather than, for either program, finalizing flat standards.

Why footprint as the attribute?

It is important that the CAFE and CO2 standards be set in a way that does not unnecessarily incentivize manufacturers to respond by selling vehicles that are less safe. Vehicle size is highly correlated with vehicle safety—for this reason, it is important to choose an attribute correlated with vehicle size (mass or some dimensional measure). Given this consideration, there are several policy and technical reasons why footprint is considered to be the most appropriate attribute upon which to base the standards, even though other vehicle size attributes (notably, curb weight) are more strongly correlated with fuel economy and tailpipe CO2 emissions.

First, mass is strongly correlated with fuel economy; it takes a certain amount of energy to move a certain amount of mass. Footprint has some positive correlation with frontal surface area, likely a negative correlation with aerodynamics, and therefore fuel economy, but the relationship is less deterministic. Mass and crush space (correlated with footprint) are both important safety considerations. As discussed below and in the accompanying PRIA, NHTSA's research of historical crash data indicates that holding footprint constant, and decreasing the mass of the largest vehicles, will result in a net positive safety impact to drivers overall, while holding footprint constant and decreasing the mass of the smallest vehicles will result in a net decrease in fleetwide safety. Properly fitted footprint-based standards provide little, if any, incentive to build smaller footprint vehicles to meet CAFE and CO2 standards, and therefore help minimize the impact of standards on overall fleet safety.

Second, it is important that the attribute not be easily manipulated in a manner that does not achieve the goals of EPCA or other goals, such as safety. Although weight is more strongly correlated with fuel economy than footprint, there is less risk of artificial manipulation (i.e., changing the attribute(s) to achieve a more favorable target) by increasing footprint under footprint-based standards than there would be by increasing vehicle mass under weight-based standards. It is relatively easy for a manufacturer to add enough weight to a vehicle to decrease its applicable fuel economy target a significant amount, as compared to increasing vehicle footprint, which is a much more complicated change that typically takes place only with a vehicle redesign.

Further, some commenters on the MY 2011 CAFE rulemaking were concerned that there would be greater potential for such manipulation under multi-attribute standards, such as those that also depend on weight, torque, power, towing capability, and/or off-road capability. As discussed in NHTSA's MY 2011 CAFE final rule,[241] it is anticipated that the possibility of manipulation is lowest with footprint-based standards, as opposed to weight-based or multi-attribute-based standards. Specifically, standards that incorporate weight, torque, power, towing capability, and/or off-road capability in addition to footprint would not only be more complex, but by providing degrees of freedom with respect to more easily adjusted attributes, they could make it less certain that the future fleet would actually achieve the projected average fuel economy and CO2 levels. This is not to say that a footprint-based system eliminates manipulation, or that a footprint-based system eliminates the possibility that manufacturers will change vehicles in ways that compromise occupant protection, but footprint-based standards achieve the best balance among affected considerations.

Several stakeholders commented on whether vehicular footprint is the most suitable attribute upon which to base standards. IPI commented that “. . . footprint-based standards may be unnecessary to respect consumer preferences, may negatively impact safety, and may be overall inefficient. Several arguments call into question the footprint-based approach, but a particularly important one is that large vehicles can impose a negative safety externality on other drivers.” [242] IPI commented, further, that the agencies should consider the relative merits of other vehicle attributes, including vehicle fuel type, suggesting that it would be more difficult for manufacturers to manipulate a flatter standard or one “differentiated by fuel type.” [243] Similarly, Michalek and Whitefoot recommended “that the agencies reexamine automaker response to the footprint-based standards to determine if adjustments should be made to avoid inducing increases to vehicle size.” [244]

Start Printed Page 24251

Conversely, ICCT commented that “the switch to footprint-based CAFE and [CO2] standards has been widely credited with diminishing safety concerns with efficiency standards. Footprint standards encourage larger vehicles with wider track width, which reduces rollovers, and longer wheelbase, which increases the crush space and reduces deceleration forces for both vehicles in a two-vehicle collision.” [245] Similarly, BorgWarner commented that “the use of a footprint standard not only provides greater incentive for mass reduction, but also encourages a larger footprint for a given vehicle mass, thus providing increased safety for a given mass vehicle,” [246] and the Aluminum Association commented footprint based standards drive “fuel-efficiency improvement across all vehicle classes,” “eliminate the incentive to shift fleet volume to smaller cars which has been shown to slightly decrease safety in vehicle-to-vehicle collisions,” and provide “an incentive for reducing weight in the larger vehicles, where weight reduction is of the most benefit for societal safety,” citing Ford's aluminum-intensive F150 pickup truck as an example.[247] NADA urged the agencies to continue basing standards on vehicle footprint, as doing so “serves both to require and allow OEMs to build more fuel-efficient vehicles across the broadest possible light-duty passenger car and truck spectrum,” [248] and UCS commented that footprint-based standards “increase consumer choice, ensuring that the vehicles available for purchase in every vehicle class continue to get more efficient.” [249] Furthermore, regarding concerns that footprint-based standards may be susceptible to manipulation, the Alliance commented that “the data above [from Novation Analytics] shows there are no systemic footprint increases (or any type of target manipulation) occurring.” [250] While FCA's comments supported this Alliance comment, FCA commented further that, lacking some utility-related vehicle attributes such as towing capability, 4-wheel-drive, and ride height, “it is clear the footprint standard does not fully account for pickup truck capability and the components needed such as larger powertrains, greater mass and frontal area,” and requested the agencies “correct LDT standards to reflect the current market preference for capability over efficiency, and introduce mechanisms into the regulation that can adjust for efficiency and capability tradeoffs that footprint standards currently ignore.” [251]

When first electing to adopt footprint-based standards, NHTSA carefully considered other alternatives, including vehicle mass and “shadow” (overall width multiplied by overall length). Compared to both of these other alternatives, footprint is much less susceptible to gaming, because while there is some potential to adjust track width, wheelbase is more expensive to change, at least outside a planned vehicle redesign. EPA agreed with NHTSA's assessment, nothing has changed the relative merits of at least these three potential attributes, and nothing in the evolution of the fleet demonstrates that footprint-based standards are leading manufacturers to increase the footprint of specific vehicle models by more than they would in response to customer demand. Also, even if footprint-based standards are encouraging some increases in vehicle size, NHTSA continues to maintain, and EPA to agree, that such increases should tend to improve overall highway safety rather than degrading it. Regarding FCA's request that the agencies adopt an approach that accounts for a wider range of vehicle attributes related to both vehicle fuel economy and customer-facing vehicle utility, the agencies are concerned that doing so could further complicate already-complex standards and also lead to unintended consequences. For example, it is not currently clear how a multi-attribute approach would appropriately balance emphasis between vehicle attributes (e.g., how much relative fuel consumption should be attributed to, respectively, vehicle footprint, towing capacity, drive type, and ground clearance). Also, basing standards on, in part, ground clearance would encourage manufacturers to increase ride height, potentially increasing the frequency of vehicle rollover crashes. Regarding IPI's recommendation that fuel type be included as a vehicle attribute for attribute-based standards, the agencies note that both CAFE and CO2 standards already account for fuel type in the procedures for measuring fuel economy levels and CO2 emission rates, and for calculating fleet average CAFE and CO2 levels.

Therefore, having considered public comments on the choice of vehicle attributes for CAFE and CO2 standards, the agencies are finalizing standards that, as proposed, are defined in terms of vehicle footprint.

3. What mathematical function should be used to specify footprint-based standards?

In requiring NHTSA to “prescribe by regulation separate average fuel economy standards for passenger and non-passenger automobiles based on 1 or more vehicle attributes related to fuel economy and express each standard in the form of a mathematical function,” EPCA/EISA provides ample discretion regarding not only the selection of the attribute(s), but also regarding the nature of the function. The CAA provides no specific direction regarding CO2 regulation, and EPA has continued to harmonize this aspect of its CO2 regulations with NHTSA's CAFE regulations. The relationship between fuel economy (and CO2 emissions) and footprint, though directionally clear (i.e., fuel economy tends to decrease and CO2 emissions tend to increase with increasing footprint), is theoretically vague, and quantitatively uncertain; in other words, not so precise as to a priori yield only a single possible curve.

The decision of how to specify this mathematical function therefore reflects some amount of judgment. The function can be specified with a view toward achieving different environmental and petroleum reduction goals, encouraging different levels of application of fuel-saving technologies, avoiding any adverse effects on overall highway safety, reducing disparities of manufacturers' compliance burdens, and preserving consumer choice, among other aims. The following are among the specific technical concerns and resultant policy tradeoffs the agencies have considered in selecting the details of specific past and future curve shapes:

  • Flatter standards (i.e., curves) increase the risk that both the size of vehicles will be reduced, potentially compromising highway safety, and reducing any utility consumers would have gained from a larger vehicle.
  • Steeper footprint-based standards may create incentives to upsize vehicles, potentially oversupplying vehicles of certain footprints beyond what consumers would naturally demand, and thus increasing the possibility that fuel savings and CO2 reduction benefits will be forfeited artificially.
  • Given the same industry-wide average required fuel economy or CO2 standard, flatter standards tend to place greater compliance burdens on full-line manufacturers.
  • Given the same industry-wide average required fuel economy or CO2Start Printed Page 24252standard, dramatically steeper standards tend to place greater compliance burdens on limited-line manufacturers (depending of course, on which vehicles are being produced).
  • If cutpoints are adopted, given the same industry-wide average required fuel economy, moving small-vehicle cutpoints to the left (i.e., up in terms of fuel economy, down in terms of CO2 emissions) discourages the introduction of small vehicles, and reduces the incentive to downsize small vehicles in ways that could compromise overall highway safety.
  • If cutpoints are adopted, given the same industry-wide average required fuel economy, moving large-vehicle cutpoints to the right (i.e., down in terms of fuel economy, up in terms of CO2 emissions) better accommodates the design requirements of larger vehicles—especially large pickups—and extends the size range over which downsizing is discouraged.

4. What mathematical functions have been used previously, and why?

Notwithstanding the aforementioned discretion under EPCA/EISA, data should inform consideration of potential mathematical functions, but how relevant data is defined and interpreted, and the choice of methodology for fitting a curve to that data, can and should include some consideration of specific policy goals. This section summarizes the methodologies and policy concerns that were considered in developing previous target curves (for a complete discussion see the 2012 FRIA).

As discussed below, the MY 2011 final curves followed a constrained logistic function defined specifically in the final rule.[252] The MYs 2012-2021 final standards and the MYs 2022-2025 augural standards are defined by constrained linear target functions of footprint, as shown below: [253]

Here, Target is the fuel economy target applicable to vehicles of a given footprint in square feet (Footprint). The upper asymptote, a, and the lower asymptote, b, are specified in mpg; the reciprocal of these values represent the lower and upper asymptotes, respectively, when the curve is instead specified in gallons per mile (gpm). The slope, c, and the intercept, d, of the linear portion of the curve are specified as gpm per change in square feet, and gpm, respectively.

The min and max functions will take the minimum and maximum values within their associated parentheses. Thus, the max function will first find the maximum of the fitted line at a given footprint value and the lower asymptote from the perspective of gpm. If the fitted line is below the lower asymptote it is replaced with the floor, which is also the minimum of the floor and the ceiling by definition, so that the target in mpg space will be the reciprocal of the floor in mpg space, or simply, a. If, however, the fitted line is not below the lower asymptote, the fitted value is returned from the max function and the min function takes the minimum value of the upper asymptote (in gpm space) and the fitted line. If the fitted value is below the upper asymptote, it is between the two asymptotes and the fitted value is appropriately returned from the min function, making the overall target in mpg the reciprocal of the fitted line in gpm. If the fitted value is above the upper asymptote, the upper asymptote is returned is returned from the min function, and the overall target in mpg is the reciprocal of the upper asymptote in gpm space, or b.

In this way curves specified as constrained linear functions are specified by the following parameters:

a = upper limit (mpg)

b = lower limit (mpg)

c = slope (gpm per sq.ft.)

d = intercept (gpm)

The slope and intercept are specified as gpm per sq. ft. and gpm instead of mpg per sq. ft. and mpg because fuel consumption and emissions appear roughly linearly related to gallons per mile (the reciprocal of the miles per gallon).

a) NHTSA in MY 2008 and MY 2011 CAFE (Constrained Logistic)

For the MY 2011 CAFE rule, NHTSA estimated fuel economy levels by footprint from the MY 2008 fleet after normalization for differences in technology,[254] but did not make adjustments to reflect other vehicle attributes (e.g., power-to-weight ratios). Starting with the technology-adjusted passenger car and light truck fleets, NHTSA used minimum absolute deviation (MAD) regression without sales weighting to fit a logistic form as a starting point to develop mathematical functions defining the standards. NHTSA then identified footprints at which to apply minimum and maximum values (rather than letting the standards extend without limit) and transposed these functions vertically (i.e., on a gallons-per-mile basis, uniformly downward) to produce the promulgated standards. In the preceding rule, for MYs 2008-2011 light truck standards, NHTSA examined a range of potential functional forms, and concluded that, compared to other considered forms, the constrained logistic form provided the expected and appropriate trend (decreasing fuel economy as footprint increases), but avoided creating “kinks” the agency was concerned would provide distortionary incentives for vehicles with neighboring footprints.[255]

b) MYs 2012-2016 Standards (Constrained Linear)

For the MYs 2012-2016 rule, potential methods for specifying mathematical functions to define fuel economy and CO2 standards were reevaluated. These methods were fit to the same MY 2008 data as the MY 2011 standard. Considering these further specifications, the constrained logistic form, if applied to post-MY 2011 standards, would likely contain a steep mid-section that would provide undue incentive to increase the footprint of midsize passenger cars.[256] A range of Start Printed Page 24253methods to fit the curves would have been reasonable, and a minimum absolute deviation (MAD) regression without sales weighting on a technology-adjusted car and light truck fleet was used to fit a linear equation. This equation was used as a starting point to develop mathematical functions defining the standards. Footprints were then identified at which to apply minimum and maximum values (rather than letting the standards extend without limit). Finally, these constrained/piecewise linear functions were transposed vertically (i.e., on a gpm or CO2 basis, uniformly downward) by multiplying the initial curve by a single factor for each MY standard to produce the final attribute-based targets for passenger cars and light trucks described in the final rule.[257] These transformations are typically presented as percentage improvements over a previous MY target curve.

c) MYs 2017 and Beyond Standards (Constrained Linear)

The mathematical functions finalized in 2012 for MYs 2017 and beyond changed somewhat from the functions for the MYs 2012-2016 standards. These changes were made both to address comments from stakeholders, and to consider further some of the technical concerns and policy goals judged more preeminent under the increased uncertainty of the impacts of finalizing and proposing standards for model years further into the future.[258] Recognizing the concerns raised by full-line OEMs, it was concluded that continuing increases in the stringency of the light truck standards would be more feasible if the light truck curve for MYs 2017 and beyond was made steeper than the MY 2016 truck curve and the right (large footprint) cut-point was extended only gradually to larger footprints. To accommodate these considerations, the 2012 final rule finalized the slope fit to the MY 2008 fleet using a sales-weighted, ordinary least-squares regression, using a fleet that had technology applied to make the technology application across the fleet more uniform, and after adjusting the data for the effects of weight-to-footprint. Information from an updated MY 2010 fleet was also considered to support this decision. As the curve was vertically shifted (with fuel economy specified as mpg instead of gpm or CO2 emissions) upwards, the right cutpoint was progressively moved for the light truck curves with successive model years, reaching the final endpoint for MY 2021.

5. Reconsidering the Mathematical Functions for Today's Rulemaking

a) Why is it important to reconsider the mathematical functions?

By shifting the developed curves by a single factor, it is assumed that the underlying relationship of fuel consumption (in gallons per mile) to vehicle footprint does not change significantly from the model year data used to fit the curves to the range of model years for which the shifted curve shape is applied to develop the standards. However, it must be recognized that the relationship between vehicle footprint and fuel economy is not necessarily constant over time; newly developed technologies, changes in consumer demand, and even the curves themselves could influence the observed relationships between the two vehicle characteristics. For example, if certain technologies are more effective or more marketable for certain types of vehicles, their application may not be uniform over the range of vehicle footprints. Further, if market demand has shifted between vehicle types, so that certain vehicles make up a larger share of the fleet, any underlying technological or market restrictions which inform the average shape of the curves could change. That is, changes in the technology or market restrictions themselves, or a mere re-weighting of different vehicles types, could reshape the fit curves.

For the above reasons, the curve shapes were reconsidered in the proposal using the newest available data from MY 2016. With a view toward corroboration through different techniques, a range of descriptive statistical analyses were conducted that do not require underlying engineering models of how fuel economy and footprint might be expected to be related, and a separate analysis that uses vehicle simulation results as the basis to estimate the relationship from a perspective more explicitly informed by engineering theory was conducted as well. Despite changes in the new vehicle fleet both in terms of technologies applied and in market demand, the underlying statistical relationship between footprint and fuel economy has not changed significantly since the MY 2008 fleet used for the 2012 final rule; therefore, EPA and NHTSA proposed to continue to use the curve shapes fit in 2012. The analysis and reasoning supporting this decision follows.

b) What statistical analyses did EPA and NHTSA consider?

In considering how to address the various policy concerns discussed above, data from the MY 2016 fleet was considered, and a number of descriptive statistical analyses (i.e., involving observed fuel economy levels and footprints) using various statistical methods, weighting schemes, and adjustments to the data to make the fleets less technologically heterogeneous were performed. There were several adjustments to the data that were common to all of the statistical analyses considered.

With a view toward isolating the relationship between fuel economy and footprint, the few diesels in the fleet were excluded, as well as the limited number of vehicles with partial or full electric propulsion; when the fleet is normalized so that technology is more homogenous, application of these technologies is not allowed. This is consistent with the methodology used in the 2012 final rule.

The above adjustments were applied to all statistical analyses considered, regardless of the specifics of each of the methods, weights, and technology level of the data, used to view the relationship of vehicle footprint and fuel economy. Table V-1, below, summarizes the different assumptions considered and the key attributes of each. The analysis was performed considering all possible combinations of these assumptions, producing a total of eight footprint curves.

Start Printed Page 24254

(1) Current Technology Level Curves

The “current technology” level curves exclude diesels and vehicles with electric propulsion, as discussed above, but make no other changes to each model year fleet. Comparing the MY 2016 curves to ones built under the same methodology from previous model year fleets shows whether the observed curve shape has changed significantly over time as standards have become more stringent. Importantly, these curves will include any market forces which make technology application variable over the distribution of footprint. These market forces will not be present in the “maximum technology” level curves: By making technology levels homogenous, this variation is removed. The current technology level curves built using both regression types and both regression weight methodologies from the MY 2008, MY 2010, and MY 2016 fleets, shown in more detail in Chapter of the PRIA, support the curve slopes finalized in the 2012 final rule. The curves built from most methodologies using each fleet generally shift, but remain very similar in slope. This suggests that the relationship of footprint to fuel economy, including both technology and market limits, has not significantly changed.

(2) Maximum Technology Level Curves

As in prior rulemakings, technology differences between vehicle models were considered to be a significant factor producing uncertainty regarding the relationship between fuel consumption and footprint. Noting that attribute-based standards are intended to encourage the application of additional technology to improve fuel efficiency and reduce CO2 emissions across the distribution of footprint in the fleet, approaches were considered in which technology application is simulated for purposes of the curve fitting analysis in order to produce fleets that are less varied in technology content. This approach helps reduce “noise” (i.e., dispersion) in the plot of vehicle footprints and fuel consumption levels and identify a more technology-neutral relationship between footprint and fuel consumption. The results of updated analysis for maximum technology level curves are also shown in Chapter of the PRIA. Especially if vehicles progress over time toward more similar size-specific efficiency, further removing variation in technology application both better isolates the relationship between fuel consumption and footprint and further supports the curve slopes finalized in the 2012 final rule.

c) What other methodologies were considered?

The methods discussed above are descriptive in nature, using statistical analysis to relate observed fuel economy levels to observed footprints for known vehicles. As such, these methods are clearly based on actual data, answering the question “how does fuel economy appear to be related to footprint?” However, being independent of explicit engineering theory, they do not answer the question “how might one expect fuel economy to be related to footprint?” Therefore, as an alternative to the above methods, an alternative methodology was also developed and applied that, using full-vehicle simulation, comes closer to answering the second question, providing a basis either to corroborate answers to the first, or suggest that further investigation could be important.

As discussed in the 2012 final rule, several manufacturers have confidentially shared with the agencies what they described as “physics-based” curves, with each OEM showing significantly different shapes for the footprint-fuel economy relationships. This variation suggests that manufacturers face different curves given the other attributes of the vehicles in their fleets (i.e., performance Start Printed Page 24255characteristics) and/or that their curves reflected different levels of technology application. In reconsidering the shapes of the proposed MYs 2021-2026 standards, a similar estimation of physics-based curves leveraging third-party simulation work form Argonne National Laboratories (Argonne) was developed. Estimating physics-based curves better ensures that technology and performance are held constant for all footprints; augmenting a largely statistical analysis with an analysis that more explicitly incorporates engineering theory helps to corroborate that the relationship between fuel economy and footprint is in fact being characterized.

Tractive energy is the amount of energy it will take to move a vehicle.[259] Here, tractive energy effectiveness is defined as the share of the energy content of fuel consumed which is converted into mechanical energy and used to move a vehicle—for internal combustion engine (ICE) vehicles, this will vary with the relative efficiency of specific engines. Data from Argonne simulations suggest that the limits of tractive energy effectiveness are approximately 25 percent for vehicles with internal combustion engines which do not possess integrated starter generator, other hybrid, plug-in, pure electric, or fuel cell technology.

A tractive energy prediction model was also developed to support today's proposal. Given a vehicle's mass, frontal area, aerodynamic drag coefficient, and rolling resistance as inputs, the model will predict the amount of tractive energy required for the vehicle to complete the Federal test cycle. This model was used to predict the tractive energy required for the average vehicle of a given footprint [260] and “body technology package” to complete the cycle. The body technology packages considered are defined in Table V-2, below. Using the absolute tractive energy predicted and tractive energy effectiveness values spanning possible ICE engines, fuel economy values were then estimated for different body technology packages and engine tractive energy effectiveness values.

Chapter 6 of the PRIA show the resultant CAFE levels estimated for the vehicle classes Argonne simulated for this analysis, at different footprint values and by vehicle “box.” Pickups are considered 1-box, hatchbacks and minivans are 2-box, and sedans are 3-box. These estimates are compared with the MY 2021 standards finalized in 2012. The general trend of the simulated data points follows the pattern of the previous MY 2021 standards for all technology packages and tractive energy effectiveness values presented in the PRIA. The tractive energy curves are intended to validate the curve shapes against a physics-based alternative, and the analysis suggests that the curve shapes track the physical relationship between fuel economy and tractive energy for different footprint values.

Physical limitations are not the only forces manufacturers face; their success is dependent upon producing vehicles that consumers desire and will purchase. For this reason, in setting future standards, the analysis will continue to consider information from statistical analyses that do not homogenize technology applications in addition to statistical analyses which do, as well as a tractive energy analysis similar to the one presented above.

The relationship between fuel economy and footprint remains directionally discernable but quantitatively uncertain. Nevertheless, each standard must commit to only one function. Approaching the question “how is fuel economy related to footprint” from different directions and applying different approaches has given EPA and NHTSA confidence that the function applied here appropriately and reasonably reflects the relationship between fuel economy and footprint.

The agencies invited comments on this conclusion and the supporting analysis. IPI raised concerns that “. . . several dozen models (mostly subcompacts and sports cars) fall in the 30-40 square feet range, which are all subject to the same standards” and that “manufacturers of these models may have an incentive to decrease footprints as a compliance strategy, since doing so would not trigger more stringent standards.” [261] NHTSA and EPA agree that, all else equal, downsizing the smallest cars (e.g., Chevrolet Spark, Ford Fiesta, Mini Cooper, Mazda MX-5, Porsche 911, Toyota Yaris) would most likely tend to degrade overall highway safety. At the same time, as discussed above, the agencies recognize that small vehicles do appear attractive to some market segments (although obviously the Ford Fiesta and Porsche 911 compete in different segments). Start Printed Page 24256Therefore, there is a tension between on one hand, avoiding standards that unduly encourage safety-eroding downsizing and, on the other, avoiding standards that unduly penalize the market for small vehicles. The agencies examined this issue, and note that the market for the smallest vehicles has not evolved at all as estimated in the analysis supporting the 2012 final rule, and attribute this more to fuel prices and consumer demand for larger vehicles than to attribute-based CAFE and CO2 standards. For example, the market for vehicles with footprints less than 40 square foot was about 45 percent smaller in MY 2017 than in MY 2010. The agencies also found that among the smallest vehicle models produced throughout MYs 2010-2017, most have become larger, not smaller. For example, while the Mazda MX-5's footprint decreased by 0.1 square foot (0.3 percent) during that time, the MY 2017 versions of the Mini Cooper, Smart fortwo, Porsche 911, and Toyota Yaris had larger footprints than in MY 2010. With the market for very small vehicles shrinking, and with manufacturers not evidencing a tendency to make the smallest vehicles even smaller, the agencies are satisfied that it would be unwise to change the target functions such that targets never stop becoming more stringent as vehicle footprint becomes ever smaller, because doing so could further impede an already-shrinking market.

B. No-Action Alternative

As in the proposal, the No-Action Alternative applies the augural CAFE and final CO2 targets announced in 2012 for MYs 2021-2025.[262] For MY 2026, this alternative applies the same targets as for MY 2025. The carbon dioxide equivalent of air conditioning refrigerant leakage credits, nitrous oxide, and methane emissions are included for compliance with the EPA standards for all model years under the no-action alternative.[263]

Start Printed Page 24257

In comments on the DEIS, CBD et al. indicated that it was appropriate for NHTSA to use the augural CAFE standards as the baseline No Action regulatory alternative.[264] However, CARB commented that the baseline regulatory alternative should include CARB's ZEV mandate, in part because EPA must consider “other regulations promulgated by EPA or other government entities,” and, according to CARB, there will be much more vehicle electrification in the future as manufacturers respond to market demand and also work to comply with the ZEV mandate.[265] Similarly, EPA's Science Advisory Board recommended—despite the action taken in the One National Program Action—that the baseline include state ZEV mandates “to be consistent with policies that would prevail in the absence of the rule change.” [266] EPA's Science Advisory Board further recommended including sensitivity analyses with different penetration rates of ZEVs.

On the other hand, arguing for consideration of standards less stringent than those proposed in the NPRM, Walter Kreucher commented that rather than using the augural standards as the baseline, “a better approach would be to assume a clean sheet of paper and start from the existing 2016MY fleet and its associated standards as the baseline using 0%/year increases for both passenger cars and light trucks for MYs 2017-2026.” [267] Similarly, AVE argued that because previously-promulgated standards for MYs 2018-2021 already present a significant challenge that “will likely require almost every automaker to continue using credits for compliance, . . . AVE believes this rulemaking should reset . . . the current compliance baseline for cars and light trucks at MY 2018 . . .” [268] BorgWarner commented similarly that “Beginning in MY 2018, standards should be reset to the levels the industry actually achieved. For MY 2018 and beyond, succeeding model year targets should be set with an annual rate of improvement defined by the slope of improvement the industry has achieved over the last six years. . . . Based on these data, our analysis suggests the most reasonable and logical rate of improvement falls between 2.0% to 2.6% for cars and trucks. Additionally, a single rate of improvement for the combined fleet should be considered.” [269]

The No-Action Alternative represents expectations regarding the world in the absence of a proposal, accounting for applicable laws already in place. Although manufacturers are already making significant use of compliance credits toward compliance with even MY 2017 standards, the agencies are obligated to evaluate regulatory alternatives against the standards already in place through MY 2025. Similarly, even though manufacturers are already producing electric vehicles, EPA and NHTSA appropriately excluded California's ZEV mandate from the No-Action alternative for the NPRM, for several reasons. First, the ZEV mandate is not Federal law; second, as described in the proposal and subsequently finalized in regulatory text, the ZEV mandate is expressly and impliedly preempted by EPCA; third, EPA proposed to withdraw the waiver of CAA preemption in the NPRM and subsequently finalized this withdrawal. Accordingly, the agencies have, therefore, appropriately excluded the ZEV mandate from the No-Action alternative. However, as discussed below, the agencies' analysis does account for the potential that under every regulatory alternative, including the No-Action Alternative, vehicle electrification could increase in the future, especially if batteries become less expensive as gasoline becomes more expensive.

C. Action Alternatives

1. Alternatives in Final Rule

Table V-5 below shows the different alternatives evaluated in today's notice.

Start Printed Page 24258

With one exception, the alternatives considered in the NPRM included the changes in stringency for the above alternatives. Alternative 3, the preferred alternative, is newly included for today's notice.[270]

Regulations regarding implementation of NEPA requires agencies to “rigorously explore and objectively evaluate all reasonable alternatives, and for alternatives which were eliminated from detailed study, briefly discuss the reasons for their having been eliminated.” [271] This does not amount to a requirement that agencies evaluate the widest conceivable spectrum of alternatives. For example, a State considering adding a single travel lane to a preexisting section of highway would not be required to consider adding three lanes, or to consider dismantling the highway altogether.

Among thousands of individual comments that mentioned the proposed standards very generally, some comments addressed the range and definition of these regulatory alternatives in specific terms, and these specific comments include comments on the stringency, structure, and particular provisions defining the set of regulatory alternatives under consideration.

As discussed throughout today's notice, the agencies have updated and otherwise revised many aspects of the analysis. The agencies have also reconsidered whether the set of alternatives studied in detail should be expanded to include standards less stringent than the proposal's preferred alternative, or to include standards more stringent than the proposal's no-action alternative. On one hand, comments from Walter Kreucher and AVE cited above indicate the agencies should consider relaxing standards below MY 2020 levels, and CEI challenged the agencies' failure to include less-stringent alternatives in the following comments on this question:

DOT failed to consider the possibility of freezing CAFE at an even more lenient standard than currently exists, nor did it consider making its proposed freeze take effect sooner than MY 2020. However, as DOT's own analysis strongly indicates, doing so would lead to even greater benefits and an even greater reduction in CAFE-related deaths and injuries. In short, DOT's failure to consider this possibility is arbitrary and capricious. It has an opportunity to remedy this in its final rule, and it should do so by selecting a standard that is even more lenient than the one it proposed. . . . It should have gone beyond its original set of alternatives and examined less stringent ones as well—until it found one that, for some reason or another, failed to produce greater safety benefits or failed to meet the statutory factors.[272]

On the other hand, a coalition of ten environmental advocacy organizations stated that the agencies should consider alternatives more stringent than those defining the baseline no action alternative, arguing that in light of CEQ guidance and the 2018 IPCC report on climate change, “the increasing danger, increasing urgency, and increasing importance of vehicle emissions all rationally counsel for strengthening emission standards.” [273] CBD et al. observe that “none of these alternatives [considered in the NPRM] increases fuel economy in comparison with the No Action Alternative, none conserves energy . . .” and go on to assert that “none represents maximum feasible CAFE standards.” [274] Similarly, EDF commented that “. . . given its clear statutory directive to maximize fuel savings, NHTSA should have considered a range of alternatives that would be more protective than the existing standards,” [275] and three State agencies in Minnesota commented that “more stringent standards are consistent with EPCA's purpose of energy conservation and the CAA's purpose of reducing harmful air pollutants.” [276] The North Carolina Department of Environmental Quality acknowledged the agencies' determination in the proposal that alternatives beyond the augural standards might be economically impracticable, but nevertheless argued that “alternatives that exceed the stringency of the current standards are consistent with EPCA's purpose” [277] In oral testimony before the agencies, the New York State Attorney General also indicated that the agencies should consider alternatives more stringent than the augural standards.[278] A coalition of States and cities commented that “at a minimum, the existing standards should be left in place, but EPA should also consider whether to make the standards more stringent, not less, just as it has done in prior proposals.” [279] More specifically, through International Mosaic, some individuals commented that the agencies must “fully and publicly consider a few options that require at least a seven annual percent [sic] improvement in vehicle fleet mileage.” [280] In comments on the DEIS, CBD, et al. went further, commenting that “NHTSA's most stringent alternative must be set at no lower than a 9 percent improvement per year.” [281] Most manufacturers who commented on stringency did not identify specific regulatory alternatives that the agencies should consider, although Honda suggested that standards be set to increase in stringency at 5 percent annually for both passenger cars and light trucks throughout model years 2021-2026.[282 283]

The agencies carefully considered these comments to expand the range of stringencies to be evaluated as possible candidates for promulgation. To inform this consideration, the agencies used the CAFE model to examine a progression of stringencies extending outside the range presented in the proposal and draft EIS, and as a point of reference, using a case that reverts to MY 2018 standards starting in MY 2021. Scenarios included in this initial screening exercise ranged as high as increasing annually at 9.5 percent during MYs 2021-2026, reaching average CAFE and CO2 requirements of 66 mpg and 120 g/mi, respectively. Results of this analysis are presented in the following tables and charts. Focusing on MY 2029, the tables show average required and achieved CAFE (as mpg) and CO2 (as g/mi) levels for each scenario, along with average per-vehicle costs (in 2018 dollars, relative to retaining MY 2017 technologies). The proposed (0%/0%), final (1.5%/1.5%), and baseline augural standards are shown in bold type. The charts present Start Printed Page 24259the same results on a percentage basis, relative to values shown below for the scenario that reverts to MY 2018 standards starting in MY 2021.

For example, reverting to the MY 2018 CAFE standards starting in MY 2021 yields an average CAFE requirement of 35 mpg by MY 2029, with the industry exceeding that standard by 5 mpg at an average cost of $1,255 relative to MY 2017 technology. Under the augural standards, the MY 2029 requirement increases to 47 mpg, the average compliance margin falls to 1 mpg, and the average cost increases to $2,770. In other words, compared to the scenario that reverts to MY 2018 stringency starting in MY 2021, the augural standards increase stringency by 34 percent (from 35 to 47 mpg), increase average fuel economy by 20 percent (from 40 to 48 mpg), and increase costs by 121 percent (from $1,255 to $2,770).

As indicated in the following two charts, the reality of diminishing returns clearly applies in both directions. On one hand, relaxing stringency below the proposed standards by reverting to MY 2018 or MY 2019 standards reduces average MY 2029 costs by only modest amounts ($54-$121). As discussed in Section VIII, the agencies' updated analysis indicates that the proposed standards would not be maximum feasible considering the EPCA/EISA statutory factors, and would not be appropriate under the CAA after considering the appropriate factors. If further relaxation of standards appeared likely to yield more significant cost reductions, it is conceivable that such savings could outweigh further foregoing of energy and climate benefits. However, this screening analysis does not show dramatic cost reductions. Therefore, the agencies did not include these two less stringent alternatives in the detailed analysis presented in Section VII.

On the other hand, increases in stringency beyond the baseline augural standards show relative costs continuing to accrue much more rapidly than relative CAFE and CO2 improvements. As discussed below in Section VIII, even the no action alternative is already well beyond levels that can be supported under the CAA and EPCA. If further stringency increases appeared likely to yield more significant additional energy and environmental benefits, it is conceivable that these could outweigh these significant additional cost increases. However, this screening analysis shows no dramatic relative acceleration of energy and environmental benefits. Therefore, the agencies did not include stringencies beyond the augural standards in the detailed analysis presented in Section VII.

Start Printed Page 24260

Start Printed Page 24261

Specific to model year 2021, some commenters argued that EPCA's lead time requirement prohibits NHTSA from revising CAFE standards for model year 2021.[284] Regarding the revision of standards for model year 2021, NHTSA did consider EPCA's lead time requirement, and determined that while the agency would need to finalize a stringency increase at least 18 months before the beginning of the first affected model year, the agency can finalize a stringency decrease closer (or even after) the beginning of the first affected model year. The agency's reasoning is explained further in Section VIII. Therefore, NHTSA did not change regulatory alternatives to avoid any relaxation of stringency in model year 2021.

The Auto Alliance stated that “the truck increase rate should be no greater than the car rate of increase and should be the `equivalent task' per fleet.” [285] Supporting these Alliance comments, FCA elaborated by commenting that “(1) in MY2017, the latest data we have available, most trucks have a larger gap to standards than cars, and (2) all of the truck segments are challenged because consumers are placing a greater emphasis on capability than fuel economy.” [286] Similarly, Ford commented that “. . . the rates of increase in the stringency of the standards should remain equivalent between passenger cars and light duty trucks.” [287] Other commenters expressed general support for equalizing the rates at which the stringencies of passenger car and light truck standards increase.[288]

For the final rule, the agencies have added an alternative in which stringency for both cars and trucks increases at 1.5 percent. This is consistent with comments received requesting that both fleets' standards increase in stringency by the same amount, and 1.5 percent represents a rate of increase within the range of rates of increase considered in the NPRM.

Throughout the NPRM, the agencies described their consideration as covering a range of alternatives.[289] The preferred alternative for this final rule, an increase in stringency of 1.5 percent for both cars and trucks, falls squarely Start Printed Page 24262within the range of alternatives proposed by the agencies.

The NPRM alternatives were bounded on the upper end by the baseline/no action alternative, and the proposed alternative on the lower end (0 percent per year increase in stringency for both cars and trucks). For passenger cars, the agencies considered a range of stringency increases between 0 percent and 2 percent per year for passenger cars, in addition to the baseline/no action alternative. For light trucks, the agencies considered a range of stringency increases between 0 percent and 3 percent per year, in addition to the baseline/no action alternative.

The agencies considered the same range of alternatives for this final rule. As with the proposal, the alternatives for stringency are bounded on the upper end by the baseline/no action alternative and on the lower end by 0 percent per year increases for both passenger cars and light trucks. Consistent with the proposal, for this final rule, the agencies considered stringency increases of between 0 and 2 percent per year for passenger cars and between 0 and 3 percent per year for light trucks, in addition to the baseline/no action alternative.

While it was not specifically modeled in the NPRM, the new preferred alternative of an increase in stringency of 1.5 percent for both cars and trucks was well within the range of alternatives considered. The proposal described the alternatives specifically modeled as options for the agencies, but also gave notice that they did not limit the agencies in selecting from among the range of alternatives under consideration.[290]

The agencies explained in the proposal that they were “taking comment on a wide range of alternatives and have specifically modeled eight alternatives.” [291] As with the proposal, for the final rule, the agencies specifically modeled the upper and lower bounds of the baseline/no action alternative and 0 percent per year stringency increases for both passenger cars and light trucks. In both the proposal and the final rule, the agencies also modeled a stringency increase of 2 percent per year for passenger cars and 3 percent per year for light trucks, as well as a variety of other specific increases between 0 and 2 percent for passenger cars and 0 and 3 percent for light trucks.

The specific alternatives the agencies modeled for the final rule reflect their consideration of public comments. As discussed above, multiple commenters expressed support for equalizing the rates at which the stringencies of passenger car and light truck standards increase. To help the agencies evaluate alternatives that include the same stringency increase for passenger cars and light trucks, three of the seven alternatives (in addition to the baseline/no action alternative) that the agencies specifically modeled for the final rule included the same stringency increase for passenger cars and light trucks. This includes the new preferred alternative of an increase in stringency of 1.5 percent for both cars and trucks. This alternative, and all others specifically modeled for the final rule, falls within the range of alternatives for stringency considered by the agencies in the proposal.

Beyond these stringency provisions discussed in the NPRM, the agencies also sought comment on a number of additional compliance flexibilities for the programs, as discussed in Section IX.

2. Additional Alternatives Suggested by Commenters

Beyond the comments discussed above regarding the shapes of the functions defining fuel economy and CO2 targets, regarding the inclusion of non-CO2 emissions, and regarding the stringencies to be considered, the agencies also received a range of other comments regarding regulatory alternatives.

Some of these additional comments involved how CAFE and CO2 standards compare to one another for any given regulatory alternative. With a view toward maximizing harmonization of the standards, the Alliance, supported by some of its members' individual comments, indicated that “to the degree flexibilities and incentives are not completely aligned between the CAFE and [CO2] programs, there must be an offset in the associated footprint-based targets to account for those differences. Some areas of particular concerns are air conditioning refrigerant credits, and incentives for advanced technology vehicles. The Alliance urges the Agencies to seek harmonization of the standards and flexibilities to the greatest extent possible. . . .” [292]

On the other hand, discussing consideration of compliance credits but making a more general argument, the NYU Institute for Policy Integrity commented that “. . . EPA is not allowed to set lower standards just for the sake of harmonization; to the contrary, full harmonization may be inconsistent with EPA's statutory responsibilities.” [293] Similarly, ACEEE argued that “any consideration of an extension or expansion of credit provisions under the [carbon dioxide] or CAFE standards program should take as a starting point the assumption that the additional credits will allow the stringency of the standards to be increased.” [294]

EPCA's requirement that NHTSA set standards at the maximum feasible levels is separate and “wholly independent” from the CAA's requirement, per Massachusetts v. EPA, that EPA issue regulations addressing pollutants that EPA has determined endanger public health and welfare.[295] Nonetheless, as recognized by the Supreme Court, “there is no reason to think the two agencies cannot both administer their obligations and yet avoid inconsistency.” [296] This conclusion was reached despite the fact that EPCA has a range of very specific requirements about how CAFE standards are to be structured, how manufacturers are to comply, what happens when manufacturers are unable to comply, and how NHTSA is to approach setting standards, and despite the fact that the CAA has virtually no such requirements. This means that while nothing about either EPCA or the CAA, much less the combination of the two, guarantees “harmonization” defining “One National Program,” the agencies are expected to be able to work out the differences.

Since tailpipe CO2 standards are de facto fuel economy standards, the more differences there are between CO2 and CAFE standards and compliance provisions, the more challenging it is for manufacturers to plan year-by-year production that responses to both, and the more difficult it is for affected stakeholders and the general public to understand regulation in this space. Therefore, even if the two statutes, taken together, do not guarantee “full harmonization,” steps toward greater Start Printed Page 24263harmonization help with compliance planning and transparency—and meet the expectations set forth by the Supreme Court that the agencies avoid inconsistencies.

The agencies have taken important steps toward doing so. For example, EPA has adopted separate footprint-based CO2 standards for passenger cars and light trucks, and has redefined CAFE calculation procedures to introduce recognition for the application of real-world fuel-saving technology that is not captured with traditional EPA two-cycle compliance testing. Detailed aspects of both sets of standards and corresponding compliance provisions are discussed at length in Section IX. The agencies never set out with the primary goal of achieving “full harmonization,” such that both sets of standards would lead each manufacturer to respond in exactly the same way in every model year.[297] For example, EPA did not adopt the EPCA requirement that domestic passenger car fleets each meet a minimum standard, or the EPCA cap on compliance credit transfers between passenger car fleets. On the other hand, EPA also did not adopt the EPCA civil penalty provisions that have allowed some manufacturers to pay civil penalties as an alternative method of meeting EPCA obligations. These and other differences provide that even if CAFE and CO2 standards are “mathematically” harmonized, for any given manufacturer, the two sets of standards will not be identically burdensome in each model year. Inevitably, one standard will be more challenging than the other, varying over time, between manufacturers, and between fleets. This means manufacturers need to have compliance plans for both sets of standards.

In 2012, recognizing that EPCA provides no clear basis to address HFC, CH4, or N2 O emissions directly, the agencies “offset” CO2 targets from fuel economy targets (after converting the latter to a CO2 basis) by the amounts of credit EPA anticipated manufacturers would, on average, earn in each model years by reducing A/C leakage and adopting refrigerants with reduced GWPs. In 2012, EPA assumed that by 2021, all manufacturers would be earning the maximum available credit, and EPA's analysis assumed that all manufacturers would make progress at the same rate. However, as discussed above, data highlighted in comments by Chemours, Inc., demonstrate that actual manufacturers' adoption of lower-GWP refrigerants thus far ranges widely, with some manufacturers (e.g., Nissan) having taken no such steps to move toward lower-GWP refrigerants, while others (e.g., JLR) have already applied lower-GWP refrigerants to all vehicles produced for sale in the U.S. Therefore, at least in practice, HFC provisions thus far continue to leave a gap (in terms of harmonization) between the two sets of standards. The proposal would have taken the additional step of decoupling provisions regarding HFC (i.e., A/C leakage credits), CH4, and N2 O emissions from CO2 standards, addressing these in separate regulations to be issued in a new proposal. As discussed above, EPA did not finalize this proposal. Accordingly, for the regulatory alternatives considered today, EPA has reinstated offsets of CO2 targets from fuel economy targets, reflecting the assumption that all manufacturers will be earning the maximum available A/C leakage credit by MY 2021.

In addition to general comments on harmonization, the agencies received a range of comments on specific provisions—especially involving “flexibilities”—that may or may not impact harmonization. With a view toward encouraging further electrification, NCAT proposed that EPA extend indefinitely the exclusion of upstream emissions from electricity generation, and also extend and potentially restructure production multipliers for PHEVs, EVs, and FCVs.[298] On the other hand, connecting its comments back to the stringency of standards, NCAT also commented that “. . . expansion of compliance flexibilities in the absence of any requirement to improve [CO2] reduction or fuel economy (as under the agencies' preferred option) could result in an effective deterioration of existing [CO2] and fuel economy performance, as well as little or no effective support for advanced vehicle technology development or deployment.” [299] Global Automakers indicated that the final rule “should include a package of programmatic elements that provide automakers with flexible compliance options that promote the full breadth of vehicle technologies,” such options to include the extension of “advanced technology” production multipliers through MY 2026, the indefinite exclusion of emissions from electricity generation, the extension to passenger cars of credits currently granted for the application of “game changing” technologies (e.g., HEVs) only to full-size pickup trucks, an increase (to 15 g/mi) of the cap on credits for off-cycle technologies, an updated credit “menu” of off-cycle technologies, and easier process for handling applications for off-cycle credits.[300] The Alliance also called for expanded sales multipliers and a permanent exclusion of emissions from electricity generation.[301] Walter Kreucher recommended the agencies consider finalizing the proposed standards but also keeping the augural standards as “voluntary targets” to “provide compliance with the statutes and an aspirational goal for manufacturers.” [302]

The agencies have carefully considered these comments, and have determined that the current suite of “flexibilities” generally provide ample incentive more rapidly to develop and apply advanced technologies and technologies that produce fuel savings and/or CO2 reductions that would otherwise not count toward compliance. The agencies also share some stakeholders' concern that expanding these flexibilities could increase the risk of “gaming” that would make compliance less transparent and would unduly compromise energy and environmental benefits. Nevertheless, as discussed in Section IX, EPA is adopting new multiplier incentives for natural gas vehicles. EPA is also finalizing some changes to procedures for evaluating applications for off-cycle credits, and expects these changes to make this process more accurate and more efficient. Also, EPA is revising its regulations to not require manufacturers to account for upstream emissions associated with electricity use for electric vehicles and plug-in hybrid electric vehicles through model year 2026; compliance will instead be based on tailpipe emissions performance only and not include emissions from electricity generation until model year 2027. As discussed below, even with this change, and even accounting for continued increases in fuel prices and reductions in battery prices, BEVs are projected in this final rule analysis to continue to account for less than 5 percent of new light vehicle sales in the U.S. through model year 2026. To the extent that this projection turns out to reflect reality, this means that the impact of upstream emissions from electricity use on the projected CO2Start Printed Page 24264reductions associated with these standards would likely remain small. Regarding comments suggesting that the augural standards should be finalized as “voluntary targets,” the agencies have determined that having such targets exist alongside actual regulatory requirements would be, at best, unnecessary and confusing.

Beyond these additional proposals, some commenters' proposals clearly fell outside authority provided under EPCA or the CAA. Ron Lindsay recommended the agencies “consider postponing the rule changes until the U.S. can establish a legally binding national and international carbon budget and a binding mechanism to adhere to it.” [303] EPCA requires NHTSA to issue standards for MY 2022 by April 1, 2020, and previously-issued EPA regulations commit EPA to revisiting MY 2021-2025 standards on a similar schedule. These statutory and regulatory provisions do not include a basis to delay decisions pending an international negotiation for which prospects and schedules are both unknown.

SCAQMD, supported by Shyam Shukla, indicated that the agencies should consider an alternative that keeps the waiver for California's CO2 standards in place.[304] NCAT and the North Carolina DEQ offered similar comments and CBD, et al. commented that “among the set of more stringent alternatives that NEPA requires the agency to consider, NHTSA must include action alternatives that retain the standards California and other states have lawfully adopted.” [305] As discussed above, the agencies recently issued a final rule addressing the issue of California's authority. NEPA does not require NHTSA to include action alternatives that cannot be lawfully realized.

International Mosiac commented that NHTSA's DEIS “is fatally flawed . . . because it does not consider any market-based alternatives (e.g., a `cap and trade' type option).” [306] While EPCA/EISA does include very specific provisions regarding trading of CAFE compliance credits, the statute provides no authority for a broad-based cap-and-trade program involving other sectors. Similarly, Michalek, et al. wrote that “a more economically efficient approach of, taxing emissions and fuel consumption at socially appropriate levels would allow households to determine whether to reduce fuel consumption and emissions by driving less, by buying a vehicle with more fuel saving technologies, or by buying a smaller vehicle—or, alternatively, not to reduce fuel consumption and emissions at all but rather pay a cost based on the damages they cause. Forcing improvements only through one mechanism (fuel-saving technologies) increases the cost of achieving these outcomes.” [307] While some economists would agree with these comments, Congress has provided no clear authority for NHTSA or EPA to implement either an emissions tax or a broad-based cap-and-trade program in which motor vehicles could participate.

3. Details of Alternatives Considered in Final Rule

a) Alternative 1

Alternative 1 holds the stringency of targets constant and MY 2020 levels through MY 2026.

Start Printed Page 24265

b) Alternative 2

Alternative 2 increases the stringency of targets annually during MYs 2021-2026 (on a gallon per mile basis, starting from MY 2020) by 0.5 percent for passenger cars and 0.5 percent for light trucks.

Start Printed Page 24266

c) Alternative 3

Alternative 3; the final standards promulgated today, increases the stringency of targets annually during MYs 2021-2026 (on a gallon per mile basis, starting from MY 2020) by 1.5 percent for passenger cars and 1.5 percent for light trucks.

Start Printed Page 24267

d) Alternative 4

Alternative 4 increases the stringency of targets annually during MYs 2021-2026 (on a gallon per mile basis, starting from MY 2020) by 1.0 percent for passenger cars and 2.0 percent for light trucks.

Start Printed Page 24268

e) Alternative 5

Alternative 5 increases the stringency of targets annually during MYs 2022-2026 (on a gallon per mile basis, starting from MY 2021) by 1.0 percent for passenger cars and 2.0 percent for light trucks.

Start Printed Page 24269

f) Alternative 6

Alternative 6 increases the stringency of targets annually during MYs 2021-2026 (on a gallon per mile basis, starting from MY 2020) by 2.0 percent for passenger cars and 3.0 percent for light trucks.

Start Printed Page 24270

g) Alternative 7

Alternative 7 increases the stringency of targets annually during MYs 2022-2026 (on a gallon per mile basis, starting from MY 2021) by 2.0 percent for passenger cars and 3.0 percent for light trucks.

EPCA, as amended by EISA, requires that any manufacturer's domestically-manufactured passenger car fleet must meet the greater of either 27.5 mpg on average, or 92 percent of the average fuel economy projected by the Secretary for the combined domestic and non-domestic passenger automobile fleets manufactured for sale in the U.S. by all manufacturers in the model year, which projection shall be published in the Federal Register when the standard for that model year is promulgated in accordance with 49 U.S.C. 32902(b).[308] Any time NHTSA establishes or changes a passenger car standard for a model year, the MDPCS for that model year must also be evaluated or re-evaluated and established accordingly. Thus, this final rule establishes the applicable MDPCS for MYs 2021-2026. Table V-22 lists the minimum domestic passenger car standards.

Start Printed Page 24271

VI. Analytical Approach as Applied to Regulatory Alternatives

A. Overview of Methods

Like analyses accompanying the NPRM and past CAFE and CAFE/CO2 rulemakings, the analysis supporting today's notice spans a range of technical topics, uses a range of different types of data and estimates, and applies several different types of computer models. The purpose of the analysis is not to determine the standards, but rather to provide information for consideration in doing so. The analysis aims to answer the question “what impacts might each of these regulatory alternatives have?”

Over time, NHTSA's and, more recently, NHTSA's and EPA's analyses have expanded to address an increasingly wide range of types of impacts. Today's analysis involves, among other things, estimating how the application of various combinations of technologies could impact vehicles' costs and fuel economy levels (and CO2 emission rates), estimating how vehicle manufacturers might respond to standards by adding fuel-saving technologies to new vehicles, estimating how changes in new vehicles might impact vehicle sales and operation, and estimating how the combination of these changes might impact national-scale energy consumption, emissions, highway safety, and public health. In addition, the EIS accompanying today's notice addresses impacts on air quality and climate. The analysis of these factors informs and supports both NHTSA's application of the statutory requirements governing the setting of “maximum feasible” fuel-economy standards under EPCA, including, among others, technological feasibility and economic practicability, and EPA's application of the CAA requirements for tailpipe emissions.

Supporting today's analysis, the agencies have brought to bear a variety of different types of data, a few examples of which include fuel economy compliance reports, historical sales and average characteristics of light-duty vehicles, historical economic and demographic measures, historical travel demand and energy prices and consumption, and historical measures of highway safety. Also supporting today's analysis, the agencies have applied several different types of estimates, a few examples of which include projections of the future cost of different fuel-saving technologies, projections of future GDP and the number of households, estimates of the “gap” between “laboratory” and on-road fuel economy, and estimates of the social cost of CO2 emissions and petroleum “price shocks.”

With a view toward transparency, repeatability, and efficiency, the agencies have used a variety of computer models to conduct the majority of today's analysis. For example, the agencies have applied DOE/EIA's National Energy Modeling System (NEMS) to estimate future energy prices, EPA's MOVES model to estimate tailpipe emission rates for ozone precursors and other criteria pollutants, DOE/Argonne's GREET model to estimate emission rates for “upstream” processes (e.g., petroleum refining), and DOE/Argonne's Autonomie simulation tool to estimate the fuel consumption impacts of different potential combinations of fuel-saving technology. In addition, the EIS accompanying today's notice applies photochemical models to estimate air quality impacts, and applies climate models to estimate climate impacts of overall emissions changes.

Use of these different types of data, estimates, and models is discussed further below in the most closely relevant sections. For example, the agencies' use of NEMS is discussed below in the portion of Section VI that addresses the macroeconomic context, which includes fuel prices, and the agencies use of Autonomie is discussed in the portion of Section VI.B.3 that addresses the agencies' approach to estimating the effectiveness of various technologies (in reducing fuel consumption and CO2 emissions).

Providing an integrated means to estimate both vehicle manufacturers' potential responses to CAFE or CO2 standards and, in turn, many of the different potential direct results (e.g., changes in new vehicle costs) and indirect impacts (e.g., changes in rates of fleet turnover) of those responses, the CAFE Model plays a central role in the agencies' analysis supporting today's notice. The agencies used the specific models mentioned above to develop inputs to the CAFE model, such as fuel prices and emission factors. Outputs from the CAFE Model are discussed in Sections VII and VIII of today's notice, and in the accompanying RIA. The EIS accompanying today's notice makes use of the CAFE Model's estimates of changes in total emissions from light-duty vehicles, as well as corresponding changes in upstream emissions. These changes in emissions are included in the set of inputs to the models used to estimate air quality and climate impacts.

The remainder of this overview focuses on the CAFE Model. The purpose of this overview is not to provide a comprehensive technical description of the model,[309] but rather to give an overview of the model's functions, to explain some specific aspects not addressed elsewhere in today's notice, and to discuss some model aspects that were the subject of significant public comment. Some model functions and related comments are addressed in other parts of today's notice. For example, the model's handling of Autonomie-based fuel consumption estimates is addressed in the portion of Section VI.B.3 that discusses the agencies' application of Autonomie. The model documentation accompanying today's notice provides a comprehensive and detailed description of the model's functions, design, inputs, and outputs.

1. Overview of CAFE Model

The basic design of the CAFE Model is as follows: The system first estimates how vehicle manufacturers might respond to a given regulatory scenario, and from that potential compliance solution, the system estimates what impact that response will have on fuel consumption, emissions, and economic externalities. A regulatory scenario involves specification of the form, or shape, of the standards (e.g., flat standards, or linear or logistic attribute-based standards), scope of passenger car and truck regulatory classes, and stringency of the CAFE and CO2 standards for each model year to be analyzed.

Manufacturer compliance simulation and the ensuing effects estimation, collectively referred to as compliance modeling, encompass numerous subsidiary elements. Compliance simulation begins with a detailed user-provided initial forecast of the vehicle models offered for sale during the simulation period. The compliance simulation then attempts to bring each manufacturer into compliance with the standards defined by the regulatory scenario contained within an input file developed by the user. For example, a regulatory scenario may define CAFE or CO2 standards that increase in stringency by 4 percent per year for 5 consecutive years.

The model applies various technologies to different vehicle models in each manufacturer's product line to simulate how each manufacturer might make progress toward compliance with the specified standard. Subject to a variety of user-controlled constraints, the model applies technologies based on Start Printed Page 24272their relative cost-effectiveness, as determined by several input assumptions regarding the cost and effectiveness of each technology, the cost of compliance (determined by the change in CAFE or CO2 credits, CAFE-related civil penalties, or value of CO2 credits, depending on the compliance program being evaluated and the effective-cost mode in use), and the value of avoided fuel expenses. For a given manufacturer, the compliance simulation algorithm applies technologies either until the manufacturer runs out of cost-effective technologies, until the manufacturer exhausts all available technologies, or, if the manufacturer is assumed to be willing to pay civil penalties, until paying civil penalties becomes more cost-effective than increasing vehicle fuel economy. At this stage, the system assigns an incurred technology cost and updated fuel economy to each vehicle model, as well as any civil penalties incurred by each manufacturer. This compliance simulation process is repeated for each model year available during the study period.

This point marks the system's transition between compliance simulation and effects calculations. At the conclusion of the compliance simulation for a given regulatory scenario, the system contains multiple copies of the updated fleet of vehicles corresponding to each model year analyzed. For each model year, the vehicles' attributes, such as fuel types (e.g., diesel, electricity), fuel economy values, and curb weights have all been updated to reflect the application of technologies in response to standards throughout the study period. For each vehicle model in each of the model year specific fleets, the system then estimates the following: Lifetime travel, fuel consumption, carbon dioxide and criteria pollutant emissions, the magnitude of various economic externalities related to vehicular travel (e.g., noise), and energy consumption (e.g., the economic costs of short-term increases in petroleum prices). The system then aggregates model-specific results to produce an overall representation of modeling effects for the entire industry.

Different categorization schemes are relevant to different types of effects. For example, while a fully disaggregated fleet is retained for purposes of compliance simulation, vehicles are grouped by type of fuel and regulatory class for the energy, carbon dioxide, criteria pollutant, and safety calculations. Therefore, the system uses model-by-model categorization and accounting when calculating most effects, and aggregates results only as required for efficient reporting.

2. Representation of the Market

As a starting point, the model needs enough information to represent each manufacturer covered by the program. As discussed below in Section VI.B.1, the MY 2017 analysis fleet contains information about each manufacturer's:

  • Vehicle models offered for sale—their current (i.e., MY 2017) production volumes, manufacturer suggested retail prices (MSRPs), fuel saving technology content and other attributes (curb weight, drive type, assignment to technology class and regulatory class);
  • Production considerations—product cadence of vehicle models (i.e., schedule of model redesigns and “freshenings”), vehicle platform membership, degree of engine and/or transmission sharing (for each model variant) with other vehicles in the fleet; and
  • Compliance constraints and flexibilities—preference for full compliance or penalty payment/credit application, willingness to apply additional cost-effective fuel saving technology in excess of regulatory requirements, projected applicable flexible fuel credits, and current credit balance (by model year and regulatory class) in first model year of simulation.

Representation of Fuel-Saving Technologies

The modeling system defines technology pathways for grouping and establishing a logical progression of technologies that can be applied to a vehicle. Technologies that share similar characteristics form cohorts that can be represented and interpreted within the CAFE Model as discrete entities. The following Table VI-1 shows the technologies available within the modeling system used for this final rule. Each technology is discussed in detail below. However, an understanding of the technologies considered and how they are defined in the model (e.g., a 6-speed manual transmission is defined as “MT6”) is helpful for the following explanation of the compliance simulation and the inputs required for that simulation.

Start Printed Page 24273

Start Printed Page 24274

These entities are then laid out into pathways (or paths), which the system uses to define relations of mutual exclusivity between conflicting sets of technologies. For example, as presented in the next section, technologies on the Turbo Engine path are incompatible with those on the HCR Engine or the Diesel Engine paths. As such, whenever a vehicle uses a technology from one pathway (e.g., turbo), the modeling system immediately disables the incompatible technologies from one or more of the other pathways (e.g., HCR and diesel).

In addition, each path designates the direction in which vehicles are allowed to advance as the modeling system evaluates specific technologies for application. Enforcing this directionality within the model ensures that a vehicle that uses a more advanced or more efficient technology (e.g., AT8) is not allowed to “downgrade” to a less efficient option (e.g., AT5). Visually, as portrayed in the charts in the sections that follow, this is represented by an arrow leading from a preceding technology to a succeeding one, where vehicles begin at the root of each path, and traverse to each successor technology in the direction of the arrows.

The modeling system incorporates twenty technology pathways for evaluation as shown below. Similar to individual technologies, each path carries an intrinsic application level that denotes the scope of applicability of all technologies present within that path, and whether the pathway is evaluated on one vehicle at a time, or on a collection of vehicles that share a common platform, engine, or transmission.

Start Printed Page 24275

Even though technology pathways outline a logical progression between related technologies, all technologies available to the system are evaluated concurrently and independently of each other. Once all technologies have been examined, the model selects a solution deemed to be most cost-effective for application on a vehicle. If the modeling system applies a technology that resides later in the pathway, it will subsequently disable all preceding technologies from further consideration to prevent a vehicle from potentially downgrading to a less advanced option. Consequently, the system skips any technology that is already present on a vehicle (either those that were available on a vehicle from the input fleet or those that were previously applied by the model). This “parallel technology” approach, unlike the “parallel path” methodology utilized in the preceding versions of the model, allows the system always to consider the entire set of available technologies instead of foregoing the application of potentially more cost-effective options that happen to reside further down the pathway.[310] This revised approach addresses comments summarized below, and allows the system to analyze all available technology options concurrently and independently of one other without having to first apply one or more “predecessor” technologies. For example, if model inputs are such that a 7-speed transmission is cost-effective, but not as cost-effective as an 8-speed transmission, the revised approach enables the model to skip over the 7-speed transmission entirely, whereas the NPRM version of the model might first apply the 7-speed transmission and then consider whether to proceed immediately to the 8-speed transmission. As such, the model's choices for evaluation of new technology solutions becomes slightly less restrictive, allowing it immediately to consider and apply more advanced options, and increasing the likelihood that the a globally optimum solution is selected.

Some commenters supported the agencies' use of such pathways in the simulation of manufacturers' potential application of technologies. As one of a dozen examples of CAFE model design elements that lead to the transparent representation of real-world factors, the Alliance highlighted “recognition of the need for manufacturers to follow `technology' pathways that retain capital and implementation expertise, such as specializing in one type of engine or transmission instead of following an unconstrained optimization that would cause manufacturers to leap to unrelated technologies and show overly optimistic costs and benefits.” [311] Similarly, Toyota commented that “the inertia of capital investments and engineering expertise dedicated to one compliance technology or set of technologies makes it unreasonable for manufacturers to immediately switch to another technology path.” [312]

Other commenters cited the use of technology pathways as inherently overly restrictive. For example, as an example of “arbitrary model constraints,” a coalition of commenters cited the fact the model “prohibit[s] manufacturers from switching vehicle technology pathways.” [313] Also, EDF, UCS, and CARB cited the combination of technology pathways, decision making criteria, and model inputs as producing unrealistic results.[314] Regarding the technology pathways, specifically, EDF's consultant argued that the technology paths are not Start Printed Page 24276transparent, and cited the potential that specific paths may not necessarily be arranged in progression from least to most cost-effective—that “NHTSA ignores the cost of the technology when developing this list.” [315] Relatedly, as EDF's consultant commented:

[T]he Volpe Model is not designed to look backwards along its technology paths. Thus, the opportunity to recover the expenditure of inefficient technology is missed. NHTSA might argue that a manufacturer will not invest in 10-speed transmissions, for example, and then return to an older design. Whether or not this is true in real life, such a view would put too much stake in the Volpe Model projections. The model simply projects what could be done, not what will be. Anyone examining the progression of technology and noting the reversion of transmission technology could easily modify the model inputs to avoid this. Also, if NHTSA evaluated combinations of technologies prior to entering them in the model piecemeal, it would automatically avoid such apparent problems.[316]

The agencies also received additional public comments on specific paths and specific interactions between paths (e.g., involving engines and hybridization). These comments are addressed below.

The agencies have carefully considered these comments and the approach summarized below reflects some corresponding revision. As mentioned above, the CAFE model now approaches the technology paths in a such way that, faced with two cost-effective technologies on the same path, the model can proceed directly to the more advanced technology if that technology is the more cost effective of the two.

However, the agencies reject assertions that the model's use of technology paths is not transparent. The agencies provided extensive explanatory text, figures, model documentation, and model source code specifically addressing these paths (and other model features). This transparency appears evident in that commenters (sometimes while claiming that a specific feature of the model is not transparent) presented analytical results involving changes to corresponding inputs that required a detailed understanding of that feature's operation.

Regarding comments that the technology paths should be arranged in order of cost-effectiveness, the agencies note that such comments presume, without merit, that costs, fuel consumption impacts, and other inputs (e.g., fuel prices) that logically impact manufacturers' decision-making are not subject to uncertainty. These inputs are all subject to uncertainty, and the CAFE Model's arrangement of technologies into several paths is responsive to these uncertainties. Nevertheless, the agencies maintain that some technologies do reflect a higher level of advancement than others (e.g., 10-speed transmissions vs. 5-speed transmissions), and while manufacturers may, in practice, occasionally revert to less advanced technologies, it is appropriate and reasonable to conduct the agencies' analysis in a manner that assumes manufacturers will continue to make forward progress. As observed by EDF's consultant's remarks, the CAFE Model “simply projects what could be done, not what will be.” While no model, much less any model relying on information that can be made publicly available, can hope to represent precisely each manufacturers' actual detailed constrains related to product development and planning, such constraints are real and important. The agencies agree that the CAFE Model's representation of such constraints—including the Model's use of technology paths—provides a reasonable means of accounting for them.

4. Compliance Simulation

The CAFE model provides a way of estimating how vehicle manufacturers could attempt to comply with a given CAFE standard by adding technology to fleets that the agencies anticipate they will produce in future model years. This exercise constitutes a simulation of manufacturers' decisions regarding compliance with CAFE or CO2 standards.

This compliance simulation begins with the following inputs: (a) The analysis fleet of vehicles from model year 2017 discussed below in Section VI.B.1, (b) fuel economy improving technology estimates discussed below in Section VI.C, (c) economic inputs discussed below in Section VI.D, and (d) inputs defining baseline and potential new CAFE or CO2 standards discussed above in Section V. For each manufacturer, the model applies technologies in both a logical sequence and a cost-optimizing strategy in order to identify a set of technologies the manufacturer could apply in response to new CAFE or CO2 standards. The model applies technologies to each of the projected individual vehicles in a manufacturer's fleet, considering the combined effect of regulatory and market incentives while attempting to account for manufacturers' production constraints. Depending on how the model is exercised, it will apply technology until one of the following occurs:

(1) The manufacturer's fleet achieves compliance [317] with the applicable standard and adding additional technology in the current model year would be attractive neither in terms of stand-alone (i.e., absent regulatory need) cost-effectiveness nor in terms of facilitating compliance in future model years;

(2) The manufacturer “exhausts” available technologies; [318] or

(3) For manufacturers assumed to be willing to pay civil penalties (in the CAFE program), the manufacturer reaches the point at which doing so would be more cost-effective (from the manufacturer's perspective) than adding further technology.

The model accounts explicitly for each model year, applying technologies when vehicles are scheduled to be redesigned or freshened and carrying forward technologies between model years once they are applied (until, if applicable, they are superseded by other technologies). The model then uses these simulated manufacturer fleets to generate both a representation of the U.S. auto industry and to modify a representation of the entire light-duty registered vehicle population. From these fleets, the model estimates changes in physical quantities (gallons of fuel, pollutant emissions, traffic fatalities, etc.) and calculates the relative costs and benefits of regulatory alternatives under consideration.

The CAFE model accounts explicitly for each model year, in turn, because manufacturers actually “carry forward” most technologies between model years, tending to concentrate the application of new technology to vehicle redesigns or mid-cycle “freshenings,” and design cycles vary widely among manufacturers and specific products. Start Printed Page 24277Comments by manufacturers and model peer reviewers strongly support explicit year-by-year simulation. Year-by-year accounting also enables accounting for credit banking (i.e., carry-forward), as discussed above, and at least four environmental organizations recently submitted comments urging the agencies to consider such credits, citing NHTSA's 2016 results showing impacts of carried-forward credits.[319] Moreover, EPCA/EISA requires that NHTSA make a year-by-year determination of the appropriate level of stringency and then set the standard at that level, while ensuring ratable increases in average fuel economy through MY 2020. The multi-year planning capability, simulation of “market-driven overcompliance,” and EPCA credit mechanisms (again, for purposes of modeling the CAFE program) increase the model's ability to simulate manufacturers' real-world behavior, accounting for the fact that manufacturers will seek out compliance paths for several model years at a time, while accommodating the year-by-year requirement. This same multi-year planning structure is used to simulate responses to standards defined in grams CO2/mile, and utilizing the set of specific credit provisions defined under EPA's program.

After the light-duty rulemaking analysis accompanying the 2012 final rule that finalized NHTSA's standards through MY 2021, NHTSA began work on changes to the CAFE model with the intention of better reflecting constraints of product planning and cadence for which previous analyses did not account. This involves accounting for expected future schedules for redesigning and “freshening” vehicle models, and accounting for the fact that a given engine or transmission is often shared among more than one vehicle model, and a given vehicle production platform often includes more than one vehicle model. These real product planning considerations are explained below.

Like earlier versions, the current CAFE model provides the capability for integrated analysis spanning different regulatory classes, accounting both for standards that apply separately to different classes and for interactions between regulatory classes. Light vehicle CAFE and CO2 standards are specified separately for passenger cars and light trucks. However, there is considerable sharing between these two regulatory classes, where a single engine, transmission, or platform can appear in both the passenger car and light truck regulatory class. For example, some sport-utility vehicles are offered in 2WD versions (classified as passenger cars for compliance purposes) and 4WD versions (classified as light trucks for compliance purposes). Integrated analysis of manufacturers' passenger car and light truck fleets provides the ability to account for such sharing and reduces the likelihood of finding solutions that could involve introducing impractical and unrealistic levels of complexity in manufacturers' product lines. In addition, integrated fleet analysis provides the ability to simulate the potential that manufacturers could earn CAFE and CO2 credits by over complying with the standard in one fleet and use those credits toward compliance with the standard in another fleet (i.e., to simulate credit transfers between regulatory classes).[320]

The CAFE model also accounts for EPCA's requirement that compliance be determined separately for fleets of domestic passenger cars and fleets of imported passenger cars. The model accounts for all three CAFE regulatory classes simultaneously (i.e., in an integrated way) yet separately: Domestic passenger cars, imported passenger cars, and light trucks. The model further accounts for two related specific statutory requirements specifically involving this distinction between domestic and imported passenger cars. First, EPCA/EISA requires that any given fleet of domestic passenger cars meet a minimum standard, irrespective of any available compliance credits. Second, EPCA/EISA requires compliance with the standards applicable to the domestic passenger car fleet without regard to traded or transferred credits.[321]

However, the CAA has no such limitation regarding compliance by domestic and imported vehicles; EPA did not adopt provisions similar to the aforementioned EPCA/EISA requirements and is not doing so today. Therefore, the CAFE model determines compliance for manufacturers' overall passenger car and light truck fleets for EPA's program.

Each manufacturer's regulatory requirement represents the production-weighted harmonic mean of their vehicle's targets in each regulated fleet. This means that no individual vehicle has a “standard,” merely a target, and each manufacturer is free to identify a compliance strategy that makes the most sense given its unique combination of vehicle models, consumers, and competitive position in the various market segments. As the CAFE model provides flexibility when defining a set of regulatory standards, each manufacturer's requirement is dynamically defined based on the specification of the standards for any simulation and the distribution of footprints within each fleet.

Given this information, the model attempts to apply technology to each manufacturer's fleet in a manner that, given product planning and engineering-related considerations, optimizes the selected cost-related metric. The metric supported by the NPRM version of the model is termed “effective cost.” The effective cost captures more than the incremental cost of a given technology; it represents the difference between their incremental cost and the value of fuel savings to a potential buyer over the first 30 months of ownership.[322] In addition to the technology cost and fuel savings, the effective cost also includes the change in CAFE civil penalties from applying a given technology and any estimated welfare losses associated with the technology (e.g., earlier versions of the CAFE model simulated low-range electric vehicles that produced a welfare loss to buyers who valued standard operating ranges between re-fueling events). Comments on this metric are discussed below, as are model changes responding to these comments.

This construction allows the model to choose technologies that both improve a manufacturer's regulatory compliance position and are most likely to be attractive to its consumers. This also means that different assumptions about future fuel prices will produce different rankings of technologies when the model evaluates available technologies for application. For example, in a high fuel price regime, an expensive but very efficient technology may look attractive to manufacturers because the value of the fuel savings is sufficiently high both to counteract the higher cost of the technology and, implicitly, to satisfy consumer demand to balance price increases with reductions in operating cost.Start Printed Page 24278

In general, the model adds technology for several reasons but checks these sequentially. The model then applies any “forced” technologies. Currently, only variable valve timing (VVT) is forced to be applied to vehicles at redesign since it is the root of the engine path and the reference point for all future engine technology applications.[323] The model next applies any inherited technologies that were applied to a leader vehicle on the same vehicle platform and carried forward into future model years where follower vehicles (on the shared system) are freshened or redesigned (and thus eligible to receive the updated version of the shared component). In practice, very few vehicle models enter without VVT, so inheritance is typically the first step in the compliance loop. Next, the model evaluates the manufacturer's compliance status, applying all cost-effective technologies regardless of compliance status.[324] Then the model applies expiring overcompliance credits (if allowed to do so under the perspective of either the “unconstrained” or “standard setting” analysis, for CAFE purposes).[325] At this point, the model checks the manufacturer's compliance status again. If the manufacturer is still not compliant (and is unwilling to pay civil penalties, again for CAFE modeling), the model will add technologies that are not cost-effective until the manufacturer reaches compliance. If the manufacturer exhausts opportunities to comply with the standard by improving fuel economy/reducing emissions (typically due to a limited percentage of its fleet being redesigned in that year), the model will apply banked CAFE or CO2 credits to offset the remaining deficit. If no credits exist to offset the remaining deficit, the model will reach back in time to alter technology solutions in earlier model years.

The CAFE model implements multi-year planning by looking back, rather than forward. When a manufacturer is unable to comply through cost-effective (i.e., producing effective cost values less than zero) technology improvements or credit application in a given year, the model will “reach back” to earlier years and apply the most cost-effective technologies that were not applied at that time and then carry those technologies forward into the future and re-evaluate the manufacturer's compliance position. The model repeats this process until compliance in the current year is achieved, dynamically rebuilding previous model year fleets and carrying them forward into the future, and accumulating CAFE or CO2 credits from over-compliance with the standard wherever appropriate.

In a given model year, the model determines applicability of each technology to each vehicle platform, model, engine, and transmission. The compliance simulation algorithm begins the process of applying technologies based on the CAFE or CO2 standards specified during the current model year. This involves repeatedly evaluating the degree of noncompliance, identifying the next “best” technology (ranked by the effective cost discussed earlier) available on each of the parallel technology paths described above and applying the best of these. The algorithm combines some of the pathways, evaluating them sequentially instead of in parallel, to ensure appropriate incremental progression of technologies.

The algorithm first finds the best next applicable technology in each of the technology pathways and then selects the best among these. For CAFE purposes, the model applies the technology to the affected vehicles if a manufacturer is either unwilling to pay penalties or if applying the technology is more cost-effective than paying penalties. Afterwards, the algorithm reevaluates the manufacturer's degree of noncompliance and continues application of technology. Once a manufacturer reaches compliance (i.e., the manufacturer would no longer need to pay penalties), the algorithm proceeds to apply any additional technology determined to be cost-effective (as discussed above). Conversely, if a manufacturer is assumed to prefer to pay penalties, the algorithm only applies technology up to the point where doing so is less costly than paying penalties. The algorithm stops applying additional technology to this manufacturer's products once no more cost-effective solutions are encountered. This process is repeated for each manufacturer present in the input fleet. It is then repeated for each model year. Once all model years have been processed, the compliance simulation algorithm concludes. The process for CO2 standard compliance simulation is similar, but without the option of penalty payment, such that technologies are applied until compliance (accounting for any modeled application of credits) is achieved. For both CAFE and CO2 standards, the model also applies any additional (i.e., beyond required for compliance) technology that “pays back” within a specified period (for the NPRM and today's analysis, 30 months).

Some commenters argued that the CAFE model applies constraints that excessively limit options manufacturers have to add technology, causing the model to overestimate costs to achieve a given level of improvement.[326] Some of these commenters further argued that the agencies should assume greater potential to apply technologies that contribute to compliance by improving air conditioner efficiency or otherwise reducing “off cycle” fuel consumption and CO2 emissions.[327] Other commenters argued that such constraints, while warranting some refinements, help the model to simulate manufacturers' decision making realistically and to estimate technology effectiveness and costs reasonably.[328 329]

Some commenters questioned the “effective cost” metric the model uses to decide among available options, claiming that the metric also causes the model to avoid selection of pathways that are not always economically optimal.[330] One of these commenters recommended the agencies modify the effective cost metric for CO2 compliance by removing the term placing a monetary value on progress toward compliance, and instead dividing the remaining net cost (i.e., the increase in technology costs minus a portion of the fuel outlays expected to be avoided) by the additional CO2 credits earned.[331] Another of these commenters claimed on one hand, that the effective cost metric “does not include a measurement of the technology's reduction in fuel consumption or CO2 emissions” and, on the other, that the metric inappropriately places a value on avoided fuel consumption.[332]

One commenter claimed that the model inappropriately allows earned Start Printed Page 24279credits (including CO2 program credits for which EPA has granted a one-time exemption from carry-forward limits) to expire while also showing undue degrees overcompliance with standards, and further proposed that the model be modified to simulate both credit “carry back” (aka “borrowing”) and credit trading between manufacturers.[333]

In addition, some commenters indicated that the agencies' analysis (impliedly, its modeling) should account for some States' mandates that manufacturers sell minimum quantities of “Zero Emission Vehicles” (ZEVs).[334 335]

Regarding the model's representation of engineering and product planning constraints, the agencies maintain that having such constraints produces more realistic potential (as mentioned above, not “predicted”) pathways forward from manufacturers' current fleets than would be the case were these constraints removed. For example, while manufacturers' product plans are protected as confidential business information (CBI), some manufacturers' public comments demonstrate year-by-year balancing such as the CAFE model emulates.[336] Also, even manufacturers that have invested in technologies such as hybrid electric powertrains and Atkinson cycle engines have commented that a manufacturers' past investments will constrain the pathways it can practicably take.[337] Therefore, the agencies have retained the model's basic structural constraints, have updated and expanded the model's technology paths (and, as discussed, the model's logic for approaching these paths), and have updated inputs defining the range of manufacturer-, technology-, and product-specific constraints. These updates are discussed below at greater length.

The agencies have also reconsidered opportunities manufacturers may have to expand the application of technologies that contribute to compliance by improving air conditioner efficiency or otherwise reducing “off cycle” fuel consumption and CO2 emissions, or to earn credit toward CO2 compliance by using refrigerants with lower global warming potential (GWP) or reducing the potential for refrigerant leaks. The version of the model used for the proposal accommodates inputs that, for each of these adjustments or credits, applies the same value to every model year. The agencies have revised the model to accommodate inputs that specify the degree of adjustment or credit separately for each model year, and have applied inputs that assume manufacturers will increase application of these improvements to the highest levels reported within the industry.

Regarding comments on the effective cost metric the model uses to compare and select among available options to add technology, the agencies have considered changes such as those mentioned above. Given the myriad of factors that manufacturers can consider, any weighing to be conducted using publicly-available information will constitute a simplified representation. Nevertheless, within the model's context, it is obvious that any weighing of options should, at a minimum, consider some measure of each option's costs and benefits. Since this aspect of the model involves simulating manufacturers' decisions, it is also clearly appropriate that these costs and benefits be considered from a manufacturer perspective rather than a social perspective.

The effective cost metric used for the NPRM version of the model represents the cost of a given option as the cost to apply a given technology to a given set of vehicles, and represents the benefit of the same option as the extent to which the manufacturer might expect buyers would be willing to pay for fuel economy (as represented by a portion of the projected fuel savings), combined with any reduction in CAFE civil penalties that the manufacturer might ultimately need to pass along to buyers. The reduction in CAFE civil penalties places a value on progress made toward compliance with CAFE standards. The CAA provides no direction regarding CO2 standards, so the model accepts inputs specifying an analogous basis for valuing changes in the quantity of CO2 credits earned from (or required by) a manufacturer's fleet. Because each of these three components (technology cost, fuel benefit, and compliance benefit) is expressed in dollars, subtracting benefits from costs produces a net cost, and after dividing net costs by the number of affected vehicles, it is logical to, at each step, select the option that produces the most negative net unit cost. This approach can be interpreted as maximizing net benefits (to the manufacturer).

As an alternative, the agencies considered a simpler metric that considers only the cost of the option and the extent to which the option increases the quantity of earned credits, and does not require input assumptions regarding how to value progress toward compliance. Such a metric is expressed in dollars per ton or dollars per gallon such that seeking options that produce the smallest (positive) values can be interpreted as maximizing cost effectiveness (of progress toward compliance). However, simply comparing technology costs to corresponding compliance improvements would implicitly assume that manufacturers do not respond at all to fuel prices. This assumption is clearly unrealistic. For example, if diesel fuel costs $5 per gallon and gasoline costs $2 per gallon, manufacturers will be reluctant to respond to stringent CAFE or CO2 standards by replacing gasoline engines with diesel engines. Manufacturers' comments credibly assert that fuel prices matter, and in the agencies' judgment, simulations of decisions between available options should continue to account for avoided fuel outlays.

On the other hand, while any metric should incorporate some measure of progress toward compliance, it is not obvious that this progress must be expressed in monetary terms. While the CAFE civil penalty provisions provide a logical basis for doing so with respect to CAFE, the recently-introduced (through EISA) option to trade credit between manufacturers adds an alternative basis that is undefined and uncertain, in part because terms of past trades are not known to the agencies. Also, as mentioned above, EPCA/EISA's civil penalty provisions are not applicable to noncompliance with CO2 standards.

Therefore, for the purpose of selecting among available options to add technology, the agencies consider it reasonable to use the degree of compliance improvement in “raw” (i.e., not monetized) form, and to divide net costs (i.e., technology costs minus a portion of expected avoided fuel outlays) by this improvement. Under a range of side-by-side tests, this change to the effective cost metric most frequently produced lower overall estimates of compliance costs. However, differences vary among manufacturers, model years, and regulatory alternatives, and also depend on other model inputs. For example, at high fuel prices, the new metric tends to select more expensive pathways than the NPRM's metric, and with the new metric, a case simulating “perfect trading” of CO2 compliance credits tends to show such trading increasing compliance costs rather than, as expected, decreasing such costs.

The version of the model used for the proposal simulates the potential that, for Start Printed Page 24280a given fleet in a given model year, a manufacturer might be able to use credits from an earlier model year or a different fleet. This version of the model did not explicitly simulate the potential that, for a given fleet in a given model year, a manufacturer might be to use credits from a future model year or a different manufacturer. However, the agencies did apply model inputs that reflected assumptions regarding possible trading of credits actually earned prior to model year 2016 (the earliest represented in detail in the agencies' analysis), and the agencies did examine a case (included in the sensitivity analysis) involving hypothetical “perfect” trading of CO2 credits among manufacturers by treating the industry as a single “manufacturer.” Although past versions of the CAFE Model had included code under development with a view toward eventually simulating one or both of these provisions, this code had never proceeded beyond preliminary experimentation, and had never been the focus of peer reviews or application in published analyses.

Nevertheless, the agencies considered expanding the model to simulate credit “carry back” (or “borrowing”) and trading (explicitly, rather than in an idealized hypothetical way). The agencies closely examined the corresponding model revisions proposed by UCS and determined that such methods would not produce repeatable results. This is because the approach proposed by UCS “randomly swaps items in list to minimize trading bias.” [338]

Even if such revisions could be modified to produce non-random results, including credit banking and trading would introduce highly speculative elements into the agencies' analysis. While manufacturers have occasionally indicated plans to carry back credits from future model years, those plans have sometimes backfired when projected credits have failed to materialize, e.g., by misjudging consumer demand for more efficient vehicles. In the agencies' judgment, it would be inappropriate to set standards based on an analysis that relies on the type of borrowing that has been known to fail. To rely also on credit trading during the model years included in the analysis would compound this undue speculation. For example, including credit borrowing and trading throughout the analysis, as some commenters proposed, would lead to an analysis that depends on the potential that, in order to comply with the MY 2022 standard for light trucks, FCA could use credits it expects to be able to buy from another manufacturer in MY 2025. Even if the agencies' analysis had knowledge of and made use of manufacturers' actual product plans, expectations about the ability to borrow others' unearned credits would necessarily be considered risky and unreliable. Within an analysis that, to provide for public disclosure, extrapolates forward many years from the most recent observed fleet, such transactions would add an unreasonable level of speculation. Therefore, the agencies have declined to introduce credit borrowing and trading into the model's logic.

The analysis presented in the proposal applied inputs reflecting potential application of credits earned earlier than the first year modeled explicitly. However, as observed by some commenters, those inputs did not fully account for the one-time exemption from the 5-year limit on the extent to which manufacturers may carry forward CO2 credits. The agencies have updated the analysis fleet to MY 2017 and, in doing so, have updated inputs specifying how credits earned to MY 2017 might be applied. These updates implement a reasonably full accounting of these “legacy” credits, including of the one-time exemption from the credit life limit.

As mentioned above, some commenters also indicated that the model is unrealistically “reluctant” to apply credits carried forward from early model years. As explained in the proposal and in the model documentation, the model's application of carried-forward credits is partially controlled by model inputs, which, for the proposal, were set to assume that manufacturers would tend to retain credits as long as possible. This assumption is entirely consistent with manufacturers' past practice and logical in a context wherein the stringency of standards is generally increasing over time. Even though using credits in some model years might seem initially advantageous, doing so means foregoing actual improvements likely to be needed in later model years.

Regarding the model's treatment of mandates and credits for the sale of ZEVs, as indicated in the model documentation accompanying the proposal, these capabilities were experimental in that version of the model. The reference case analysis for today's notice, like that for the proposal, does not simulate compliance with ZEV mandates.[339]

For the NPRM, the CAFE model was exercised with inputs extending this explicit simulation of technology application through MY 2032, as the agencies anticipated this was sufficiently beyond MY 2026 that nearly all multiyear planning attributable to MY 2026 standards should be accounted for, and any compliance credits carried forward from MY 2026 would have expired. The analysis met this expectation, and the agencies presented analysis of the resultant estimated impacts over the useful lives of vehicles produced prior to MY 2030. The agencies invited comment on all aspects of the analysis, and relevant to this aspect of the analysis—i.e., its perspective and temporal span—EDF stated that that these led the agencies to overstate the proposal's positive impacts on safety, in part because by explicitly representing vehicle model years only through 2032, the agencies had failed to account for the impact of distant model years prices and fuel economy levels on the retention and scrappage of vehicles produced through MY 2029.[340] For example, some vehicles produced in MY 2026 will likely still be on the road during calendar years (CY) 2033-2050 and the rates at which these MY 2026 vehicles will be scrapped during CYs 2033-2050 will be impacted by the prices and fuel economy levels of vehicles produced during MYs 2033-2050.

The agencies have addressed this comment by expanding model inputs to extend the explicit simulation of technology application through MY 2050. Most of these expanded model inputs involve the analysis fleet and inputs defining the cost and availability of various fuel-saving technologies. These inputs are discussed below. The agencies also made minor modifications to the model in order to extend model outputs to cover this wider span and to carry forward each regulatory alternative's standards automatically through the last year to be modeled (e.g., extending standards without change from MY 2032 through MY 2050). The model documentation discusses these Start Printed Page 24281minor changes.[341] In addition, although the agencies published detailed model output files documenting all estimated annual impacts through calendar year 2089, the notice and PRIA both emphasized the above-mentioned “model year” perspective, as in past regulatory analyses supporting CAFE and CO2 standards. Recognizing that an alternative “calendar year” perspective is of interest to EDF and, perhaps other stakeholders, the agencies have expanded the presentation of results in today's notice and FRIA by presenting some physical impacts (e.g., fuel consumption and CO2 emissions) as well as monetized benefits, costs, and net benefits for each of CYs 2017-2050. All of these results appear in the model output files published with today's notice, as do corresponding results for more specific impacts (e.g., year-by-year components of monetized social costs).[342]

5. Calculation of Physical Impacts

Once it has completed the simulation of manufacturers' potential application of technology in response to CAFE/CO2 standards and fuel prices, the CAFE Model calculates impacts of the resultant changes in new vehicle fuel economy levels and prices. This involves several steps.

The model calculates changes in the total quantity of new vehicles sold in each model year as well as the relative shares passenger cars and light trucks comprise of the overall new vehicle market. The agencies received many comments on the estimation of sales impacts, and as discussed below, today's analysis applies methods and corresponding estimates that reflect careful consideration of these comments. Related to these calculations, the model now operates in an iterated fashion with a view toward obtaining sales impacts that are balanced with changes in vehicle prices and fuel economy levels. This involves solving for compliance, calculating sales impacts, re-solving for compliance, and repeating these steps as many times as specified in model inputs. For today's analysis, the agencies operated the model with four iterations, as early testing suggested three iterations should be sufficient for fleetwide results to converge between iterations. The model documentation describes the procedures for iteration in detail.

The impacts on outlays for new vehicles occur coincident with the sale of these vehicles so the model can simply calculate and record these for each model year included in the analysis. However, virtually all other impacts result from vehicle operation that extends long after a vehicle is produced. Like other models (including, e.g., NEMS), the CAFE Model includes procedures (sometimes referred to as “stock models” or as models of fleet turnover) to estimate annual rates at which new vehicles are used and subsequently scrapped. The agencies received many comments on procedures for estimating vehicle scrappage and on procedures for estimating annual quantities of highway travel, accounting for the elasticity of travel demand with respect to per-mile costs for fuel. Below, Section VI.D.1 discusses these comments and reviews procedures and corresponding estimates that also reflect careful consideration of these comments.

For each vehicle model in each model year, these procedures result in estimates of the number of vehicles remaining in service in each calendar year, as well as the annual mileage accumulation (i.e., vehicle miles traveled, or VMT) in each calendar year. As mentioned above, most of the physical impacts of interest derive from this vehicle operation. Also discussed above, the simulated application of technology results in “initial” and “final” estimates of the cost, fuel type, fuel economy, and fuel share (for, in particular, PHEVs that can run on gasoline or electricity) applicable to each vehicle model in each model year. Together with quantities of travel, and with estimates of the “gap” between “laboratory” and “on-road” fuel economy, these enable calculation of quantities of fuel consumed in each year during the useful life of each vehicle model produced in each model year.[343] The model documentation provides specific procedures and formulas implementing these calculations.

As for the NPRM, the model calculates emissions of CO2 and other air pollutants, reporting emissions both from vehicle tailpipes and from upstream processes (e.g., petroleum refining) involved in producing and supplying fuels. Section VI.D.3 below reviews methods, models, and estimates used in performing these calculations. The model also calculates impacts on highway safety, accounting for changes in travel demand, changes in vehicle mass, and continued past and expected progress in vehicle safety (through, e.g., the application of new crash avoidance systems). Section VI.D.2 discusses methods, data sources, and estimates involved in estimating safety impacts, comments on the same, and changes included in today's analysis. In response to the NPRM, some comments urged the agencies also to quantify different types of health impacts from changes in air pollution rather than only accounting for such impacts in aggregate estimates of the social costs of air pollution. Considering these comments, the agencies added such calculations to the model, as discussed in Section VI.D.3.

6. Calculation of Benefits and Costs

Having estimated how technologies might be applied going forward, and having estimated the range of resultant physical impacts, the CAFE Model calculates a variety of private and social benefits and costs, reporting these from the consumer, manufacturer, and social perspectives, both in undiscounted and discounted present value form (given inputs specifying the corresponding discount rate and present year). Estimates of regulatory costs are among the direct outputs of the simulation of manufacturers' potential responses to new standards. Other benefits and costs are calculated based on the above-mentioned estimates of travel demand, fuel consumption, emissions, and safety impacts. The agencies received many comments on the NPRM's calculation of benefits and costs, and Section VI.D.1 discusses these comments and presents the methods, data sources, and estimates used in calculating benefits and costs reported here.

7. Structure of Model Inputs and Outputs

All CAFE Model inputs and outputs described above are specified in Microsoft Excel format, and the user can define and edit all inputs to the system. Table VI-3 describes (non-exhaustively) which inputs are contained within each input file and Table VI-4 describes which outputs are contained in each output file. This is important for three reasons: (1) Each file is discussed throughout the following sections; (2) several commenters conflated aspects of the model with its inputs; and (3) several commenters seemed confused about where to find specific information in the output files. This information was described in detail in the NPRM CAFE Model Documentation, but is reproduced here for quick reference. When specifically referencing the input Start Printed Page 24282or output file used for the NPRM or final rule in the following discussion, NPRM or FRM, respectively, will precede the file name.

A catalog of the Argonne National Laboratory Autonomie fuel economy technology effectiveness value output files are reproduced in the following Table VI-5 as well. The left column shows the terminology used in this text to refer to the file, while the right column describes each file. NPRM or FRM, respectively, may precede the terminology in the text as appropriate.

Start Printed Page 24283

Finally, Table VI-6 lists the terminologies used to refer to other model-related documents which are referred to frequently throughout the text. NPRM or FRM, respectively, may precede the terminology in the text as appropriate.

Start Printed Page 24284

B. What inputs does the compliance analysis require?

1. Analysis Fleet

The starting point for the evaluation of the potential feasibility of different stringency levels for future CAFE and CO2 standards is the analysis fleet, which is a snapshot of the recent vehicle market. The analysis fleet provides a baseline from which to project what and how additional technologies could feasibly be applied to vehicles in a cost-effective manner to raise those vehicles' fuel economy and lower their CO2 emission levels.[344] The fleet characterization also provides a reference point with data for other factors considered in the analysis, including environmental effects and effects estimated by the economic modules (i.e., sales, scrappage, and labor utilization). When the scope of the analysis widens, another piece of data must be included for each vehicle in the analysis fleet to map a given element of the fleet appropriately onto an analysis module.

For the analysis presented in this final rule, the analysis fleet includes information about vehicles that is essential for each analysis module. The first part of projecting how additional technologies could be applied to vehicles is knowing which vehicles are produced by which manufacturers, the fuel economies of those vehicles, how many of each are sold, whether they are passenger cars or light trucks, and their footprints. This is important because it improves understanding of the overall impacts of different levels of CAFE and CO2 standards; overall impacts that result from industry's response to standards, and industry's response, is made up of individual manufacturer responses to the standards in light of the overall market and their individual assessment of consumer acceptance. Establishing an accurate representation of manufacturers' existing fleets (and the vehicle models in them) that will be subject to future standards helps in predicting potential individual manufacturer responses to those future standards in addition to potential changes in those standards.

Another part of projecting how additional fuel economy improving technologies could be applied to vehicles is knowing which fuel saving technologies manufacturers have equipped on which vehicles. In many cases, the agencies also collect and reference additional information on other vehicle attributes to help with this process.[345] Accounting for technologies already applied to vehicles helps avoid “double-counting” the value of those technologies, by assuming they are still available to be applied to improve fuel economy and reduce CO2 emissions. It also promotes more realistic determinations of what additional technologies can feasibly be applied to those vehicles: If a manufacturer has already started down a technological path to fuel economy or performance improvements, the agencies do not assume it will completely abandon that path because doing so would be unrealistic and fails to represent accurately manufacturer responses to standards. Each vehicle model (and configurations of each model) in the analysis fleet, therefore, has a comprehensive list of its technologies, which is important because different configurations may have different technologies applied to them.[346] In addition, to properly account for technology costs, the agencies assign each vehicle to a technology class and an engine class. Technology classes reference each vehicle to a set of full vehicle simulations, so that the agencies may project fuel efficiency with combinations of additional fuel saving equipment and hybrid and electric vehicle battery costs.

Yet another part of projecting which vehicles might exist in future model years is developing reasonable real-world assumptions about when and how manufacturers might apply certain technologies to vehicles. The analysis fleet accounts for links between vehicles, recognizing vehicle platforms will share technologies, and the vehicles that make up that platform should receive (or not receive) additional technological improvements together. Shared engines, shared transmissions, and shared vehicle platforms for mass reduction technology are considered. In addition, each vehicle model/configuration in the analysis fleet also has information about its redesign Start Printed Page 24285schedule, i.e., the last year it was redesigned and when the agencies expect it to be redesigned again. Redesign schedules are a key part of manufacturers' business plans, as each new product can cost more than $1B, and involve a significant portion of a manufacturer's scarce research, development, and manufacturing and equipment budgets and resources.[347] Manufacturers have repeatedly told the agencies that sustainable business plans require careful management of resources and capital spending, and that the length of time each product remains in production is crucial to recouping the upfront product development and plant/equipment costs, as well as the capital needed to fund the development and manufacturing equipment needed for future products. Because the production volume of any given vehicle model varies within a manufacturer's product line, and varies among different manufacturers, redesign schedules typically vary for each model and manufacturer. Some (relatively few) technological improvements are small enough that they can be applied in any model year; a few other technological improvements may be applied during a refreshening (when a few additional changes are made, but well short of a full redesign), but others are major enough that they can only be cost-effectively applied at a vehicle redesign, when many other things about the vehicle are already changing. Ensuring the CAFE model makes technological improvements to vehicles only when it is feasible to do so also helps the analysis better represent manufacturer responses to different levels of standards.

Finally, the agencies restrict the applications of some technologies on some vehicles upon determining the technology is not compatible with the functional and performance requirements of the vehicle, or if the manufacturers are unlikely to apply a specific technology to a specific vehicle for reasons articulated with confidential business information that the agencies found credible.

Other data important for the analysis that are referenced to the analysis fleet include baseline economic, environmental, and safety information. Vehicle fuel tank size is required to estimate range and refueling benefit while curb weights and safety class assignments help the agencies consider how changes in vehicle mass may affect safety. The agencies identify the final assembly location for each vehicle, engine, and transmission, as well as the percent of U.S. content to support the labor impact analysis. In addition, the aforementioned accounting for first-year vehicle production volumes (i.e., the number of vehicles of each new model sold in MY 2017, for this analysis) is the foundation for estimating how future vehicle sales might change in response to different potential standards.

The input file for the CAFE model characterizing the analysis fleet, referred to as the “market inputs” file or “market data” file, accordingly includes a large amount of data about vehicles, their technological characteristics, the manufacturers and fleets to which they belong, and initial prices and production volumes, which provide the starting points for projection (by the sales model) to ensuing model years. In the Draft TAR (which utilized a MY 2015 analysis fleet) and NPRM (which utilized a MY 2016 analysis fleet), the agencies needed to populate about 230,000 cells in the market data file to characterize the fleet. For this final rule (which utilized a MY 2017 analysis fleet), the agencies populated more than 400,000 cells to characterize the fleet. While the fleet is not actually much more heterogeneous in reality,[348] the agencies have provided and collected more data to justify the characterization of the analysis fleet, and to support the functionality of modules in the CAFE model.

A solid characterization of a recent model year as an analytical starting point helps realistically estimate ways manufacturers could potentially respond to different levels of standards, and the modeling strives to simulate realistically how manufacturers could progress from that starting point. While manufacturers can respond in many ways beyond those represented in the analysis (e.g., applying other technologies, shifting production volumes, changing vehicle footprint), such that it is impossible to predict with any certainty exactly how each manufacturer will respond, it is still important to establish a solid foundation from which to estimate potential costs and benefits of potential future standards. The following sections discuss aspects of how the analysis fleet was built for this analysis, and includes discussion of the comments on fleet that the agencies received on the proposed rule.

a) Principles on Data Sources Used To Populate the Analysis Fleet

The source data for vehicles in the analysis fleet and their technologies is a central input for the analysis. The sections below discuss pros and cons of different potential sources and what the agencies used for this analysis, and responds to comments the agencies received on data sources in the proposal.

(1) Use of Confidential Business Information Versus Publicly-Releasable Sources

Since 2001, CAFE analysis has used either confidential, forward-estimating product plans from manufacturers, or publicly available data on vehicles already sold as a starting point for determining what technologies can be applied to what vehicles in response to potential different levels of standards. The use of either data source requires certain tradeoffs. Confidential product plans comprehensively represent what vehicles a manufacturer expects to produce in coming years, accounting for plans to introduce new vehicles and fuel-saving technologies and, for example, plans to discontinue other vehicles and even brands. This information can be very thorough and can improve the accuracy of the analysis, but cannot be publicly released. This makes it difficult for public commenters to reproduce the analysis for themselves as they develop their comments. Some non-industry commenters have also expressed concern about manufacturers having an incentive in the submitted plans to underestimate (deliberately or not) their future fuel economy capabilities and overstate their expectations about, for example, the levels of performance of future vehicle models in order to affect the analysis. Accordingly, since 2010, EPA and NHTSA have based analysis fleets almost exclusively on information from commercial and public sources, starting with CAFE compliance data and adding information from other sources.

An analysis fleet based primarily on public sources can be released to the public, solving the issue of commenters being unable to reproduce the overall analysis. However, industry commenters have argued such an analysis fleet cannot accurately reflect manufacturers' actual plans to apply fuel-saving technologies (e.g., manufacturers may apply turbocharging to improve not just fuel economy, but also to improve vehicle performance) or manufacturers' plans to change product offerings by introducing some vehicles and brands and discontinuing other Start Printed Page 24286vehicles and brands, precisely because that information is typically confidential business information (CBI). A fully-publicly-releasable analysis fleet holds vehicle characteristics unchanged over time and lacks some level of accuracy when projected into the future. For example, over time, manufacturers introduce new products and even entire brands. On the other hand, plans announced in press releases do not always ultimately bear out, nor do commercially available third-party forecasts. Assumptions could be made about these issues to improve the accuracy of a publicly releasable analysis fleet, but concerns include that this information would either be largely incorrect, or, if the assumptions were correct, information would be released that manufacturers would consider CBI.

Furthermore, some technologies considered in the rulemaking are difficult to observe in the analysis fleet without expensive teardown study and time-consuming benchmarking. Not giving credit for these technologies puts the analysis at significant risk of double-counting the effectiveness of these technologies, as manufacturers cannot equip technologies twice to the same vehicle for double the fuel economy benefit. As discussed in the Draft TAR, the agencies assigned little (if any) technology application in the baseline fleet for some of these technologies.[349] For the NPRM MY 2016 fleet development process, the agencies again offered the manufacturers the opportunity to volunteer CBI to the agencies to help inform the technology content of the analysis fleet, and many manufacturers did. The agencies were able to confirm that many manufacturers had already included many hard-to-observe technologies in the MY 2016 fleet (which they were not properly given credit for in the characterization of the MY 2014 and MY 2015 fleets presented in Draft TAR) so the agencies reflected this new information in the NPRM analysis and in the analysis presented today.

In addition, many manufacturers provided confidential comment on the potential applicability of fuel-saving technologies to their fleet. In particular, many manufacturers confidentially identified specific engine technologies that they will not use in the near term, either on specific vehicles, or at all. Reasons varied: Some manufacturers cited intellectual property concerns, and others stated functional performance concerns for some engine types on some vehicles. Other manufacturers shared forward-looking product plans, and explained that it would be cost prohibitive to scrap significant investments in one technology in favor of another. This topic is discussed in more detail in Section VI.B.1.b)(6), below.

The agencies sought comment on how to address this issue going forward, recognizing both the competing interests involved and the typical timeframes for CAFE and CO2 standards rulemakings.

Many commenters expressed concern with the agencies using any CBI as part of the rulemaking process. Some commenters expressed concern that use of CBI would make the CAFE model subject to inaccuracies because manufacturers would only provide additional information in situations in which a correction to the agencies' baseline assumptions would favor the manufacturers.[350] The agencies recognize this as a reasonable concern, but the analysis presented in the Draft TAR consistently assumed very little (if any) technology had been applied in the baseline. In addition, many manufacturers shared information on advanced technologies that were not yet in production in MY 2017, but could be used in the future; manufacturer contributions helped the agencies better model many advanced engine technologies and to include them in today's analysis, and inclusion of these technologies (and costs) in the analysis sometimes lowered the projected cost of compliance for stringent alternatives. Other commenters expressed concern that automakers would supply false or incomplete information that would unduly restrict what technologies can be deployed.[351] When possible, the agencies sought independently to verify manufacturer CBI (or claims made by other stakeholders) through lab testing and benchmarking.[352] The agencies found no evidence of misrepresentation of engineering specifications in the MY 2017 fleet in manufacturer CBI; instead, the agencies were able to verify independently many CBI submissions, and confirm the credibility of information provided from those sources.

Some commenters requested that more CBI be used in the analysis. For instance, some commenters suggested that the agencies should return to the use of product plans and announcements regarding future fleets because manufacturers had already committed investments to bring announced products to market.[353] However, if the agencies were to assume that these commitments were already in the baseline, the agencies would underestimate the cost of compliance for stringent alternatives. Moreover, while upfront investments to bring technologies to market are significant, the total marginal costs of components are typically large in comparison over the entire product life-cycle, and these costs have not yet been realized in vehicles not yet produced.

The agencies did make use of some forward-looking CBI in the analysis. The agencies received many comments from manufacturers on the technological feasibility, or functional applicability of some fuel saving technologies to certain vehicles, or certain vehicle applications, and the agencies took this information into consideration when projecting compliance pathways. These cases are discussed generally in Section VI.B.1.b)(6), below, and specifically for each technology in those technology sections. Some commenters expressed that the use of CBI for future product plans would be acceptable, but only if the agencies disclosed the CBI affecting all vehicles through MY 2025 at the time of publication.[354] Functionally, this is not possible. Manufacturer's confidential product plans cannot be made public, as prohibited under NHTSA's regulations at 49 CFR part 512, and if the information meets the requirements of section 208(c) of the Clean Air Act. If the agencies disclosed confidential information, it would not only violate the terms on which the agencies obtained the CBI, but it is unlikely that manufacturers would continue to offer CBI, which in turn would likely degrade the quality of the analysis. The agencies believe that the use of CBI in the NPRM and final rule analysis—to confirm, reference, or to otherwise modify aspects of the analysis that can be made public—threads the needle between a more accurate but less transparent analysis (using more CBI) and a less accurate but more transparent analysis (using less CBI).

Start Printed Page 24287

(2) Source Data and Vintage Used in the Analysis

Based on the assumption that a publicly-available analysis fleet continued to be desirable, manufacturer compliance submissions to EPA and NHTSA were used as a starting point for the NPRM and final rule analysis fleets. Generally, manufacturer compliance submissions break down vehicle fuel economy and production volume by regulatory class, and include some very basic product information (typically including vehicle nameplate, engine displacement, basic transmission information, and drive configuration). Many different trim levels of a product are typically rolled up and reported in an aggregated fashion, and these groupings can make decomposition of different fuel-saving, road load reducing technologies extremely difficult. For instance, vehicles in different test weight classes, with different tires or aerodynamic profiles may be aggregated and reported together.[355] A second portion of the compliance submission summarizes production volume by vehicle footprints (a key compliance measure for standard setting) by nameplate, and includes some basic information about engine displacement, transmission, and drive configuration. Often these production volumes by footprint do not fit seamlessly together with the production volumes for fuel economy, so the agencies must reconcile this information.

Information from the MY 2016 fleet was chosen as the foundation for the NPRM analysis fleet because, at the time the rulemaking analysis was initiated, the 2016 fleet represented the most up-to-date information available in terms of individual vehicle models and configurations, production technology levels, and production volumes. If MY 2017 data had been used while this analysis was being developed, the agencies would have needed to use product planning information that could not be made available to the public until a later date.

The NPRM analysis fleet was initially developed with 2016 mid-model year compliance data because final compliance data was not available at that time, and the timing provided manufacturers the opportunity to review and comment on the characterization of their vehicles in the fleet. With a view toward developing an accurate characterization of the 2016 fleet to serve as an analytical starting point, corrections and updates to mid-year data (e.g., to production estimates) were sought, in addition to corroboration or correction of technical information obtained from commercial and other sources (to the extent that information was not included in compliance data), although future product planning information from manufacturers (e.g., future product offerings, products to be discontinued) was not requested, as most manufacturers view such information as CBI. Manufacturers offered a range of corrections to indicate engineering characteristics (e.g., footprint, curb weight, transmission type) of specific vehicle model/configurations, as well as updates to fuel economy and production volume estimates in mid-year reporting. After following up on a case-by-case basis to investigate significant differences, the analysis fleet was updated.

Sales, footprint, and fuel economy values with final compliance data were also updated if that data was available. In a few cases, final production and fuel economy values were slightly different for specific MY 2016 vehicle models and configurations than were indicated in the NPRM analysis; however, other vehicle characteristics (e.g., footprint, curb weight, technology content) important to the analysis were reasonably accurate. While some commenters have, in the past, raised concerns that non-final CAFE compliance data is subject to change, the potential for change is likely not significant enough to merit using final data from an earlier model year reflecting a more outdated fleet. Moreover, even ostensibly final CAFE compliance data is frequently subject to later revision (e.g., if errors in fuel economy tests are discovered), and the purpose of the analysis was not to support enforcement actions but rather to provide a realistic assessment of manufacturers' potential responses to future standards.

Manufacturers integrated a significant amount of new technology in the MY 2016 fleet, and this was especially true for newly-designed vehicles launched in MY 2016. While subsequent fleets will involve even further application of technology, using available data for MY 2016 provided the most realistic detailed foundation for analysis that could be made available publicly in full detail, allowing stakeholders to reproduce the analysis presented in the proposal independently. Insofar as future product offerings are likely to be more similar to vehicles produced in 2016 than to vehicles produced in earlier model years, using available data regarding the 2016 model year provided the most realistic, publicly releasable foundation for constructing a forecast of the future vehicle market for this proposal. Many comments responding to the Draft TAR, EPA's Proposed Determination, EPA's 2017 Request for Comment, and the NPRM preceding today's notice stated that the most up-to-date analysis fleet possible should be used, because a more up-to-date analysis fleet will better capture how manufacturers apply technology and will account better for vehicle model/configuration introductions and deletions.[356 357]

On the other hand, some commenters suggested that because manufacturers continue improving vehicle performance and utility over time, an older analysis fleet should be used to estimate how the fleet could have evolved had manufacturers applied all technological potential to fuel economy rather than continuing to improve vehicle performance and utility.[358] Because manufacturers change and improve product offerings over time, conducting analysis with an older analysis fleet (or with a fleet using fuel economy levels and CO2 emissions rates that have been adjusted to reflect an assumed return to levels of performance and utility typical of some past model year) would miss this real-world trend. While such an analysis could project what industry could do if, for example, manufacturers devoted all technological improvements toward raising fuel economy and reducing CO2 emissions (and if consumers decided to purchase these vehicles), the agencies do not believe it would be consistent with a transparent examination of what effects different levels of standards would have Start Printed Page 24288on individual manufacturers and the fleet as a whole.

All else being equal, using a newer analysis fleet will produce more realistic estimates of impacts of potential new standards than using an outdated analysis fleet. However, among relatively current options, a balance must be struck between input freshness, and input completeness and accuracy.[359] During assembly of the inputs for the NPRM analysis, final compliance data was available for the MY 2015 model year but not, in a few cases, for MY 2016. However, between mid-year compliance information and manufacturers' specific updates discussed above, a robust and detailed characterization of the MY 2016 fleet was developed. While information continued to develop regarding the MY 2017 and, to a lesser extent MY 2018 and even MY 2019 fleets, this information was—even in mid-2017—too incomplete and inconsistent to be assembled with confidence into an analysis fleet for modeling supporting deliberations regarding the NPRM analysis.

Manufacturers requested that the baseline fleet supporting the final rule incorporate the MY 2018 or most recent information available.[360] Other commenters expressed desire for multiple fleets of various vintages to compare the updated model outputs with those of previous rule-makings. Specifically, some commenters requested that older fleet vintages (MY 2010, for instance) be developed in parallel with the MY 2017 fleet so that those too may be used as inputs for the model.[361]

Between the NPRM and this final rule, manufacturers submitted final compliance data for the MY 2017 fleet. When the agencies pulled together information for the fleet for the final rule, the agencies decided to use the highest-quality, most up-to-date information available. Given that pulling this information together takes some time, and given that “final” compliance submissions often lag production by a few years, the agencies decided to use 2017 model year as the base year for the analysis fleet, as the agencies stated in the NPRM.[362] While the agencies could have used preliminary 2018 data or even very early 2019 data, this information was not available in time to support the final rulemaking. Likewise, the agencies chose not to revert to a previous model year (for instance 2016 or 2012) because many manufacturers have incorporated fuel savings technologies over the last few years, realized some benefits for fuel economy, and adjusted the performance or sales mix of vehicles to remain competitive in the market. Also, using an earlier model year would provide less accurate projections because the analysis would be based on what manufacturers could have done in past model years and would have estimated the fuel economy improvements instead of using known information on the technologies that were employed and the actual fuel economy that resulted from applying those technologies.

Some additional information (about off-cycle technologies, for instance) was often not reported by manufacturers in MY 2017 formal compliance submissions in a way that provided clear information on which technologies were included on which products. As part of the formal compliance submission, some manufacturers voluntarily submitted additional information (about engine technologies, for instance). While this data was generally of very high quality, there were some mistakes or inconsistencies with publicly available information, causing the agencies to contact the manufacturers to understand and correct identified issues. In most cases, however, the formal compliance data was very limited in nature, and the agencies collected additional information necessary to characterize fully the fleet from other sources, and scrutinized additional information submitted by manufacturers carefully, independently verifying when possible.

Specifically, the agencies downloaded and reviewed numerous marketing brochures and product launch press releases to confirm information submitted by manufacturers and to fill in information necessary for the analysis fleet that was not provided in the compliance data. Product brochures often served as the basis for the curb weights used in the analysis. This publicly available manufacturer information sometimes also included aerodynamic drag coefficients, information about steering architecture, start-stop systems, pickup bed lengths, fuel tank capacities, and high-voltage battery capacities. The agencies recorded vehicle horsepower, compression ratio, fuel-type, and recommended oil weight rating from a combination of manufacturer product brochures and owner's manuals. The product brochures, as well as online references such as Autobytel, informed which combinations of fuel saving technologies were available on which trim levels, and what the manufacturer suggested retail price was for many products. Overall this information proved helpful for assigning technologies to vehicles, and for getting data (such as fuel tank size [363] ) necessary for the analysis. These reference materials have been included in the rulemaking documentation.[364]

The agencies elected not to develop fleets of previous model year vintages that could be used in parallel as an input to the CAFE model. Developing a detailed characterization of the fleet of any vintage would be a huge undertaking with few benefits. As the scope has increased, and as additional modules are added, going back in time to re-characterize a previous fleet in a format that works with CAFE model updates can be time- and resource-prohibitive for the agencies, even if that work is adapting a fleet that was used in previous rule-making analysis. Doing so also offers little value in determining what potential fuel saving technology can be added to a more recent fleet during the rulemaking timeframe.

The MY 2017 manufacturer-submitted data, verified and supplemented by the agencies with publicly-available information, therefore presented the fullest, most up-to-date data set that the agencies could have used to support this analysis.Start Printed Page 24289

b) Characterizing Vehicles and Their Technology Content

The starting point for projecting what additional fuel economy improving technologies could feasibly be applied to vehicles is knowing what vehicles are produced by which manufacturers and what technologies exist on those vehicles. Rows in the market data file are the smallest portion of the fleet to which technology may be applied as part of a projected compliance pathway. For the analysis presented in this final rule, the agencies, when possible, attempted to include vehicle trim level information in discrete rows. A manufacturer, for example GM, may produce one or more vehicle makes (or brands), for example Chevrolet, Buick and others. Each vehicle make may offer one or more vehicle models, for example Malibu, Traverse and others. And each vehicle model may be available in one or more trim levels (or standard option levels), for example “RS,” “Premier” and others, which have different levels of standard options, and in some cases, different engines and transmissions.

Manufacturer compliance submissions, discussed above, were used as a starting point to define working rows in the market data file; however, often the rows needed to be further disaggregated to correctly characterize vehicle information covered in the scope of the analysis, and analysis fleet. Manufacturers often grouped vehicles with multiple trim levels together because they often included the same fuel-saving technologies and may be aggregated to simplify reporting. However, the manufacturer suggested retail prices of different trim levels are certainly different, and other features relevant to the analysis are occasionally different.

As a result of further disaggregating compliance information, the number of rows in the market data file increased from 1,667 rows used in the NPRM to 2,952 rows for this final rule analysis. The agencies do not have data on sales volumes for each nameplate by trim level, and used an approach that evenly distributed volume across offered trim levels, within the defined constraints of the compliance data.[365] Evenly distributing the volume across trim levels is a simplification, but this action should (1) highlight some difficulties that could be encountered when acquiring data for a full-vehicle consumer choice model should the agencies pursue developing one in the future (discussed further, below), and (2) lower the average sales volume per row in the market data file, thereby allowing the application of very advanced electrification technologies in smaller lumps. The latter effect is responsive to comments (discussed below) that suggested electrification technologies could be more cost-effectively deployed in lower volumes, and that the CAFE model artificially constrains cost effective technologies that may be deployed, resulting in higher costs and large over-compliance.

(1) Assigning Vehicle Technology Classes

While each vehicle in the analysis fleet has its list of observed technologies and equipment, the ways in which manufacturers apply technologies and equipment do not always coincide perfectly with how the analysis characterizes the various technologies that improve fuel economy and reduce CO2 emissions. To improve how the observed vehicle fleet “fits into” the analysis, each vehicle model/configuration is “mapped” to the full-vehicle simulation modeling by Argonne National Laboratory that is used to estimate the effectiveness of the fuel economy-improving/CO2 emissions-reducing technologies considered. Argonne produces full-vehicle simulation modeling for many combinations of technologies, on many types of vehicles, but it did not simulate literally every single manufacturer's vehicle model/configuration in the analysis fleet because it would be impractical to assemble the requisite detailed information—much of which would likely only be provided on a confidential basis—specific to each vehicle model/configuration and because the scale of the simulation effort would correspondingly increase by at least two orders of magnitude. Instead, Argonne simulated 10 different vehicle types corresponding to the “technology classes” generally used in CAFE analysis over the past several rulemakings (e.g., small car, small performance car, pickup truck, etc.). Each of those 10 different vehicle types was assigned a set of “baseline characteristics” to which Argonne added combinations of fuel-saving technologies and then ran simulations to determine the fuel economy achieved when applying each combination of technologies to that vehicle type given its baseline characteristics.

Start Printed Page 24290

In the analysis fleet, inputs assign each specific vehicle model/configuration to a technology class, and once there, map to the simulation within that technology class most closely matching the combination of observed technologies and equipment on that vehicle. This mapping to a specific simulation result most closely representing a given vehicle model/configuration's initial technology “state” enables the CAFE model to estimate the same vehicle model/configuration's fuel economy after application of some other combination of technologies, leading to an alternative technology state.

(2) Assigning Vehicle Technology Content

As explained above, the analysis fleet is defined not only by the vehicles it contains, but also by the technologies on those vehicles. Each vehicle in the analysis fleet has an associated list of observed technologies and equipment that can improve fuel economy and reduce CO2 emissions.[366] With a portfolio of descriptive technologies arranged by manufacturer and model, the analysis fleet can be summarized and project how vehicles in that fleet may increase fuel economy over time via the application of additional technology.

In many cases, vehicle technology is clearly observable from the 2017 compliance data (e.g., compliance data indicates clearly which vehicles have turbochargers and which have continuously variable transmissions), but in some cases technology levels are less observable. For the latter, like levels of mass reduction, the analysis categorized levels of technology already used in a given vehicle. Similarly, engineering judgment was used to determine if higher mass reduction levels may be used practicably and safely for a given vehicle.

Either in mid-year compliance data for MY 2016, final compliance data for MY 2017, or separately and at the agencies' invitation prior to the NPRM or in comments in responses to the NPRM, most manufacturers provided guidance on the technology already present in each of their vehicle model/configurations. This information was not as complete for all manufacturers' products as needed for the analysis, so, in some cases, information was supplemented with publicly available data, typically from manufacturer media sites. In limited cases, manufacturers did not supply information, and Start Printed Page 24291information from commercial and publicly available sources was used.

The agencies continued to evaluate emerging technologies in the analysis. In response to comments,[367] and given recent product launches for MY 2020, and some very recently announced future product offerings, the agencies elevated some technologies that were discussed in the NPRM to the compliance simulation. As a result, several additional engine technologies, expanded levels of mass reduction technology, and some additional combinations of engines with plug-in hybrid, or strong hybrid technology are available in the compliance pathways for the final rule analysis.

In addition, some redundant technologies, or technologies that were inadvertently represented on the technology tree as being available to be applied twice, have been consolidated. For instance, previous basic versions of engine friction reduction were layered on top of basic engine maps, but the efficiency in many modern engine maps already include the benefits of that engine friction reduction technology. The following Table VI-8 lists the technologies considered in the final rule analysis, with the data sources used to map those technologies to vehicles in the analysis fleet.

Start Printed Page 24292

Start Printed Page 24293

Start Printed Page 24294

Industry commenters generally stated the MY 2016 baseline technology content presented in the NPRM as an improvement over previous analyses because it more accurately accounted for technology already used in the fleet.[368 369] In contrast, some commenters expressed preference for EPA's baseline technology assignment assumptions presented in the Draft TAR for mass reduction, tire rolling resistance, and aerodynamic drag because those assumptions projected very few technology improvements were present in the baseline fleet. In assessing the comments, the agencies found that Start Printed Page 24295using the EPA Draft TAR approach would lead to projected compliance pathways with overestimated fuel economy improvements and underestimated costs.[370]

Many of those assumptions were neither scientifically meritorious, nor isolated examples. For instance, for the EPA Draft TAR and Proposed Determination analyses, the BMW i3, a vehicle with full carbon fiber bodysides and downsized, mass-reduced wheels and tires (some of the most advanced mass reducing technologies commercialized in the automotive industry), was assumed to have 1.0 percent mass reduction (a very minor level of mass reduction). Similarly, previous analyses assigned the Chevrolet Corvette, a performance vehicle that has long been a platform for commercializing advanced weight saving technologies,[371] with zero mass reduction. For aerodynamic drag, previous EPA analysis assumed that pickup trucks could achieve the aerodynamic drag profile typical of a sedan, with little regard for form drag constraints or frontal area (and headroom, or ground clearance) considerations. These assumptions commonly led to projections of a 20 percent improvement in mass, aerodynamic drag, and tire rolling resistance, even when a large portion of those improvements had either already been implemented, or were not technologically feasible. On the other hand, in the Draft TAR, NHTSA presented methodologies to evaluate content for mass reduction technology, aerodynamic drag improvements, and rolling resistance technologies that better accounted for the actual level of technologies in the analysis fleet. Throughout the rulemaking process, the agencies reconciled these differences, jointly presented improved approaches in the NPRM similar to what NHTSA presented in the Draft TAR, and again used those reconciled approaches in today's analysis.[372]

Many commenters correctly observed that the analysis fleet in the NPRM recognized more technology content in the baseline than in the Draft TAR (with higher penetration rates of tire rolling resistance and aerodynamic drag improvements, for instance), but also that the fuel economy values of the fleet had not improved all that much from the previous year. Some commenters concluded that the NPRM baseline technology assignment process was arbitrary and overstated the technology content already present in the baseline fleet.[373 374] The agencies agree that there was a large increase in the amount of road load technology credited in the baseline fleet between EPA's Draft TAR and the jointly produced NPRM, and clarify that this change was largely due to a recognition of technologies that were actually present in the fleet, but not properly accounted for in previous analyses. The change in penetration rates of road load technologies (after accounting for glider share updates, which is discussed in more detail in the mass reduction technology section) between the NPRM and today's analysis is relatively small.

Many commenters noted that the different baseline road load assumptions (and other technology modeling) materially affect compliance pathways, and projected costs.[375] ICCT commented that the agencies should conduct sensitivity analyses assuming every vehicle in the analysis fleet is set to zero percent road load technology improvement, to demonstrate how the technology content of the analysis fleet affected the compliance scenarios.[376]

While the agencies have clearly described the methods by which initial road load technologies are assigned in Section VI.C.4 Mass Reduction, Section VI.C.5 Aerodynamics, and Section VI.C.6 Tire Rolling Resistance below, the agencies considered a sensitivity case that assumed no mass reduction, rolling resistance, or aerodynamic improvements had been made to the MY 2017 fleet (i.e., setting all vehicle road levels to zero—MRO, AERO and ROLL0). While this is an unrealistic characterization of the initial fleet, the agencies conducted a sensitivity analysis to understand any affect it may have on technology penetration along other paths (e.g. engine and hybrid technology). Under the CAFE program, the sensitivity analysis shows a slight decrease in reliance on engine technologies (HCR engines, turbocharge engines, and engines utilizing cylinder deactivation) and hybridization (strong hybrids and plug-in hybrids) in the baseline (relative to the central analysis). The consequence of this shift to reliance on lower-level road load technologies is a reduction in compliance cost in the baseline of about $300 per vehicle (in MY 2026). As a result, cost savings in the preferred alternative are reduced by about $200 per vehicle. Under the CO2 program, the general trend in technology shift is less dramatic (though the change in BEVs is larger) than the CAFE results. The cost change is also comparable, but slightly smaller ($200 per vehicle in the baseline) than the CAFE program results. Cost savings under the preferred alternative are further reduced by about $100. With the lower technology costs in all cases, the consumer payback periods decreased as well. These results are consistent with the approach taken by manufacturers who have already deployed many of the low-level road load reduction opportunities to improve fuel economy.

Some commenters preferred that the agencies develop a different methodology based on reported road load coefficients (“A,” “B” and “C” coastdown coefficients) to estimate levels of aerodynamic drag improvement and rolling resistance in the baseline fleet that did not rely on CBI.[377] The agencies considered this, but determined that using CBI to assign baseline aerodynamic drag levels and rolling resistance values was more accurate and appropriate. Estimating aerodynamic drag levels and rolling resistance levels from coastdown coefficients is not straightforward, and to do it well would require information the agencies do not have (much of which is also CBI). For instance, rotational inertias of wheel, tire, and brake packages can affect coastdown, so mass of the vehicle is not sufficient. The frontal area of the vehicles, a key component for calculating aerodynamic drag, is rarely known, and often requires manufacturer input to get an accurate value. Other important vehicle features like all-wheel-drive should also be accounted for, and the agencies would struggle to correctly identify improvements in rolling resistance, low-drag brakes, and secondary axle disconnect, because all of these technologies would present similar signature on a coast down test. All of these technologies are represented as technology pathways in today's analysis. Manufacturers acknowledged the possibility of using road load coefficients to estimate rolling resistance and aerodynamic features, but warned that the process “required Start Printed Page 24296various assumptions and is not very accurate,” and stated that the use of CBI to assess aerodynamic and rolling resistance technologies is an “accurate and practical solution” to assign these difficult to observe technologies.[378]

(3) Assigning Engine Configurations

Engine technology costs can vary significantly by the configuration of the engine. For instance, adding variable valve lift to each cylinder on an engine would cost more for an engine with eight cylinders than an engine with four cylinders. Similarly, the cost of adding a turbocharger to an engine and downsizing the engine would be different going from a naturally aspirated V8 to a turbocharged V6 than going from a naturally aspirated V6 to a turbocharged I4. As discussed in detail in the engine technology section of this document, the cost files for the CAFE model account for instances such as these examples.

Information in the analysis fleet enables the CAFE model to reference the intended engine costs. The “Engine Technology Class (Observed)” lists the architecture of the observed engine. Notably, the analysis assumes that nearly all turbo charged engines take advantage of downsizing to optimize fuel efficiency, minimize the cost of turbo charging, and to maintain performance (to the extent practicable) with the naturally aspirated counterpart engine. Therefore, engines observed in the fleet that have already been down-sized must reference costs for a larger basic engine, which assumes down-sizing with the application of turbo technology. In these cases, the “Engine Technology Class” which is used to reference costs will be larger than the “Engine Technology Class (Observed).”

This is the same process agencies used in the NPRM, and it corrects a previous error in the Draft TAR analysis, which incorrectly underestimated turbocharged engine costs.[379] Some commenters expressed confusion and disagreement with this correction, with some even commenting that the analysis baselessly inflated costs of turbocharging technologies between the Draft TAR and the NPRM.[380] To be clear, this was a correction so that the costs used to calculate turbocharged engine costs accurately reflected the total costs for a turbocharged engine.

(4) Characterizing Shared Vehicle Platforms, Engines, and Transmissions

Another aspect of characterizing vehicle model/configurations in the analysis fleet is based on whether they share a “platform” with other vehicle model/configurations. A “platform” refers to engineered underpinnings shared on several differentiated products. Manufacturers share and standardize components, systems, tooling, and assembly processes within their products (and occasionally with the products of another manufacturer) to manage complexity and costs for development, manufacturing, and assembly.

The concept of platform sharing has evolved over time. Years ago, manufacturers rebadged vehicles and offered luxury options only on premium nameplates (and manufacturers shared some vehicle platforms in limited cases). Today, manufacturers share parts across highly differentiated vehicles with different body styles, sizes, and capabilities that may share the same platform. For instance, the Honda Civic and Honda CR-V share many parts and are built on the same platform. Engineers design chassis platforms with the ability to vary wheelbase, ride height, and even driveline configuration. Assembly lines can produce hatchbacks and sedans to cost-effectively utilize manufacturing capacity and respond to shifts in market demand. Engines made on the same line may power small cars or mid-size sport utility vehicles. In addition, although the agencies' analysis, like past CAFE analyses, considers vehicles produced for sale in the U.S., the agency notes these platforms are not constrained to vehicle models built for sale in the U.S.; many manufacturers have developed, and use, global platforms, and the total number of platforms is decreasing across the industry. Several automakers (for example, General Motors and Ford) either plan to, or already have, reduced their number of platforms to less than 10 and account for the overwhelming majority of their production volumes on that small number of platforms.

Vehicle model/configurations derived from the same platform are so identified in the analysis fleet. Many manufacturers' use of vehicle platforms is well documented in the public record and widely recognized among the vehicle engineering community. Engineering knowledge, information from trade publications, and feedback from manufacturers and suppliers was also used to assign vehicle platforms in the analysis fleet.

When the CAFE model is deciding where and how to add technology to vehicles, if one vehicle on the platform receives new technology, other vehicles on the platform also receive the technology as part of their next major redesign or refresh.[381] Similar to vehicle platforms, manufacturers create engines that share parts. For instance, manufacturers may use different piston strokes on a common engine block, or bore out common engine block castings with different diameters to create engines with an array of displacements. Head assemblies for different displacement engines may share many components and manufacturing processes across the engine family. Manufacturers may finish crankshafts with the same tools to similar tolerances. Engines on the same architecture may share pistons, connecting rods, and the same engine architecture may include both six and eight cylinder engines. One engine family may appear on many vehicles on a platform, and changes to that engine may or may not carry through to all the vehicles. Some engines are shared across a range of different vehicle platforms. Vehicle model/configurations in the analysis fleet that share engines belonging to the same platform are also identified as such.

It is important to note that manufacturers define common engines differently. Some manufacturers consider engines as “common” if the engines shared an architecture, components, or manufacturing processes. Other manufacturers take a narrower definition, and only assume “common” engines if the parts in the engine assembly are the same. In some cases, manufacturers designate each engine in each application as a unique powertrain. For example, a manufacturer may have listed two engines separately for a pair that share designs for the engine block, the crank shaft, and the head because the accessory drive components, oil pans, and engine calibrations differ between the two. In practice, many engines share parts, tooling, and assembly resources, and manufacturers often coordinate design updates between two similar engines. Engine families, designated in the analysis using “engine codes,” for each manufacturer were tabulated and assigned based on data-driven criteria. If engines shared a common cylinder count and configuration, displacement, valvetrain, and fuel type, those engines Start Printed Page 24297may have been considered together. In addition, if the compression ratio, horsepower, and displacement of engines were only slightly different, those engines were considered the same for the purposes of redesign and sharing.

Vehicles in the analysis fleet with the same engine family will, therefore, adopt engine technology in a coordinated fashion. Specifically, if such vehicles have different design schedules (i.e., refresh and redesign schedules), and a subset of vehicles using a given engine add engine technologies during of a redesign or refresh that occurs in an early model year (e.g., 2018), other vehicles using the same engine “inherit” these technologies at the soonest ensuing refresh or redesign. This is consistent with a view that, over time, most manufacturers are likely to find it more practicable to shift production to a new version of an engine than to continue production of both the new engine and a “legacy” engine indefinitely. By grouping engines together, the CAFE model controls future engine families to ensure reasonable powertrain complexity. This means, however, that for manufacturers that submitted highly atomized engine and transmission portfolios, there is a practical cap on powertrain complexity and the ability of the manufacturer to optimize the displacement of (i.e., “right size”) engines perfectly for each vehicle configuration. This concept is discussed further in Section VI.B.4.a), below.

Like with engines, manufacturers often use transmissions that are the same or similar on multiple vehicles. Manufacturers may produce transmissions that have nominally different machining to castings, or manufacturers may produce transmissions that are internally identical, except for the final gear ratio. In some cases, manufacturers sub-contract with suppliers that deliver whole transmissions. In other cases, manufacturers form joint ventures to develop shared transmissions, and these transmission platforms may be offered in many vehicles across manufacturers. Manufacturers use supplier and joint-venture transmissions to a greater extent than they do with engines. To reflect this reality, shared transmissions were considered for manufacturers as appropriate. Transmission configurations are referred to in the analysis as “transmission codes.” Like the inheritance approach outlined for engines, if one vehicle application of a shared transmission family upgraded the transmission, other vehicle applications also upgraded the transmission at the next refresh or redesign year. To define common transmissions, the agencies considered transmission type (manual, automatic, dual-clutch, continuously variable), number of gears, and vehicle architecture (front-wheel-drive, rear-wheel-drive, all-wheel-drive based on a front-wheel drive platform, or all-wheel-drive based on a rear-wheel-drive platform). If vehicles shared these attributes, these transmissions were grouped for the analysis. Vehicles in the analysis fleet with the same transmission configuration will adopt transmission technology together, as described above.

Having all vehicles that share a platform (or engines that are part of a family) adopt fuel economy-improving/CO2 emissions-reducing technologies together, subject to refresh/redesign constraints, reflects the real-world considerations described above, but also overlooks some decisions manufacturers might make in the real world in response to market pull. Accordingly, even though the analysis fleet is incredibly complex, it is also over-simplified in some respects compared to the real world. For example, the CAFE model does not currently attempt to simulate the potential for a manufacturer to shift the application of technologies to improve performance rather than fuel economy. Therefore, the model's representation of the “inheritance” of technology can lead to estimates a manufacturer might eventually exceed fuel economy standards as technology continues to propagate across shared platforms and engines. While the agencies have previously seen examples of extended periods during which some manufacturers exceeded one or both CAFE and/or CO2 standards, in plenty of other examples, manufacturers chose to introduce (or even reintroduce) technological complexity into their vehicle lineups in response to buyer preferences. Going forward, and recognizing the recent trend for consolidating platforms, it seems likely manufacturers will be more likely to choose efficiency over complexity in this regard; therefore, the potential should be lower that today's analysis turns out to be oversimplified compared to the real world.

Manufacturers described shared engines, transmissions, and vehicle platforms as “standard business practice” and they were encouraged that the NHTSA analysis in the Draft TAR, and the jointly issued NPRM placed realistic limits on the number of unique engines and transmissions in a powertrain portfolio.[382] In previous rulemakings, stakeholders pointed out that shared parts and portfolio complexity should be considered (but were not), and that the proliferation of unique technology combinations resulting from unconstrained compliance pathways would jeopardize economies of scale in the real world.[383]

HD Systems acknowledged that previous rulemakings did not appropriately consider part sharing, but contended that in today's global marketplace, manufacturers have flexibility to compete in new ways that break old part sharing rules.[384] The agencies acknowledge that some transmissions are now sourced through suppliers, and that economies of scale could, in the future be achieved at an industry level instead of a manufacturer level; however, even when manufacturers outsource a transmission, recent history suggests they apply that transmission to multiple vehicles to control assembly plant and service parts complexity, as they would if they were making the transmission themselves. Similarly, even for global platforms, or global powertrains, there is little evidence that manufacturers fragment powertrain line-ups for a vehicle, or a set of vehicles that have typically used the same engine. The agencies will continue to consider how to capture more accurately the ways vehicles share engines, transmissions, and platforms in future rulemakings, but the part-sharing and modeling approach presented in the NPRM and this final rule represents a marked improvement over previous analysis.

(5) Characterizing Production Design Cycles

Another aspect of characterizing vehicles in the analysis fleet is based on when they can next be refreshed or redesigned. Redesign schedules play an important role in determining when new technologies may be applied. Many technologies that improve fuel economy and reduce CO2 emissions may be difficult to incorporate without a major product redesign. Therefore, each vehicle model in the analysis fleet has an associated redesign schedule, and the CAFE model uses that schedule to implement significant advances in some technologies (like major mass reduction) to redesign years, while allowing manufacturers to include minor advances (such as improved tire rolling Start Printed Page 24298resistance) during a vehicle “refresh,” or a smaller update made to a vehicle, which can happen between redesigns. In addition to refresh and redesign schedules associated with vehicle model/configurations, vehicles that share a platform subsequently have platform-wide refresh and redesign schedules for mass reduction technologies.

To develop the refresh/redesign cycles used for the NPRM vehicles in the analysis fleet, information from commercially available sources was used to project redesign cycles through MY 2022, as was done for NHTSA's analysis for the 2016 Draft TAR.[385] Commercially available sources' estimates through MY 2022 are generally supported by detailed consideration of public announcements plus related intelligence from suppliers and other sources, and recognize that uncertainty increases considerably as the forecasting horizon is extended. For MYs 2023-2035, in recognition of that uncertainty, redesign schedules were extended considering past pacing for each product, estimated schedules through MY 2022, and schedules for other products in the same technology classes. As mentioned above, potentially confidential forward-looking information was not requested from manufacturers; nevertheless, all manufacturers had an opportunity to review the estimates of product-specific redesign schedules. A few manufacturers provided related forecasts and, for the most part, that information corroborated the estimates.

Some commenters suggested supplanting these estimated redesign schedules with estimates applying faster cycles (e.g., four to five years), and this approach was considered for the analysis. Some manufacturers tend to operate with faster redesign cycles and may continue to do so, and manufacturers tend to redesign some products more frequently than others. However, especially considering that information presented by manufacturers largely supports estimates discussed above, applying a “one size fits all” acceleration of redesign cycles would not improve the analysis; instead, assuming a fixed, shortened redesign schedule across the industry would likely reduce consistency with the real world, especially for light trucks, which are redesigned, on average, no less than every six years (see Table VI-9, below). Moreover, if some manufacturers accelerate redesigns in response to new standards, doing so would likely involve costs (greater levels of stranded capital, reduced opportunity to benefit from “learning”-related cost reductions) greater than reflected in other inputs to the analysis.

As discussed in the NPRM, manufacturers use diverse strategies with respect to when, and how often they update vehicle designs. While most vehicles have been redesigned sometime in the last five years, many vehicles have not. In particular, vehicles with lower annual sales volumes tend to be redesigned less frequently, perhaps giving manufacturers more time to recoup the investment needed to bring the product to market. In some cases, manufacturers continue to produce and sell vehicles designed more than a decade ago.

Start Printed Page 24299

Each manufacturer may use different strategies throughout their product portfolio, and a component of each strategy may include the timing of refresh and redesign cycles. Table VI-10 summarizes the average time between redesigns, by manufacturer, by vehicle technology class. Dashes mean the manufacturer has no volume in that vehicle technology class in the MY 2017 analysis fleet. Across the industry, manufacturers average 6.6 years between product redesigns.

Trends on redesign schedules identified in the NPRM remain in place for today's analysis. Pick-up trucks have much longer redesign schedules than small cars. Some manufacturers redesign vehicles often, while other manufacturers redesign vehicles less often. Even if two manufacturers have similar redesign cadence, the model years in which the redesigns occur may still be different and dependent on where each of the manufacturer's products are in their life cycle.

Table VI-11 summarizes the average age of manufacturers' offering by vehicle technology class. A value of “0.0” means that every vehicle for a manufacturer in the vehicle technology class, represented by the MY 2017 analysis fleet was new in MY 2017. Across the industry manufacturers redesigned MY 2017 vehicles an average of 3.5 years earlier, meaning the average MY 2017 vehicle was last redesigned in approximately MY 2013, also on average near a midpoint in their product lifecycle.

Start Printed Page 24300

Some commenters cited examples of vehicles in the NPRM analysis fleet where the redesign years were off by a year here or there in the 2017-2022 timeframe relative to the most recent public announcements, or that the extended forecasts were too rigid.[386] The CAFE model structurally requires an input for the redesign years, and the agencies worked to make these generally representative without disclosing precise CBI product plans. Many of the redesign schedules were carried over from the NPRM, with a few minor updates.

Some commenters contended that the agencies should not look at the historical data to project the timing between redesigns (“business as usual”), but should instead adopt a “policy case” with an accelerated pace of redesigns and refreshes.[387] Some commenters suggested that the agencies use a standard 5 or 6 year redesign schedule for all manufacturers and all products as a way to lower projected costs.[388] Other stakeholders commented that the entire industry should be modeled with the ability to redesign everything at one time in the near term because that would not presuppose precisely how manufacturers may adjust their fleet.[389]

If the agencies were to implement any such approaches, the agencies would need to more precisely account for tooling costs, research and development costs, and product lifecycle marketing costs, or risk missing “hidden costs” of a shortened cadence. To account properly for these, the CAFE model would require major changes, and would require specific inputs that are currently covered generically under the retail price equivalency (RPE) factor.[390] The agencies considered these comments, and decided the process for refresh and redesign outlined in the NPRM was a reasonable and realistic approach to characterize product changes. The agencies conducted sensitivity analysis with compressed redesign and refresh schedules, though these ignore the resulting compressed amortization schedules, missing important costs that are incorporated in the current RPE assumptions.

Some commenters claimed that the agency had extraordinarily extended redesign schedule of 17.7 years for FCA between 2021-2025, and an average redesign time of 25.8 years for Ford between 2022-2025.[391] The agencies found these claims inaccurate and without basis. Table VI-10, “Summary of Sales Weighted Average Time Start Printed Page 24301between Engineering Redesigns, by Manufacturer, by Vehicle Technology Class” summarizes the data used in today's analysis (which is very similar to the information used in the NPRM, with some minor adjustments and updates to the fleet), and the detailed information vehicle-by-vehicle is reported in the “market data” file. The agencies recognize that the natural sequence of redesigns for some manufacturers and some products is not ideal to meet stringent alternatives, which is part of the consideration for economic practicability and technological feasibility. Manufacturers commented supportively on the idea of vehicle specific redesign schedules, and the redesign cadence used in the NPRM, as these contribute to realistic assessments of new technology penetration within the fleet, and acknowledge the heterogeneity in the product development approaches and business practices for each manufacturer.[392] One commenter recognized that redesign and refresh schedules represented a vast improvement over phase-in caps to model the adoption of mature technologies.[393]

Other commenters argued that the structural construct of technologies only being available at redesign or at refresh (via inheritance) did not reflect real world actions and was not supported by any actual data.[394] Other commenters acknowledged the inheritance of engine and transmission technologies at refresh as an important, positive feature of the CAFE model.[395] HD Systems argued that an engine or transmission package available in other markets on a global platform could be imported to the U.S. market during refresh, and did not require a “leader” at redesign in the U.S. market to seed adoption. HDS cited a few examples where manufacturers have introduced strong hybrid powertrains on an existing vehicle a year or two after the product launch, not associated with any particular vehicle redesign or refresh.

The agencies carefully considered these comments, and observed that some relatively low volume hybrid options may appear after launch, or that some transmissions were quickly replaced shortly after a major redesign. In many of these cases, launch delays, warranty claims, or other external factors contributed to, at least in part, an atypically timed introduction of fuel saving technology to the fleet.[396] At this point, this does not appear to be a mainstream, or preferred industry practice. However, the agencies will continue to evaluate this. For future rulemaking, the agencies may consider engine refresh and redesign cycles for engines and transmissions. These may be separate from vehicle redesign and refresh schedules because the powertrain product lifecycles may be longer on average than the typical vehicle redesign schedules. This approach, if researched and implemented in future analysis, could provide some opportunity for manufacturers to introduce new powertrain technologies independent of the vehicle redesign schedules, in addition to inheriting advanced powertrain technology as refresh as already modeled in the NPRM and today's analysis.

For today's analysis, the agencies, with a few exceptions based on updated publicly available information, carried over redesign cadences for each vehicle nameplate as presented in the NPRM. The agencies do not claim that the projected redesign years will perfectly match what industry does—notably because refresh and redesign information is CBI and the agencies have applied more generalized schedules to protect the CBI. Also, what any individual manufacturer may choose to do today could be completely different than what it chooses to do tomorrow due to changing business circumstances and plans—but the agencies have worked to ensure the timing of redesigns will be roughly correct (especially in the near term), and that the time between redesigns will continue forward for each manufacturer as it has based on recent history. The agencies have also increased the frequency of refreshes in response to comments about the proliferation of some engine and transmission families through manufacturers' product portfolios.

Also for today's analysis, the agencies now explicitly model CAFE compliance pathways out through 2050. For the model to work as intended, the agencies must project refresh and redesign schedules out through 2050. The agencies recognize that the accuracy of predictions about the distant future, particularly about refresh and redesign cycles through the 2030-2050 timeframe, are likely to be poor. If historical evolution of the industry continues, many of the nameplates carried forward in the fleet are likely to be out of production, and new nameplates not considered in the analysis are sure to emerge. Still, carrying forward the MY 2017 fleet with the current refresh and redesign cadences is consistent with the current analysis, and imposing an alternative schedule on the fleet, or making up new nameplates and retiring older nameplates without a clear basis, would lack proper foundation.

Start Printed Page 24302

(6) Defining Technology Adoption Features

In some circumstances, the agencies may reference full vehicle simulation effectiveness data for technology combinations that are not able to be, or are not likely to be applied to all vehicles. In some cases, a specific technology as modeled only exists on paper, and questions remain about the technological feasibility of the efficiency characterization.[397] Or, a technology may perform admirably on the test cycle, but fail to meet all functional, or performance requirements for certain vehicles.[398] In other cases, the intellectual property landscape may make commercialization of one technology risky for a manufacturer without the consent of the intellectual property owner.[399] In such cases, the agencies may not allow a technology to be applied to a certain vehicle. The agencies designate this in the “market data” file with a “SKIP” for the technology and vehicle. The logic is explained technology by technology in this document, as the logic was explained in the PRIA for this rule.

Some commenters argued that the restrictions of technologies on a case-by-case basis required case-by-case explanation (and not objective specification defined cut-offs), and that the use of CBI for performance considerations was unacceptable unless fully disclosed.[400] As discussed above, the agencies are not able to disclose CBI. Stakeholders have had plenty of opportunities to comment on the applicability of technologies, including the few that have used SKIP logic restrictions for a portion of the fleet.

Other commenters suggested an optimistic and wholly unfounded approach to manufacturer innovation, arguing that costs would continue to come down (beyond what is currently modeled with cost learning), and the list of fuel-saving technologies would continually regenerate itself (even if the technological mechanism for fuel saving technologies was not yet identified).[401] Therefore, the argument goes that people will figure out new ways to improve fuel saving technologies to increase their applicability, and the current technology characterization should be enabled for selection with no restriction—not because the commenter knows how the technology will be adapted, but that the commenter believes the technology could, eventually, within the timeline of the rulemaking, be adapted, brought to market, and be accepted by consumers. While the agencies recognize the improvements that many manufacturers Start Printed Page 24303have achieved in fuel saving technologies, some of which were difficult to foresee, the agencies have an obligation under the law to be judicious and specific about technological feasibility, and to avoid speculative conclusions about technologies to justify the rulemaking.

c) Other Analysis Fleet Data

(1) Safety Classes

The agencies referenced the mass-size-safety analysis to project the effects changes in weight may have on crash fatalities. That analysis, discussed in more detail in Section VI.D.2, considers how weight changes may affect safety for cars, crossover utility vehicles and sport utility vehicles, and pick-up trucks. To consider these effects, the agencies mapped each vehicle in the analysis fleet to the appropriate “Safety Class.”

(2) Labor Utilization

The analysis fleet summarizes components of direct labor for each vehicle considered in the analysis. The labor is split into three components: (1) Dealership hours worked on sales functions per vehicle, (2) direct assembly labor for final assembly, engine, and transmission, and (3) percent U.S. content.

In the MY 2016 fleet for the NPRM, the agencies catalogued production locations and plant employment, reviewed annual reports from the North American Dealership Association to estimate dealership employment (27.8 hours per vehicle sold), and estimated the industry average labor hours for final assembly of vehicles (30 hours per vehicle produced), engine machining and assembly (4 hours per engine produced), and transmission production (5 hours per transmission produced).

Today's analysis fleet carries over the estimated labor coefficients for sales and production, but references the most recent Part 583 American Automobile Labeling Act Report for percent U.S. content and for the location of vehicle assembly, engine assembly, and transmission assembly.[402]

(3) Production Volumes for Sales Analysis

A final important aspect of projecting what vehicles will exist in future model years and potential manufacturer responses to standards is estimating how future sales might change in response to different potential standards. If potential future standards appear likely to have major effects in terms of shifting production from cars to trucks (or vice versa), or in terms of shifting sales between manufacturers or groups of manufacturers, that is important for the agencies to consider. For previous analyses, the CAFE model used a static forecast contained in the analysis fleet input file, which specified changes in production volumes over time for each vehicle model/configuration. This approach yielded results that, in terms of production volumes, did not change between scenarios or with changes in important model inputs. For example, very stringent standards with very high technology costs would result in the same estimated production volumes as less stringent standards with very low technology costs. For this analysis, as in the proposal, the CAFE model begins with the first-year production volumes (i.e., MY 2017 for today's analysis) and adjusts ensuing sales mix year by year (between cars and trucks, and between manufacturers) endogenously as part of the analysis, rather than using external forecasts of future car/truck split and future manufacturer sales volumes. This leads the model to produce different estimates of future production volumes under different standards and in response to different inputs, reflecting the expectation that regulatory standards and other external factors will, in fact, impact the market.

(4) Comments on Other Analysis Fleet Data

Some commenters suggest that the CAFE model should run as a full consumer choice model (and this idea is discussed in more detail in Section VI.D.1). While this sounds like a reasonable request on the surface, such an approach would place enormous new demands on the data characterized in the fleet (and preceding fleets, which may be needed to calibrate a model properly). For instance, some model concepts may depend on a bevy of product features, such as interior cargo room, artistic appeal of the design, and perceived quality of the vehicle. But product features alone may not be sufficient. Additional information about dealership channels, product awareness and advertising effectiveness, and financing terms also may be required. Such information could dramatically increase the scope of work needed to characterize the analysis fleet for future rulemakings. As described in Section VI.D.1.b)(2)(d) Using Vehicle Choice Models in Rulemaking Analysis. Accordingly, the agencies decided not to develop such a model for this rulemaking.

2. Treatment of Compliance Credit Provisions

Today's final rule involves a variety of provisions regarding “credits” and other compliance flexibilities. Some recently introduced regulatory provisions allow a manufacturer to earn “credits” that will be counted toward a vehicle's rated CO2 emissions level, or toward a fleet's rated average CO2 or CAFE level, without reference to required levels for these average levels of performance. Such flexibilities effectively modify emissions and fuel economy test procedures, or methods for calculating fleets' CAFE and average CO2 levels. Such provisions are discussed below in Section VI.B.2. Other provisions (for CAFE, statutory provisions) allow manufacturers to earn credits by achieving CAFE or average CO2 levels beyond required levels; these provisions may hence more appropriately be termed “compliance credits.”

EPCA has long provided that, by exceeding the CAFE standard applicable to a given fleet in a given model year, a manufacturer may earn corresponding “credits” that the same manufacturer may, within the same regulatory class, apply toward compliance in a different model year. EISA amended these provisions by providing that manufacturers may, subject to specific statutory limitations, transfer compliance credits between regulatory classes, and trade compliance credits with other manufacturers. The CAA provides EPA with broad standard-setting authority for the CO2 program, with no specific directives regarding either CO2 standards or CO2 compliance credits.

EPCA also specifies that NHTSA may not consider the availability of CAFE credits (for transfer, trade, or direct application) toward compliance with new standards when establishing the standards themselves.[403] Therefore, this analysis, like that presented in the NPRM, considers 2020 to be the last model year in which carried-forward or transferred credits can be applied for the CAFE program. Beginning in model year 2021, today's “standard setting” analysis for NHTSA's program is conducted assuming each fleet must comply with the CAFE standard separately in every model year.

The “unconstrained” perspective acknowledges that these flexibilities exist as part of the program, and, while not considered by NHTSA in setting standards, are nevertheless important to consider when attempting to estimate the real impact of any alternative. Under Start Printed Page 24304the “unconstrained” perspective, credits may be earned, transferred, and applied to deficits in the CAFE program throughout the full range of model years in the analysis. The Final Environmental Impact Analysis (FEIS) accompanying today's final rule, like the corresponding Draft EIS analysis, presents results of “unconstrained” modeling. Also, because the CAA provides no direction regarding consideration of any CO2 credit provisions, today's analysis, like the NPRM analysis, includes simulation of carried-forward and transferred CO2 credits in all model years.

Some commenters took issue broadly with this treatment of compliance credits. Michalek and Whitefoot wrote that “we find this requirement problematic because the automakers use these flexibilities as a common means of complying with the regulation, and ignoring them will bias the cost-benefit analysis to overestimate costs.” [404]

Counter to the above general claim, the CAFE model does provide means to simulate manufacturers' potential application of some compliance credits, and both the analysis of CO2 standards and the NEPA analysis of CAFE standards do make use of this aspect of the model. As discussed above, NHTSA does not have the discretion to consider the credit program—in fact, the agency is prohibited by statute from doing so—in establishing maximum feasible standards. Further, as discussed below, the agencies also continue to find it appropriate for the analysis largely to refrain from simulating two of the mechanisms allowing the use of compliance credits.

The model's approach to simulating compliance decisions accounts for the potential to earn and use CAFE credits as provided by EPCA/EISA. The model similarly accumulates and applies CO2 credits when simulating compliance with EPA's standards. Like past versions, the current CAFE model can be used to simulate credit carry-forward (a.k.a. banking) between model years and transfers between the passenger car and light truck fleets but not credit carry-back (a.k.a. borrowing) from future model years or trading between manufacturers.

Regarding the potential to carry back compliance credits, UCS commented that, although past versions of the CAFE model had “considered this flexibility in its approach to multiyear modeling,” NHTSA had, without explanation, “abruptly discontinued support of this method of compliance,” such that “manufacturers are generally incentivized to over comply, regardless of whether carrying forward a deficit to be compensated by later overcompliance would be a more cost-effective method of compliance.” [405] Citing the potential that manufacturers could make use of carried back credits in the future, UCS also stated that “NHTSA's decision to constrain it in the model is unreasonable and arbitrary.” [406] UCS effectively implies that the agencies should base standards on analysis that presumes manufacturers will take full theoretical advantage of provisions allowing credits to be borrowed.

The agencies have carefully considered these comments, and while EPA's decisions regarding CO2 standards can consider the potential to carry back compliance credits from later to earlier model years, and NHTSA's “unconstrained” evaluation could also do so, past examples of failed attempts to carry back CAFE credits (e.g., a MY2014 carry back default leading to a civil penalty payment) underscore the riskiness of such “borrowing.” Recent evidence indicates manufacturers are disinclined to take such risks,[407] and both agencies find it reasonable and prudent to refrain from attempting to simulate such “borrowing” in rulemaking analysis.

Unlike past versions, the NPRM and current versions of CAFE model provide a basis to specify (in model inputs) CAFE credits available from model years earlier than those being explicitly simulated. For example, with this analysis representing model years 2017-2050 explicitly, credits earned in model year 2012 are made available for use through model year 2017 (given the current five-year limit on carry-forward of credits). The banked credits are specific to both the model year and fleet in which they were earned.

In addition to the above-mentioned comments, UCS also cited as “errors” that “the model does not accurately reflect the one-time exemption from the EPA 5-year credit life for credits earned in the MY 2010-2015 timeframe” and “NHTSA assumes that there will be absolutely no credit trading between manufacturers.”

As discussed below, in the course of updating the analysis fleet from MY 2016 to MY 2017, the agencies have updated and expanded the manner in which the model accounts for credits earned prior to MY 2017, including credits earned as early as MY 2009. In order to increase the realism with which the model transitions between the early model year (MYs 2017-2020) and the later years that are the subject of this action, the agencies have accounted for the potential that some manufacturers might trade some of these pre-MY 2017 credits to other manufacturers. However, as with the NPRM, the analysis refrains from simulating the potential that manufacturers might continue to trade credits during and beyond the model years covered by today's action. The agencies remain concerned that any realistic simulation of such trading would require assumptions regarding which specific pairs of manufacturers might actually trade compliance credits, and the evidence to date makes it clear that the credit market is far from fully “open.” With respect to the FCA comment cited above, the agencies also remain concerned that to set standards based on an analysis that presumes the use of program flexibilities risks making the corresponding actions mandatory. Some flexibilities—credit carry-forward (banking) and transfers between fleets in particular—involve little risk, because they are internal to a manufacturer and known in advance. As discussed above, credit carry-back involves significant risk, because it amounts to borrowing against future improvements, standards, and production volume and mix—and anticipated market demand for fuel efficient vehicles often fail to materialize. Similarly, credit trading also involves significant risk, because the ability of manufacturer A to acquire credits from manufacturer B depends not just on manufacturer B actually earning the expected amount of credit, but also on manufacturer B being willing to trade with manufacturer A, and on potential interest by other manufacturers. Manufacturers' compliance plans have already evidenced cases of compliance credit trades that were planned and subsequently aborted, reinforcing the agencies' judgment that, like credit banking, credit trading involves too much risk to be included in an analysis that informs decisions about the stringency of future standards. Nevertheless, recognizing that some manufacturers have actually been trading credits, the agencies have, as in the NPRM, included in the sensitivity analysis a case that simulates “perfect” trading of compliance credits, focusing Start Printed Page 24305on CO2 standards to illustrate the hypothetical maximum potential impact of trading. The FRIA summarizes results of this and other cases included in the sensitivity analysis.

As discussed in the CAFE model documentation, the model's default logic attempts to maximize credit carry-forward—that is, to “hold on” to credits for as long as possible. If a manufacturer needs to cover a shortfall that occurs when insufficient opportunities exist to add technology in order to achieve compliance with a standard, the model will apply credits. Otherwise the manufacturer carries forward credits until they are about to expire, at which point it will use them before adding technology that is not considered cost-effective. The model attempts to use credits that will expire within the next three years as a means to smooth out technology application over time to avoid both compliance shortfalls and high levels of over-compliance that can result in a surplus of credits. Although it remains impossible precisely to predict manufacturer's actual earning and use of compliance credits, and this aspect of the model may benefit from future refinement as manufacturers and regulators continue to gain experience with these provisions, this approach is generally consistent with manufacturers' observed practices.

NHTSA introduced the CAFE Public Information Center to provide public access to a range of information regarding the CAFE program,[408] including manufacturers' credit balances. However, there is a data lag in the information presented on the CAFE PIC that may not capture credit actions across the industry for as much as several months. Furthermore, CAFE credits that are traded between manufacturers are adjusted to preserve the gallons saved that each credit represents.[409] The adjustment occurs at the time of application rather than at the time the credits are traded. This means that a manufacturer who has acquired credits through trade, but has not yet applied them, may show a credit balance that is either considerably higher or lower than the real value of the credits when they are applied. For example, a manufacturer that buys 40 million credits from Tesla may show a credit balance in excess of 40 million. However, when those credits are applied, they may be worth only 1/10 as much—making that manufacturer's true credit balance closer to 4 million than 40 million.

For the NPRM, the agencies reviewed then-recent credit balances, estimated the potential that some manufacturers could trade credits, and developed inputs that make carried-forward credits available in each of model years 2011-2015, after subtracting credits assumed to be traded to other manufacturers, adding credits assumed to be acquired from other manufacturers through such trades, and adjusting any traded credits (up or down) to reflect their true value for the fleet and model year into which they were traded.[410] For today's analysis, an additional model year's data was available in mid-2019, and the agencies updated these inputs, as summarized in Table VI-12, Table VI-13, and Table VI-14. While the CAFE model will transfer expiring credits into another fleet (e.g., moving expiring credits from the domestic car credit bank into the light truck fleet), some of these credits were moved into the initial banks to improve the efficiency of application and both to reflect better the projected shortfalls of each manufacturer's regulated fleets and to represent observed behavior. For context, a manufacturer that produces one million vehicles in a given fleet, and experiences a shortfall of 2 mpg, would need 20 million credits, adjusted for fuel savings, to offset the shortfall completely.

Start Printed Page 24306

Start Printed Page 24307

In addition to the inclusion of these existing credit banks, the CAFE model also updated its treatment of credits in the rulemaking analysis. EPCA requires that NHTSA set CAFE standards at maximum feasible levels for each model year without consideration of the program's credit mechanisms. However, as recent NHTSA CAFE/EPA tailpipe CO2 emissions rulemakings have evaluated effects of standards over longer time periods, the early actions taken by manufacturers required more nuanced representation. Accordingly, the CAFE model now provides for a setting to establish a “last year to consider credits.” This adjustment is set at the last year for which new standards are not being considered (MY 2020 in this analysis). This allows the model to replicate the practical application of existing credits toward compliance in early years but also to examine the impact of proposed standards based solely on fuel economy improvements in all years for which new standards are being considered.

Regarding the model's simulation of manufacturers' potential earning and application of compliance credits, UCS commented that the model “inexplicably lets credits expire” because “all technologies which pay for themselves within the assumed payback period are applied to all manufacturers, regardless of credit status.” UCS also claimed that “NHTSA did not accurately reflect unique attributes of EPA's credit bank,” that “credits are not traded between manufacturers,” and that “NHTSA does not model credit carryback for compliance.” [411] Relatedly, as discussed above, UCS attributes modeling outcomes to the “effective cost” metric used to select from among available fuel-saving technologies.[412] As discussed in Section VI.B.1, the agencies expect that manufacturers are likely to improve fuel economy voluntarily insofar as doing so “pays back” economically within a short period (30 months), and the agencies note that periods of regulatory stability have, in fact, been marked by CAFE levels exceeding requirements. As discussed above, the agencies have excluded simulation of credit trading (except in MYs prior to those under consideration, aside from an idealized case presented in the sensitivity analysis) and likewise excluded simulation of potential “carryback” provisions. The agencies have excluded modeling these scenarios not just because of the analytical complexities involved (and rejecting, for example, the random number generator analysis suggested by UCS), but also because the agencies agree that the actual provisions regarding trading and borrowing of compliance credits create too much risk to be used in the analysis underlying consideration of standards. However, as discussed above, the agencies have revised the “metric” used to prioritize available options to apply fuel-saving technologies. As discussed below, the agencies have revised model inputs to include the large quantity of “legacy” compliance credits EPA has made available under its CO2 standards.

The CAFE model has also been modified to include a similar representation of existing credit banks in EPA's CO2 program. While the life of a CO2 credit, denominated in metric tons of CO2, has a five-year life, matching the lifespan of CAFE credits, such credits earned in the early MY 2009-2011 years of the EPA program, may be used through MY 2021.[413] The CAFE model was not modified to allow Start Printed Page 24308exceptions to the life-span of compliance credits, and, to reflect statutory requirements, treated them as if they may be carried forward for no more than five years, so the initial credit banks were modified to anticipate the years in which those credits might be needed. MY 2016 was simulated explicitly in the NPRM analysis to prohibit the inclusion of banked credits in MY 2016 (which could be carried forward from MY 2016 to MY 2021), and thus underestimated the extent to which individual manufacturers, and the industry as a whole, could rely on these early credits to comply with EPA standards between MY 2016 and MY 2021. However, as indicated in the NPRM, the final rule's model inputs updated the analysis fleet's basis to MY 2017, such that these additional banked credits can be included. The credit banks with which the simulations in this analysis were conducted are presented in the following Tables:

Start Printed Page 24309

While the CAFE model does not simulate the ability to trade credits between manufacturers, it does simulate the strategic accumulation and application of compliance credits, as well as the ability to transfer credits between fleets to improve the compliance position of a less efficient fleet by leveraging credits earned by a more efficient fleet. The model prefers to hold on to earned compliance credits within a given fleet, carrying them forward into the future to offset potential future deficits. This assumption is consistent with observed strategic manufacturer behavior dating back to 2009.

From 2009 to present, no manufacturer has transferred CAFE credits into a fleet to offset a deficit in the same year in which they were earned. This has occurred with credits acquired from other manufacturers via trade but not with a manufacturer's own credits. Therefore, the current representation of credit transfers between fleets—where the model prefers to transfer expiring, or soon-to-be-expiring credits rather than newly earned credits—is both appropriate and consistent with observed industry behavior.

This may not be the case for CO2 standards, though it is difficult to be certain at this point. The CO2 program seeded the industry with a large quantity of early compliance credits (earned in MYs 2009-2011) [414] prior to the existence formal CO2 standards. Early credits from MYs 2010 and 2011, however, do not expire until 2021. Thus, for manufacturers looking to offset deficits, it is more sensible to exhaust credits that were generated during later model years (which are set to expire within the next five years), rather than relying on the initial bank of credits from MYs 2010 and 2011. The first model year for which earned credits outlive the initial bank is MY 2017, for which final manufacturer CO2 performance data (and hence, banked credits) has not yet been released. However, considering that under the CO2 program manufacturers simultaneously comply with passenger car and light truck fleets, to more accurately represent the CO2 credit system the CAFE model allows (and encourages) intra-year transfers between regulated fleets for the purpose of simulating compliance with the CO2 standards.

a) Off-Cycle and A/C Efficiency Adjustments to CAFE and Average CO2 Levels

In addition to more rigorous accounting of CAFE and CO2 credits, the model now also accounts for air conditioning efficiency and off-cycle adjustments. NHTSA's program considers those adjustments in a manufacturer's compliance calculation starting in MY 2017, and the NPRM version of the model used the adjustments claimed by each manufacturer in MY 2016 as the starting point for all future years. Because air conditioning efficiency and off-cycle adjustments are not credits in NHTSA's program, but rather adjustments to compliance fuel economy (much like the Flexible Fuel Vehicle adjustments due to phase out in MY 2019), they may be included under either a “standard setting” or “unconstrained” analysis perspective.

The manner in which the CAFE model treats the EPA and CAFE A/C efficiency and off-cycle credit programs is similar, but the model also accounts for A/C leakage (which is not part of NHTSA's program). When determining the compliance status of a Start Printed Page 24310manufacturer's fleet (in the case of EPA's program, PC and LT are the only fleet distinctions), the CAFE model weighs future compliance actions against the presence of existing (and expiring) CO2 credits resulting from over-compliance with earlier years' standards, A/C efficiency credits, A/C leakage credits, and off-cycle credits.

Another aspect of credit accounting, implemented in the NPRM version of the CAFE model, involved credits related to the application of off-cycle and A/C efficiency adjustments, which manufacturers earn by taking actions such as special window glazing or using reflective paints that provide fuel economy improvements in real-world operation but do not produce measurable improvements in fuel consumption on the 2-cycle test.

NHTSA's inclusion of off-cycle and A/C efficiency adjustments began in MY 2017, while EPA has collected several years' worth of submissions from manufacturers about off-cycle and A/C efficiency technology deployment. Currently, the level of deployment can vary considerably by manufacturer, with several claiming extensive Fuel Consumption Improvement Values (FCIV) for off-cycle and A/C efficiency technologies, and others almost none. The analysis of alternatives presented here (and in the NPRM) does not attempt to project how future off-cycle and A/C efficiency technology use will evolve or speculate about the potential proliferation of FCIV proposals submitted to the agencies. Rather, this analysis uses the off-cycle credits submitted by each manufacturer for MY 2017 compliance, and, with a few exceptions, carries these forward to future years. Several of the technologies described below are associated with A/C efficiency and off-cycle FCIVs. In particular, stop-start systems, integrated starter generators, and full hybrids are assumed to generate off-cycle adjustments when applied to vehicles to improve their fuel economy. Similarly, higher levels of aerodynamic improvements are assumed to include active grille shutters on the vehicle, which also qualify for off-cycle FCIVs.

The NPRM analysis assumed that any off-cycle FCIVs that are associated with actions outside of the technologies discussed in Section VI.C (either chosen from the pre-approved “pick list,” or granted in response to individual manufacturer petitions) remained at the levels claimed by manufacturers in MY 2017. Any additional A/C efficiency and off-cycle adjustments that accrued as the result of explicit technology application calculated dynamically in each model year for each alternative. The NPRM version of the CAFE model also represented manufacturers' credits for off-cycle improvements, A/C efficiency improvements, and A/C leakage reduction in terms of values applicable across all model years.

Recognizing that application of these improvements thus far varies considerably among manufacturers, such that some manufacturers have opportunities to earn significantly more of the corresponding adjustments over time, the agencies have expanded the CAFE model's representation of these credits to provide for year-by-year specification of the amounts of each type of adjustment for each manufacturer, denominated in grams CO2 per mile,[415] as summarized in the following table:

Start Printed Page 24311

Start Printed Page 24312

Start Printed Page 24313

In addition to these refinements to the estimation of the quantities of adjustments earned over time by each manufacturer, the agencies revised the Start Printed Page 24314CAFE model to apply estimates of the corresponding costs. For today's analysis, the agencies applied estimates developed previously by EPA, adjusting these values to 2019 dollars. The following table summarizes inputs through model year 2030:

The model currently accounts for any off-cycle adjustments associated with technologies that are included in the set of fuel-saving technologies explicitly simulated as part of this proposal (for example, start-stop systems that reduce fuel consumption during idle or active grille shutters that improve aerodynamic drag at highway speeds) and accumulates these adjustments up to the 10 g/mi cap. As a practical matter, most of the adjustments for which manufacturers are claiming off-cycle FCIV exist outside of the technology tree, so the cap is rarely reached during compliance simulation. The agencies have considered the potential to model their application explicitly. However, doing so would require data regarding which vehicle models already possess these improvements as well as the cost and expected value of applying them to other models in the future. Such data is currently too limited to support explicit modeling of these technologies and adjustments.

b) Alternative Fuel Vehicles

When establishing maximum feasible fuel economy standards, NHTSA is prohibited from considering the availability of alternatively fueled vehicles,[417] and credit provisions related to AFVs that significantly increase their fuel economy for CAFE compliance purposes. Under the “standard setting” perspective, these technologies (pure battery electric vehicles and fuel cell vehicles) [418] are not available in the compliance simulation to improve fuel economy. Under the “unconstrained” perspective, such as is documented in the DEIS and FEIS, the CAFE model considers these technologies in the same manner as other available technologies, and may apply them if they represent cost-effective compliance pathways. However, under both perspectives, the analysis continues to include dedicated AFVs that already exist in the MY 2017 fleet (and their projected future volumes). Also, because the CAA provides no direction regarding consideration of alternative fuels, the final rule's analysis includes simulation of the potential that some manufacturers might introduce new AFVs in response to CO2 standards. To represent the compliance benefit from such a response fully, NHTSA modified the CAFE model to include the specific provisions related to AFVs under the CO2 standards. In particular, the CAFE model now carries a full representation of the production multipliers related to electric vehicles, fuel cell vehicles, plug-in hybrids, and CNG vehicles, all of which vary by year through MY 2021.

EPCA also provides that CAFE levels may, subject to limitations, be adjusted upward to reflect the sale of flexible fuel vehicles (FFVs). Although these adjustments end after model year 2020, the final rule's analysis, like the NPRM's, includes estimated potential use through MY 2019, as summarized below:

Start Printed Page 24315

For its part, EPA has provided that manufacturers selling sufficient numbers of PHEVs, BEVs, and FCVs may, when calculating fleet average CO2 levels, “count” each unit of production as more than a single unit. The CAFE model accounts for these “multipliers.” As for the NPRM, the final rule's analysis applies the following multipliers:

For example, under EPA's current regulation, when calculating the average CO2 level achieved by its MY 2019 passenger car fleet, a manufacturer may treat each 1,000 BEVs as 2,000 BEVs. When calculating the average level required of this fleet, the manufacturer must use the actual production volume (in this example, 1,000 units). Similarly, the manufacturer must use the actual production volume when calculating compliance credit balances.

There were no natural gas vehicles in the baseline fleet, and the analysis did not apply natural gas technology due to cost effectiveness. The application of a 2.0 multiplier for natural gas vehicles for MYs 2022-2026 would have no impact on the analysis because given the state of natural gas vehicle refueling infrastructure, the cost to equip vehicles with natural gas tanks, the outlook for petroleum prices, and the outlook for battery prices, we have little basis to project more than an inconsequential response to this incentive in the foreseeable future.

For the final rule's analysis, the CAFE model can be exercised in a manner that simulates these current EPA requirements, or that simulates two alternative approaches. The first includes the above-mentioned multipliers in the calculation of average requirements, and the second also includes the multipliers in the calculation of credit balances. The central analysis reflects current regulations. The sensitivity analysis presented in the FRIA includes a case Start Printed Page 24316applying multipliers to the calculation of achieved and required average CO2 levels, and calculation of credit balances.

c) Civil Penalties

Throughout the history of the CAFE program, some manufacturers have consistently achieved fuel economy levels below applicable standards, electing instead to pay civil penalties as specified by EPCA. As in previous versions of the CAFE model, the current version allows the user to specify inputs identifying such manufacturers and to consider their compliance decisions as if they are willing to pay civil penalties for non-compliance with the CAFE program. As with the NPRM, the civil penalty rate in the current analysis is $5.50 per 1/10 of a mile per gallon, per vehicle manufactured for sale.

NHTSA notes that treating a manufacturer as if it is willing to pay civil penalties does not necessarily mean that it is expected to pay penalties in reality. Doing so merely implies that the manufacturer will only apply fuel economy technology up to a point, and then stop, regardless of whether or not its corporate average fuel economy is above its standard. In practice, the agencies expect that many of these manufacturers will continue to be active in the credit market, using trades with other manufacturers to transfer credits into specific fleets that are challenged in any given year, rather than paying penalties to resolve CAFE deficits. The CAFE model calculates the amount of penalties paid by each manufacturer, but it does not simulate trades between manufacturers. In practice, some (possibly most) of the total estimated penalties may be a transfer from one OEM to another.

Although EPCA, as amended in 2007 by the Energy Independence and Security Act (EISA), prescribes these specific civil penalty provisions for CAFE standards, the Clean Air Act (CAA) does not contain similar provisions. Rather, the CAA's provisions regarding noncompliance prohibit sale of a new motor vehicle that is not covered by an EPA certificate of conformity, and in order to receive such a certificate the new motor vehicle must meet EPA's Section 202 regulations, including applicable emissions standards. Therefore, inputs regarding civil penalties—including inputs regarding manufacturers' potential willingness to treat civil penalty payment as an economic choice—apply only to simulation of CAFE standards. On the other hand, some of the same manufacturers recently opting to pay civil penalties instead of complying with CAFE standards have also recently led adoption of lower-GWP refrigerants, and the “A/C leakage” credits count toward compliance only with CO2 standards, not CAFE standards. The model accounts for this difference between the programs.

When considering technology applications to improve fleet fuel economy, the model will add technology up to the point at which the effective cost of the technology (which includes technology cost, consumer fuel savings, consumer welfare changes, and the cost of penalties for non-compliance with the standard) is less costly than paying civil penalties or purchasing credits. Unlike previous versions of the model, the current implementation further acknowledges that some manufacturers experience transitions between product lines where they rely heavily on credits (either carried forward from earlier model years or acquired from other manufacturers) or simply pay penalties in one or more fleets for some number of years. The model now allows the user to specify, when appropriate for the regulatory program being simulated, on a year-by-year basis, whether each manufacturer should be considered as willing to pay penalties for non-compliance. This provides additional flexibility, particularly in the early years of the simulation. As discussed above, this assumption is best considered as a method to allow a manufacturer to under-comply with its standard in some model years—treating the civil penalty rate and payment option as a proxy for other actions it may take that are not represented in the CAFE model (e.g., purchasing credits from another manufacturer, carry-back from future model years, or negotiated settlements with NHTSA to resolve deficits).

For the NPRM, NHTSA relied on past compliance behavior and certified transactions in the credit market to designate some manufacturers as willing to pay CAFE penalties in some model years. The full set of NPRM assumptions regarding manufacturer behavior with respect to civil penalties is presented in Table VI-21, which shows all manufacturers were assumed to be willing to pay civil penalties prior to MY 2020. This was largely a reflection of either existing credit balances (which manufacturers will use to offset CAFE deficits until the credits reach their expiration dates) or inter-manufacturer trades assumed likely to happen in the near future, based on previous behavior. The manufacturers in the table whose names appear in bold all had at least one regulated fleet (of three) whose CAFE was below its standard in MY 2016. Because the NPRM analysis began with the MY 2016 fleet, and no technology could be added to vehicles that are already designed and built, all manufacturers could generate civil penalties in MY 2016. However, once a manufacturer is designated as unwilling to pay penalties, the CAFE model will attempt to add technology to the respective fleets to avoid shortfalls.

Start Printed Page 24317

Several of the manufacturers in Table VI-21 that were presumed to be willing to pay civil penalties in the early years of the program have no history of paying civil penalties. However, several of those manufacturers have either bought or sold credits—or transferred credits from one fleet to another to offset a shortfall in the underperforming fleet. As the CAFE model does not simulate credit trades between manufacturers, providing this additional flexibility in the modeling avoids the outcome where the CAFE model applies more technology than needed in the context of the full set of compliance flexibilities at the industry level. By statute, NHTSA cannot consider credit flexibilities when setting standards, so most manufacturers (those without a history of civil penalty payment) are assumed to comply with their standards through fuel economy improvements for the model years being considered in this analysis. The notable exception to this assumption is Fiat Chrysler Automobiles (FCA), which could still satisfy the requirements of the program through a combination of credit application and civil penalties through MY 2025 before eventually complying exclusively through fuel economy improvements in MY 2026.

As mentioned above, the CAA does not provide civil penalty provisions similar to those provisions specified in EPCA/EISA, and the above-mentioned corresponding inputs apply only to simulation of compliance with CAFE standards.

Some stakeholders offering comments related to the analytical treatment of civil penalties indicated that NHTSA should tend toward assuming manufacturers will take advantage of this EPCA provision as an economically attractive alternative to compliance. Other commenters implied that NHTSA should tend toward not relying on compliance flexibilities in the analysis used to determine the maximum feasible stringency of CAFE standards. For example, New York University's Institute for Policy Integrity (IPI) offered the following comments:

NHTSA assumes that most manufacturers will be unwilling to pay penalties based in part on the fact that most manufacturers have not paid penalties in recent years. The Proposed Rule cites the statutory prohibition on NHTSA considering credit trading as a reason to assume manufacturers without a history of paying penalties will comply through technology alone, whatever the cost. But this is an arbitrary assumption and is in no way dictated by the statute. NHTSA knows as much, since elsewhere in the proposed rollback, the agency explains “EPCA is very clear as to which flexibilities are not to be considered” and NHTSA is allowed to consider off-cycle adjustments because they are not specifically mentioned. But considering penalties are not mentioned as off-limits for NHTSA in setting the standards either. Instead, the prohibition focuses on credit trading and transferring. The penalty safety valve has existed in EPCA for decades, and Congress clearly would have known how to add penalties to the list of trading and transferring. The fact that Congress did not bar NHTSA from considering penalties as a safety valve means that NHTSA must consider manufacturer's efficient use of penalties as a cost minimizing compliance option. Besides, NHTSA does consider penalties for some of the manufacturers making its statutory justification even less rational.[419]

On the other hand, in more general comments about NHTSA's analytical treatment of program flexibilities, FCA stated that “when flexibilities are considered while setting targets, they cease to be flexibilities and become simply additional technology mandates.” [420]

NHTSA agrees with IPI that EPCA does not expressly prohibit NHTSA, when conducting analysis supporting determinations of the maximum feasible stringency of future CAFE standards, from including manufacturers' potential tendency to pay civil penalties rather than complying with those standards. However, EPCA also does not require NHTSA to include this tendency in its analysis. NHTSA also notes, as does IPI, that EPCA does prohibit NHTSA from including credit trading, transferring, or the availability of credits in such Start Printed Page 24318analysis (although NHTSA interprets this prohibition to apply only to the model years for which standards are being set). This statutory difference is logical based on the way credits and penalties function differently under EPCA. Because credits help manufacturers achieve compliance with CAFE standards, absent the statutory prohibition, credits would be relevant to the feasibility of a standard.[421] Penalties, on the other hand, do not enable a manufacturer to comply with an applicable standard; penalties are for noncompliance.[422] When Congress added credit trading provisions to EPCA in 2007, NHTSA anticipated that competitive considerations would make manufacturers reluctant to engage in such trades. Since that time, manufacturers actually have demonstrated otherwise, although the reliance on trading—especially between specific pairs of OEMs—appears to vary widely. At this time, NHTSA considers it most likely that manufacturers will shift away from paying civil penalties and toward compliance credit trading. Consequently, for NHTSA to include civil penalty payment in its analysis would increasingly amount to using civil penalty payment as an analytical proxy for credit trading. Having further considered the question, NHTSA's current view is, therefore, that including civil penalty payment beyond MY 2020 would effectively subvert EPCA's prohibition against considering credit trading. Therefore, for today's announcement, NHTSA has modified its analysis to assume that BMW, Daimler, FCA, JLR, and Volvo would consider paying civil penalties through MY 2020, and that all manufacturers would apply as much technology as would be needed in order to avoid paying civil penalties after MY 2020.

3. Technology Effectiveness Values

The next input required to simulate manufacturers' decision-making processes for the year-by-year application of technologies to specific vehicles is estimates of how effective each technology would be at reducing fuel consumption. In the NPRM, the agencies used full-vehicle modeling and simulation to estimate the fuel economy improvements manufacturers could make to a fleet of vehicles, considering those vehicles' technical specifications and how combinations of technologies interact. Full-vehicle modeling and simulation uses computer software and physics-based models to predict how combinations of technologies perform as a full system under defined conditions.

A model is a mathematical representation of a system, and simulation is the behavior of that mathematical representation over time. In this analysis, the model is a mathematical representation of an entire vehicle,[423] including its individual components such as the engine and transmission, overall vehicle characteristics such as mass and aerodynamic drag, and the environmental conditions, such as ambient temperature and barometric pressure. The agencies simulated the model's behavior over test cycles, including the 2-cycle laboratory compliance tests (or 2-cycle tests),[424] to determine how the individual components interact. 2-cycle tests are test cycles that are used to measure fuel economy and emissions for CAFE and CO2 compliance, and therefore are the relevant test cycles for determining technology effectiveness when establishing standards. In the laboratory, 2-cycle testing involves sophisticated test and measurement equipment, carefully controlled environmental conditions, and precise procedures to provide the most repeatable results possible with human drivers. Measurements using these structured procedures serve as a yardstick for fuel economy and CO2 emissions.

Full-vehicle modeling and simulation was initially developed to avoid the costs of designing and testing prototype parts for every new type of technology. For example, if a truck manufacturer has a concept for a lightweight tailgate and wants to determine the fuel economy impact for the weight reduction, the manufacturer can use physics-based computer modeling to estimate the impact. The vehicle, modeled with the proposed change, can be simulated on a defined test route and under a defined test condition, such as city or highway driving in warm ambient temperature conditions, and compared against the baseline reference vehicle. Full-vehicle modeling and simulation allows the consideration and evaluation of different designs and concepts before building a single prototype. In addition, full vehicle modeling and simulation is beneficial when considering technologies that provide small incremental improvements. These improvements are difficult to measure in laboratory tests due to variations in how vehicles are driven over the test cycle by human drivers, variations in emissions measurement equipment, and variations in environmental conditions.[425]

Full-vehicle modeling and simulation requires detailed data describing the individual technologies and performance-related characteristics. Those specifications generally come from design specifications, laboratory measurements, and other subsystem simulations or modeling. One example of data used as an input to the full vehicle simulation are engine maps for each engine technology that define how much fuel is consumed by the engine technology across its operating range.

Using full-vehicle modeling and simulation to estimate technology efficiency improvements has two primary advantages over using single or limited point estimates. An analysis using single or limited point estimates may assume that, for example, one fuel economy improving technology with an effectiveness value of 5 percent by itself and another technology with an effectiveness value of 10 percent by itself, when applied together achieve an additive improvement of 15 percent. Single point estimates generally do not provide accurate effectiveness values because they do not capture complex relationships among technologies. Technology effectiveness often differs significantly depending on the vehicle type (e.g., sedan versus pickup truck) and how the technology interacts with other technologies on the vehicle, as different technologies may provide different incremental levels of fuel economy improvement if implemented alone or in tandem with other technologies. Any oversimplification of these complex interactions leads to less accurate and often overestimated effectiveness estimates.

In addition, because manufacturers often implement several fuel-saving Start Printed Page 24319technologies simultaneously when redesigning a vehicle, it is difficult to isolate the effect of individual technologies using laboratory measurement of production vehicles alone. Modeling and simulation offers the opportunity to isolate the effects of individual technologies by using a single or small number of baseline vehicle configurations and incrementally adding technologies to those baseline configurations. This provides a consistent reference point for the incremental effectiveness estimates for each technology and for combinations of technologies for each vehicle type. Vehicle modeling also reduces the potential for overcounting or undercounting technology effectiveness.

An important feature of this analysis is that the incremental effectiveness of each technology and combinations of technologies be accurate and relative to a consistent baseline vehicle. The absolute fuel economy values of the full vehicle simulations are used only to determine incremental effectiveness and are never used directly to assign an absolute fuel economy value to any vehicle model or configuration for the rulemaking analysis.

For this analysis, absolute fuel economy levels are based on the individual fuel economy values from CAFE compliance data for each vehicle in the baseline fleet. The incremental effectiveness from the full vehicle simulations performed in Autonomie, a physics-based full-vehicle modeling and simulation software developed and maintained by the U.S. Department of Energy's Argonne National Laboratory, are applied to baseline fuel economy to determine the absolute fuel economy of applying the first technology change. For subsequent technology changes, incremental effectiveness is applied to the absolute fuel economy level of the previous technology configuration.

For example, if a Ford F150 2-wheel drive crew cab and short bed in the baseline fleet has a fuel economy value of 30 mpg for CAFE compliance, 30 mpg will be considered the reference absolute fuel economy value. A similar full vehicle model in the Autonomie simulation may begin with an average fuel economy value of 32 mpg, and with incremental addition of a specific technology X its fuel economy improves to 35 mpg, a 9.3 percent improvement. In this example, the incremental fuel economy improvement (9.3 percent) from technology X would be applied to the F150's 30 mpg absolute value.

For this analysis, the agencies determined the incremental effectiveness of technologies as applied to the 2,952 unique vehicle models in the analysis fleet. Although, as mentioned above, full-vehicle modeling and simulation reduces the work and time required to assess the impact of moving a vehicle from one technology state to another, it would be impractical—if not impossible—to build a unique vehicle model for every individual vehicle in the analysis fleet. Therefore, as explained further below, vehicle models are built in a way that maintains similar attributes to the analysis fleet vehicles, which ensures key components are reasonably represented.

We received a wide array of comments regarding the full-vehicle modeling and simulation performed for the NPRM, but there was general agreement that full-vehicle modeling and simulation was the appropriate method to determine technology effectiveness.[426] Stakeholders commented on other areas, such as full vehicle simulation tools, inputs, and assumptions, and these comments will be discussed in the following sections. For this final rule, the agencies continued to use the same full-vehicle simulation approach to estimate technology effectiveness for technology adoption in the rulemaking timeframe. The next sections will discuss the details of the explicit input specifications and assumptions used for the final rule analysis.

a) Why This Rulemaking Used Autonomie Full-Vehicle Modeling and Simulation To Determine Technology Effectiveness

The NPRM and final rule analysis use effectiveness estimates for technologies developed using Autonomie, a physics-based full-vehicle modeling and simulation software developed and maintained by the U.S. Department of Energy's Argonne National Laboratory.[427] Autonomie was designed to serve as a single tool to meet requirements of automotive engineering throughout the vehicle development process, and has been under continuous improvement by Argonne for over 20 years. Autonomie is commercially available and widely used in the automotive industry by suppliers, automakers, and academic researchers (who publish findings in peer reviewed academic journals).[428] DOE and manufacturers have used Autonomie and its ability to simulate a large number of powertrain configurations, component technologies, and vehicle-level controls over numerous drive cycles to support studies on fuel efficiency, cost-benefit analysis, and carbon dioxide emissions,[429] and other topics.

Autonomie has also been used to provide the U.S. government with data to make decisions about future research, and is used by DOE for analysis supporting budget priorities and plans for programs managed by its Vehicle Technologies Office (VTO), and to support decision making among competing vehicle technology research and development projects.[430] In addition, Autonomie is the primary vehicle simulation tool used by DOE to support its U.S. DRIVE program, a government-industry partnership focused on advanced automotive and related energy infrastructure technology research and development.[431]

Autonomie is a MathWorks-based software environment and framework for automotive control-system design, simulation, and analysis.[432] It is designed for rapid and easy integration of models with varying levels of detail (low to high fidelity), abstraction (from subsystems to systems and entire architectures), and processes (e.g., calibration, validation). By building models automatically, Autonomie allows the quick simulation of many component technologies and powertrain configurations, and, in this case, to assess the energy consumption of advanced powertrain technologies. Autonomie simulates subsystems, Start Printed Page 24320systems, or entire vehicles; evaluates and analyzes fuel efficiency and performance; performs analyses and tests for virtual calibration, verification, and validation of hardware models and algorithms; supports system hardware and software requirements; links to optimization algorithms; and supplies libraries of models for propulsion architectures of conventional powertrains as well as hybrid and electric vehicles.

With hundreds of pre-defined powertrain configurations along with vehicle level control strategies developed from dynamometer test data, Autonomie is a highly capable tool for analyzing advantages and drawbacks of applying different technology options within each technology family, including conventional, parallel hybrid, power-split hybrid electric vehicles (HEVs), plug-in hybrid electric vehicles (PHEVs), battery electric vehicles (BEV) and fuel cell vehicles (FCVs). Autonomie also allows users to evaluate the effect of component sizing on fuel consumption for different powertrain technologies as well as to define component requirements (e.g., power, energy) to maximize fuel displacement for a specific application.[433] To evaluate properly any powertrain-configuration or component-sizing influence, vehicle-level control models are critical, especially for electric drive vehicles like hybrids and plug-in hybrids. Argonne has extensive expertise in developing vehicle-level control models based on different approaches, from global optimization to instantaneous optimization, rule-based optimization, and heuristic optimization.[434]

Autonomie has been developed to consider real-world vehicle metrics like performance, hardware limitations, utility, and drivability metrics (e.g., towing capability, shift busyness, frequency of engine on/off transitions), which are important to producing realistic estimates of fuel economy and CO2 emission rates. This increasing realism has, in turn, steadily increased confidence in the appropriateness of using Autonomie to make significant investment decisions. Autonomie has also been validated for a number of powertrain configurations and vehicle classes using Argonne's Advanced Mobility Technology Laboratory (AMTL) (formerly Advanced Powertrain Research Facility, or APRF) vehicle test data.[435]

Argonne has spent several years developing, applying, and expanding the means to use distributed computing to exercise its Autonomie full-vehicle simulation tool over the scale necessary for realistic analysis to provide data for CAFE and CO2 standards rulemaking. The NPRM and PRIA detailed how Argonne used Autonomie to estimate the fuel economy impacts for roughly a million combinations of technologies and vehicle types.[436 437] Argonne developed input parameters for Autonomie to represent every combination of vehicle, powertrain, and component technologies considered in this rulemaking. The sequential addition of more than 50 fuel economy-improving technologies to ten vehicle types generated more than 140,000 unique technology and vehicle combinations. Running the Autonomie powertrain sizing algorithms to determine the appropriate amount of engine downsizing needed to maintain overall vehicle performance when vehicle mass reduction is applied and for certain engine technology changes (discussed further, below) increased the total number of simulations to more than one million. The result of these simulations is a useful dataset identifying the impacts of combinations of vehicle technologies on energy consumption—a dataset that can be referenced as an input to the CAFE model for assessing regulatory compliance alternatives.

The following sections discuss the full-vehicle modeling and simulation inputs and data assumptions, and comments received on the NPRM analysis. The discussion is necessarily technical, but also important to understand the agencies' decisions to modify (or not) the Autonomie analysis for the final rule.

(1) Full-Vehicle Modeling, Simulation Inputs and Data Assumptions

The agencies provided extensive documentation that quantitatively and qualitatively described the over 50 technologies considered as inputs to the Autonomie modeling.[438 439] These inputs consisted of engine technologies, transmission technologies, powertrain electrification, light-weighting, aerodynamic improvements, and tire rolling resistance improvements.[440] The PRIA provided an overview of the sub-models for each technology, including the internal combustion engine model, automatic transmission model, and others.[441] The Argonne NPRM model documentation expanded on these sub-models in detail to show the interaction of each sub-model input and output.[442] Start Printed Page 24321For example, as shown in Figure VI-2, the input for Autonomie's driver model (i.e., the model used to approximate the driving behavior of a real driver) is vehicle speed, and outputs are accelerator pedal, brake pedal, and torque demand.

Effectiveness inputs for the NPRM and the final rule analysis were specifically developed to consider many real world and compliance test cycle constraints, to the extent a computer model could capture them. Examples include the advanced engine knock model discussed below, in addition to other constraints like allowing cylinder deactivation to occur in ways that would not negatively impact noise-vibration-harshness (NVH), and similarly optimizing the number of engine on/off events (e.g., from start/stop 12V micro hybrid systems) to balance between effectiveness and NVH.

One major input used in the effectiveness modeling that the agencies provided key specifications for in the PRIA are engine fuel maps that define how an engine equipped with specific technologies operates over a variety of engine load (torque) and engine speed conditions. The engine maps used as inputs to the Autonomie modeling portion of the analysis were developed by starting with a base map and then modifying that base map, incrementally, to model the addition of engine technologies. These engine maps, developed using the GT-Power modeling tool by IAV, were based off real-world engine designs. Simulated operation of these engines included the application of an IAV knock model, also developed from real-world engine data.[443 444] Using this process, which incorporated real-world data, ensured that real-world constraints were considered for each vehicle type. Although the same type of engine map is used for all technology classes, the effectiveness varies based on the characteristics of each vehicle type. For example, a compact car with a turbocharged engine will have different fuel economy and performance values than a pickup truck with the same engine technology type. The engine map specifications are discussed further in Section VI.C.1 of this preamble and Section VI of FRIA.

The agencies also provided key details about input assumptions for various vehicle specifications like transmission gear ratios, tire size, final drive ratios, and individual component weights.[445] Each of these assumptions, to some extent, varied between the ten technology classes to capture appropriately real-world vehicle specifications like wheel mass or fuel tank mass. These specific input assumptions were developed based on the latest test data and current market fleet information.[446] The agencies relied on default assumptions developed by the Autonomie team, based on test data and technical publication review, for other model inputs required by Autonomie, such as throttle time response and shifting strategies for different transmission technologies. The Autonomie modeling tool did not simulate vehicle attributes determined to have minimal impacts, like whether a vehicle had a sun roof or hood scoops, as those attributes would have trivial impact in the overall analysis.

Because the agencies model ten different vehicle types to represent the 2,952 vehicles in the baseline fleet, improper assumptions about an advanced technology could lead to errors in estimating effectiveness. Autonomie is a sophisticated full-vehicle modeling tool that requires extensive technology characteristics based on both physical and intangible data, like proprietary software. With a few technologies, the agencies did not have publicly available data, but had received confidential business information confirming such technologies potential availability in the market during the rulemaking time frame. For such technologies, including advanced cylinder deactivation, the agencies adopted a method in the CAFE model to represent the effectiveness of the technology, and did not explicitly simulate the technologies in the Autonomie model. For this limited set of technologies, the agencies determined that effectiveness could reasonably be represented as a fixed value.[447] Effectiveness values for technologies not explicitly simulated in Autonomie are discussed further in the individual technology sections of this preamble.

The agencies sought comments on all effectiveness inputs and input assumptions, including the specific data used to characterize the technologies, Start Printed Page 24322such as data to build the technology input, data representing operating range of technologies, and data for variation among technology inputs. The agencies also sought comment on the effectiveness values used for technologies not explicitly defined in Autonomie.

Meszler Engineering Services, commenting on behalf of the Natural Resources Defense Council, and ICCT questioned the accuracy of the effectiveness estimates in the Argonne database, and as an example Meszler analyzed the fuel economy impacts of a 10-speed automatic transmission relative to a baseline 8-speed automatic transmission, concluding that the widely ranging effectiveness estimates were unexpected. ICCT questioned the accuracy of the IAV engine maps that serve as an input to the Autonomie effectiveness modeling, and asked whether those could “reasonably stand as a foundation for automotive developments and technology combinations” discussed elsewhere in their comments. ICCT also questioned whether Autonomie realistically and validly modeled synergies between technologies, using the effectiveness values from CEGR and transmissions as an example. Meszler stated that the agencies have an obligation to validate the Autonomie estimates before using them to support the NPRM or any other rulemaking. The agencies also received comments on the specific effectiveness estimates generated by Autonomie; however, those comments will be discussed in each individual technology section, below.

Despite these criticisms, Meszler stated that the critiques of the Autonomie technology database were not meant to imply that the Autonomie vehicle simulation model used to develop the database was fundamentally flawed, or that the model could not be used to derive accurate fuel economy impact estimates. Meszler noted that, as with any model, estimates derived with Autonomie are only valid for a given set of modeling parameters and if those parameters are well defined, the estimates should be accurate and reliable. Conversely, if those parameters are not well defined, the estimates would be inaccurate and unreliable. Meszler stated that the agencies must make the full set of modeling assumptions used for the Autonomie database available for review and comment.

We agree with Meszler that, in general, when inputs to a model are inaccurate, output effectiveness results may be too high or too low. The technology effectiveness estimates from modeling results often vary with the type of vehicle and the other technologies that are on that vehicle.[448] The Autonomie output database consists of permutations of over 50 technologies for each of the ten technology classes simulated by the CAFE model. A wide range of effectiveness is expected when going from a baseline technology to an advanced technology across different technology classes because there are significant differences in how much power is required from the powertrain during 2-cycle testing across the ten vehicle types. This impacts powertrain operating conditions (e.g., engine speed and load) during 2-cycle testing. Fuel economy improving technologies have different effectiveness at each of those operating conditions so vehicles that have higher average power demands will have different effectiveness than vehicles with lower average power demands. Further, the differences in effectiveness at higher power and lower power vary by technology so the overall relationship is complex. Large-scale full-vehicle modeling and simulation account for these interactions and complexities.

Before conducting any full-vehicle modeling and simulation, the agencies spent a considerable amount of time and effort developing the specific inputs used for the Autonomie analysis. The agencies believe that these technology inputs provide reasonable estimates for the light-duty vehicle technologies the agencies expect to be available in the market in the rulemaking timeframe. As discussed earlier, these inputs vary in effectiveness due to how different vehicles, like compact cars and pickup trucks, operate on the 2-cycle test and in the real world. Some technologies, such as 10-speed automatic transmissions (AT10) relative to 8-speed automatic transmissions (AT8), can and should have different effectiveness results in the analysis between two different technology classes.[449] These unique synergistic effects can only be taken into account through conducting full-vehicle modeling and simulation, which the agencies did here.

With regards to Meszler's comment that the agencies have an obligation to validate the Autonomie estimates before using them to support the NPRM or any other rulemaking, the agencies would like to point Meszler to the description of the Argonne Autonomie team's robust process for vehicle model validation that was contained in the PRIA.[450] To summarize, the NPRM and final rule analysis leveraged extensive vehicle test data collected by Argonne National Laboratory.[451] Over the past 20 years, the Argonne team has developed specific instrumentation lists and test procedures for collecting sufficient information to develop and validate full vehicle models. In addition, the agencies described the Argonne team's efforts to validate specific component models as well, such as the advanced automatic transmission and dual clutch transmission models.[452]

The agencies also described the process for validating inputs used to develop the IAV engine maps,[453 454] another input to the Autonomie simulations. As discussed in the PRIA, IAV's engine model development relied on a collection of sub-models that controlled independent combustion characteristics such as heat release, combustion knock, friction, heat flow, and other combustion optimization tools. These sub-models and other Start Printed Page 24323computational fluid dynamics models were utilized to convert test data for use in the IAV engine map development. Specific combustion parameters, like from test data for the coefficient of variation for the indicated mean effective pressure (COV of IMEP), which is a common variable for combustion stability in a spark ignited engine, was used to assure final engine models were reasonable. The assumptions and inputs used in the modeling and validation of engine model results leveraged IAV's global engine database, which included benchmarking data, engine test data, single cylinder test data and prior modeling studies, and also technical publications and information presented at conferences. The agencies referenced in the PRIA that engine maps were validated with engine dynamometer test data to the maximum extent possible.[455] Because the NPRM and the final rule analysis considered some technologies not yet in production, the agencies relied on technical publications and engine modeling by IAV to develop and corroborate inputs and input assumptions where engine dynamometer test data was not available.

In addition, as described earlier in this section, the full set of NPRM modeling assumptions used for the Autonomie database were available for review and comment in the docket for this rulemaking.[456] The full set of modeling assumptions used for the final rule are also available in the docket.[457]

Both ICCT and Meszler also commented on the availability of technologies within the Autonomie database, with Meszler stating that with limited exceptions, technologies were not included in the NPRM CAFE model if they were not included in the simulation modeling that underlay the Argonne database, and accordingly if a combination of technologies was not modeled during the development of the Argonne database, that package (or combination) of technologies was not available for adoption in the CAFE model. Meszler stated that these constraints limited the slate of technologies available to respond to fuel economy standards, and independently expanding the model to include additional technologies or technology combinations is not trivial.

ICCT gave specific examples of key efficiency technologies that it stated Autonomie did not include, like advanced DEAC, VCR, Miller Cycle, e-boost, and HCCI. ICCT argued that this was especially problematic as the agencies appeared to have available engine maps from IAV on advanced DEAC, VCR, Miller Cycle, E-boost (and from advanced DEAC, VCR, Miller Cycle, E-boost, HCCI from EPA) that Argonne or the agencies have been unable to or opted not to include in their modeling. ICCT stated that the agencies must disclose how Autonomie had been updated to incorporate “cutting edge” 2020-2025 automotive technologies to ensure they reflect available improvements.[458]

The agencies have updated the final rule analysis to include additional technologies. In the NPRM, the agencies presented the engine maps for all of the technologies that ICCT listed, except HCCI, and sought comment on the engine maps, technical assumptions and the potential use of the technologies for the final rule analysis. Based on the available technical information and the ICCT and Meszler comments, for the final rule analysis, VCR, Miller Cycle (VTG), and e-boost (VTGe with 48V BISG) technologies have been added and included in the Autonomie modeling and simulations, and advanced DEAC technology has been added using fixed point effectiveness estimates in the CAFE model analysis. The agencies disagree with ICCT's assessment of HCCI and do not believe it will be available for wide-scale application in the rulemaking timeframe, and therefore have not included it as a technology. HCCI technology has been in the research phase for several decades, and the only production applications to date use a highly-limited version that restricts HCCI combustion to a very narrow range of engine operating conditions.[459 460 461] Additional discussion of how Autonomie-modeled and non-modeled technologies are incorporated into the CAFE Model is located in Section VI.B.3.c), below.

ICCT and Meszler also commented that the agencies overly limited the availability of several technologies in the NPRM analysis. In response, the agencies reconsidered the restrictions that were applied in the NPRM analysis, and agree with the commenters for several technologies and technology classes. Many technologies identified by the commenters are now in production for the MY2017 as well as MY2018 and MY2019. The agencies also think that the baseline fleet compliance data reflects adoption of many of these technologies. For the final rule analysis, the agencies have expanded the availability of several technologies. In the CAFE model, the agencies are now allowing parallel hybrids (SHEVP2) to be adopted with high compression Atkinson mode engines (HCR0 and HCR1). In addition, as mentioned above, the Autonomie full-vehicle modeling included Variable Compression Ratio engine (VCR), Miller Cycle Engine (VTG), E-boost (VTGe) technologies, and cylinder deactivation technologies (DEAC) to be applied to turbocharged engines (TURBO1). As these changes relate to the technology effectiveness modeling, the CAFE model analysis now includes effectiveness estimates based on full vehicle simulations for all of these technology combinations.

We disagree with comments stating the agencies should allow every technology to be available to every vehicle class.[462] Discussed earlier in this section, Autonomie models key aspects of vehicle operation that are most relevant to assessing fuel economy, vehicle performance and certain aspects of drivability (like EPA 2-cycle tests, EPA US06 cycle tests, gradability, low speed acceleration time from 0-to-60 mph, passing acceleration time from 50 to 80 mph, and number of transmission shifts). However, there are other critical aspects of vehicle functionality and operation that the agencies considered beyond those criteria, that cannot necessarily be reflected in the Autonomie modeling. For example, a pickup truck can be modeled with a Start Printed Page 24324continuously variable transmission (CVT) and show improvements on the 2-cycle tests. However, pickup trucks are designed to provide high load towing utility.[463] CVTs lack the torque levels needed to provide that towing utility, and would fail mechanically if subject to high load towing.[464] The agencies provided discussions of some of these technical considerations in the PRIA, and explained why the agencies had limited technologies for certain vehicle classes, such as limiting CVTs on pickups as in the example above. These and other limitations are discussed further in the individual technology sections.

The agencies also received a variety of comments that conflated aspects of the Autonomie models with technology inputs and input assumptions. For example, commenters expressed concern about the transmission gear set and final drive values used for the NPRM analysis, or more specifically, that the gear ratios were held constant across applications.[465] In this case, both the inputs (gear set and final drive ratio) and input assumption (ratios held constant) were discussed by the commenters. Because these comments are actually about technology inputs to the Autonomie model, for these and similar cases, the agencies are addressing the comments in the individual technology sections which discuss the technology inputs and input assumptions that impact the effectiveness values for those technologies.

For the NPRM analysis, the agencies prioritized using inputs that were based on data for identifiable technology configurations and that reflected practical real world constraints. The agencies provided detailed information on the NPRM analysis inputs and input assumptions in the NPRM Preamble, PRIA and Argonne model documentation for engine technologies, transmission technologies, powertrain electrification, light-weighting, aerodynamic improvements, tire rolling resistance improvements, and other vehicle technologies. Comments and the agencies' assessment of comments for each technology are discussed in the individual technology sections below. Through careful consideration of the comments, the agencies have updated analytical inputs associated with several technologies, and as discussed above, have included several advanced technologies for which technical information was included in the NPRM. However, for most technologies, the agencies have determined that the technology inputs and input assumptions that were used in the NPRM analysis remain reasonable and the best available for the final rule analysis.

(2) How The Agencies Defined Different Vehicle Types in Autonomie

As described in the NPRM, Argonne produced full-vehicle models and ran simulations for many combinations of technologies, on many types of vehicles, but it did not simulate literally every single vehicle model/configuration in the analysis fleet because it would be impractical to assemble the requisite detailed information—much of which would likely only be provided on a confidential basis—specific to each vehicle model/configuration and because the scale of the simulation effort would correspondingly increase by orders of magnitude. Instead, Argonne simulated 10 different vehicle types, corresponding to the five “technology classes” generally used in CAFE analysis over the past several rulemakings, each with two performance levels and corresponding vehicle technical specifications (e.g., small car, small performance car, pickup truck, performance pickup truck, etc.).

Technology classes are a means of specifying common technology input assumptions for vehicles that share similar characteristics. Because each vehicle technology class has unique characteristics, the effectiveness of technologies and combinations of technologies is different for each technology class. Conducting Autonomie simulations uniquely for each technology class provides a specific set of simulations and effectiveness data for each technology class. Like the Draft TAR analysis, there are separate technology classes for compact cars, midsize cars, small SUVs, large SUVs, and pickup trucks. However, new for the NPRM analysis and carried into this final rule analysis, each of those vehicle types has been split into “low” (or “standard”) performance and a “high” performance versions, which represent two classes with similar body styles but different levels of performance attributes (for a total of 10 technology classes). The separate technology classes for high performance and low performance vehicles better account for performance diversity across the fleet.

NHTSA directed Argonne to develop a vehicle assumptions database to capture vehicle attributes that would comprise the full vehicle models. For each vehicle technology class, representative vehicle attributes and characteristics were identified from publicly available information and automotive benchmarking databases like A2Mac1,[466] Argonne's Downloadable Dynamometer Database (D[3] ),[467] and EPA compliance and fuel economy data,[468] EPA's guidance on the cold start penalty on 2-cycle tests.[469] The resulting vehicle assumptions database consists of over 100 different attributes like vehicle frontal area, drag coefficient, fuel tank weight, transmission housing weight, transmission clutch weight, hybrid vehicle component weights, and weights for components that comprise engines and electric machines, tire rolling resistance, transmission gear ratios and final drive ratio. Each of the 10 different vehicle types was assigned a set of these baseline attributes and characteristics, to which combinations of fuel-saving technologies were added as inputs for the Autonomie simulations. For example, the characteristics of the MY 2016 Honda Fit were considered along with a wide range of other compact cars to identify representative characteristics for the Autonomie simulations for the base compact car technology class. The simulations determined the fuel economy achieved when applying each combination of technologies to that vehicle type, given its baseline characteristics.

For each vehicle technology class and for each vehicle attribute, Argonne estimated the attribute value using statistical distribution analysis of publicly available data and data obtained from the A2Mac1 benchmarking database.[470] Some Start Printed Page 24325vehicle attributes were also based on test data and vehicle benchmarking, like the cold-start penalty for the FTP test cycle and vehicle electrical accessories load. The analysis of vehicle attributes used in the NPRM was discussed in the Argonne model documentation,[471] and values for each vehicle technology class were provided with the NPRM for public review.[472]

The agencies did not believe it was appropriate to assign one single engine mass for each vehicle technology class in the NPRM analysis. To account for the difference in weight for different engine types, Argonne performed a regression analysis of engine peak power versus weight, based on attribute data taken from the A2Mac1 benchmarking database. For example, to account for weight of different engine sizes like 4-cylinder versus 8-cylinder, Argonne developed a relationship curve between peak power and engine weight based on the A2Mac1 benchmarking data. For the NPRM analysis, this relationship was used to estimate mass for all engine types regardless of technology type (e.g., variable valve lift and direct injection). Secondary weight reduction associated with changes in engine technology was applied by using this linear relationship between engine power and engine weight from the A2Mac1 benchmarking database. When a vehicle in the analysis fleet with an 8-cylinder engine adopted a more fuel efficient 6-cylinder engine, the total vehicle weight would reflect the updated engine weight with two less cylinders based on the peak power versus engine weight relationship. The impact of engine mass reduction on effectiveness is accounted for directly in the Autonomie simulation data through the application of the above relationship. Engine mass reduction through downsizing is, therefore, appropriately not included as part of vehicle mass reduction technology that is discussed in Section VI.C.4 because doing so would result in double counting the impacts. As discussed further below, for the final rule the agencies improved upon the precision of engine weights by creating two curves to separately represent naturally aspirated engine designs and turbocharged engine designs.

In addition, certain attributes were held at constant levels within each technology class to maintain vehicle functionality, performance and utility including noise, vibration, and harshness (NVH), safety, performance and other utilities important for customer satisfaction. For example, in addition to the vehicle performance constraints discussed in Section VI.B.3.a)(6), the analysis does not allow the frontal area of the vehicle to change, in order to maintain utility like ground clearance, head-room space, and cargo space, and a cold-start penalty is used to account for fuel economy degradation for heater performance and emissions system catalyst light-off.[473] This allows us to capture the discrete improvement in technology effectiveness while maintaining vehicle attributes that are important vehicle utility, consumer acceptance and compliance with criteria emission standards, and considering these constraints similar to how manufacturers do in the real world.

The agencies sought comment on the analytical approach used to determine vehicle attributes and characteristics for the Autonomie modeling. In response, the agencies received a wide variety of comments on vehicle attributes ranging from discussions of performance increase from technology adoption (e.g., if a vehicle adopting an electrified powertrain improved its time to accelerate from 0-60 mph), to comments on vehicle attributes not modeled in Autonomie, like heated seats and cargo space.

Toyota and the Alliance commented that the inclusion of performance vehicle classes addressed the market reality that some consumers will purchase vehicles for their performance attributes and will accept the corresponding reduction in fuel economy. Furthermore, Toyota commented that some gain in performance is more realistic, and that “dedicating all powertrain improvements to fuel efficiency is inconsistent with market reality.” Toyota “supports the agencies' inclusion of performance classes in compliance modeling where a subset of certain models is defined to have higher performance and a commensurate reduction in fuel efficiency.” [474] Also, in support of the addition of performance vehicle classes, the Alliance commented that “vehicle categories have been increased to 10 to better recognize the range of 0-60 performance characteristics within each of the 5 previous categories, in recognition of the fact that many vehicles in the baseline fleet significantly exceeded the previously assumed 0-60 performance metrics. This provides better resolution of the baseline fleet and more accurate estimates of the benefits of technology.” [475]

UCS commented that the CAFE model incorporates technology improvements to each vehicle by applying the effectiveness improvement of the average vehicle in the technology class, leading to discrete “stepped” effectiveness levels for technologies across the different vehicle types. UCS stated that in contrast, the OMEGA model takes into account a vehicle's performance characteristics through response-surface modeling based on relative deviation from the class average modeled in ALPHA.[476]

Although differences between the ALPHA and Autonomie models are discussed in more detail below, for the NPRM vehicle simulation analysis the agencies expanded the number of vehicle classes from the five classes used in the Draft TAR to ten classes, to represent better the diversity of vehicle characteristics across the fleet. Each of these ten vehicle technology classes are empirically built from benchmarking data and other information from various sources, amounting to hundreds of vehicle characteristics data points to develop each vehicle class. The agencies expand on these vehicle classes and characteristics in Section VI.B.3.(a)(2) Vehicle Types in Autonomie and Section VI.B.3.(a)(3) How Vehicle Models are Built in Autonomie and Optimized for Simulation. The agencies believe that the real-world data used to define vehicle characteristics for each of the ten vehicle classes, in addition to the ten vehicle technology classes themselves, ensures the analysis reasonably accounts for the diversity in vehicle characteristics across the fleet.

The agencies believe that UCS's characterization of how technology improvements are applied in the analysis is a misleading oversimplification. While the analysis approach in the final rule uses a representative effectiveness value, the value is not linked solely to the vehicle technology class, as the UCS implies. The entire technology combination, or technology key, which includes the vehicle technology class, is used to Start Printed Page 24326determine the value for the platform being considered. Within each vehicle class, the interactions between the added technology and the full vehicle system (including other technologies and substantial road load characteristics) are considered in the effectiveness values calculated for each technology during compliance modeling. As discussed under each of the technology pathways sections, the effectiveness for most technologies is reported as a range rather than a single value. The range exists because the effectiveness for each technology is adjusted based on the technologies it is coupled with and the major road load characteristics of the full vehicle system. This approach, in combination with using the baseline vehicle's initial performance values as a starting point for performance improvement, results in a widely variable level of improvement for the system, dependent on individual vehicle platform characteristics. As a result, the application of a response-surface approach would likely result in minimal improvement in accuracy for the Autonomie and CAFE model analysis approach.

For the final rule analysis, the agencies used the same process to obtain the vehicle attributes and characteristics for the vehicle technology classes. Data was acquired from publicly available sources, Argonne D3, EPA compliance and fuel economy data, and A2mac1 benchmarking data. Accordingly, the attributes and characteristics of the modeled vehicles reflect actual vehicles that meet customer expectations and automakers' capabilities to manufacture the vehicles. In addition, for the final rule, the agencies improved the NPRM analysis by updating some of the attribute values to account for changes in the fleet. For example, the agencies have updated vehicle electrical accessory load on the test cycle to reflect higher electrical loads associated with contemporary vehicle features.

(3) How This Rulemaking Builds Vehicle Models for Autonomie and Optimize Them for Simulation

Before any simulation is initiated in Autonomie, Argonne must “build” a vehicle by assigning reference technologies and initial attributes to the components of the vehicle model representing each technology class.[477] The reference technologies are baseline technologies that represent the first step on each technology pathway used in the analysis. For example, a compact car is built by assigning it a baseline engine, a baseline 6-speed automatic transmission (AT6), a baseline level of aerodynamic improvement (AERO0), a baseline level of rolling resistance improvement (ROLL0), a baseline level of mass reduction technology (MR0), and corresponding attributes from the Argonne vehicle assumptions database like individual component weights.[478] A baseline vehicle will have a unique starting point for the simulation and a unique set of assigned inputs and attributes, based on its technology class.

The next step in the process is to run a powertrain sizing algorithm that ensures the built vehicle meets or exceeds defined performance metrics, including low-speed acceleration (i.e., time required to accelerate from 0-60 mph), high-speed passing acceleration (time required to accelerate from 50-80 mph), gradeability (e.g. the ability of the vehicle to maintain constant 65 miles per hour speed on a six percent upgrade), and towing capacity. Together, these performance criteria are widely used by industry as metrics to quantify vehicle performance attributes that consumers observe and that are important for vehicle utility and customer satisfaction.

In the compact car example used above, the agencies assigned an initial specific engine design and engine power, transmission, AERO, ROLL, and MR technologies, and other attributes like vehicle weight. If the built vehicle does not meet all the performance criteria in the first iteration, then the engine power is increased to meet the performance requirement. This increase in power is from higher engine displacement, which could involve an increase in number of cylinders, leading to an increase in the engine weight. The iterative process continues to check whether the compact car with updated engine power, and corresponding updated engine weight, meets its defined performance metrics. The loop stops once all the metrics are met, and at this point, a compact car technology class vehicle model becomes ready for simulation. For further discussion of the vehicle performance metrics, see Section VI.B.3.(a).

Autonomie then adopts a single fuel saving technology to the baseline vehicle model, keeping everything else the same except for that one technology and the attributes associated with it. For example, the model would apply an 8-speed automatic transmission in place of the baseline 6-speed automatic transmission, which would lead to either an increase or decrease in the total weight of the vehicle based on the technology class assumptions. At this point, Autonomie confirms whether performance metrics are met for this new vehicle model through the previously discussed sizing algorithm. Once a technology has been assigned to the vehicle model and the resulting vehicle meets its performance metrics, those vehicle models will be used as inputs to the full vehicle simulations. So, in the example of the 6-speed to 8-speed automatic transmission technology update, the agencies now have the initial ten vehicle models (one for each technology class), plus the ten new vehicle models with the updated 8-speed automatic transmission, which adds up to 20 different vehicle models for simulation. This permutation process is conducted for each of the over 50 technologies considered, and for all ten technology classes, which results in more than one million optimized vehicle models.

Figure VI-3 shows the process for building vehicles in Autonomie for simulation.

Start Printed Page 24327

Some of the technologies require extra steps for optimization before the vehicle models are built for simulation; for example, the sizing and optimization process is more complex for the electrified vehicles (i.e., HEVs, PHEVs) compared to vehicles with internal combustion engines, as discussed further, below. Throughout the vehicle building process, the following items are considered for optimization:

  • Vehicle weight is decreased or increased in response to switching from one type of technology to another for the technologies for which the agencies consider weight, such as different engine and transmission types;
  • Vehicle performance is decreased or increased in response to the addition of mass reduction technologies when switching from one vehicle model to another vehicle model for the same engine;
  • Vehicle performance is decreased or increased in response to the addition of a new technology when switching from one vehicle model to another vehicle model for the same hybrid electric machine; and
  • Electric vehicle battery size is decreased or increased in response to the addition of mass, aero and/or tire rolling resistance technologies when switching from one vehicle model to another vehicle model.

Every time a vehicle adopts a new technology, the vehicle weight is updated to reflect the new component weight. For some technologies, the direct weight change is easy to assess. For example, in the NPRM the agencies designated weights for transmissions so, when a vehicle is updated to a higher geared transmission, the weight of the original transmission is replaced with the corresponding transmission weight (e.g., the weight of a vehicle moving from a 5-speed automatic transmission to an 8-speed automatic transmission will be updated based on the 8-speed transmission weight).

For other technologies, like engine technologies, assessing the updated vehicle weight is much more complex. Discussed earlier, modeling a change in engine technology involves both the new technology adoption and a change in power (because the reduction in vehicle weight leads to lower engine loads, and a resized engine). When a new engine technology is adopted on a vehicle the agencies account for the associated weight change to the vehicle based on the earlier discussed regression analysis of weight versus power. For the NPRM engine weight regression analysis, the agencies considered 19 different engine technologies that consisted of unique components to achieve fuel economy improvements. This regression analysis is technology agnostic by taking the approach of using engine peak power versus engine weight because it removed biases to any specific engine technology in the analysis. Although the agencies do not estimate the specific weight for each individual engine technology, such as VVT and SGDI, this process provides a reasonable estimate of the weight differences among engine technologies.

Start Printed Page 24328

For the final rule analysis, the agencies used the same process to assign initial weights to the original 19 engines, plus the added engines. However, the agencies improved upon precision of the weights by creating two separate curves separately to represent naturally aspirated engine designs and turbocharged engine designs.[479] This update resulted in two benefits. First, small naturally aspirated 4-cylinder engines that adopted turbocharging technology reflected the increased weight of associated components like ducting, clamps, the turbocharger itself, a charged air cooler, wiring, fasteners, and a modified exhaust manifold. Second, larger cylinder count engines like naturally aspirated 8-cylinder and 6-cylinder engines that adopted turbocharging and downsized technologies would have lower weight due to having fewer engine cylinders. For example, a naturally aspirated 8-cylinder engine that adopts turbocharging technology when downsized to a 6-cylinder turbocharged engine appropriately reflects the added weight of turbocharging components, and the lower weight of fewer cylinders.

As with conventional vehicle models, electrified vehicle models were built from the ground up. For the NPRM analysis, Argonne used data from the A2mac1 database and vehicle test data to define different attributes like weights and power. Argonne used one electric motor specific power for each type of hybrid and electric vehicle.[480] For MY2017, the U.S. market has an expanded number of available hybrid and electric vehicle models. To capture appropriately the improvements for electrified vehicles for the final rule analysis, the agencies applied the same regression analysis process that considers electric motor weight versus electric motor power for vehicle models that have adopted electric motors. Benchmarking data for hybrid and electric vehicles from the A2Mac1 database was analyzed to develop a regression curve of electric motor peak power versus electric motor weight.[481]

(4) How Autonomie Sizes Powertrains for Full Vehicle Simulation

The agencies maintain performance neutrality of the full vehicle simulation analysis by resizing engines, electric machines, and hybrid electric vehicle battery packs at specific incremental technology steps. To address product complexity and economies of scale, engine resizing is limited to specific incremental technology changes that would typically be associated with a major vehicle or engine redesign.[482] Manufacturers have repeatedly told the agencies that the high costs for redesign and the increased manufacturing complexity that would result from resizing engines for small technology changes preclude them from doing so. It would be unreasonable and unaffordable to resize powertrains for every unique combination of technologies, and exceedingly so for every unique combination of technologies across every vehicle model due to the extreme manufacturing complexity that would be required to do so. The agencies reiterated in the NPRM that the analysis should not include engine resizing with the application of every technology or for combinations of technologies that drive small performance changes so that the analysis better reflects what is feasible for manufacturers.[483]

When a powertrain does need to be resized, Autonomie attempts to mimic manufacturers' development approaches to the extent possible. Discussed earlier, the Autonomie vehicle building process is initiated by building a baseline vehicle model with a baseline engine, transmission, and other baseline vehicle technologies. This baseline vehicle model (for each technology class) is sized to meet a specific set of Start Printed Page 24329performance criteria, including acceleration and gradeability.

The modeling also accounts for the industry practice of platform, engine, and transmission sharing to manage component complexity and the associated costs.[484] At a vehicle refresh cycle, a vehicle may inherit an already resized powertrain from another vehicle within the same engine-sharing platform that adopted the powertrain in an earlier model year. In the Autonomie modeling, when a new vehicle adopts fuel saving technologies that are inherited, the engine is not resized (the properties from the baseline reference vehicle are used directly and unchanged) and there may be a small change in vehicle performance. For example, in Figure VI-3, Vehicle 2 inherits Eng01 from Vehicle 1 while updating the transmission. Inheritance of the engine with new transmission may change performance. This example illustrates how manufacturers generally manage manufacturing complexity for engines, transmissions, and electrification technologies.

Autonomie implements different powertrain sizing algorithms depending on the type of powertrain being considered because different types of powertrains contain different components that must be optimized.[485] For example, the conventional powertrain resizing considers the reference power of the conventional engine (e.g., Eng01, a basic VVT engine, is rated at 108 kilowatts and this is the starting reference power for all technology classes) against the power-split hybrid (SHEVPS) resizing algorithm that must separately optimize engine power, battery size (energy and power), and electric motor power. An engine's reference power rating can either increase or decrease depending on the architecture, vehicle technology class, and whether it includes other advanced technologies.

Performance requirements also differ depending on the type of powertrain because vehicles with different powertrain types may need to meet different criteria. For example, a plug-in hybrid electric vehicle (PHEV) powertrain that is capable of traveling a certain number of miles on its battery energy alone (referred to as all-electric range, or AER, or as performing in electric-only mode) is also sized to ensure that it can meet the performance requirements of a US06 cycle in electric-only mode.

The powertrain sizing algorithm is an iterative process that attempts to optimize individual powertrain components at each step. For example, the sizing algorithm for conventional powertrains estimates required power to meet gradeability and acceleration performance and compares it to the reference engine power for the technology class. If the power required to meet gradeability and acceleration performance exceeds the reference engine power, the engine power is updated to the new value. Similarly, if the reference engine power exceeds the gradeability and acceleration performance power, it will be decreased to the lower power rating. As the change in power requires a change design of the engine, like increasing displacement (e.g., going from a 5.2-liter to 5.6-liter engine, or vice versa) or increasing cylinder count (e.g., going from an I4 to a V6 or vice versa), the engine weight will also change. The new engine power is used to update the weight of the engine.

Next, the conventional powertrain sizing algorithm enters an acceleration algorithm loop to verify low-speed acceleration performance (time it takes to go from 0 mph to 60 mph). In this step, Autonomie adjusts engine power to maintain a performance attribute for the given technology class and updates engine weight accordingly. Once the performance criteria are met, Autonomie ends the low-speed acceleration performance algorithm loop and enters a high-speed acceleration (time it takes to go from 50 mph to 80 mph) algorithm loop. Again, Autonomie might need to adjust engine power to maintain a performance attribute for the given technology, and it exits this loop once the performance criteria have been met. At this point, the sizing algorithm is complete for the conventional powertrain based on the designation for engine type, transmissions type, aero type, mass reduction technology and low rolling resistance technology.

Figure VI-5 below shows the sizing algorithm for conventional powertrains.

Start Printed Page 24330

Depending on the type of powertrain considered, the sizing algorithms may also size to meet different performance criteria in different order. The powertrain sizing algorithms for electrified vehicles are considerably more complex, and are discussed in further detail in Section VI.C.3, below.

(5) How the Agencies Considered Maintaining Vehicle Attributes

For this rulemaking analysis, consistent with past CAFE and CO2 rulemakings, the agencies have analyzed technology pathways manufacturers could use for compliance that attempt to maintain vehicle attributes, utility, and performance. Using this approach allows the agencies to assess costs and benefits of potential standards under a scenario where consumers continue to get the similar vehicle attributes and features, other than changes in fuel economy. The purpose of constraining vehicle attributes is to simplify the analysis and reduce variance in other attributes that consumers value across the analyzed regulatory alternatives. This allows for a more streamlined accounting of costs and benefits by not requiring the values of other vehicle attributes that trade off with fuel economy.

Several examples of vehicle attributes, utility and performance that could be impacted by adoption of fuel economy improving technology include the following.

Start Printed Page 24331

Consequences for the agencies not fully considering or accounting for potential changes in vehicle attributes, utility, and performance are degradation in vehicle attributes, utility, and performance that lead to consumer acceptance issues without accounting for the corresponding costs and/or not accounting for the costs of technology designs that maintain vehicle attributes, utility, and performance. The agencies incorporated changes in the NPRM analysis and that are carried into this final rule that address deficiencies in past analyses, including the Draft TAR and Proposed Determination analyses. These changes were discussed in the NPRM and are repeated in the discussion of individual technologies in this Preamble, the FRIA, and supporting documents. The following are several examples of technologies that did not maintain vehicle attributes, utility, and performance in the Draft TAR and Proposed Determination analyses.

For the EPA Draft TAR and Proposed Determination analyses, HCR engine and downsized and turbocharged engine technologies effectiveness was estimated using Tier 2 certification fuel, which has a higher octane rating compared to regular octane fuel.[486 487] This does not maintain functionality because consumers would incur higher costs for using premium fuel in order to achieve the modeled fuel economy improvements, compared to baseline engines that were replaced, which operated on lower cost regular octane fuel. By not maintaining the fuel octane functionality and vehicle attributes, the EPA Draft TAR and Proposed Determination analyses applied higher effectiveness for these technologies than could be achieved had regular octane fuel been assumed for the HCR and downsized turbocharged engines. The Draft TAR and Proposed Determination analyses also did not account for the higher costs that would be incurred by consumers to pay for high octane fuel. These issues were addressed in the Start Printed Page 24332NPRM and this final rule analysis, and account for some of the effectiveness and cost differences between the Draft TAR/Proposed Determination and the NPRM/final rule.[488]

Another example is mass reduction technology. As background, the agencies characterize mass reduction as either primary mass reduction or secondary mass reduction. Primary mass reduction involves reducing mass of components that can be done independently of the mass of other components. For example, the mass of a hood (e.g., replacing a steel hood with an aluminum hood) or reducing the mass of a seat are examples of primary mass reduction because each can be implemented independently. When there is a significant level of primary mass reduction, other components that are designed based on the mass of primary components, may be redesigned and have lower mass. An example of secondary mass reduction is the brake system. If the mass of primary components is reduced sufficiently, the resulting lighter weight vehicle could maintain braking performance and attributes, and safety with a lighter weight brake system. Mass reduction in the brake system is secondary mass reduction because it requires primary mass reduction before it can be incorporated. For the EPA Draft TAR and Proposed Determination analyses, secondary mass reduction was applied exclusively based on cost, with no regard to whether sufficient primary mass reduction was applied concurrently. The analyses did not account for the degraded functionality of the secondary components and systems and also understated the costs for lower levels of mass reduction.[489] These issues were addressed in the NPRM and this final rule analysis, and account for some of the cost differences between the Draft TAR/Proposed Determination and the NPRM/final rule.

The agencies note that for some technologies it is not reasonable or practicable to match exactly the baseline vehicle's attributes, utility, and performance. For example, when engines are resized to maintain acceleration performance, if the agencies applied a criterion that allowed no shift in performance whatsoever, there would be an extreme proliferation of unique engine displacements. Manufacturers have repeatedly and consistently told the agencies that the high costs for redesign and the increased manufacturing complexity that would result from resizing engines for small technology changes preclude them from doing so. It would be unreasonable and unaffordable to resize powertrains for every unique combination of technologies, and exceedingly so for every unique combination technologies across every vehicle model due to the extreme manufacturing complexity that would be required to do so.[490] For the NPRM and final rule analyses, engine resizing is limited to specific incremental technology changes that would typically be associated with a major vehicle or engine redesign to address product complexity and economies of scale considerations. The EPA Draft TAR and Proposed Determination analyses adjusted the effectiveness of every technology combination assuming performance could be held constant for every combination, and the analysis did not recognize or account for the extreme complexity nor the associated costs for that impractical assumption. The NPRM and final rule analyses account for these real-world practicalities and constraints, and doing so explains some of the effectiveness and cost differences between the Draft TAR/Proposed Determination and the NPRM/final rule.

The subsections for individual technologies discuss the technology assumptions and constraints that were considered to maintain vehicle attributes, utility, and performance as closely as possible. The agencies believe that any minimal remaining differences, which may directionally either improve or degrade vehicle attributes, utility and performance are small enough to have de minimis impact on the analysis.

(6) How the Agencies Considered Performance Neutrality

The CAFE model examines technologies that can improve fuel economy and reduce CO2 emissions. An improvement in efficiency can be realized by improving the powertrain that propels the vehicle (e.g., replacing a 6-cylinder engine with a smaller, turbocharged 4-cylinder engine), or by reducing the vehicle's loads or burdens (e.g., lowering aerodynamic drag, reducing vehicle mass and/or rolling resistance). Either way, these changes reduce energy consumption and create a range of choices for automobile manufacturers. At the two ends of the range, the manufacturer can choose either:

(A) To design a vehicle that does same the amount of work as before but uses less fuel.

For example, a redesigned pickup truck would receive a turbocharged V6 engine in place of the outgoing V8. The pickup would offer no additional towing capacity, acceleration, larger wheels and tires, expanded infotainment packages, or customer convenience features, but would achieve a higher fuel economy rating (and correspondingly lower CO2 emissions).

(B) To design a vehicle that does more work and uses the same amount of fuel as before.

For example, a redesigned pickup truck would receive a turbocharged V6 engine in place of the outgoing V8, but with engine efficiency improvements that allow the same amount of fuel to do more work. The pickup would offer improved towing capacity, improved acceleration, larger wheels and tires, an expanded (heavier) infotainment package, and more convenience features, while maintaining (not improving) the fuel economy rating of the previous year's model.

In other words, automakers weigh the trade-offs between vehicle performance/utility and fuel economy, and they choose a blend of these attributes to balance meeting fuel economy and emissions standards and suiting the demands of their customers.

Historically, vehicle performance has improved over the years. The average horsepower is the highest that it has ever been; all vehicle types have improved horsepower by at least 49 percent compared to the 1975 model year, and pickup trucks have improved by 141 percent.[491] Since 1978, the 0-60 acceleration time of vehicles has improved by 39-47 percent depending on vehicle type.[492] Also, to gain consumer acceptance of downsized turbocharged engines, manufacturers have stated they often offer an increase in performance.[493] Fuel economy has also improved, but the horsepower and acceleration trends show that not 100 percent of technological improvements have been applied to fuel savings. While future trends are uncertain, the past trends suggest vehicle performance is unlikely to decrease, as it seems reasonable to assume that customers Start Printed Page 24333will at a minimum demand vehicles that offer the same utility as today's fleet.

For this rulemaking analysis, consistent with past CAFE and CO2 rulemakings, the agencies have analyzed technology pathways manufacturers could use for compliance that attempt to maintain vehicle attributes, utility and performance. NHTSA's analysis in the Draft TAR used the same approach for performance neutrality as was used for the NPRM and is being carried into this final rule. This approach is described throughout this section and further in FRIA Section VI. For the Draft TAR and Proposed Determination, the EPA analyses used an approach that maintained 0-60 mph acceleration time for every technology package. However, that approach did not account for the added development, manufacturing, assembly and service parts complexity and associated costs that would be incurred by manufacturers to produce the substantial number of engine variants that would be required to achieve those CO2 improvements.[494] Using the NPRM approach, which is carried into this final rule, allows the agencies to assess costs and benefits of potential standards under a scenario where consumers continue to get the same vehicle attributes and features, other than changes in fuel economy (approaching the scenario in example “A” above). This approach also eliminates the need to assess the value of changes in vehicle attributes and features. As discussed later in this section, while some small level of performance increase is unavoidable when conducting this type of analysis, the added technology results almost exclusively in improved fuel economy. This allows the cost of these technologies to reflect almost entirely the cost of compliance with standards with nearly neutral vehicle performance.

The CAFE model maintains the initial performance and utility levels of the analysis vehicle fleet, while considering real world constraints faced by manufacturers.

To maintain performance neutrality when applying fuel economy technologies, it is first necessary to characterize the performance levels of each of the nearly 3000 vehicle models in the MY 2017 baseline fleet. As discussed in Section VI.B.1.b) Assigning Vehicle Technology Classes, above, each individual vehicle model in the analysis fleet was assigned to one of ten vehicle “technology classes”—the class that is most similar to the vehicle model. The technology classes include five standard class vehicles (compact car, midsize car, small SUV, midsize SUV, pickup) plus five “performance” versions of these same body styles.[495] Each vehicle class has a unique set of attributes and characteristics, including vehicle performance metrics, that describe the typical characteristics of the vehicles in that class.

The analysis used four criteria to characterize vehicle performance attributes and utility:

  • Low-speed acceleration (time required to accelerate from 0-60 mph)
  • High-speed acceleration (time required to accelerate from 50-80 mph)
  • Gradeability (the ability of the vehicle to maintain constant 65 miles per hour speed on a six percent upgrade)
  • Towing capacity

Low-speed and high-speed acceleration target times are typical of current production vehicles and range from 6 to 10 seconds depending on the vehicle class; for example, the midsize SUV performance class has a low- and high-speed acceleration target of 7 seconds.[496] The gradeability criterion requires that the vehicle, given its attributes of weight, engine power, and transmission gearing, be capable of maintaining a minimum of 65 mph while going up a six percent grade. The towing criterion, which is applicable only to the pickup truck and performance pickup truck vehicle technology classes, is the same as the gradeability requirement but adds an additional payload/towing mass (3,000 lbs. for pickups, or 4,350 lbs for performance pickups) to the vehicle, essentially making the vehicle heavier.

In addition, to maintain the capabilities of certain electrified vehicles in the 2017 baseline fleet, the analysis required that those vehicles be capable of achieving the accelerations and speeds of certain standard driving cycles. The agencies use the US06 “aggressive driving” cycle and the UDDS “city driving” cycle to ensure that core capabilities of BEVs and PHEVs, such as driving certain speeds and/or distances in electric-only mode, are maintained. In addition to the four criteria discussed above, the following performance criteria are applied to these electrified vehicles:

  • Battery electric vehicles (BEV) are sized to be capable of completing the US06 “aggressive driving” cycle.
  • Plug-in hybrid vehicles with 50 mile all-electric range (PHEV50) are sized to be capable of completing the US06 “aggressive driving” cycle in electric-only mode.
  • Plug-in hybrid vehicles with 20 mile all-electric range (PHEV20) are sized to be capable of completing the UDDS “city driving” cycle in electric-only (charge depleting) mode.[497]

Together, these performance criteria are widely used by industry as metrics to quantify vehicle performance attributes that consumers observe and that are important for vehicle utility and customer satisfaction.[498]

When certain fuel-saving technologies are applied that affect vehicle performance to a significant extent, such as replacing a pickup truck's V8 engine with a turbocharged V6 engine, iterative resizing of the vehicle powertrain (engine, electric motors, and/or battery) is performed in the Autonomie simulation such that the above performance criteria is maintained. For example, if the aforementioned engine repla